jf.ti. PIBLH MOKI.S PROGRAM i:valHali«R »f Ikr \alional Imparls «f Ikf local Pvklir Marks Program U.S. DEPARTMENT OF COMMERCE • Economic Development Administration LOCAL PIBLIf WORKS PROGRAM Report G Evaluation of the National Impacts of the Local Public Works Program Pre pare i k\ i hasr EetRtMflrir Assoriales. Inc. for Ihf Of f if r tf homcslic ErtMmir Polin Coordination **TESO** U.S. DEPARTMENT OF COMMERCE Philip M. Klutznick, Secretary Robert T. Hall, Assistant Secretary for Economic Development DECEMBER 1980 Digitized by the Internet Archive in 2012 with funding from LYRASIS Members and Sloan Foundation http://archive.org/details/localpublicworksOOchas FOREWORD This report and the supporting evaluation studies present the major results of the $6 billion Local Public Works Program (LPW) which the Economic Development Administration (EDA) implemented in 1977. The program had two phases: LPW I at $2 billion was initiated by Congress in the closing days of the Ford Administration, and LPW II at $4 billion was a keystone in the economic stimulus program initiated by President Carter at t-he onset of his Administration. Given the magnitude of the funds and the policy issues involved in the use of public works as a tool for stimulating the economy, it was incumbent upon EDA to carry out a thorough evaluation of the program and publicly report its findings. The Local Public Works Program — Final Report (Report A) summarizes those findings, which are presented in detail in this study (Report G) and in the following evaluations conducted by EDA staff or outside researchers working for the Department of Commerce: o Analysis of Construction Employment Generation by Project Types in the Local Public Works Program (Report B) o Report on the Characteristics of LPW Employees (Report C) o On-Site Data Collection and Analysis in Support of a Microeconomic Evaluation of the Local Public Works Program (Report D) o The Timing of the Construction Employment Impacts of the LPW Program Relative to Local Area Employment Cycles (Report E) o The Net Cost of the Local Public Works Program (Report F) Our purpose in these documents is to report on our stewardship of significant public resources and to provide guidance to future policy makers. It is EDA's practice to preface the final reports of studies conducted by outside researchers with a set of Agency comments on the content of the report. A set of EDA staff comments follows this foreword. LPW, the largest countercyclical public works program under- taken since the depression, had three basic objectives: providing a stimulus to the national economy and distressed local economies; generating employment opportunities in construction and related industries; and constructing or renovating needed public facilities. As these reports document, the program was largely successful in meeting these objectives, as well as several secondary goals. In addition, the LPW experience revealed valuable insights about a more effective structure for both an accelerated public works program like LPW and a countercyclical public works program like that advocated by the House in 1979 and 1980 as part of the expansion and extension of EDA's legislation. I will use this "Foreword" as a forum for discussing these insights, as well as outstanding issues, in the context of examining the utility of public works as a countercyclical tool, a subject that has been hotly debated for more than a decade. At the outset, it is important to note that much of the criticism of public works as a countercyclical instrument stemming from the LPW Program is misdirected in that it was based on LPW's failure to meet objectives that were not set forth for the program. LPW, particularly LPW II which I helped design and had responsi- bility for administering, was not directed at soaking up large numbers of the long-term unemployed, nor was it conceived as a skill training mechanism. Moreover, since both rounds of the program were initiated long after the trough of the recession, they clearly could not perform as countercyclical measures in the classic sense. One of the factors that is often overlooked in discussing this subject is that public works is not the only instrument in the panoply of economic or employment stimulus tools. Tax cuts, public service employment, expansion of training programs, extended unemployment insurance, and fiscal assistance aid are among the other measures which are or could be brought to bear when such micro-economic policies are called for. There are, however, some "open" issues which need to be addressed in the use of public works as a countercyclical tool. First and foremost is the utility of traditional economic stimulus programs in the changed economic circumstances characterized as "stagflation." Certainly the advent of persistent inflationary pressures simultaneous with declining or lagging economic growth and employment must now be considered in determining the appropriateness of public works as a countercyclical effort. In addition, there have been three historically persistent problems in using public works as a countercyclical tool, each of which is related to the difficulty in obtaining the anticipated impact within the desired timeframe. One involves the ability of Federal, State, and local governments to propose, review and initiate construction on useful public works projects within a short period of time. Under the LPW Program, it was demonstrated that this can be accomplished. A second relates to the length of time required to complete projects. Again, the LPW experience revealed that there are certain types of projects that are completed more rapidly, such as repair and renovation projects involving less than $500,000. A third problem relates to the timing of program initiation. Too often in the past, the programs have begun so late that the major effects are produced after the economy has turned around or at least is well on its way to recovering. Indeed, some of the impact of such programs may well have exacerbated inflationary pressures which are latent in a strong economic growth period. LPW II, for example, may well have had such an effect given the enactment and implementation of the program so many months after the 1975 «, recessionary trough. To counteract such lags, proposals for "auto- matic trigger" devices for standby public works programs have been made. The translation of this concept into an appropriate legislative and administrative mechanism, however, is fraught with difficulties. No single or group of "trigger" measurement devices can address the myriad economic considerations (e.g., inflationary factors and size of the budget deficit) that must be assessed in determining the efficacy of mounting a countercyclical public works program. Thus, even a mechanism that accurately predicts a reces- sion before it ensues, cannot be used in isolation to trigger such a program. The objectives of a countercyclical public works program might also be reexamined. Goals other than the traditional ones of generating a timely stimulus, creating short-term employment and providing useful facilities can be identified. For example, more labor intensive projects with greater impact on the general unemployed and long-term unemployed might be sought. More timely completion of projects might also be pursued to maximize the stimulus effort and reduce any procyclical effect. In addition, more return on the expenditures for public infrastructure might be achieved if greater emphasis were given to the rehabilitation of deteriorating public infrastructure, thereby assisting communities by bringing that infrastructure to "like new" condition and delaying full capital replacement costs. Another set of policy issues involve the multiplicity of sub- objectives which may be appended to a countercyclical public works program, such as minority business involvement, the use of American- made materials, employment for veterans, etc. All of these are desirable and legitimate public policy objectives, but the admin- istrative implications for a rapid implementation effort are sub- stantial, and should be carefully weighed before such requirements are added. As a result of my experience with LPW and other stimulus programs, I believe it would behoove the framers of any possible future countercyclical public works program to explore the possi- bilities of increasing the program's impact on the long-term unemployed and the rehabilitation needs of deteriorating public infrastructure. Indeed, a labor intensive public works program primarily directed at the rehabilitation and weather ization of existing public facilities which had hiring linkages with the long- term unemployed could be considered not only as a countercyclical device, but as an ongoing, regular program. Organized training could be structured as part of the program, and the work would largely be performed through competitive bids by contractors. Only a portion of each contractor's workforce would have to be drawn from the long-term unemployed, and they would be paid a training wage. As a result of these features, such a program would continue the objectives of upgrading infrastructure and providing meaningful private sector employment experiences for the long-term unemployed. These as well as other characteristics give it obvious advantages over the Public Service Employment Program. In closing, I would be remiss not to cite the unparalleled public administration effort involved in the implementation of the LPW Program — an effort not explicitly documented in this report and the supporting evaluation studies. Special recognition must go to the hundreds of dedicated men and women who worked long hours and weeks in processing and managing over 10,000 public works projects. Since public praise for the civil servant is scarce, the achievements of the LPW staff have not received the attention warranted by the resulting benefits to so many communities and individuals throughout the country. To each of them, I extend my own deep appreciation and a "well done." A^«^Uc Robert T. Hall Assistant Secretary for Economic Development EDA COMMENTS ON THE EVALUATION OF THE NATIONAL IMPACT OF THE LOCAL PUBLIC WORKS PROGRAM by Chase Econometric Associates, Inc. (CEAI) As part of an expanded program evaluation effort of the Economic Development Administration (EDA) , the Agency is now including an introductory section containing EDA comments and responses in appropriate evaluation reports. This innovation will ensure that the insights gained and issues raised as a result of evaluations are addressed by the Agency. It will also inform those reading a particular evaluation report of the Agency's position on it, and the response to the conclusions and recommendations. For this particular evaluation of the national impact of EDA's Local Public Works (LPW) program, most of the EDA comments address themselves to data and/or econometric modelling problems. Specifically, several of the main substudies of this evaluation effort are based on data which are not LPW-specific and/or contain projections of anticipated trends. The use of such data within a macroeconomic model framework compounds the probability that the impact estimates will be substantially different from estimates based on actual program reporting. 1. Timing of Expenditures Substudy CEAI used an incomplete set of LPW expenditure data as well as EDA disbursement data to estimate a pattern of expenditures over time. Subsequent EDA research has shown that the timing of disbursements is more closely related to the requirements of the local jurisdictional budgetary structures than to actual expenditure timing patterns. Further, since actual disbursement data were only available through the start of 1978 at the time of CEAI ' s study effort, CEAI projections were used as proxies for disbursement patterns after that date. It is unlikely that the use of such estimates and projections will provide an accurate picture of actual countercyclical impacts. 2. Employment Impacts Substudy CEAI estimated direct employment impacts using Bureau of Labor Statistics (BLS) labor requirements coefficients. Although EDA was in the process of collecting direct employment data through the use of the Local Public Works Payroll Reporting Form (ED-746) , the time frame for the CEAI study was too short to allow the provision of ED-746 data when required for the analysis schedule. The 746-based labor coefficients differ from the BLS labor requirements coefficients. A more accurate estimate of LPW direct employment is contained in the Local Public Works Program Final Report. 3 . Impact on State and Local Governments The assessment of the impact of LPW on state and local governments, the subsequent analysis of fiscal substitution, and, thus, the estimation of net employment impacts were based on data obtained from a non-random, non-stratified sample of 50 jurisdictions. This sample size, as noted on page 52 of this report, "is too small for effective statistical analysis." Abt Associates, Inc., under contract to EDA's Program Evaluation Division (PED) , has undertaken a similar assessment based on a stratified sample of 100 jurisdictions, and the results are significantly different. 4 . Industry, Regional, and Cyclical Impacts Substudies Since the analyses of industrial, regional, and cyclical impacts were based on the estimated expenditure patterns and employment impacts discussed above, the conclusions of these studies are subject to the same limitations. EXECUTIVE SUMMARY The Local Public Works program was one of several countercyclical policies undertaken to alleviate the effects of the 1973-1975 recession. These policies- including countercyclical revenue sharing and employ- ment and training programs— were designed to be complementary in the sense that each policy was designed to concentrate on one or more of the specific areas of distress which arose from the recession. The primary purpose of the LPW program was to create employment in the severely distressed construction industry and in the industries supplying materials and services to the construction industry. The increase in employment was expected to be targeted on areas of high unemployment. Another objective was to increase the stock of public facilities. It was hoped that the growth in public infrastructure, which had been severely curtailed in the recent past, would provide continuing benefits to the localities in which the projects were situated. Finally, the program was intended to provide increased aid to the hard pressed state and local governments through which the funds were channeled. To accomplish these goals, $2 billion was authorized under the Local Public Works Capital Development and Investment Act of 1976, and $4 billion was authorized under the Public Works Employment Act of 1977. Both Acts specified that an evaluation of the impacts of these expenditures be undertaken. The Department of Commerce, in line with this mandate, is conducting a comprehensive three-part evaluation of the Local Public Works Program, consisting of an administrative evaluation, a microeconomic evalua- tion of the immediate impacts, and a macroeconomic evaluation of the national impact. The macroeconomic evaluation, carried out by Chase Econometric Associates, Inc., under contract to the Department of Commerce, is the subject of this report. The following major issues are addressed in this report: • How large are the employment impacts? • What industries are the primary beneficiaries? • To what extent are states and localities assisted? • What are the regional impacts? • How effective is the LPW program as a countercyclical policy? The data used in this study included results of special surveys of locality officials and contractors, results of BLS survey-based studies on construction labor requirements, LPW-specific data gathered by the Department of Commerce, (consisting of appli- cation and awards data and accounting data on actual disbursements), locality budget documents, and public data on commodity flows, construction activity, state and local fiscal conditions, and economic conditions. The analysis also employed several econometric models, particularly the CEAI Macro- economic Model and the CEAI INFORUM (input/output) Model. This study, as with all research efforts, is constrained by the assumptions used in formulating the questions to be analyzed (implicitly and explicitly noted in the Request for Proposal), the assumptions and limitations of the model structure used in the analysis, and any limitations inherent in the data. These constraints are not unique among research efforts. However, the reader should be cognizant of the limitations and caveats resulting from such constraints as noted in the study report. The basic findings of the study are presented in this summary statement. More detailed discussions of both the findings and the methodologies are presented in the body of the report. 'The Congressional Budget Office has estimated that total anti-recession employment-creation expenditure for the period FY 1974 through FY 1980— including the Comprehensive Employment and Training Act (Title II and IV); the Job Opportunities Program; countercyclical revenue sharing; the Jobs Tax Credit; and the LPW program— will be approximately $39 billion. Timing of Expenditures The LPW program was implemented in two rounds. As part of the 1976 Act (Round I) $2 billion was appropriated in October of 1976, and significant levels of expenditures began in January of 1977. Expenditure of the additional $4 billion authorized as part of the 1977 Act (Round II) began approximately three quarters later. CEAI estimates that the following pattern of construction expenditures will occur as a result of the combined authorizations. • 20 percent ($1.16 billion) in 1977 • 61 percent ($3.60 billion) in 1978 • 18 percent ($1.03 billion) in 1979 • 1 percent ($0.07 billion) in 1980 Thus LPW expenditures appear to be occurring in a far more protracted fashion than was envisioned in the legislation and by most commentators. Employment Impacts CEAI estimates that the LPW program will create a total addition to employment— including employment induced by the increase in macroeconomic activity, but net of substitution and crowding out effects— of approximately 192,000 person-years of work. The cost per person-year of this employmnt is $31,000, which is equivalent to a net increase in employment of 3.2 person-years per $100,000 of expenditure. This total increase in employment is estimated to be approximately equally distributed among construction industry (direct) employment, materials supply industries (indirect) employment, and induced employment. CEAI estimates that within the con- struction industry 63,000 person-years of employ- ment (or 1 .1 person-years for every $100,000 of LPW expenditure) will be created. The cost per person- year of direct employment in the construction industry is thus estimated to be just under $95,000. These employment estimates are net of estimated substitution and crowding out. They also reflect adjustments of BLS data to allow for the effects on labor requirements of price and productivity changes over time. Based primarily upon the results of our survey, CEAI estimates that between 25 and 30 percent of LPW funds were substituted for local funds that would have been spent anyway. The LPW program, however, also induced additional con- struction expenditures by localities. When adjusted for these expenditures, CEAI's estimate of net substitution in LPW program construction expend- itures falls to 20 percent of LPW funding CEAI's estimate of the crowding out of private activity (derived from macroeconomic model simulations) is equal to approximately 9 percent of LPW funding. The combined effect of substitution and crowding out is a reduction in the gross employment impact of LPW expenditures of 29 percent. Gross of these adjust- ments, direct employment impacts were estimated to be 88,640 person-years, and total employment im- pacts were estimated to be 270,170 person-years. In addition, both rising wage rates and productivity over the life of the LPW program act to reduce the employment per dollar. CEAI estimates that this reduced LPW employment, as measured by the 1977 BLS labor requirements figures, by an additional 1 1 percent. Expenditures for 1977 reflect actual disbursements, for which data are available. Expenditures for other years are projected from 1977 disbursement patterns. ^The appropriation legislation stated that Round I would run through September 30, 1 977, and Round II would run through December 31 , 1 978. 4 Most public accounts of the LPW program imply that the program— or at least each round— is a one-year program. Thus, media description of Round II typically states that $4 billion will be spent in 1 978. This is approximately true of the LPW program overall, but only because nearly $900 million of Round I expenditures are occurring in calendar 1978. However, it is estimated that only about two-thirds of Round'll expenditures will occur during this one-year period. 5 Due to a lack of available primary data on direct employment impacts of the LPW program, CEAI estimated direct construction employment impacts using Bureau of Labor Statistics labor requirements data. Indirect materials supply industry employment was estimated using the Chase/INFORUM input/output model's relationships. An employment multiplier to measure induced employment was derived from a macroeconomic model simulation. Both the BLS and input/output labor requirements data were adjusted for the changes in prices and productivity estimated to occur during the course of the LPW program. CEAI's estimate of the employment impact of the LPW program is significantly lower than most other employment estimates. Nevertheless, CEAI believes that its estimates are more likely to overstate than to understate the employment impacts/ CEAI attributes the differences between its estimates and those of the earlier studies to the lack of adjustment of other studies for one or more of the following factors: productivity and price changes since the time of earlier studies, substitution, and crowding out effects. Industry Impact The primary industry impact of LPW expenditures is, of course, on the construction industry. The total $6 billion stimulus to the construction industry was equivalent to 3.7 percent of a single year's (1976) construction expenditures. In addition, nearly 60 percent of LPW expenditures are transmitted from the construction industry to other industries through the purchases of goods and services by contractors. The total expenditure impact of the LPW program on supply industries— including the effects at earlier stages of production— was nearly as large as LPW expenditures themselves. The net indirect employ- ment impact is estimated to be approximately 65,000 person-years, which is slightly larger than the direct employment impact. The major industry groups benefiting most from these estimated indirect impacts are: • Services: 37 percent of indirect expenditures ($2,141 million) and 50 percent of indirect employment (45,946 person-years) • Fabricated Metal Products: 14 percent of indirect expenditures ($797 million) and 11 percent of indirect employment (10,195 person- years) • Stone, Clay, and Glass: 12 percent of indirect expenditures ($671 million) and 11 percent of indirect employment (10,195 person-years) • Metals: 10 percent of indirect expenditures ($598 million) and 7 percent of indirect employ- ment (6,406 person-years) • Machinery: 7 percent of indirect expenditures ($372 million) and 6 percent of indirect employ- ment (5,591 person-years) • Lumber and Wood Products: 5 percent of indirect expenditures ($293 million) and 5 percent of indirect employment (4,344 person-years) Relative to the size of the affected industries, and to their cyclic distress, the LPW program was extremely small. The impact of the LPW program in its peak year (1978) exceeded 0.4 percent of annual shipments for only two industries. In both of these cases the impact was approximately 1 percent of industry shipments. In only four cases was the total stimulus of the LPW program greater than 10 percent of the decline in output suffered in the 1973-1975 recession, and in no case was the total stimulus higher than 17 percent. Impact on State and Local Governments A variety of trends and cyclical factors suggests that state and local governments were in need of assistance at the time of the inception of the LPW program. Resistance to state and local government expenditures, and to the tax burden that financed them, appears to have been growing for several years. The 1973-1975 recession hit state and local governments especially hard, and this distress lasted nearly two years after most sectors of the economy had begun to recover. Local fiscal distress was further exacerbated by well publicized financial difficulties in a few cities. Because of size, inherent nature, and/or the flexibility that exists in locality budgeting processes, construction expenditures are likely to be the type of locality expenditure most adversely affected by fiscal distress. In actuality, real state and local construction expenditures had been steadily declining since 1968. 'Some adjustments— especially a rounding up of the employment multiplier— were made to allow for possible error. Morever, an alternative calculation of direct employment impact, based on CEAI's macroeconomic simulations for the LPW program, suggests that direct LPW employment was about 75% of the level estimated by CEAI and reported here. 8 Cement and Clay Products and Structural Metal Products. 9 Electrical Machinery, NEC (16.9%); Structural Metal Products (16.7%); Plastic Products (1 1.9%); and Cement and Clay Products (11.0%). The pattern of construction expenditure financed by LPW funds differs from that of normal locality expenditure in a manner which suggests that in many cases it is unlikely that LPW projects would have received local funding in the absence of the LPW program. In general, LPW funds appear to be used disproportionately for types of projects which tended to be: • without obvious alternative sources of funding • relatively easy to defer • not visible enough to sustain active public support five times the ratio for counties. Because of declining student populations, the small share of funds provided to school districts may be appropriate. Moreover, the fiscal distress of cities— particularly certain individual cities— does appear to have been the most severe of any type of locality. This, for example, is probably a signficant factor in the relatively slow growth rate in city expenditures. Nevertheless, the evidence on relative locality distress does not provide a clear indication that cities as a class faced fiscal pressures that were several times as great as those faced by counties. In CEAI's judgment, therefore, the data indicate that counties as a class may have received disproportionately little LPW funding relative to cities. The LPW program's assistance to local govern- ments was substantial, as compared with the baseline pattern of construction expenditures. CEAI estimates that LPW funding was equal to approximately 30 to 40 percent of one year's (1976) expenditure on construction by surveyed localities. It represented a substantially larger share of one year's estimated discretionary capital expenditure— about 75 percent for cities, and 35 to 45 percent for counties and school districts. The estimated 20 percent of all LPW funds which were substituted for local funds acted to reduce the financial distress of localities by approximately $1.2 billion. Among the specific questions addressed in this study was the issue of the amount of LPW funding for different types of localities relative to their fiscal distress. In order to assess the overall relative fiscal impacts, CEAI compared LPW funding with total (capital and operating) expenditures by type of locality. Per dollar of normal expenditures, cities received three times as much LPW funding as counties and four times as much as school districts. If the growth of expenditure is considered, the differential is even greater; the ratio of share of LPW funds to growth rate of expenditures for cities is nearly five times the ratio for school districts and over Regional Impacts LPW funds were allocated primarily on the basis of unemployment measures of state and sub-state areas. In absolute dollar amounts LPW program expenditures are concentrated in the industrialized states of the North Atlantic and Great Lakes area and in the Southwest. Much of this variation, however, corresponds to great variations in population and in economic activity among regions. A major policy issue is what regional impacts of LPW are relative to the regional contributions toward funding of the program. Regional personal and income tax payments were therefore used as a standard of comparison against which the regional impacts of the LPW expenditures were measured. All parts of the country except the Great Lakes and the Mid-South Atlantic regions received a somewhat larger share of LPW awards than their shares of personal income tax payments. This excess of LPW share over tax share was the greatest in the North Pacific and Midwest regions (the plains and mountain states of the West). The Great Lakes and Mid-South Atlantic states received only about 70 percent of their tax share of LPW awards. Consideration of the regional impacts is incomplete without including the effects of inter-regional flows of I °CEAI's assessment of fiscal distress is based on an analysis of the joint movement of a number of variables, including locality expenditures (and construction expenditures in particular), revenues, surpluses, changes in locality interest rates, and level of locality interest rates relative to other types of interest rates. I I As part of the analysis, CEAI defined a new set of regions, based on relative cyclical unemployment characteristics. Both the definitions and names of these regions are somewhat different from those with which the reader will be familiar. The definitions of regions and the precise results are presented in Chapter 7. expenditures that resulted from the LPW program. CEAI's analysis suggests that these flows were confined almost exclusively to materials purchases from manufacturers by distributors, and that they are relatively small— averaging about 5 percent of the LPW expenditures on a regional basis. Two factors appear substantially to have determined the size of the net flows from a region. • States that are relatively industrialized tended to experience net inflows of expenditures (i.e. net materials exports to other regions) • States that received shares of LPW funds that were larger than their shares of shipments of manufactured goods tended to experience net outflows of expenditures (i.e. net materials imports from other regions) The net impact of materials expenditure flows was generally to make the regional shares of LPW program impact more nearly equal to the region's personal income tax shares than are the shares of awards. In particular, the Great Lakes region was a net exporter of materials which raised the relative impact of the LPW program from 70 to 85 percent of the region's tax shares. The ratio of share of net LPW impact (after inter-regional expenditure flows) to share of taxes for each region is as follows: • 1.14 for the North Atlantic Region were both slight net beneficiaries of the LPW programs. On a relative basis, the North Pacific states were the greatest net beneficiaries. The Great Lakes states and mid-south were— at least on the basis of personal income tax— substantial net contributors to the LPW program. Cyclical Impact of the LPW Program The LPW program was conceived as a counter- cyclical program. In order to assess its potential countercyclical efficacy if implemented one quarter after identification of the recession, CEAI used macroeconomic simulations to analyze the effects of the LPW program under the assumption that it had been started in 1 974, rather than in 1 977. Under these assumptions, it was found that a timely construction grants program could have been an effective countercyclical program for the 1973-1 975 recession. In the first place, more jobs would have been created. CEAI's simulations suggest that about 9 percent more construction employment (approximately 5,500 person-years) and over 13 percent more overall employment (approximately 25,800 person-years) would have resulted from the earlier timing of the program. Moreover, these jobs would have been created in the depth of the recession. The actual LPW program, however, was not implemented in this timely fashion. LPW expenditures commenced: • Thirteen quarters after the 1973 peak (i.e. after the beginning of the recession); • 0.85 for the Great Lakes Region • 0.60 for the Mid-South Atlantic Region • 0.98 for the South Atlantic Region • 0.97 for the Midwest Region • 1 .05 for the Southwest Region • 1 .39 for the North Pacific Region The major results with respect to regional income distribution are: The Northeast and the Southwest • Nearly eight quarters after the recovery began in 1975; • Nearly two quarters after the state and local government recovery began; and • Up to a year after most affected materials supply industries had regained their previous peak levels of production. The maximum level of impact of the LPW program occurred in mid-1978— about six quarters after expenditures began. By the time this maximum level of expenditure was achieved: 1 ^A value of unity for this ratio indicates that a region benefited from net LPW impact to the same degree that it contributed to the program's financing in the form of personal income taxes. A value greater than unity (e.g. 1.14) indicates that the region's beneficial impact was greater (in this example 14% greater) than the personal income tax contribution. A value of less than unity (e.g. 0.85) indicates that a smaller benefit (in this example 15% smaller) than the tax contribution. • The overall unemployment rate had already fallen by a third; • The unemployment rate in construction had fallen by well over half; and • Output of the affected materials supply industries averaged 7 percent above previous peak levels. The LPW program impacts will taper off during 1979, with only 19 percent of expenditure estimated to occur after 1 978, and only about 6 percent after the first half of 1979. During the period: • Overall unemployment is forecast to rise; • Unemployment in construction, which has already begun rising, is forecast to rise further; and • The economy is given nearly a 50 percent chance of experiencing a recession. Thus, the pattern of LPW expenditure is one of starting and building up after the recovery is well under way, reaching its highest levels when the economy is strong, and terminating as the economy is expected to weaken. This pattern is strongly pro- cyclical, rather than countercyclical. This finding points to the need to encourage expeditious implementation of possible future pro- grams of this type, which are initiated for counter- cyclical purposes. Means of expediting implementa- tion must be explored with great care, however, since they may present problems of their own. CEAI's research and survey experience on this program suggests, for example, that a stand-by program of this nature— or even a reasonable certainty that such a program would be enacted on a discretionary basis- may lead to very high levels of substitution. Although this might not compromise the countercyclical efficacy of such a program, it could greatly reduce the net aid given to state and local government and to the construction industry and related industries. TABLE OF CONTENTS Page Executive Summary i Table of Contents vii Table of Tables viii Table of Figures xii Introduction xiii Chapter 1 Chapter 2 Chapter 3 Chapter 4 Chapter 5 Chapter 6 Chapter 7 Chapter 8 Chapter 9 Appendix A Appendix B Appendix C Appendix D Appendix E Summary of the LPW Program 1 Methodology 10 The Path of LPW Funds 37 Employment Creation Impacts 61 Industry Impacts 74 Impacts Upon Local Government 97 Inter-Regional Impacts 132 Other Policy Issues 142 Data Deficiencies and Suggestions for Future Research . . .160 Project Type and Construction Category Classifications , .165 Selected Survey Information 171 Substitution Cross-Tab Results 176 Chase INFORUM Results 186 Simulation of Two "Normal" Recovery Patterns 193 TABLE OF TABLES Page Table 1-1 L.P.W. Chronology 2 Table 1-2 Provisions of Round I and Round II of the Public Works Employment Act . . 4 Table 2-1 Local Public Works Sample 13 Table 2-2 Quarterly LPW Expenditures 21 Table 2-3 Direct & Indirect Material Requirements Per Dollar of Construction. ... 28 Table 3-1 Funding Alternatives for the Sample of LPW Projects, by Project 43 Table 3-2 Funding Alternatives for the Sample of LPW Projects, by Obligations ... 44 Table 3-3 Funding Plans for Rejected Projects, by Project 45 Table 3-4 Funding Plans for Rejected Projects by Application Request 46 Table 3-5 Level of Overall Substitution, by Locality 53 Table 4-1 Net Employment Estimates (Person-Years) 63 Table 4-2 Macroeconomic Simulation Estimates of Employment (Persons Employed) ... 65 Table 4-3 Labor Cost Estimates (Person Years Per $1,000,000) 70 Table 5-1 Classification of New Construction Categories for Materials Requirements Study 75 Table 5-2 Timing of Material Supply Demand for Buildings 76 vm TABLE OF TABLES (Continued) Paqe Table 5-3 LPW Impact on Industries, by Quarter *. . 80 Table 5-4 Indirect Employment, by Industry (Person-Years) 81 Table 5-5 Shares of Indirect LPW Impact by Industry Group 82 Table 5-6 Construction Related Industrial Production Indices 85 Table 5-7 Timing of Industry Cycles 86 Table 5-8 Amplitude of Industrial Cycles 87 Table 5-9 Construction Activity Cycle: Peak, March 1966--Trouqh, March 1967. ... 88 Table 5-10 LPW Impacts on Industry Outputs 92 Table 5-11 Distress and Stimulus 93 Table 5-12 Stimulus Needed to Produce Measurable Change 94 Table 6-1 State & Local Revenue & Expenditure, by Quarter 105 Table 6-2 State & Local Government Surplus (1972 Dollars) 107 Table 6-3 Interest Rate Comparisons Ratio of Interest Rates 109 Table 6-4 Comparison of Cycles 110 Table 6-5 Comparison of LPW Funding and Total Construction Expenditures 114 Table 6-6 Comparison of LPW Expenditures to Constructing Spending (NIPA Basis). . . 115 TABLE OF TABLES (Continued) Page Table 6-7 Comparison of LPW Expenditures to Locally Financed Portions of Their Capital Budgets Locality Perception Basis (31 Sampled Localities). . . . 118 Table 6-8 Comparison of LPW Expenditures to Discretionary Capital Expenditure Locality Perception Basis (31 Localities) 119 Table 6-9 Comparison of LPW Expenditures to Discretionary Construction Expenditures All Localities (NIPA Basis) 120 Table 6-10 Comparison of Census Trends & LPW Distribution by Budget Share, Localities. .122 Table 6-11 Comparison of Census Trend & LPW Distribution by Budget Share Municipality. .123 Table 6-12 Comparison of Round I and Round II Budget Shares, All Localities 125 Table 6-13 Selected Implicit Interest Rates by Class of Locality 128 Table 6-14 Locality Expenditures by Class of Government 130 Table 6-15 LPW Expenditures by Class of Government 131 Table 7-1 LPW Expenditure, Impact, and Flows of Expenditure by State 133 Table 7-2 LPW Expenditure, Impact, and Flows of Expenditure by Region 135 Table 7-3 Trade Flows & Shares of LPW Obligations & Shipments for Selected States . . 136 Table 7-4 Shares of Awards, Net Impact and Financing of LPW, by Region 138 Table 7-5 Comparison of Awards, Net Impact, and Financing of LPW, by Region 140 Table 8-1 LPW Chronology 144 TABLE OF TABLES (Continued) Page Table 8-2 Time Requirements of LPW Program . . . , 145 Table 8-3 Timing of Materials Supply Industry Recession and Recovery 148 Table 8-4 Change in Construction Industry Employment (Persons Employed) 151 Table 8-5 Change in Total Employment (Persons Employed) 152 Table 8-6 Duration of Post-War Recessions and Recoveries to Peak 154 Table 8-7 Sample of Davis-Bacon Effects. 156 TABLE OF FIGURES Page Figure 2-1 Time Profile of LPW Expenditures 18 Figure 2-2 Round I and Round II and Total LPW Expenditures by Month 22 Figure 2-3 Local Public Works Survey Sites 34 Figure 4-1 Unemployment Rates for the Economy As a Whole & the Construction Industry . 62 Figure 5-1 Constant Dollar GNP Showing NBER Reference Cycles 84 Figure 6-1 State and Local Government Tax Burden 98 Figure 6-2 State and Local Expenditures Share of Total Expenditures 100 Figure 6-3 Selected State and Local Government Construction Expenditures 101 Figure 6-4 State and Local Revenue and Expenditure 104 Figure 6-5 State and Local Government Surplus 106 Figure 6-6 Interest Rate Comparisons Ratio of Interest Rates 108 INTRODUCTION This report is orqanized by chapter and prefaced by an executive summary. Chapter 1 summarizes the legislative history and intent, of the LPW program and provides a chronology of events leading up to and including the implementation of the program. A brief discussion of the policy issues to be addressed in the study is also presented. The methodoloay used to assess theimpacts of the LPW program is addressed in Chapter 2. The first part of the chapter discusses the general methodology utilized in the study. Documentation of the data used is presented, and the survey procedure by which some data were collected is outlined. A description of the models employed in the analysis is included. The second half of the chapter discusses the specific methodologies used to evaluate the timina of the expenditure; to estimate direct, indirect, and induced employment impacts; to assess the impacts on state and local governments and on industry; and to determine the effect of inter-regional flows of expenditures. Chapter 3 traces the path of the LPW funds, describing the uses, restric- tions, and the side effects of the expenditures. The chapter focuses particu- larly on the two major side effects of LPW expenditures -- substitution and crowding out. The question of how much federal money was substituted for local funds, and for what purpose it was used, are addressed. Also discussed here are the implications for counter-cyclical policy of inter-temporal substitution. The extent to which other economic activity is displaced by the injections of government expenditure into the economy is assessed in the analysis of crowding out, and an estimate of the extent of crowding out is presented. Estimates of the employment impacts of the program are reported in Chapter 4. Total employment created is divided into three categories: direct, (con- struction industry employment), indirect, and induced employment. These esti- mates are compared to estimates presented in other studies. Chapter 5 discusses the impact of LPW expenditures on industry, particu- larly on material supply industries. The effect of the expenditures on each major supply industry is analyzed in terms of the price and employment impacts. The estimated timing of the expenditures is also scrutinized with respect to the position of the industrial cycle. Finally, the size of the LPW program impact is compared to the size of industry output and its fluctuations. The effects of the LPW program on local governments is the topic of dis- cussion in Chapter 6. An assessment of the relative fiscal situations of differ- ent types of local governments at the time of the implementation of LPW is offered. In addition, the estimated timina of LPW expenditures is comDared to the cyclic position of local qovernments in order to determine whether the timing of the LPW program was optimal for local governments. The question of of the funds by local governments is addressed, and the impact by type of local itity if dicusssed. The chapter concludes with a comparison of the rela- tive degrees of fiscal distress and benefits from the program of different types of localities. The geographical distribution of the impacts of government expenditure programs have important policy ramifications involving issues such as regional economic development, equity, and the reduction of differentials in the level of unemployment. The inter-regional impacts of LPW expenditures are sutdied in Chapter 7. Figures on the incidence of LPW expenditure, the net inflow of materials expenditure, and the net impact of LPW expenditure are presented for each state. In order to qain a picture of the net redistributional effects of LPW expenditures by -region, reqional shares of total LPW awards, and total awards minus net flows of materials are compared to reqional shares of total U.S. income taxes. The efficacy of tarqetinq funds toward small pockets of unemployment is discussed, and policy implications are presented. Chapter 8 presents discussions of several varied issues important to the study of LPW impacts. Local Public Works as a counter-cyclical program is assessed, an analysis of the Dotential counter-cyclic effect of an LPW program is discussed, and the inflationary impact is considered. Chapter 9, which concludes the report, focuses on the condition and avail- ability of relevant data. Areas for further resect ch are also suggested. CHAPTER 1 : SUMMARY OF THE LPW PROGRAM The Local Public Works (LPW) program is really a combination of two programs. The first program, generally known as Round I of the LPW program, was authorized by the Public Works Employment Act of 1976. The second program, known as "Round II of the LPW program, was authorized by the Public Works Employment Act of 1977. As a closed end grant public works program, LPW was designed to reduce unemploy- ment (especially in the construction industry and in distressed regions), to in- crease the nation's infrastructure, and to aid local governments. Together the two Acts provided $6 billion of direct federal grants to state and local jurisdic- tions. In Round I $2 billion of appropriations were used to fund 2,010 projects. In Round II 8,542 additional projects were funded for $4 billion. Legislative History- of the Program On July 23, 1976, the Congress overrode President Ford's veto of the Public Works Employment Act of 1976. This Act authorized $2 billion of counter-cyclical assistance in the form of construction grants to state and local governments for public works projects to be used in the period ending September 30, 1977. On May 13, 1977 President Carter signed the Public Works Employment Act of 1977 into law. Title I of this law expanded the total authorization of public works funds by an additional $4 billion and permitted expenditures through the period ending December 31, 1978. A chronology of the events relating to the legislative and early administrative history of the LPW program is supplied in Table 1-1. Im plementation of the Act To implement the LPW program, the Secretary of Commerce, through the Economic Development Administration (EDA), is authorized to make disbursements to any state or local government for projects planned prior to the grant and meeting the basic criteria set forth in the Act. The bills provide the Secretary with the flexi- bility to tailor the programs to local needs. In order to increase the timeliness of the aid, the Round I projects are required to have "on-site labor begin within ninety days of project approval."' Also, to minimize delay, the legislation re- quires that "the secretary shall make a final determination with respect to each application for a grant submitted to him under the Act no later than the sixtieth day after the date he receives such application. "2 In order to "target" the pro- gram toward the areas of greatest need, the 1976 Act requires that in geographic- ally allocating funds "the Secretary shall consider other factors: (1) the sever- ity and duration of unemployment in proposed project areas, (2) the income levels and extent of unemployment in proposed project areas, and (3) the extent to which proposed projects will contribute to the reduction of unemployment."^ More specifi- cally, the law requires seventy percent of LPW funds to be allocated to "applications Public Law 94-369, Title I, Sec. 106(d) Ibid, Sec. 107. TABLE 1-1 L.P.W. CHRONOLOGY 2/13/76 2/19/76 4/13/76 5/13/76 7/6/76 7/21/76 7/22/76 10/7/76 10/26/76 12/21/76 2/24/77 3/10/77 5/13/77 5/13/77 9/30/77 Ford vetoed HR 5247 (a $2.5 billion public works bill) House overrode but Senate sustained the veto Senate passed S 3201 (a $2 billion public works bill) House passed Ford vetoed Senate overrode the veto House overrode the veto; the bill became PL 94-369 Public Works Employment Act of 1976. The Public Works Appropriation Act of 1976 (PL 94-447) appropriating the $2 billion enacted. EDA accepted Round I applications Round I Project Selection ended House passed HR 11 ($4 billion public works bill) Senate passed Carter signed the bill, making it PL 95-28, the Public Works Employment Act of 1977 Appropriations bill became PL 95-29, Public Works Appropriations Act of 1977 Round II Project Selection ended submitted by states or local governments having unemployment rates for the three most recent consecutive months in excess of the national unemployment rate" (if that exceeds 6^ percent). Round II also requires that on-site labor must begin within ninety days and that applications must be acted uuon quickly. In Round II the definition of the unemployment rate was adjusted to be defined as that appli- cable for the area in which the project was to be built instead of using state- wide unemployment rate as in Round I. The time span for the unemployment rate under consideration was extended to the previous 12 month period. Whereas' in Round I, the selection of the projects to be funded was made by EDA based upon the criteria noted above, in Round II Localities that had submitted Round I applications were selected for funding. The selected localities then chose the projects, up to their funding limit, for which funding was desired. These localities were also allowed to submit additional applications and to fund these projects instead of those applied for under Round I. Regarding employment and materials requirements, Round II legislated payment of Davis-Bacon wages to subcontractors and prohibited the use of state employees on projects. Also the Round II provisions required use of mate- rials either mined or produced in the United States. The differences between imple- menting the provisions of Round I and Round II are shown in Table 1-2. Under LPW the federal government agreed to pay for up to 100% of the construc- tion costs of the authorized projects. Money could be used for the local share of projects receiving some other type of federal support. The government would cover only the requested cost of the project. If the locality was required to contribute to a project due to some local law, the federal government would make up the remainder. Legislative Intent In passing the Public Works Employment Acts, Congress attempted to create jobs, provide additional economic stimulus to help the economy out of a recession and to build public infrastructure. During the debates on these two proposals there was agreement on the purpose of the legislation but some controversy surrounding the proper timing of the proposals, the magnitude of the expected impact, and the possi- ble adverse inflationary impact of the program. Indeed, when President Ford vetoed the LPW program, his major objectives centered around the jobs inflation tradeoff. President Ford argued that "the best and most effective way to create new jobs is to pursue balanced economic policies that encourage the growth of the private sector without risking a new round of inflation." 5 The primary objective of LPW was the creation of "new job opportunities for millions of unemployed citizens." 6 In particular, the public works program was 4 — Ibid., Sec. 108(c) and (d). The detail of this formula were slightly changed by the 1977 Act, but the goal of "targeting" relatively depressed regions remains unchanged. Message from the President of the United States transmitting without my approval the Bill (S. 3201) entitled "Public Works Employment Act of 1976", July 6, 1976. Page IV. Senator Randolph, Congressional Record, July 21, 1976, S12121. Provisions of Round I and Round II of the Public Works Employment Act Round I Round II FUNDING AND SET ASIDES Authorized $2 billion for public works projects Added $4 billion to the funding, thus increasing the authorization to $6 billion through 1978. Provides a 2^% set aside for Indian Tribes and Alaska Native Villages Provides a $70 million set aside for grants for projects not received, not considered or rejected as a result of an error made by an employee or official of the United States. These projects must comply with all other standards to be eligible for consideration. ELIGIBILITY CRITERIA AND APPLICATION PROCEDURE The application scoring procedure is as follows: 30% on the number of unemployed; 25% on the rate of un- employment; 30% on the labor intensity of the project (with maximum points given projects with labor intensities between 35% and 80%); 15% on per capital income; 5% bonus for projects' relation- ship to ongoing plans; 10% for long-term benefits of the project; and 5% bonus for general purpose local governments (includ- ing school districts, or 3% bonus for special districts). The time span for the unemployment rate under consideration was the 3 months prior to the application. Funds were to be distributed via the "70-30 split," signifying that 70% of the funds were to be given to areas with an unemployment rate above the national average and the remainder would be given to those localities with unem- ployment rates of equal or greater than 6.5%. The application scoring procedure is as follows: 40% of the score would be based on the rate of unemployed in the project area; 40% on the number of unem- ployed in the project area; 10% on the per capital income of the applicant jurisdiction, and up to 10% priority if the applicant is in a general purpose local government Extends the time period to the previous 12 months. Eliminated this provision and stipulated that: (1) States and local governments submitting more than one project must list their priorities and submit this in the application; (2) States receiving the minimum funding may, with the permis- sion of the Secretary, locate some of their projects in areas with an unemploy- ment rate of less than 6.5% in order to Table 1-2 (continued; Round I Round II (ELIGIBILITY CRITERIA AND APPLICATION PROCEDURE) No state shall receive less than 1% or more than 12.5% of the total funds. insure the widest distribution of funds; (3) 65% of the funds are to be allocated based on the number of unem- ployed and 35% based on the relative severity of unemployment (if their rate exceeds 6.5%) with 85% of funds going to high unemployment areas. No state shall receive less than .75% or more than 12.5% of the total funds. States were only permitted to use state- wide per Capital unemployment rates when applying for LPW funding. Allows the Secretary to consider the unemployment rate of the neighborhood from which the labor force will be drawn. Priority given to local governments. Allows the Secretary to consider the unemployment rate of neighboring areas. Allowed states to use the per capital unemployment rate for the area in which the project was to be built. This cor- rected the problem inherent in Round I where use of state-wide unemployment rates resulted in states receiving very little of the LPW funding. Repeals this provision. Gives priority to state-sponsored proj- ect if the local governments specify that this project is a priority of theirs also. Provides that the project must be within the area whose unemployment rate is used in the project application. The Secretary must only consider applica- tions submitted after the enactment of the Act and on or before December 2, 1976, except for those: (1) from the trust territory of the Pacific Isles; (2) from Indian Tribes and Alaska Native Villages; (3) that arrive late if all of the allocated funds are not distributed. Requires the Secretary to give preference to pending applications which would result in energy conservation. Round II (ELIGIBILITY CRITERIA AND APPLICATION PROCEDURE) The Architectural and Transportation Barriers Compliance Board must ascer- tain that the projects considered will be designed in accordance with the standards of accessibility to the handi- capped and elderly as provided by the Act of August 12, 1968. Prohibits Water Projects from consideration (those projects involving damming, dredging or diversion) . Allows for the funding of drought and disaster relief projects. Monies already allocated for LPW projects can be used for drought and disaster relief projects if the new project's cost does not exceed the cost of the old project. The Secretary's permission is required for this substitution. EMPLOYEE AND MATERIALS REQUIREMENTS Provides that Davis-Bacon wages be paid to contractors. State governments are prohibited from using their own employees for the con- struction of the LPW project. All LPW contracts are to be distributed on the basis of competitive bidding. Contractors must provide assurances that they will not employ illegal aliens. Extends this provision to subcontractors. Special consideration will be given for projects which employ veterans. At least 10% of the grant must be ex- pended for minority business enterprises. The Secretary will determine the correct level of minority participation if the locality has a minority population of less than 5% of the total population. The projects are to be constructed with materials mined or produced in America unless the Secretary waives this as costly or inefficient. expected to provide jobs for construction industry workers, the need for which was viewed as particularly great. During the debate on the LPW program it was noted the unemployment rate of these workers was "approximately double the national average and in some cities approximately half of all construction workers were without jobs."' The Congress was also cognizant of the indirect effect of the LPW program on the materials producing sectors fo the economy. For example, it was noted that the LPW program would "also have a multiplier effect on the building materials industry, hardware retailers, forestry and lumbering occupation's as well as other, fields." 8 The expected impetus in terms of job creation was expected to be less than, but comparable with, other employment programs. As Senator Gravel argued, "although the direct job creation per dollar of outlay may not be as great as in other strategies, the multiplier effect of public works may be greater than that of other job creation measures. " y A presumed advantage of using a program like LPW to create jobs was that it was targetable. For example Senator Humphrey said the acts would create jobs in "precisely those regions, cities and sectors of the economy that are experiencing the most severe unemployment problems."^ The construction of public infrastructure was a second objective of the legis- lation. Having observed recent cutbacks in state and local government expenditures, proponents of the legislation argued the nation's stock of public infrastructure was in need of replenishment. The use of an LPW type program to generate employment would therefore comprise a second goal to the extent that the funds would be used to construct community facilities badly needed by individual communities with a backlog of capital construction projects. A secondary advantage of constructing public facilities suggested by LPW proponents was that these facilities would "in turn create staffing and maintenance jobs in the locality and (that) income. . .(would be) generated and recycled on the local level." 11 The third intent of the LPW legislation was to provide an increase in Federal aid to state and local governments which, it was argued, were suffering greatly as a result of the recession. There was a recognition that high unemployment brought increased demands on local public services, at a time when receipts from taxes were declining. Also, the Congress did not want state and local government budget related actions to undermine efforts by the Federal government to stimulate the economy. The financial problems of New York City had been, of course, well publicized at the time and the Congress felt that "state and local government economic health was essential to national economic properity." ' Report of the Committee on Public Works, United States Senate, Public Works Employ- ment Act of 1975. Page 2, July 21, 1975. No. 94-285. 9 Senator Gravel, Congressional Record, January 25, 1977. Page S1358. Senator Humphrey, Congressional Record, July 21, 1976. Page S12122. Senator Gravel, Congressional Record, July 21, 1976. Page S12122. Policy Issues to be Addressed Since the primary goal of the LPW program was job creation, the net number of jobs created must be the primary criterion for evaluating the usefulness of the program. The measurement of employment is somewhat complex, since there are three distinct types of employment effects -- direct construction jobs, indirect (materials supply industry) jobs, and jobs "induced" by the multiplier effects of increase in economic activity. ( A fourth potential type of employment is the continuing post-LPW jobs created due to follow-on maintenance and program needs. Employment of this type is discussed separately.) Unfortunately not all LPW grants can be translated into direct jobs. Two mitiqatina factors exist in the path from expenditures to increased economic activity. These two factors are substitution and crowding out. Substitution is the use of federal funds to replace local funds on a specific project. The result of course is a reduction in the net increase in economic activity. Crowding out refers to the situation which occurs when the undertaking of a publicly financed project reduces the level of private economic activity by competing with private activities in the factor markets. Crowding out can occur either nationally or locally (or both) in response to competition in the capital, labor, and materials markets. The timing of the employment impacts is another crucial issue for evaluation. It is important to investigate the degree to which jobs were created during the recession, so that the impact was counter-cyclical as originally envisioned. Just as important to the evaluation, however, is the extent to which the LPW program may have been too late to be counter-cyclical. Finally the degree to which an improved LPW-type program could be counter-cyclical should be addressed. The inter-regional impacts of the local public works programs deal with the geographic distribution of the impacts of the expenditures associated with the program. The issues center around the fact that LPW is a "targetable" program. Issues such as regional economic development, equity, and the reduction of differ- entials in the level of unemployment all become major policy issues. Expenditure flows between regions must therefore be investigated in order to determine the final location of the economic activity generated by LPW monies. The Role of Chase Econometric Associates, Inc. Since both the magnitude and the nature of the impacts of a counter-cyclical public works program were far from certain, the Congress mandated that an evaluation of the LPW program be undertaken, to measure the extent to which the program did meet its legislated goals. In partial fulfillment of this mandate the Office of Budget and Program Evaluation of the Office of the Secretary, Department of Commerce contracted with Chase Econometric Associates, Inc. (CEAI) to undertake a macro- economic evaluation of the LPW program. Also being carried out under this mandate are a microeconomic evaluation of the program, undertaken by the Program Analysis Division of the Economic Development Administration, and an administrative evaluation of the program, which is also being supervised by the Office of Budget and Program Evaluation. This report contains the procedures and results of CEAI's evaluation of the effectiveness of the LPW program as a counter-cyclical tool. As a macroeconomic evaluation, it concentrates en the impacts upon employment and economic activity at the national level. However, the evaluation also addresses questions concern- ing the ability of the program to meet almost all of its policy objectives, in- cluding questions concerning regional targeting, the impact upon state and local governments, and the effectiveness of the program in reducing unemployment of particular labor groups. CHAPTER 2: METHODOLOGY The focus of the analysis of the LPW program carried out by Chase Econometric Associates, Inc. is on the impacts that correspond to macroeconomic policy objectives. These impacts include employment generation, effects on local governments, industry impacts, inter-regional effects, and related ques- tions such as the efficacy of LPW as a counter-cyclical program. In assessing these impacts, CEAI has used a variety of analytical techniques, including sur- veying and the simulation of various models of the economy. The purpose of this chapter is to outline these procedures. First, individual methodologies are dis- cussed. Then the sequence of procedures used to analyze each of the major im- pacts is outlined. The result of this analysis are presented in the subsequent sections of this report. Data and Procedures Data Analysis of several of the impacts required data on variables such as indus- trial production, unemployment, and state and local government fiscal behavior. These data are regularly published by agencies of the federal government such as the Bureau of Labor Statistics (BLS) and the Federal Reserve Board, (FRB) and thus need no more comment. In some cases, such as in the analysis of local government revenues and expenditures large amounts of unpublished government (in this case Census) data were also examined. Several sets of data, however, merit discussion because of their relative specificity and the key role they played in the analysis. These data include LPW awards data, Economic Development Administration (EDA) accounting data on actual disbursements, Commodity Transpor- tation Survey (CTS) shipments data, BLS construction labor requirements data, and EDA LPW employment data collected on ED-746 forms. The LPW awards data contain specific information about individual projects and jurisdictions. These data were compiled by EDA from the LPW applications and were updated according to the disposition of the application and the progress of the project. The data served as the basis for: (a) selecting CEAI's sample for surveying; (b) providing the information necessary for classifying projects for the analysis of employment and industry impacts; (c) determining expenditure pat- terns by jurisdiction type, region, and type of capital good; and (d) providing other useful descriptive information for all phases of the analysis. This information includes the project size (measured in dollars), the type of structure, the type of construction activity, a verbal description of the proj- ect, the name of the jurisdiction receiving the funds, the type of governmental unit, the size of the jurisdiction, and locational information about the juris- diction and the project. EDA accounting data were used as the basic information on the actual timing of LPW expenditures. These data record the date of the reimbursement to local jurisdictions of monies that they have spent on the LPW project. These disburse- ments occurred at intervals of up to three months (depending on the billing pro- cedures to the jurisdiction), and were estimated to lag the actual expenditure by an average of about a month. The data were available only through the start of 1978. The Commodity Transportation Survey (CTS) was carried out as part of the 1972 Census of Transportation. Based on a random sample of bills of lading from a stratified sample of manufacturing plants, the CTS provides data on shipments to each state by Transportation Commodity Classification by state of origin. The CTS provided the primary data base for estimation of inter-state and inter- regional flows of materials purchases. Over the years the BLS has conducted several studies of the labor require- ments^ of various types of construction. Although some of these studies are over ten years old, when updated they still represent the best generally avail- able data on the labor requirements of specific type of construction. In some form or other they serve as the primary input for most efforts to estimate the employment generated by construction projects. Each year the BLS updates these estimates based upon changes in labor productivity and the price of each type of structure. The results, in terms of person-hours and person years per unit of expenditure, are available from the BLS in unpublished form. These labor require- ments coefficients 3 were used to estimate construction industry and indirect employment impacts of the LPW program. As part of the evaluation of Round I of the LPW program, EDA has been collecting payroll data from contractors and subcontractors on actual employment. These represent the only primary data on LPW employment. These data, however, were not available in time to be included in the analysis presented in this re- port. Labor requirements coefficients are given for both the construction industry (direct employment) and other industries (indirect employment). Construction industry requirements are further divided into onsite and offsite. Other indus- tries are divided into manufacturing; trade, transportation and services; and mining and all other. The construction categories that were relevant to LPW are private multi-family housing, private single-family housing, general hospitals, elementary and second- ary schools, federally aided highways, sewer works (lines), sewer works (plants), civil works (land), public housing, and federal office buildings. A coefficient for direct labor requirements in utilities construction was developed from data on cost and labor requirements per kw of capacity in a joint DOL/DOE report on labor and capital requirements for electric power plants. Less than 4 percent of employment was generated by this type of construction, however, so BLS will here- after be referred to exclusively as the source of these labor requirements coefficients. Surveys A major portion of the research involved the conducting of two surveys, one of local government officials in recipient jurisdictions, the other of LPW project contractors. Since the intention of the Department of Commerce is to follow-up these surveys with a more extensive data gathering effort designed to evaluate the LPW program from a microeconomic viewpoint, the size of the survey was limited to less than the number of observations which would be sufficient to allow stratification for rigorous statistical analysis. Nevertheless the surveys were large enough to cover a wide range of localities, project types, regions and contractors. The results of this effort showed some fairly clear Datterns and provided information that was extremely useful in many areas of the research. The primary survey was a survey of officials of 50 local jurisdictions receiving LPW funds. These jurisdictions .were chosen to be representative with respect to the population universe of type of governmental unit, population of LPW recipients, total LPW awards to individual jurisdictions, number of LPW proj- ects received, and regional distribution of LPW funds. ^ The sample is presented in Table 2-1. The survey contained two parts, a written questionnaire and an orally admin- istered questionnaire. The written questionnaire dealt primary with current and recent budget and financial management. One written questionnaire was sent to each jurisdiction. The oral questionnaire covered a range of topics, including capital budgets and financial planning, general economic conditions, the decision making process that led to LPW applications, actual and potential local funding for both LPW-funded and rejected projects, potential difficulties with contrac- tors, and community attitudes toward the LPW project. Some of these questions were designed to elicit perceptual (rather than necessarily factual) responses. In each jurisdiction a range of officials was interviewed, including (where pos- sible) the LPW coordinator, policy makers, financial officials, and elected officials. Survey sites were restricted to counties, cities, and school districts. Other types of jurisdictions are to heterogeneous and few in number to provide mean- ingfully interpretable results in a survey of this size. Cities receiving large amounts of LPW funding and large numbers of LPW projects were sampled at two to three times the proportions appearing in the population. Two-thirds of the sample jurisdictions were located in SMSAs, so as to over-sample urban jurisdictions. 12 C\J r— r— i — CM LO C\J . — W i — LDi — i — i — O OJ CJ Ol 0) oo en U3 O CO o <— r— IT) O CsJ to O O cr> O "* ud ro c\j O CO U3 W i — W Ol Oi o w o OO O LD _ r— CO r— CT. r— =3 <; C3 S Z l^ C3 C! at _i oj tr> jc 3 ai -D v at i~ r - i- cr> r— at <— -a -i- a» •— c -— (_> CJ a v: s: 2: 2: z o. iX) vo r-^ r~- CO CTi O 1 — . _ ro «* to N M !\J CM in en >3- w in CM -c^" 00 CM .— -— i— CM CM OO ■— CM d) O" QJ QJ m O) O) OJ OJ OJ § I O CO O CO r— IT) vo CO O i— F— r— UO LO O CO O CO o oi id oi n LnCMOOCMLDCTiO^CMO'lO'i ' CO r— CM i— r— x: _c (D J) < -I 4-> +J 2 4J +J oo o o z oo ■z. -z. z d) cu <: 01 01 c oj cu a) •«- Q. 3 -r- QJ CM CM CM OO VX3 P^ CO CTi O r~ 00 The survey of local government officials served many purposes. The most im- portant of these was to provide data for measuring and analyzing the amount of funding substitution. Survey results were also used in the analyses of inter- regional flows of labor, crowding out, 5 and several other specific issues. More- over, the survey generally provided a background familiarity with the workings of the L.PW program at the local level. The second survey covered 27 contractors, all of whom were working on projects included in the survey of local government officials. This guestionnaire-was admin- istered orally to the contractor or official of the contracting firm. It included guestions on the employment qenerated by the LPW job, employment alternatives, sources of labor and materials supply, and any recent or potential changes in use of materials, labor or technology. This last line of questions was designed to deter- mine the possible need for updating non-current input/output tables for analysis of industry impacts. (No need was found.) The other primary area of application was in estimating possible inter-state and inter-renional flows of labor and contractors. The results of this survey were also used in the analysis of crowding out and of em- ployment impacts. Althouqh the sample was too -small for statistical rigor, the uni- formity of many of the results of this survey suggests that they can be used with considerable confidence in addressing a number of important--if limited—questions. Econometric Models Econometric models were used as an integral part of the analysis in many parts of the research. Two models used particularly heavily were the CEAI Macro- economic Model and the Chase/INFORUM input/output Model. The effects of the LPW program on the economy as a whole were estimated by a simulation of the CEAI Macroeconomic Model. 6 First, a baseline simulation was run using actual historical values of variables through 1978.2 and CEAI standard forecast values from 1978.3 to 1980.4 of all variables. Second, a policy impact simulation was prepared deducting LPW expenditures from this baseline. The officials from 7 extremely isolated jurisdictions receiving over $2,000,000 in LPW funds. These officials were asked questions pertaining to the impact on the local economy and the sources of contractors, labor, and materials. The CEAI Macroeconomic Model is a regression-based model of the United States economy. It analyses the consequences of economic and political events on approximately 350 economic indicators ten quarters into the future (ten years, on an annual basis). The data base is continually updated, and the forecasts are updated each month to reflect revisions in the data and changes in the economic environment. LPW expenditures, by quarter, were subtracted from federal grants-in-aid to state and local governments. State and local expenditures were reduced by the amount of LPW expenditure net of estimated substitution. Direct materials sup- ply industry impacts estimated from the Chase/INFORUM Model were also deducted, since state and local construction does not presently impact the materials sup- ply industries in the CEAI Macroeconomic Model. difference in results was attributed to the LPW program. This procedure was used to obtain estimates of construction industry employment, total employment, crowding out, and spillover effects. Another set of simulations was run to evaluate the effects the LPW program would have had, had it been initiated in the second quarter of 1974 -- 90 days after the earliest possible identification of the latest recession. The Chase/INFORUM Model was used in several contexts to determine the impact of the LPW program on materials supply industries. The first step in this application was to classify the LPW projects into two dozen construction categories. From the input/output inverse coefficients the materials require- ments of expenditure on each type of construction were determined. Labor require- ments to produce these materials were also determined, based on measures of pro- ductivity. By this means estimates of the indirect employment effects were derived. The direct input/output coefficients were also used to provide estimates of the direct materials requirements to estimate the first round materials flows between states. The INFORUM Model was also used in assessing industry impacts in the LPW program. Analysis of Ma jo r Issues Timing of Expenditure Underlying the analysis of both the timing of impacts of the LPW program and the size of the impact at any point in time is the pattern of LPW expenditures over time. EDA accounting data on disbursements were used to measure this. Unfortunately, however, these data were only available through very early 1978. Thus the expenditure for 1978 and beyond had to be predicted. The first step was to determine the time pattern of expenditure on completed Round I projects. For each project, disbursements were smoothed to a stream of daily expenditures y and then aggregated to monthly expenditures and seasonally adjusted. These expenditures were dated one month earlier than the disbursements to allow for a lag in f 11 inq for reimbursement. The percent completion of each project was then calculated'^, and a subset containing all completed Round I The Chase/INFORUM Model is a version of the University of Maryland's inter- industry forecasting model which uses a combination of input/output analysis and econometric modeling to determine annual industry outputs based on forecasts for all product markets. It is design linked to the CEAI macroeconomic model and provides projections of the activity levels of 200 industries and detailed fore- casts of industrial output and employment. Q Each disbursement was assumed to have been spent in equal daily amounts over the period since the previous disbursement. For the first disbursement, the expen- diture was allocated evenly over the previous month. The percent completion was measured by the ratio of the cumulative total of dis- bursements to EDA obligations for the project. projects was selected. Classifications based on the type of construction were also explored, but no significant or consistent differences in the timing of expenditures were found with respect to the size adjusted types of construction. 3 For each size class, a monthly cumulative percent of expenditure was then computed. Finally, values were interpolated to produce a function of cumulative percent of expenditure relative to cumulative percent of time for the projects completed in the period covered by the data. Figure 2-1 shows the graph of this function for the -smallest size class. The non-linear pattern of expenditure reflects three factors: • Expenditure on a given project is highest in the middle of the period because of gradual start-up at the beginning and phasing out at the end of a project. • Differences in the time of initiation of individual projects cause the aggregate level of expenditure to rise initially over a period of months. • Different lengths of time required to complete each project cause aggregate expenditures to fall during the latter half, of the period. The next step was to predict the timing of future LPW expenditures, based on the actual timing of expenditures on completed projects. In making these pro- jections it was assumed that the time pattern of expenditures -- in terms of per- cent of expenditures relative to percent of time -- would be the same for all projects as it was for completed projects. The percent of expenditures on all Round I projects through December 1977 was calculated for each size class. Next the corresponding percentage of time that the first twelve months represented for that size class of Round I projects was derived from the percentage expenditure/time 11 12 Completed projects were defined as those for which 99 to 101 percent of obligated funds had been spent. The size classes were $1 to $199,999; $200,000 to $499,999; $500,000 to $999,999; $1,000,000 to $2,999,999; and $3,000,000 and over. There were eight distinct construction categories with a share of LPW funds over about 4 percent. Upon visual inspection of graphs of exoenditure (month- by-month, or cumulative), it was apparent that only two or three of these dif- fered from the average timing pattern. When standard deviations of the quar- terly cumulative percentages of expenditure were calculated, the standard devi- ation within each construction category was found to be at least twice the size of the difference between means of construction categories. FIGURE 2-1 TIME PROFILE OF LPW EXPENDITURES ($0-199,999 SIZE CLASS) cumulative % expenditure 1.0 cumulative 1.0 % of time Based on this percentage, the total time (in months) for completion of the size class as a whole was calculated. '^ Finally monthly expenditures for the size class were derived from the percentage expenditure/time function. The results were summed over size classes for each month and aggre- gated on a quarterly and annual basis. - For Round II expenditures only four months of data existed, which was not enough to obtain any independent estimate of timing. Moreover CEAI's survey experience suggested that the exceptionally hard winter had substantially delayed initiation of many projects. Aggregate state and local construction data, how- ever, suggested a boom in the second quarter of 1978. Thus Round II expenditures were predicted in the following manner: For the first quarter of Round II (October-December 1977) historical data were used. For the next quarter expendi- tures were assumed to have maintained the December 1977 level. For the third quarter the cumulative shortfall from the first six months of the Round I pattern was added to the expenditures for each class. Beginning with the tenth month (July 1978), each class of Round expenditure of each size class was assumed to follow the same pattern (in percentage terms) as Round I expenditure. As in the case of Round I, the monthly percentages of expenditure were multi- plied by size class expenditure totals, summed across size classes, and aggregated to quarterly and annual values. 14 There were not enough completed projects in the largest three size classes to calculate stable functions relating percentage expenditure to percentage time. Thus the time pattern for the second smallest size class adjusted for the size of the projects was used for these size classes. 15 This computation of the total time required for completion is equivalent to solving for "Months to completion" in the formula: Expenditure through December 1977 12 Total Obligations Months to Completion where December 1977 is the 12th month of the project. The estimated number of months to complete the five size classes of projects (smallest to largest) were 19, 20, 23, 28, and 36, respectively. Through December 1977 actual expenditure data were used. For subsequent months, the time through that month was taken as a percentage of total months, the corresponding cumulative percent of expenditure was determined from the function, and the cumulative percent of expenditure for the previous month was subtracted from this total. The remainder was then multiplied by total expenditure for the size class. Expenditures for Round I, Round II, and the LPW program as a whole are shown in Table 2-2 on a quarterly basis and in Figure 2-2 on a monthly basis. These expenditures streams form the basis of the analysis of employment creation over time (Chapter 4), Industry impact over time (Chapter 5), and evaluation of LPW as a counter-cyclical policy (Chapter 8). Estimation of Employment Impacts Construction industry expenditures create employment at three levels: Direct employment (in the construction industry itself), indirect employment (employment in the materials supply industries and other related employment), and induced employment (non-specific employment brought about by a higherjevel of aggregated demand). CEAI used several approaches in estimating these impacts, including the application of BLS labor requirements coefficients, input/output table coefficients, and macroeconomic simulation. This section deals. with these methodologies. Actual results are discussed in Chapter 4 and are presented in summary form in Tables 4-1 and 4-2. Both direct employment and indirect employment were estimated from BLS labor requirements. The first step in this analysis was to classify the LPW projects by BLS construction categories. In most cases this was done on the basis of the structure codes contained in the awards data, although the activity codes were also determining factors for some types of projects. Several miscel- laneous structure types also had to be classified by project description. The exact classification of work-type codes is shown in Appendix A. Projects in each construction category were then disaggregated by round and size class, and quar- terly expenditures were computed for each size class. Expenditures were reaggre- gated by quarter over the ten round/size classes for each construction category.'' Once the expenditure streams for each construction category were determined, the expenditures could be multiplied by the corresponding employment requirements coefficients. Here, however, a problem arose because the LPW expenditures were spread over more than three years. During this time significant changes in prices and productivity could be expected to take place. 18 The BLS makes these adjust- ments using a variety of price and productivity indices, but they are not avail- able on a forecast basis. CEAI therefore assumed that the percentage change in the BLS labor requirements between 1976 and 1977 will also occur between 1977 and 1979, and 1979 and 1980. These adjusted coefficients are shown in Appendix A. These adjusted annual labor requirements coefficients were then multiplied by the corresponding quarterly construction category expenditure streams to produce quarterly employment estimates. This process was done both for the BLS construction industry labor requirements and the BLS "other industry" labor requirements to diture by the percent of round/size class expenditure measured or estimated for that quarter, as described above. The disaggregation was made necessary by variation among construction categories of the proportions of expenditure in each round and size class. Both price and productivity increases will decrease the employment generated per dollar. Price increases will mean that more dollars are required to build a given structure; productivity increases will mean fewer person-hours are re- quired. Because of these factors, the BLS estimates of labor requirements per dollar fell by an average of over 12 percent between 1976 and 1977. 20 TABLE 2-2 QUARTERLY LPW EXPENDITURES ($ 1000s) Quarter Round I Round II Total 1977.1 72,864 72,864 1977.2 200,422 200,422 1977.3 326,462 737 327,199 1977.4 323,299 240,393 563,692 1978.1 311,784 469,114 780,898 1978.2 237,330 1,058,427 1,295,758 1978.3 167,328 664,809 832,137 1978.4 102,157 587,779 689,936 1979.1 66,219 418,172 484,391 1979.2 34,425 259,862 294,287 1979.3 18,709 142,431 161,140 1979.4 6,431 83,128 89,559 1980.1 38,866 38,866 1980.2 20,507 20,507 1980.3 7,049 7,049 9>. i <$,000't$> 22 produce, respectively, estimates of direct and indirect employment. This direct employment estimate was incorporated into CEAI's results. The indirect employment estimate was supplemented by the following procedure. Indirect employment effects were also measured using the Chase/INFORUM (input/output) Model. The procedure was quite similar to that used with the BLS labor requirements coefficients, but it was more complex because there are about twice as many relevant input/output construction categories as BLS cate- gories. First the projects were classified into input/output categories, based upon structure codes, activity codes, and project descriptions. The exact'clas- sifications are shown in Appendix A. Next the expenditure stream was computed by round/size class for each input/output category and agareqated over the round/ size classes annually. During this process some adjustments were made for varia- tions in timing of use for some of the major materials.^ The result of this process was a stream of annual expenditures for each input/output category. 20 The coefficients used were the inverse requirements coefficients of the Chase/INFORUM Model, aggregated to 57 sectors. The labor requirements of each of these sectors were adjusted to an annual basis for Dredicted productivity and 19 20 Estimates of these variations were obtained throuqh conversations with trade associations which were familiar with the stage of construction requirements of different types of construction. The direct coefficients in an input/output table measure the labor and mate- rials inputs required to produce a given material. The inverse requirements matrix, measures the materials requirements at the first and all previous stages of production. Like other input/output coefficients, these coefficients represent average labor requirements. It is arguable that marginal labor requirements should be used for analysis of individual programs. (See, for example, Georges Vernez, et . al . , Regional Cycles and Employment Effects of Public Works Investment . R-20520-EDA' (SantaMonica : Rand, 1977.) It is doubtful, however, whether mar- ginal labor requirements can be estimated of marginal labor productivity (which is the reciprocal of labor requirements) range from negative values to more than 40 times as large as average productivity. Many of these values are not even remotely plausible. Moreover it is doubtful that marginal labor require- ments differ much from average requirements in this case. As a long run propo- sition at least, the marginal cost of production has generally found to be con- stant -- and equal to averaae cost -- in the relevant range. The issue thus be- comes one of short-run fluctuations. Yet cyclic chanqes in manufacturing produc- tivity have usually been observed to be quite small, even on a quarterly basis. Even during the period 1973.4-1975.4, the period of the worst post-war recession, productivity fell only slightly below the trend line maintained in subsequent ex- pansionary periods. If there were large changes in the marginal productivity of labor, inventories could (and would) be used to equalize the value of output per employee. Furthermore, one would expect substantially larger cyclic variations in pricing policies than are commonly observed. Finally, even if marginal labor productivity (and requirements differed from averaqe in booms and recessions, the LPW program is occurring during a period which might well be described as one of "average" economic activity. Output is well above recession levels, but below the level at which shortage would be a danger. Thus even if cyclical fluctuations were important, the economy is in an intermediate situation where marginal produc- tivity can be expected to be close Ui averaqe productivity. price changes, using CEAI forecast values. Each sector of industry labor require- ments was then multiplied by the stream of expenditures for the corresponding input/output construction category. The results of this operation were summed over industries and over input/output construction categories to obtain annual em- ployment estimates. For comparability with other quarterly estimates a quarterly employment series was generated by allocating the annual employment creation to the quarters in the same proportion as estimated LPW-generated expenditure on mate- rials. CEAI judges these results to be more accurate than the 8LS based indirect employment estimates, because of the greater detail of the Chase/INFORUM Model and availability of forecast adjustments. CEAI also used its Macroeconomic Model to forecast LPW employment. This simu- lation utilized the estimated quarterly stream of net LPW expenditures and of expen- ditures on materials. The output of the simulation included quarterly employment impacts in the construction industry (direct) and on the economy as a whole (the sum of direct, indirect, and induced). In addition, the simulation showed some negative effects on expenditure in the residential and non-residential construction sectors. These were interpreted as crowding out effects, and were converted to employment estimates using the adjusted BLS labor requirements coefficients. Simulation of econometric models is the only available technique for the estimation of total employment. Unfortunately the results are not directly com- parable with labor requirements coefficients results. The labor coefficient tech- nique produces results in terms of person-hours or person-year equivalents. The macroeconomic simulation results are in terms of the number of people who are "employed." Since most construction work was extremely short-term, 2T much of this work may have been done by people who were already "employed." This portion of the effect would not tend to be counted in the simulation results. To overcome this difficulty, the simulation results were transformed into an employment multi- plier. The employment multiplier was used to expand the direct employment esti- mates" based on labor requirements in order to produce an estimate of total employ- ment that was comparable with the BLS-based or INFORUM-based estimates of direct and indirect employment. 21 EDA estimates of PWIP employment suggest that the average job is slightly less than a month in duration. Preliminary ED-746 results confirm this estimate. The employment multiplier is defined labor required in direct employment. CEAI actually multiplied the employment multi- plier by the average of estimated direct and indirect employment, which were approximately equal overall. This was done to reflect differences in timing of the two types of employment rather than simply to project the timing pattern of direct employment. 24 Estimation of Impacts Upon Governments The analysis of impacts upon local governments involved two major issues. The first of these was the definition and identification of the trend and cyclic positions of state and local governments (including the effects of economic con- ditions on local governments and their responses to them, the degree of locality distress, and differences among types of localities). The second was an analysis of the importance of LPW expenditures relative to this cyclic position. The pro- cedure used was generally interpretive analysis and comparison of data series, although some econometric estimation was also used. Findings on these issues are discussed in Chapter 5. Both the concept of "trend" and of "cyclic position" are inherently judg- mental. However, efforts were made to date and to suggest the importance of both factors. The concept of cyclic position used in the analysis was the effect that general economic conditions have on the state and local fiscal situation. ^ Locality trend was estimated based upon expenditure and tax burden" data. Exploring the impact of cyclic conditions on localities and the localities' responses was of course more complex, and many time series were examined and interpreted, including: • State and local government expenditures • State and local government capital expenditures, by type of structure • State and local taxes • State and local surpluses the effect of state and local net expenditure on the economy as a whole. The measure used was a ratio of state and local taxes to the sum of personal income, minus total federal personal taxes, minus transfer payments. Specific measures included: • The ratio of composite state and local interest rates to federal bond interest rates t The ratio of state and local interest rates to composite corporate bond rates • The ratio of municipal bond interest rates to Moody's AAA corporate bond rates 25 • A measure of state and local tax burden Construction activity These series were analyzed for trends and cyclical movements. Significant turning points were compared with NBER reference cycles. Trends were compared with infor- mation on demographic and political developments over the last two decades. The relationship among the different variables themselves, of course, was a key focus of the analysis."* The insights gained from this analysis were consolidated into a hypothesis about the timing and flexibility of response of localities to general economic conditions and a regression equation to predict state and local govern- ment construction expenditures. Further analysis was then carried out on individual classes of government. Measures of their fiscal situation -- including interest rates, bond rating classes, debt load", surplus, and welfare expenditures -- were examined and com- pared to establish relative needs for counter-cyclical assistance or these classes of government. The analysis concluded with a comparison of LPW expenditure with state and local government spending. First LPW expenditures were compared with capital bud- gets^ by locality class. These comparisons were related to the relative need of locality classes. Next projections were made of state and local capital expendi- ture adjustments during the LPW period based upon the decline from the 1974 local maximum. LPW expenditures were then compared to these adjustments. 26 29 The ratio of state and local BAA bond rates relative to state and local AAA bond rates The first two of these measures were designed to show the fiscal tightness of state and local governments relative to other sectors of the economy. The last two measures were designed to reflect the risk factor in state and local govern- ment finances. The measures used were ratios of property tax receipts per house to the mean value of new houses, and to the median value of new houses. Some of the variables, taken individually, had ambiguous interpretations. A declining surplus, for example, could reflect an uncontrollable recessionary reduction of taxes and resultant tightening fiscal situation, or it could re- flect intentional increases of expenditures due to a comfortably recovering fiscal situation. Estimates of construction budgets were used as an alternate measure. 26 For this analysis, considerable difficulty was encountered simply defining state and local capital budgets. 30 The problem extends far beyond differing accounting procedures. Formal procedures for allocating and transferring funds differ widely, as do the relationships between different funds in the budgets and the treatment of non-local funds. The differences between informal ad hoc procedures and formal ones is perhaps even more important. As a result CEAI used two concepts of capital budget: The first was Bureau of the Census defini- tions, which are comparable to National Income and Product Accounts concepts. The second was CEAI survey responses, which reflect locality perceptions of capital budgets. Estimation of Industry Impact s Assessment of industry impacts involved four basic stages. First the affected industries were identified. Next their cyclic positions and distress were analyzed. Third, the impacts of LPW expenditures on these industries were calculated. Finally, the impacts were compared with the state of the industry at the time of the LPW pro- gram. For the most part the analysis involved applying the data bases employed with, and the relationships embodied in, the Chase/INFORUM and CEAI short term industry models. Simulations of the economy were also used. Results are discussed in Chapter 5. Identification of affected industries was done using the Chase/INFORUM (input/ output) model. This model was aggregated to 57 J supplying industries, which generally correspond to the 3-digit, and often the 4-digit SIC code level of detail. Construction was disaggregated into 24 types, using Bureau of Economic Analysis (BEA) categories. The industries, construction categories, and direct requirements coefficients are shown in Table 2-3. Analysis of the cyclic position and distress of industries entailed evaluation of both the cycles themselves and their relation to the NBER reference cycles. The period of analysis was 1957-1977. Analysis was carried out for two dozen major in- dustries on the 2-digit and 3-digit level of detail. Industrial production indices were used to reflect the status of the industry. The data used included: • Peak and trough values • Peak and trough lags behind NBER reference cycle turning points • Percent changes from peak to trough and trough to peak t Trend growth rates means other than judgmental assumptions. The exact number of industries varies from procedure to procedure. This is partly due to the fact that some industries are not sufficiently important to be of interest for certain analyses. It also reflects differences between the definitions of industries in the input/output classifications (used in INFORUM) and FRB industrial production indices. 27 saanddns iviaaivw 28 saanddns iviasiw 29 SHBIIddflS 1VIH31VW Cyclical characteristics of different industries were compared, and the stability of the cycle of each industry over the entire study period was evaluated. The share of output of each industry whose output is used in construction -- and thus the potential for a program such as LPW to assist the industry -- was also analy- zed. A measure of industry distress was defined as the percentage decline of the production index from peak to trough during a recession. 32 Finally, in order to establish further the "normal" pattern of recovery, the CEAI Macroeconomic and Industry Models were simulated for the 1970 and 1974 recessions under the assump- tion that federal counter-cyclic monetary and fiscal policy had not taken place. -" Impacts of LPW expenditures on industries were traced by applying data and projections on expenditure to the Chase/INFORUM Model. CEAI's estimated quarterly expenditure streams, disaggregated by type of construction and adjusted for time phasing of major inputs, were mul tipl ied times the Chase/INFORUM total requirements coefficients. 34 The results of this procedure were summed by quarter and industry to produce a quarterly stream of value of shipments, in constant dollars, for each industry. Comparison of LPW impacts with the state of the industry involved three pro- cedures. First the impact of LPW in 1978^5 was computed as a percent of the pre- viously derived measure of industry distress (percent decline of the production index from the 1973-74 peak to the following trough). Next the value and movement of each production index in 1977-78 was compared to the previous peak. Finally a computation was made of the amount of LPW expenditure that would have been required to produce a 20 percent reduction in distress for each industry. These results were then analyzed to determine the efficacy of. the amount and timing of LPW expenditures in reliving industry distress. 32 Rates of return were also explored as a measure of distress. They were found to be of very limited value, however, because the firms for which data are available are typically active in several industries. Because of the mixed motives behind most fiscal measures, the identification of "counter-cyclical" policy measures is necessarily based on informed judgment. The specific modifications used are: For 1970 . The phase-out of the income tax sur- charge was revised, as the Phase II (1971) personal tax cut. The investment tax credit was not re-instituted. A steady 4% growth in the non-borrowed monetary base less currency was assumed. For 1975 . The March tax reduction act affecting 1975 and 1976 was eliminated. Transfer payments to persons in 1975 and 1976 were reduced by approximately $4.5 billion. Grants-in-aid were reduced by an amount which grew steadily from $5 billion to $18 billion, with 50% of the change being reflected in state and local expenditures. Federal direct purchases in 1976 were reduced by $3 billion. A steady 7.5% growth in the non-borrowed monetary base less currency was assumed. These coefficients include both direct requirements of construction and require- ments of earlier stages of production. This represents the maximum level of impact of LPW, since over 60 percent of expenditures took place in 1978. Estimation of Inter-Regional Impacts An assessment of inter-regional flows of expenditures were carried out to determine the effects that materials purchases had in altering the regional dis- tribution of LPW impacts and to gain insights into the feasibility of geographical targeting of expenditures. First, a suitable set of regions was defined based upon economic homogeneity. Next the flows between states and regions were esti- mated. Finally these flows were analyzed to determine the net regional impact of LPW expenditures. A variety of techniques was used, including correlation analy- sis, input/output analysis, and the compilation of survey information. Findings are discussed in Chapter 7. As a preliminary step to defining the regions, the Census regions were -, fi analyzed for cyclic homogeneity. Variables were devised to measure cyclic timing, severity, 3 ^ duration, 38 and volatility 3 ^ of unemployment in the component states of each region. Statistical tests were made for the 1971 and 1974 recessions to deter- mine if there were significant differences between regions for these variables. Few cases of significant difference were found. The degree of correlation of these variables from recession to recession for individual states was also computed and found to be low. It was therefore concluded that neither the Census regions nor individual states provided a particularly useful definition of regions for evalua- tion of a counter-cyclical program. In defining reaions, the most important policy characteristic is that they be homogeneous with respect to unemployment. For reasons of data availability, the regions were constrained not to subdivide states. Pursuant to these criteria, CEAI computed the correlation coefficient between states for the monthly unemploy- ment rates from January 1965 to July 1977, inclusive. As a preliminary rule of thumb, all adjacent states with a correlation coefficient of 0.8 or greater were grouped together. This procedure classified approximately two-thirds of the states into distinct regions. Additional judgment had to be used in classifying two types of cases. Some states on the border of a reqion had unemployment rates that were highly correlated with that of states in both adjacent regions. This was apparently due to the dual nature of those states' economies. Other states M ng unemployment (using relative weights of 1 , 2, and 1). The absolute unemployment rate was used for troughs. For peaks the cyclical deviations from trend of the unemployment rate were used. Severity is measured in two ways. The first is the difference between the maximum unemployment rate and the coinciding value of the secular (trend) level of unem- ployment. The second is the cumulative amount (integral) of unemployment above the secular level during the recession. Duration is the time in months between a peak in unemployment and the previous trough (see "timing" for these definitions). Volatility is the ratio of the cyclical maximum of unemployment to the number of months elapsed from the beginning of the recession. 32 were not particularly highly correlated with any adjacent states. In several cases (plains states) they were grouped together as a separate region. In other cases they were judgmentally added to adjacent regions. The resulting regions are shown in Figure 2-3. The projections of flows of materials was done on a state-to-state basis using the 1972 Census of Transportation Commodity Transportation Survey (CTS.) results. First LPW projects were disaggregated by state of location and aggre- gated into the construction categories used in the analysis of industry impacts. Using the direct Chase/INFORUM materials requirement coefficients, 40 materials requirements by state were calculated. Adjustments were then made for local sources of supply and definitional problems with structural steel 4 ! on t ne 57. industry level of detail. The CTS results were then aggregated to the 57-industry 40 Tracing materials flows father back into the production process was judged to be beyond the scope of this study, since each stage in the production process would have to be analyzed in the same manner as direct materials requirements. This would require successive applications of the input/output model and CTS data. Furthermore, in the earlier stages of production much of the material was undoubtedly intra-firm shipments. Thus the use of CTS data would provide a misleading view of the inter-regional flows. Using the total requirements coefficients as a shortcut was judged to be an inferior technique, since it would implictly assume that all phases of production originated at the project site, which is clearly incorrect. Also, spot checks revealed that in most cases the earlier stages of production occurred in the same region as the "final" materials stage. The CTS data do not include shipments of less than 25 miles. Locally supplied materials were thus excluded from the subsequent calculations. "Locally sup- plied materials" were assumed to comprise agricultural production, mining (primarily stone, clay & gravel), and all services. This involved nine indus- tries. The problem with structural steel is that its "fabrication" is really distribution rather than manufacture of products. Thus, supplies of fabri- cated structural steel were traced one step farther back in the production process. This reduced the number of industries to 56. These matrices were then converted to percentage form, by state of destination. That is, for each state of destination a sector was computed that showed the .. percent of the material shipped to that state which was shipped from each state. These coefficients were then multiplied by the LPW materials requirements for each state, and the results were summed over materials. The result was a single origin- destination matrix of values of shipments of materials, on a state-to-state basis. The results were further aggregated into a value of net imports of materials for each state as a result of the LPW expenditures. Finally net material imports were deducted from total state LPW awards to obtain a measure of net LPW impact. ^ The above procedures do not deal with three possible sources of titer-state flows of expenditure: (a) because the CTS data deal with shipments from manufac- tures to distributors, they do not allow for possible purchases by contractors from out-of-state distributors; (b) contractors themselves may come from a state other than that in which the project is located, in which case the expenditure flows across state lines; and (c) labor may be hired from out of state. In the case of contractor's materials purchases, it was assumed that conven- ience to the site and transportation costs would tend to make most purchases local. Conversations with contractors confirmed this assumption. Analysis of the 42 Most of the CTS data are at the level of detail of 3-digit or 4-digit SIC indus- tries. These have to be aggregated to match the input/output industries, which are at the 2-digit and 3-digit level of detail. In some cases, however, CTS data were at a much higher degree of aggregation. When this occurred, it could not be included in the subsequent calculations. This generally occurred in non- industrial states, for which the CTS data were rather sparse anyway. Results involving states with such data problems were computed and reported separately, as well as being included in the overall results. In order to assess the sensi- tivity of results to the omission of the highly aggregated CTS results, the entire estimation of flows was repeated on a more aggregated level, which in- cluded most of these observations. These results varied only slightly from the basic estimates. At this stage the note 40). There were thus 47 matrices of size 48x48 (for the contiguous states) . The original CTS data were arrayed by state of origin. The purpose of this pro- cedure, therefore, was simply to interpret the data from the other side - by state of destination. Equivantly, the net ii that state, plus net exports of materials. type and size of projects strongly suggested that contractors would tend to be .-, local, and the type of labor requirements also strongly suggested local labor. It was assumed, therefore, that materials suppliers, contractors, and labor were all in-state. CEAI's survey of contractors and subcontractors found very few ex- ceptions to this pattern. Only three types of interstate flows were noted: • Highly specialized materials or labor sometimes came from out of state • When states were yery small (primarily New England) materials, con- tractors, or labor sometimes came from an adjacent state e When states were very sparsely settled (primarily in the plains and mountains of the West) some inputs came from an adjacent state None of these exceptions appeared to involve significant inter-regional flows. CEAI therefore allowed the assumption of in-state materials, contractors, and labor to stand. 4 ° Thus the estimate of inter-regional flows of expenditure consists entirely of materials flows. Estimates of inter-regional flows of materials were made by aggregating the estimates of inter-state flows. The net impact of LPW, by region, was computed in the same manner used for analysis of states. Net imports were also computed as a percentage of LPW grants to the region. Finally the impact of the LPW program, both gross and net of materials shipments, was compared with the regional shares of personal income tax to assess the inter-regional distributinal effects of the LPW program. These results are discussed in Chapter 7. 48 from any distance. The fact that they were local public works projects tended to give a competitive edge to contractors who were familiar with public-sector construction and with whom local officials were used to dealing. Such contrac- tors tended to be local. Construction labor generally tends to be less mobile than other labor. Unskilled labor is similarly relatively immobile. Davis-Bacon requirements tended to re- quire relatively high wages. If the work was done by a union contractor, the union tended to be a barrier to mobility. If the work was done by a non-union contractor, the highly paid skilled labor tended to be a regular employee of the contractor, and thus local if the contractor was local. This result is strongly influenced by CEAI's definition of regions. Since regions were established based upon homogeneity in unemployment there is little tendency for contractors, labor, or materials to lie near regional boundaries. Census Bureau regions, by contrast, cut through many urban areas, and had those defini- tions been used a considerable amount of inter-regional movement of contractors and labor might be expected. CHAPTER 3: THE PATH OF LPW FUNDS Before policy issues concerning the impact of LPW funds can be addressed it is important to understand what actually happened to the funds which were appropriated for the LPW program. The uses of these funds, the restrictions on them, and their side effects are the subject of this chapter. First tbe path of the funds is traced. Then two major side effects of LPW expenditures -- substitution and crowding out -- are discussed and evaluated. The results pre- sented in this chapter are an "intermediate good" in the sense that the policy re- sults presented in Chapters 4-8 of this report are based upon the understandinq of the path of LPW expenditures as described in this chapter. Uses and Restrictions of LPW Funds The funds authorized for the Local Public Works program were to be used as categorical grants to state and local governments for the construction, demoli- tion, renovation, or repair of public facilities.' The concept of construction includes architectural and engineering planning, but excludes acquisition of real property or maintenance. Further requirements stipulate that contracts shall be awarded by competitive bidding to private sector contractors, and that materials shall be domestically produced. These requirements and restrictions substantially determine the nature of the industry impacts of the program. LPW grants may be used to make up the state or local share that matches Federal public works grant under another program, or the local share to match a similar state grant. The LPW funds may not, however, be used unless a project has not been initiated because of a lack of non-Federal funds. The level of construction industry employment generated by the LPW program is primarily determined by the types of construction projects undertaken and the method of construction. In Round I the estimated labor intensity of the project was included as a criterion for selection of the project. The quality of these estimates were uncertain, however, and selection of individual projects for Round II was left up to the localities. In Round I priority was given to projects and localities on the basis of unemployment. 3 In Round II this priority was assigned on a state level only. Geographical targeting of LPW funds consisted primarily of The funds were authorized as part of the "Public Works Employment Act of 1976" (Public Law 94-369) and "Public Works Employment Act of 1977" (Public Law 95-28). See Chapter 1 for a full description and chronology of the LPW program. "Other" construction related improvements were also presented. Estimates were made by the applicant jurisdictions. They were not checked, and the methodology is unknown. Because "the extent to which proposed projects will contribute to the reduction of unemployment" was known to be a criterion in proj- ect selection, there is a priori reason to believe that the estimates were biased upward. Both the unemployment rate and the absolute level of unemployment were used in the selection process. targeting grants toward localities or states with high unemployment. The authori- zation also contained a floor and ceiling on the amount that each state was to re- ceive. In both cases these requirements influenced the distribution of jobs rather than the size of the effect on unemployment. Many provisions in the authorization are designed to speed up the timing of impact. The most important of these was, of course, the date of appropriation. Additional provisions included requirements that (a) specific rules for project selection were to be promulgated within thirty days;(b) applications were to be ruled upon within 60 days of receipt by the federal government; and (c) on-site labor was to begin within 90 days of project approval. No federal or state funds from other programs were to be used unless they were "immediately available." Round I funds were to have been spent by September 30, 1977, and Round II funds by December 31, 1978. Such provisions, of course, were subject to the weather, and to the time required ex post to complete work on the projects. Timing of Expenditures Significant Round I LPW expenditures began in January, 1977, although a few small projects were initiated in late 1976. Round II expenditures began about nine months later, in October, 1977. The largest of the LPW projects are esti- mated to take up to 36 months for completion. Some small level of Round II LPW expenditures are therefore expected to continue through 1980.3. Of the total of both rounds, 19.9 percent took place in 1977; 61.4 percent is estimated to take place in 1978; 17.6 percent in 1979; and 1.1 percent in 1980. The median expen- diture, with respect to time, was estimated to have occurred in June, 1978. The largest levels of LPW expenditures are also estimated to have taken place at about the same time. The time pattern of expenditure is presented in more detail in Table 2-2 and Figure 2-2. The timing of LPW expenditures plays a crucial role in the LPW program's efficacy as a counter-cyclical policy, a subject discussed further in Chapter 8. The length of the program also has an effect on the employment impact, since the expenditures in the latter years are subject to significant crowding out effects and reduced leverage due to rising productivity and prices. Considering that the Round I appropriation was "for the period ending September 30, 1977," and the Round II appropriation was "for the period ending December 31, 1978," the LPW program does not appear to have had the quick, concentrated impact on the economy that it was designed to have had. Substitution The Concept of Substitution Substitution is the use of federal funds to replace local funds. The effect of complete substitution is that the Federal expenditures produces no net increases in government expenditures. There are, however, two further elements in the definition of substitution. The first is the issue of how the replaced local funds are used. The second is the timing, relative to the Fed- eral program, of the local expenditures that were replaced. Local funds that are replaced must be used for some other purpose. The funds may be used for alternative government efforts. These efforts may be other types of capital budget expenditures such as the purchase of land or, in fact any expenditure within the local budget (capital or operating). The alter- native use may also be implicit, as when a locality uses the funds to reduce a budgetary deficit or returns them to the private sector by reducing taxes. Each of these uses will produce an impact on the economy. However, except for tfie first use -- other construction activities -- this effect will be different in targeting, and probably in size, than the impact that the federal grants would have had, if substitution had not occurred. The uses to which replaced local funds may be put deleniates a typology of substitution. CEAI has found it most useful to define types of substitution in terms of the broadest identifiable category for which replaced funds are not used CEAI 's types are: t Substitution denotes the situation which occurs when federal monies replace local monies on a specific project. • Project Substitution denotes the situation which occurs when federal monies replaced local funds on a specific project and the funds are used for other construction activities. • Construction Substitution denotes a case in which replaced local funds are not used for construction expenditures. 5 t Capital Substitution denotes a case in which replaced local funds are not retained in the capital budget (i.e. in which they are used for operating purposes, deficit reduction, or are returned to the private sector). • Locality Expenditure Substitution denotes a case in which replaced local funds are not explicitly spent by the locality, but are used to reduce the deficit and/or are rebated to the private sector. Most information on alternate uses of local funds is somewhat incomplete. It is often limited to distinguishing the sorts of uses to which funds would not be put (e.g. locality expenditure) from those to which they might be put (e.g. reduced deficits or rebates to the private sector). In such cases the exact use of funds (e.g. reduced taxes or lower borrowing) may not be known. CEAI thus believes that the most effective typology of substitution focuses on the known categories for which funds would not be used, rather than the specific use to. which funds will be put. However, we have made an exception to this typology in our definition of proj- ect substitution both to avoid having two terms, "substitution" and "project substi- tution" for the same phenomenon and to prevent confusion which might arise from the use of the phrase "project substitution" by CEAI in a manner different from that of EDA. Any of these types of substitution may be partial or complete. From the point of view of the employment generating impacts of the LPW program, the most critical issue is the amount of net new construction generated, or concisely, the amount of construction substitution. Project substitution alone does not alter the nature of expenditure nor (except for minor differences in the employment intensity of dif- ferent projects) the impact of the expenditure. There are three possible patterns of substitution with respect to time. They are: • Concurrent Substitution : A case in which the replaced local expen- ditures would have taken place at the same time as the federal expenditures. t Project Acceleration : A case in which the replaced local expendi- tures would have taken place later than the federal expenditures. • Project Delay : A case in which the replaced local expenditures would have taken place earlier than the federal expenditures. The latter two patterns constitute intertemporal substitution. Project delay can be further subdivided into two cases. Short-run project delay is typically a speculative postponement of a project in the expectation that funds will be avail- able from a specific program at some point in the near future. Long-run project delay is the indefinite postponement of a project under the general expectation that some type of funding will be available sooner or later. Intertemporal substitution is of particular interest because of its poten- tial for pro-cyclical or counter-cyclical effects outside as well as within the time frame of a counter-cyclical grants program. Project acceleration is likely to be counter-cyclical, since the replaced expenditure would have taken place at a later point in the business cycle when output and employment would have been higher. Short-run project delay resulting from a contemplated counter-cyclical program, however, is almost certain to be pro-cyclical. The expectations of re- ceiving specific counter-cyclical funds are formed between the time of the pro- posal and implementation of the program -- a time when the recession is typically at its worst. Under these circumstances there is a high probability that expenditure would be transferred from a trough to a recovery. The cyclic effects of long-run project delay are more difficult to assess. However, should a recession develop, it is most likely that the general expecta- tion of counter-cyclic funding for public works activities is likely to have pro- cyclical effect. Also, it is clear that general expectations of future local pub- lic works financing will substantially increase the level of all types of substi- tution. Estimation Procedures The primary source for estimation of the magnitude of substitution in the LPW program was CEAI's survey of local government officials. The survey provided at least three types of information: direct answers to survey questions (both factual and perceptual); additional observations by surveyors on the specific locality; and the general insights of CEAI analysts' into the workings of LPW at the locality level . A second source of data, locality financial documents, proved much less use- and provided only general insights into the locality budgetary process. The problem was that data were incomplete and too full of incomparabilities° to allow either quantitative analysis or the tracking of flows of funds. A third possible source of information, estimates of substitution from other research, proved vir- tually non-existant for the LPW program. 9 Alan Fechter of the Urban Institute aptly characterized the conclusions of previous efforts in the area in a comment that substitution was definitely "between zero and 100 percent." A direct aggregate estimate of the level of overall substitution was based on the survey results. Separate analyses were conducted for projects that were funded under LPW and those that were not. The final analysis used responses on whether the project would have been built, when it would have been built, 10 how fully it would have been funded, and from what source. In the course of the survey and 7 So as to gain first-hand knowledge of both the survey and the program, all CEAI project analysts, including the Project Manager, participated in the fieldwork. Areas of incomparability -- both between localities and between budgets of one locality over time -- include categorization of funds and modes of transfer between funds, documentation of revenues and expenditures, treatment of trans- fers from other government units, and distinctions between capital and operating expenditures. The last issue is particularly troublesome, because it reflects not just accounting procedures but wide differences in perceptions about what constitutes a capital expenditure. Many of the incomparabil ities also reflect a wide variety of procedures in budgeting and allocating funds. Only one scholarly article was found dealing with any form of substitution in the LPW program: Edward Gramlich, "State and Local Budgets the Day After It Rained: Why is the Surplus So High?" Brooki ngs Papers on Economi c Acti vi ty , 1:1978, pp. 191-216 Gramlich does pioneering work in identifying the concept of project delay intertemporal substitution. However, his empirical methodol- ogy is so badly flawed (e.g. he forecasts using an equation with an R 2 = 0.06) that his results must be considered theoretical only. In some cases the response was that a project would have been built piecemeal over a period of time. Some apparently promising lines of questioning proved impossible to interpret with- out further detailed information. The question of whether funding of a project had been attempted before, for example, had ambiguous interpretations. A positive response could have meant that a project had a high priority, but that circumstances (possibly including project delay substitution) had prevented its funding. Alterna- tively, a positive response could have meant that the project had not been funded because it was of ~\o\n priority, but that the "attempt" had been made to satisfy some political group. The former case entails high substitution; the latter very little, if any. 41 analysis a further factor arose -- the issue of how certain it was when a project would be built. In many cases respondents replied that it could not be predicted when a project would be built, even with local funding available. In others, the construction was conditional on receiving some type of funding, and the respondent could not say when this might occur. Of perhaps the most importance CEAI surveyors found that most jurisdictions had a substantial "wish list" of projects that had been proposed, and for which some planning work -- possibly including an attempt at funding -- had occurred. For many of these projects the acquisition of complete funding appeared doubtful for many reasons including the severe competition from other "wish list" projects. In these cases, respondents appeared wery reluctant to state flatly that such projects would not be built but rather, in CEAI's judgment, described these projects as ones which would be built with either an uncertain start date or with a start date that was beyond realistic planning horizons. The distribution of the sampled LPW projects and by the existence, timing, certainty of timing, degree, and type of alternate funding expected is presented in Tables 3-1 and 3-2. These distributions for rejected LPW applications in the sampled localities are shown in Tables 3-3 and 3-4. CEAI's analytical procedure resulted in a set of coefficients, or percen- tages of substitution, applied to each project. Where no specific data exist, the value of 0.5 is, of course, intuitively appealing as the midpoint of the range of possibilities. However, the estimated values of these coefficients were determined 'only after an extensive debriefing of the surveyors as to the appropriate percentage of substitution to be attributed to each relevant factor. The values assigned to the coefficients also tended to be confirmed by the (extremely scanty) quantitative evidence available. Where the surveyors had more specific information — either as a quantitative response or from their observa- tions of response to related and/or following questions -- this was used instead of the standard coefficient. The coefficients can be summarized as follows: t Projects that reportedly would not be built without LPW funding were assigned a substitution coefficient of 0.0. • Projects that would otherwise have been funded from non-local ,p sources (grants) were assigned a substitution coefficient of 0.0. An exception is revenue-sharing funds, which were treated as local funds. • Projects that would have been funded partly from local sources and partly from non-local sources were assigned a substitution coeffi- cient of 0.5. § Projects for which the alternative funding was described as "partial" were assigned a substitution coefficient of 0.5. 12 Legislation authorizing the LPW program is concerned only with substitution of local funds. Moreover, other Federal funds -- especially categorical grants -- will generally be spent on similar projects. Thus no more than project substi- tution will occur. 42 43 2<° |SS ^1 ".3* | r a gas fc_a || 2 a -s ?1 II 'III ll tffj 2 * ?:i? i ! r Q — I -H 0) -H j-i y_i re ro 4-i 03 4-1 (0 A3 -H 0) -H IT) -H 0> t#P 4-1 >H#4J 1 dP 4-> X O J-i O l-i ro O Li O fO -DO fl O <0 +)h a C H D^ W rH Qj -( ro C >H U-l fC h inop-u C O U J c o T5 >h dit: C C C C C C c c c c c c c ■H 3 •H 3 ■H 3 •H 3 •H W -H 3 m -h 3 T3 U-l T3 M-l T3 M-i T3 t <4-l (13 .. M-l (0 (0 -H -h a; -h C -H Q) <#> 4J >H c*> JJ 1 dP JJ >M <& 4J ■H lT) c*> -U (T3 <*> -P >H O U O U ro o u o n (T3 O U DO U O (0 TJO (C O (0 in o (0 4J CO (5 £ O (0 -p rH a CHD. W H Oi ,h a, VJ -H H & oh a CO 0) -C o o 03 CD O J-> a; •H C -H fa CO >H o D £ a. Projects whose construction start date could not be estimated by the respondent ("timing uncertain") were assigned a substitution coefficient of 0.5. Projects with reported start dates beyond a five year time horizon were assigned substitution coefficients of 0.0, unless the surveyor had add it} be built. 1 Repair projects (primarily streets) involving a number of distinct sites were assigned a substitution coefficient of at least 0.75, regardless of other factors. '^ 15 All the above substitution coefficients, except those for repair projects involving a number of distinct sites, were assumed to interact mul tipl icatively For each project the overall substitution coefficient was multiplied by the LPW obligations to obtain a dollar estimate of substitution. Based on analysis of LPW- funded projects, the aggregate gross estimate of overall substitution was 26.6 per- cent of the sampled obligations. For purposes of assessing the impact of the LPW program, it is not the level of overall substitution that is of primary interest but the level of construction substitution. Project substitution itself does not materially alter the impact of LPW expenditures. Unfortunately the survey results of the more specific types of substitution are not \/ery clear. In general, locality officials had little idea exactly how and when the "saved" funds would be spent. However, the respondents were able to provide some estimate of where the "saved" money would have come from. The responses indicate that, in the absence of the LPW program, just over 25 per- cent of alternative financing would have come from general revenues, and the other 75 percent would have come from bonds or other private sector contributions. Re- spondents generally did not distinguish, however, between financing from capital funds or operating funds. Indeed the lack of clear (much less consistent) dis- tinctions between capital and operating funds, and the prevalence of transfers among funds that was found in our analysis of the financial documents suggests that such a distinction may be impossible for projects that would have been financed in future fiscal years. 13 In most of these exceptions, the project was a type of municipal service that would be required because of local growth in the forseeable future. The indi- vidual substitution coefficient was judgmental ly estimated by the surveyor in such cases. This type of project regularly appears in budgets of larger jurisdictions. The surveyors therefore judged that this case was much more likely simply to be project acceleration than was the average project. Thus, for example, a project that would reportedly be partially funded (with certainty) by a mixture of local and grant funds was assigned a substitution co- efficient of 0.25. If, in addition, the timing was uncertain, the substitution coefficient was 0.125. All projects with a substitution coefficient of 0.0 as a result of any factor therefore had an overall substitution coefficient of 0.0. Lacking actual data on alternative financing from capital budgets, CEAI has made a rough estimate of project substitution based upon the assumption that 25 percent of the "saved" general revenues would be spent in the capital budget. This results in an estimate of oroject substitution via the capital budget of only 1.7 percent of LPW obligations. 16 Project substitution financed through unallocated bond funds would make this figure somewhat higher. A second adjustment must be made for increases in local construction expendi- tures that were generated by the LPW program. Survey responses indicated that the total cost of the LPW projects in the sample were 8.1 percent higher than the sum of total costs shown on EDA records. ' 7 This complementary expenditure must be sub- tracted from the above estimate of overall substitution (26.6 percent) to obtain a net value of substitution.' 8 An important analytical issue is how to treat jointly these two adjustments (1.7 percent project substitution and 8.1 percent complementary expenditure). It would be double-counting to add local dollars released to the budget and spent on capital items (project substitution) to local capital funds used for complementary expenditure; the latter funds represent reduced expenditure on other construction. Only to the extent that funds released to local capital budgets exceed complemen- tary expenditure from that source is there an additional effect. According to the survey responses, about 24 percent of complementary expenditure (1.9 percent of LPW obligations) came from capital budgets. If we subtract the 1.9 percent of LPW obligations which would have been spent anyway from the 8.1 percent of complementary expenditure we have net construction complementary expenditures of 6.2 percent. Adding the 1.7 percent project substitution from capital budgets calculated earlier yields net "project substitution"' 9 of 7.9 percent. At first one might think that by simply subtracting this 7.9 percent from the estimate of gross substitution (26.6 percent) would yield the amount of net construction substitution. However two problems remain. First, some localities have "bond funds" which result from the floating of a large construction bond and gradual use of these funds for con- struction projects as the need arises. The total amount of complementary expendi- tures coming from this type of funds is estimated to be 0.5 percent of LPW expen- ditures. The amount of construction resulting from the replenishment of these 16 Project substitution (as a percent of LPW obligations) equals overall substitu- tion (as a percent of LPW obligations), times the fraction of substitution in- volving local government revenues, times the share of the local budget devoted to capital expenditure, or: (0.266)x(0.25)x(0.25)=l .7 Surveyors found two primary reasons for this difference. One was cost overruns that were financed out of local funds. The second was an occasional local decision to enlarge the project substantially beyond what had been reported on the application. In a few cases, this involved using LPW funds as seed money to encourage private funds. It may be useful to think of this complementary expenditure as a type of project substitution. Technically, the two are the same only if local funds that would have been used on one project in the absence of the LPW program are used on another project in the same jurisdiction. Thus, in the aggregate the effect is much like that of project substitution. See Footnote 18. funds through substitution is unknown but is probably small. Second, in several cases CEAI surveyors noted that the complementary expenditures went for activi- ties other than construction such as the purchase of real estate. For the pur- poses of this study, CEAI has assumed that the amount of addition construction in the LPW period resulting from the replenishment of the bond funds was equal to the amount of complementary expenditures from these funds. Also it was assumed that the amount of non-construction complementary expenditures was approximately 1 percent. Combining these factors -- and ignoring the spurious precision implied by the decimals in the intermediate calculations -•- yields an estimate of net con- struction expenditures of 20 percent.^ As part of the analysis of substitution, CEAI analyzed many of the character- istics of a locality in an attempt to identify influences on the levels of substi- tution. Levels of substitution were computed for each locality in the survey, and these were compared with total obligations to the locality, locality size, type of locality, SMSA status, region, a composite measure of difficulty of bonding, *T and the ratio of LPW obligations to a measure of local capital expenditure." The results were not notably successful. Instead it appears that the factors deter- mining the amount of substitution were generally individualistic and depended in large part upon the particular locality officials conincidental confluence of needs and the availability of the LPW funds. CEAI also analyzed inter-temporal substitution in terms of the planned con- struction date (in the absence of the LPW program) given in the responses. The greatest difficulty in this procedure is defining a realistic definition of time periods. The difficulty stems from the fact that the LPW program involves sub- stantial expenditure over a three year period. This is a longer period than the realistic planning horizon of many localities. Moreover, the majority of LPW ex- penditures occur in the second year of the program, so that project acceleration could not logically be said to occur unless the project would otherwise have been built in the third year or later. This, in fact, was the definition of project acceleration that was used. No direct evidence of project delay was found. Two additional inter-temporal effects are noted: project acceleration when the alter- native funding source is grants, and possible project delay of the non-LPW funded projects. Neither of these, however, is substitution per se . A major concern in estimation of substitution is the potential limitations inherent in any survey. Three major potential limitations exist: (a) the accuracy of responses; (b) limitations stemming from the size of the sample; and, (c) the problem which arises when some information is either unknown or unknowable. 20 Gross substitution (26.6 percent) minus net induced expenditures (7.9) plus in- duced non-construction expenditure (1 percent) yields the estimate of construc- tion of 19.7 percent, or approximately 20 percent. This measure was compiled from questions on the date, type, and fate of the last bond issue, and on community attitudes on floating bonds or raising taxes (see Appendix B). The localities were classified judgmental ly as the evidence of fiscal constraints being "none," "some," or "fairly conclusive." Local capital expenditure was measured as the 1976 capital expenditure reported in the written questionnaire, times the fraction of capital expenditure reported to be financed from local sources. 49 The survey procedure involved several means of checking the answers. Inter- viewing several respondents in a locality made it possible to distinguish between facts and perceptions in most cases and to identify discrepancies that needed fur- ther reconciliation. The oral format put each answer in contect, and allowed re- phrasing or followup if an answer seemed inconsistent with other responses. The internal pattern of answers also suggested that answers were satisfactorily re- sponsive. 3 Some difficulties were encountered with individual officials who were not particularly knowledgeable about the LPW projects, but these respondents were easily identified. Assimilation and interpretation of direct responses to ques- tions and other observations required judgment of the surveyors. However, all of CEAI's surveyors were full-time professionals who were quite familiar with the issues and problems to be addressed. It is CEAI's judgment that the survey data used in the estimate of substitution are without substantial bias. The sample itself is limited in size to 50 jurisdictions. It was carefully chosen for representativeness with respect to region, jurisdiction type, jurisdic- tion population, overall level of LPW funding, and SMSA status. ^ As such, the aggregate estimate of substitution is representative. Two problems arise, however, in any attempt to analyze the causes of substitution. First, the sample size is small enough that disaggregation by values of any variable reduces the number of observations in any cell to such a low level that precise statistical analysis becomes virtually impossible. Second, CEAI's analysis suggests that some of the most important influences were not, and could not be, among the variables used in sample selection. 25 The result is that any attempted statistical relationship is plagued by a large proportion of special case outliers, and analysis quickly evolves into individual analysis of these observations. Perhaps the most important limitation of a survey effort such as this is that some of the information is inherently hypothetical rather than observable. One 24 Although substitution is clearly illegal, respondents stated that a project would not have been built without LPW funds in fewer than 25 percent of the cases. Moreover, although there were some cases of an official's giving identi- cal answers for a number of projects, the general pattern of response was a va- riety of answers for the different projects within a locality- together with some degree of volunteered explanation of the circumstances of the project. It therefore appears that these questions were, on the whole, answered invididually and candidly. See Appendix B for the exact distribution. The sample was also representative of project size, round, and number of projects in a locality. (As noted, localities inside SMSA's were purposely oversampled. ) Selecting a sample of 50 localities that was precisely representative of project-specific, as well as locality-specific, variables would have been exceedingly complex. One potential problem that did arise was that a high proportion of both projects and funds came from the two largest localities (Philadelphia and Detroit). The average level of substitution of these two cities, however, approximately equaled the average level of substitution of the sample as a whole. Thus the potential dominance of these two localities does not appear to have imparted any signifi- cant bias to the main findings of the survey. 25 These influences included local fiscal constraints, local grantsmanship tactics, and the timing of LPW funds relative to the appearance of local specific needs and local budgeting procedures. specific difficulty is that the universe is not neatly divided between projects that would not be built and projects that were "previously appropriated or planned." As noted earlier, most localities have a substantial "wish list" of partially planned projects, whose probability of future construction can only be estimated. Second, any question of what would have been done in the absence of LPW funds is purely hypothetical. This applies to project delay as well as proj- ect acceleration. Thus direct objective questions pertaining to project ..del ay inter- temporal substitution cannot be asked. Substitution Estimates Based on the above analytical procedures, CEAI has made the following quan- titative estimates of substitution: t Based on analysis of LPW- funded projects, overall gross substi- tution is estimated to be 26.6 percent of LPW obligations. • Additional expenditure on LPW projects financed from local funds is estimated to be 8.1 percent of LPW obligations. 2 ^ • Locality expenditure substitution was estimated to be 74.6 percent of overall substitution, or 19.8 percent of LPW obligations. • Based on analysis of funded projects, capital budget substitution (which is equivalent, in this case, to substitution net of com- plementary expenditures) is estimated to be 18.5 percent of LPW obligations. • Based on analysis of LPW-funded projects, construction substitu- tion 27 is estimated to be 20 percent of LPW obligations. • Project acceleration is esti overall (gross) substitution. t LPW funds are estimated to have accelerated 71.9 percent of expen- diture that would have been financed by other grants. 28 This measure corresponds to EDA's concept of fiscal substitution. Of these expenditures, 67.0 percent were reported as having been raised especi- ally for the project, 23.8 percent as coming from capital budgets, 6.4 percent as coming from operating budgets, and 2.8 percent wre not identified by source. Because these proportions are roughly the same as the identifiable ultimate uses replaced funds (see Footnote 26), CEAI estimates that locality expenditure substitution is also about 70 percent of construction substitution, or 14 per- cent of LPW obligations. Project acceleration is defined as expenditure made in the first two years of the LPW program (1977-1978) that would otherwise have been made at a later date (1979 and beyond). The estimated level of substitution varied greatly among sample localities, as is shown in Table 3-5. CEAI therefore sought to explain this variation in terms of the variables used in sample selection. Summary tables of these dis- tributions of the sample observations are presented in Appendix C. CEAI also explored a measure of local fiscal constraint, the timing of LPW funds, and local grantmanship strategies as possible explanations. Analysis was greatly complica- ted by several factors: • The sample is too small for effective statistical analysis. • No one or two explanations seem to determine the level of sub- stitution for the sample as a whole. Instead, each explanation seems to dominate a relatively small number of cases. • Different influences appear to interact in complex ways. The result of these factors is that any attempt' to determine the relationship between any potential cause and the level of .substitution requires individual inspection of outlying observations. The results obtained from the analysis of individual localities can be summa- rized as follows: • Type of locality, level of total obligations, and SMSA status appear to have little relationship to substitution. • A slight regional pattern is observed (the Great Lakes and South Atlantic Regions had levels of substitution that were substantially below average). This could be largely explained, however, in terms of other factors.^ 9 • There is a distinct -- but by no means unanimous -- tendency for large and/or older industrial cities to have relatively low sub- stitution. • Difficulty of access to capital funds appears to be related to sub- stitution in a rather wide range of cases. (This factor is perhaps the only generally applicable explanation of the level of substitu- tion.) For the most part this tended to produce a low level of sub- stitution. However, when combined with a project so urgent that it The localities in the South Atlantic Region all had some combination of high demand for public capital; extreme poverty; and/or intense resistance to bonding. The Great Lakes Region was dominated by large cities (which may it- self have been a factor), several of which faced poor access to capital markets. Cleveland, Detroit, Milwaukee, and San Antonio all are estimated to have sub- stitution far below average. Philadelphia and San Diego are major exceptions. TABLE 3-5 Level of Overall Substitution, by Locality Average Level L ocality No. Locality Name No. of P rojects of Substit ution 1 Alton 2 Bedford Co 3 Bethlehem 4 Brewton 5 Chicago 6 Cleveland 7 Collier Co. 8 Connel sville 9 Deland 10 Detroit 11 Durham 12 Elkins 13 Erie 14 Evanston 15 Fairfield 16 Great Falls 17 Greenbel t 18 Henderson 19 Iron Co. 20 Kingman 21 Mars 22 Milwaukee 23 Muskogee 24 Nashua 25 Passiac 26 Philadelphia 27 Portland 28 Providence 29 Renton 30 Roanoke 31 Rome 32 Roy 33 Salem 34 San Antonio 35 San Diego 36 Saratoga 37 Savannah 38 Sea ford 39 Sedalia 40 Sequin 41 Silver City 42 Slidell 43 South Bend 44 Springfield, Ky 45 SDrinqfield, Ma 46 Tucson 47 Uinta Co. 48 Vernon 49 Watervl iet 50 Worcester 2 1 1 100% 1 5 25% 4 16% ' 3 9% 3 100% 1 17 10% 1 50% 2 33% 5 31% 2 12% 1 50% 2 88% 1 100% 5 8% 1 1 50% 1 25% 10 15% 1 5 40% 3 50% 35 49% 9 87% 4 14% 3 35% 5 95% 4 5% 1 25% 2 84% 4 8 61% 1 1 3 50% 2 39% 1 50% 1 100% 1 2 41% 2 3 2% 1 100% 2 25% 1 1 75% 1 would have had to be built anyway, difficulty of access to capital funds seemed to guarantee high levels of substitution. • The funding of a small number of pro.iects tended to be associated with extreme levels of substitution. 32 This suggests that the degree of substitution was related to project characteristics as well as locality characteristics. t In a significant number of cases the size of the LPW grant is sub- stantially larger than the locally funded capital budget. In such cases there appears to be a fairly strong tendency for substitution to be quite low. • The particular selection and grantsmanship tactics used by local officials appears to play a significant role in the level of sub- stitution. These tactics probabiv, but not necessarily, tended to lower the level of substitution. 3 ^ (Some localities virtually never financed construction from their own funds. Such cases usually had some combination of difficulty raising funds, little or no growth, and/or small size. As part of the analysis, CEAI analyzed a sample of projects applied for but not funded by the LPW program. The resulting estimate of potential "substitution" (i.e. actual subsequent funding of these projects) was 40.7 percent, which is 50 percent higher than the estimate based on LPW- funded projects. CEAI believes that the difference in estimates of overall substitution is primarily a reflection of an inherent bias in the sample of non-LPW projects. Potential "substitution" in rejec- ted projects is relatively hiahly concentrated (over 25 percent) in projects funded 3l Although "urgency" obviously comes in various degrees, CEAI surveyors found a number of projects that were being constructed as a result of such events as availability of matching grants, bridge collapses, sewer breaks, EPA orders, court injunctions, and tornado damage. If LPW funds became available after the need arose but before local funding had been arranged, substitution was highly probable. 33 had only one project. In order to maximize local capital funding, some local officials appeared to pursue a policy of using LPW funds for projects that would be difficult to fund locally. CEAI's analysis of construction category variations between Round I and Round II suggested that local officials tended to select a higher proportion of local municipal buildings and other general facilities that they might have an interest in, but which had no particular constituency. In the survey sample there were several projects that were actually highly controversial. Surveyors also noted a tendency of local officials to search out grants to fill needs that they perceived would be difficult to finance locally. Such tactics would tend to produce low levels of substitution if local officals' perceptions are accurate. this year from general revenues. This suggests that local officials tended to initiate their highest priority projects anyway, and save lower priority projects (with correspondingly lower probabilities of substitution) for possible (and actual) funding in Round II. 34 35 Public finance literature tends to suggest that larger older cities will have relatively high levels of substitution. This prediction is based on a num- ber of urban characteristics, including: Relatively service-oriented (rather than capital-intensive) local functions A relatively large existing capital stock Complex decision-making processes Fairly ready locality access to capital funds High absolute levels of public spending Stable population The finding that a majority of the large and/or old cities have low substitution appears to contradict the theoretical result. Closer inspection reveals, however, that most of the exceptions have relatively difficult access to capital markets (e.g. Chicago, Milwaukee, Cleveland, Detroit), or ambitious urban-renewal programs "34 There is also a possibility that the sample and responses of rejected projects is biased. In the first place the selection process depended somewhat on the respondents' familiarity with the projects in cases where more than five rejec- ted applications had been made for a locality. If the respondent was familiar with a project because it was being built, this would bias the estimate of sub- stitution upwards. Secondly, the wording of the rejected-projects question asked about actual plans, rather than hypothetical plans in the absence of LPW. If the LPW grant had been large, the resulting upward shift of the local capital budget constraint could have enabled more projects to be done, which would also have biased upward the estimate of substitution. The other most likely explanation of the tendency to build rejected projects is provided by Gramlich, who argues that speculative project delay affected proj- ects not funded by LPW as well as resulting in substitution proper. By this line of reasoning, the burst of construction of rejected projects merely reflects local decisions to go ahead with these projects after they were rejected. The problem with this argument is that it treats Round I and Round II asymmetrically. At the time of local budget formation, Round II was farther along, more certain and larger than Round I had been one year previously. This should have been an even greater incentive to await Round II funds than Round I funds. Yet there are very few rejected projects definitely planned in the immediate future (2 to 5 years), now that the outcome of Round II allocations is known. This clearly contradicts Gramlich's hypothesis. 35 A standard source is: Richard A. Musgrave and Peggy B. Musgrave, Public Finance in Theory and Practice (New York: McGraw-Hill, 1973), pp. 614-620. that, entail replacement of substantial public capital (e.g. Detroit). A city that almost exactly matches the hypothesized characteristics of the "larger older cities" (Philadelphia) was estimated to have a hiuh level of substitution. No particular evidence was found of the sort of project delay posited by Gramlich, in which localities defer projects in anticipation of funding, and then build them anyway if they are not funded. This, of course, is not actually substi- tution. The issue was therefore not part of the research effort. Nevertheless the data collected do not tend to support this hypothesis (see Footnote 35). Macro- economic data are not yet available, because analysis requires data on a considerable period after the LPW awards. The preliminary BEA state and local government capital expenditure data for the time since Round II began are ambiguous and inconclusive. They do indeed show a sharp increase in these expenditures during mid 1978 (as Gramlich would predict). Yet this upturn is not so large that it cannot mostly be explained by recovery from an abnormally severe winter, plus the LPW program itself. It is CEAI's judgment that this project delay effect, if it existed, would have in- volved a delay of approximately Si to $2 billion of state and local expenditures for no more than one or two quarters. CEAI has used as its basic measure of substitution the estimate of construc- tion substitution equal to 20 percent. Analysis of potential determinants of sub- stitution suggested that the sample was too small to obtain separate estimates of substitution on the basis of round, time, project type, or similar variable. CEAI therefore applied the estimate of 20 percent to all LPW expenditure and to indirect and induced effects. The estimate of construction substitution equal to 20 percent applies to the existing LPW program. CEAI's field experience and the history of programs such as revenue sharing, however, strongly suggest that local government officials have a substantial learning curve. A similar "one-shot" program, such as an LPW Round III, or an ongoing program of support for local public works, could be expected to have substantially higher levels of substitution -- particularly short-run and long-run project delay. Crowding Out The Concept Crowding out is the displacement of other economic activity as a result of unused government expenditures. Crowding out characteristically occurs through a market process. Either through price rationing or through the development of phy- sical shortages of materials, labor, or capital, other demands are prevented from generating real output and employment. This can occur either because markets are generally tight or because of specific bottlenecks, although it is generally much more serious when the economy isstrong than in recessions. In eithercase, the effects of crowding out must be deducted from the employment generated by a program such as LPW to obtain a net impact. Es timation Pr ocedures Estimation of crowding out effects of the LPW program was done using the CEAI Macroeconomic Model. This model was simulated with and without the LPW expenditures. Simulation results are shown at the sectoral level and aggregated to the entire economy. Isolation and measurement of crowding out effects is complicated, how- ever, by two characteristics of the results. First the simulation results show only the net impact of an expenditure without distinguishing between the positive and negative effects. Second the LPW program is so small relative to the economy that most sectoral impacts are rounded to zero or have a high margin or rounding error. In order to deal with these constraints, CEAI divided the analysisof crowding out into two parts: direct crowding out and other crowding out. Direct crowding out was estimated to be the quarterly net reduction of expen- diture in all components of the construction industry. 37 This change in expenditure based upon: (a)responses to the contractor survey, (b) the survey of local offi- cials in localities with populations under 10,000 (9 sites); and (c) telephone interviews with officials of seven of the most isolated localities receiving LPW funds of over $2,000,000. Only isolated instances of possible crowding out were observed. They tended to be in rapidly growing localities (mostly related to the energy boom). If there was a victim of crowding out, it was probably residential construction. There was also reported to be a shortage of concrete, although this was a regional problem (at least in the first half of 1978 when the analysis was undertaken). In general any difficulty in locating local labor, .contractors or materials simply meant that those inputs would come from as far afield as neces- sary. In sparsely settled areas the construction market is already geographically large, and the LPW programs were readily absorbed with hardly a ripple of crowding out. Furthermore, in many cases contractor's seemed to regard LPW work as less desirable than private sector work. Thus, even in cases where contractors were scarse the result tended to be higher or fewer bids on the LPW project rather than the crowding out of private sector projects. The lack of perceived crowding out on the local level can easily be reconciled with the national results. First, the LPW program involved thousands of locali- ties, so that the demands for inputs must have interacted in the aggregate. The lack of perceptible crowding out at the local level thus does not imply the ab- sence of crowding out at the aggregate level. Second, the local observations were based on experience through 1978.2. Until after this period, the macroeconomic simulation suggests overall crowding out was minimal. 37 A very small portion of LPW expenditure was on housing. Thus the net effect on residential construction (single-family and multi-family housing) is made up al- most entirely of crowding out. As expected, this net impact is negative through- out the LPW period, but it is not substantial until late 1978. Since the primary effect on non-residential construction is the LPW expenditure itself, crowding out as defined above (i.e. a negative impact) is not observed until 1980. Thus crowding out is understated in this case, although if the pattern is the same as in residential construction, the omission is probably serious only for 1979. was multiplied by the corresponding labor requirements coefficients to obtain a measure of employment lost in the construction industry due to crowding out. 39 The effects were then traced through the economy by assuming that the crowding out loss in each quarter was the same fraction of employment generated for in- direct and induced employment as for direct (construction) employment. All other crowding out effects are implicitly measured by the use of the employment multiplier 40 ^ derived from the CEAI Macroeconomic Model simulation in estimating total employment. 41 This simulation of the actual LPW program was com- pared with a parallel simulation that assumed that the LPW expenditure 4 ^ had taken place beginninq in 1974.2. The total and indirect employment impacts of the actual LPW simulation are substantially smaller than those of the alternative scenario. Had the LPW exoenditure taken place earlier in the recessionary period, the total employment generated would have been an estimated 13 percent higher than actually occurred, and non-construction employment would have been 16 percent higher. Much of this difference is due to crowding out. 40 41 BLS coefficients were used, adjusted on an annual basis for price and produc- tivity changes. For non-residential construction an unweighted average of coefficients was used. Because this estimate than simulation estimates of employment, the results are comparable with the BLS-based and INFORUM-based estimates of employment in person-years (see Foot- note 41 ) . The employment multiplier is the ratio of total employment generated to direct (construction) employment generated. Because the simulation and the resulting estimate of construction employment are net of direct crowding out as defined and derived above, these effects are not double-counted in this implicit mea- sure of other crowding out effects. The employment multiplier is estimated from the simulation to be just under 3.0. This is multiplied by an average of direct and indirect employment (which are approximately equal in size) in order to obtain a better estimate of the timing pattern of the change in total employment. Since the employment multiplier re- flects other crowding out, this estimate of total employment is net of all but direct crowding out. Expenditure was simulated to be the same in real terms as the actual LPW expen- diture, and it was assumed to follow the same time pattern once initiated. Ad- justments were also made for productivity changes. Results Estimates of construction crowding out, by quarter, are shown in Column 1 of Table 3-6. The loss of construction employment is estimated to be 7976 person- years, or 9.0% percent of gross direct employment generated by the LPW program. Of this amount 4,949 person-years is estimated to have occurred in single-family residential construction; 1530 person-years in multi -family residential construc- tion; and 1497 person-years in non-residential construction. Estimates of" indirect, induced, and total employment losses due to crowding out in the construction industry are also included in Table 3-6. The total job loss due to construction crowding out is estimated to be 24,189 person-years, or 9.0 percent of total LPW job creation. All other crowding out is implicitly included in the projections of total employment generation reported in Chapter 4. As noted above, this level of crowding out is estimated to be larger -- by 16 percent of all job creation (14,710 person-years) -- than would have existed had the LPW program been iniated in mid-1974. TABLE 3-6 EMPLOYMENT LOSS DUE TO CROWDING OUT IN THE CONSTRUCTION INDUSTRY Quarter Direct Indirect Induced Total Employment Employment Employment Employment Loss Due to Loss Due to Loss Due to Loss Due to Crowding Out Crowding Out Crowding Out Crowding Out 1977.1 72 91 82 245 1977.2 87 108 98 293 1977.3 182 228 205 615 1977.4 182 212 197 591 1978.1 313 336 325 974 1978.2 466 473 470 1409 1978.3 651 585 618 1854 1978.4 659 616 639 1914 1979.1 661 671 666 1998 1979.2 712 706 709 2127 1979.3 830 621 726 2177 1979.4 799 766 783 2348 1980.1 796 879 838 2513 1980.2 773 855 815 2443 1980.3 793 999 896 2688 TOTAL 7976 24189 60 CHAPTER 4: EMPLOYMENT CREATION IMPACTS Employment creation -- and particularly the creation of employment in the construction industry -- is a primary goal of the local public works program.' In the 1974-75 recession the overall unemployment rate had risen as high as 9.1 percent, and unemployment in the construction industry peaked at 22.1 percent (see Figure 4-1). Within months of these peaks the first public works bill, had been passed by the Congress, but it was vetoed. Although the unemployment rate did decline from its 1975.2 peak, it remained high. In early 1977, as Round II was being debated for passage, the overall employment rate was still about 7.5 percent -- well above levels usually considered acceptable. Unemployment in construction was still about 15 percent, and according to organized labor 1 - unem- ployment in some areas and in some trades exceeded 50 percent. The purpose of this chapter is to report on CEAI's findings on the impact of the LPW program as a means of employment creation. Particular emphasis will be devoted to the employment creation impacts within the construction industry. First the results are presented and assessed. Next they are compared with other employment estimates and prior expectations for the LPW program. The chapter concludes with a brief discussion of policy implications. The specific issue of timing -- or counter-cyclical efficacy -- of the employment impact is discussed further in Chapter 8. CEAI FINDINGS Employment Estimates CEAI's estimate of the number of person-years of employment created by the LPW program is presented, by guarter and type of employment, in Table 4-1. Esti- mated direct (construction industry) employment is shown in Column 1 of Table 4-1. Direct employment was estimated by multiplying projected LPW expenditure streams for each type or project by BLS labor requirements coefficients, which had been adjusted for productivity and price changes over time. Estimated indirect (mate- rials supply industry) employment is shown in Column 2 of Table 4-1. Indirect employment was estimated by multiplying LPW expenditure streams for each type and size of project by the labor requirements coefficients of the inverted Leontieff matrix of the CEAI INFORUM (input/output) model. Adjustments were made for fore- cast changes in productivity and (in some cases) for the phasing of materials re- quirements in the production process. Estimated induced employment is shown in Column 3 of Table 4-1. Induced employment was estimated by means of an employment multiplier (the ratio of induced to direct employment) derived from a simulation of the CEAI Macroeconomic Model. This employment multiplier was, in essense, multiplied times the estimate of direct employment to obtain an estimate of in- duced employment. Estimated total employment, which is shown in Column 4 of Table 4-1, is simply the sum of the other three types of employment, by quarter. These methodologies are discussed in greater detail in Chapter 2. Cf. Public Law 94-369, Public Works Employment Act of 1976; Section 201. 2 George Meany, "Statement before the House Committee on Public Works, on H.R. 11 and Related Measures to Increase Authorization for the Local Public Works Capital Development and Investment Act of 1976," February 1, 1977, p. 2. 61 I o°>- 5^ \ % — / y / ""■ \ «>. / > J s \ s / * •— v ) -* """ ( *s ./ < "* I ^* J i" ( \ ^ \ s J J / • / X ( «*" ) / j "**"—• \ **" """* "*» A v. v^,^^ """ *" ~* "* -*» ^^ v. \ ^s > S \ <^ \ \ V s ( *■"• «^ j s 1 / I J ) e J <: — """*" i **• N / > ( < ) 1 1 1 | 1 1 1 1 1 1 1 1 1 1 1 j 1 1 1 1 1 1 1* I 1 ! 1 I 1 1 ! I 1 1 1 CM CM CM 00 CO CM CO 31Va !N3WA01dN3Nn NET EMPLOYMENT ESTIMA1 (Person-Years) "ES „ Quarter Direct Employment Indirect Employment Induced Employment Total Employment 1977.1 868 1,103 984 2,955 1977.2 3,782 4,686 4,234 12,702 1977.3 4,104 5,142 4,624 13,870 1977.4 7,145 8,321 7,716 23,182 1978.1 8,621 9,162 9,026 26,809 1978.2 14,367 14,581 14,473 43,421 1978.3 10,026 9,003 9,515 28,544 1978.4 7,680 7,204 7,440 22,325 1979.1 4,373 4,431 4,411 13,215 1979.2 2,226 2,207 2,211 6,654 1979.3 1,340 1,004 1,171 3,515 1979.4 147 141 143 431 1980.1 -430 -474 -453 -1,357 1980.2 -582 -642 -612 -1,836 1980.3 -735 -925 -832 -2,492 TOTAL 62,932 64,944 64,062 191,938 Person Years Per $100,000 1.06 a a 3.23 Cost Per Person Year $94,685 b 46,952 a $31,045 a) Not computed separately b) Cost per person-year for direct and indirect employment The estimates of all four types of employment shown in Table 4-1 are further adjusted for substitution and crowding out. For substitution a simple multipli- cative adjustment was made to all employment estimates to reflect the level of substitution of LPW funds for local funds which was estimated to be twenty per- cent of LPW expenditures. Crowding out adjustments were based on the degree of crowding out in the (residential) construction industry shown in a Macroeconomic simulation of the LPW program. It was assumed that crowding out of indirect and induced employment was equi proportional to crowding out of direct employment. The methodologies used for substitution and crowding out are discussed in greater detail in Chapter 3. CEAI's estimates that the LPW program will have created 191,938 person-years of employment. Of that amount, 62,932 person-years is in the construction indus- try, and 64,994 person-years is in the materials supply industries. ^ Slightly over 27 percent of the employment impact occurred in 1977, 63 percent is in 1978, and 12 percent will be in 1979. In 1980 CEAI estimates that crowding out will actually produce a slight net reduction4 in employment. ^ The proportions of con- struction employment are slightly less in 1977 and slightly more in subsequent years, primarily because certain material expenditures are concentrated in the early phases of construction. The employment is created at an estimated cost of approximately $95,000 per person-year for the construction industry employment, and $31,000 per person-year for overall employment (including the induced imapct). As part of its analysis, CEAI estimated LPW employment on the basis of a simulation of the CEAI Macroeconomic Model. The resulting estimates of direct and total employment are shown in Table 4-2. This measure of employment is in persons 3 The industry employment impact estimates are disaggregated further in Chapter 5. 4 Because of this phenomenon, the percentages of employment in previous years sum to 102 percent. 5 That the net employment impact should be negative for the final quarters of the program is certainly counter-intuitive. This result nevertheless flows logically from the combination of several factors. First, crowding out occurs through in- direct as well as direct channels -- i.e. not only does the LPW orogram hire away resources, but other construction is also reduced by such factors as inflation and rising interest rates. These indirect effects are particularly important when the economy is relatively tight, as during the latter stages of the LPW program. Second, the crowding out effects tend to lag the positive LPW impact. The indirect crowding out in particular tends to build up a cumulative effect which is especially significant because of the length of the program. Third, the LPW expenditures are estimated to taper off to a relatively low level -- especially in the last year of the program's impact. Thus even small crowding out effects resulting from large expenditures in earlier periods can exceed the small positive employment impact of the LPW program in its final stages. Indeed, this is exactly what is estimated to occur: Once crowding out adjustments are made, small negative employment impact does result in the last three quarters. TABLE 4-2 MACROECONOMIC SIMULATION ESTIMATES OF EMPLOYMENT (Persons Employed) Quarter Net Direct Employment Net Total Employment 1977.1 275 275 1977.2 1,158 2,046 1977.3 1,755 4,126 1977.4 3,054 8,126 1978.1 4,665 12,612 1978.2 7,079 17,094 1978.3 6,930 19,402 1978.4 6,836 19,492 1979.1 6,023 17,394 1979.2 4,641 13,094 1979.3 2,763 8,755 1979.4 1,440 6,587 1980.1 2,446 2,205 1980.2 1980.3 TOTAL 47,065 131,397 employed per year and is not comparable with CEAI's other estimates. b Nevertheless the simulation estimate -- equal to 75 percent of direct employment estimated by labor requirements -- may serve as a useful lower bound. Assessment of CEAI Employment Estimates The single most important factor in an assessment of the CEAI evaluation of LPW employment impacts is the fact that CEAI has made ex ante estimates, not an ex post evaluation. This is a necessary result of the timing of the evaluation. Except for some variables in the Macroeconomic Model simulation, data ai the timing of expenditure in 1977 and CEAI's own surveys, no primary data on LPW employment or other impacts were available to CEAI. Even these data covered only about the first 30 percent of LPW expenditures. The necessity of estimation undoubtedly results in a higher potential for error than does ex post measurement. In making its estimates of employment impact, CEAI has generally had a choice of methodologies or of quantitative estimates within a methodology. CEAI has based its choice of procedure or estimate on: (a) degree of detail; (b) judgmental assessment of accuracy; (c) the comparability of employment estimates; and (d) judgmental assessment of appropriateness to the conditions surrounding the LPW pro- gram. CEAI has also made several other adjustments to the estimates of employment impact not found in other estimates. They are: • Estimated adjustments of labor requirements coefficients for future years • Adjustments in employment estimates for substitution • Adjustments in employment estimates for crowding out The net effect of each of these adjustments was to lower the estimate of net employment. The CEAI Macroeconomic Model measures the number of people "employed" (in a National Income Accounts sense) during the quarter. Yet the average LPW job is estimated to be about a month, and construction work is notoriously erratic. Thus it is possible that an LPW worker, who otherwise had only little employment, would be counted as fully employed even if he worked only a few days in the quar- ter. In this case, the average month of LPW work for each worker would not count as an increase in employment, and the macroeconomic measure of employment would be biased downwards, relative to a measure of person-years. Aside from the possi- bility of biases, person-years is a preferable measure of employment from the point of view of standardization and comparability with other estimates. 66 Besides the necessity of using estimation procedures instead of measurement, the most important potential source of error in the results is data limitaions. Data were often less complete than was desirable, which led to simplifying assump- tions and approximations. 7 It is not clear, however, that improvement to the re- sults would justify upgrading the data base. Data collection is costly, 8 can be fraught with problems, and some of the information may be virtually unknowable. 9 Some data -- especially the BLS labor reguirements estimates -- are so widely, used that more intensive and extensive research would appear justified. CEAI believes that the person-year estimates of employment impact are accu- rate within a margin of 5 to 10 percent. Nevertheless, if there is a systematic error, CEAI is confident that the estimates overstate, rather than understate, the employment impact. '^ The two major alternative measures of employment -- the Macroeconomic Model simulation is about 25 percent lower than CEAI ' s person-year estimates, and may serve as a lower bound to the estimate of employment impact. Nevertheless, CEAI believes that the downward bias of estimates based directly upon this source is larger than the potential error of CEAI ' s person-year esti- mates. The following are representative examples of the problems: (a) lack of comprehen- sive data on labor mobility necessitated the assumption that labor did not cross state (or at least regional) boundaries in the LPW program. Fortunately this assumption was substantially born out by CEAI's limited survey; (b) the limited number of construction categories for which BLS labor requirements estimates ex- isted made classification of LPW projects difficult, and sometimes highly judg- mental ; (c) even with a survey, estimation of the degree of substitution in cases where partial and/or uncertain construction would have occurred in the absence of LPW could only be done on an aggregated, somewhat judgmental basis. ^Improvement of the CTS data for relatively non-industral ized states, for example, would have required resurveying those areas. Comparability then might require re- doing the entire survey. 9 It is highly doubtful, for example, that the probability of construction or the exact source of funding for a local project that is scheduled for initiation four or five years in the future can be assessed objectively. mates: (a) CEAI's estimate of substitution is substantially lower than other esti- mates (albeit they were made for other programs). If CEAI's estimate of substitu- tion is inaccurate it probably understates substitution and thus overstated employ- ment impact; (b) similarly, to the extent that marginal labor requirements differ from average labor requirements, marginal requirements are often thought to be lower than average requirements. Thus if CEAI's assumption of equal marginal and average labor requirements is incorrect, it is likely to have resulted in an over- statement of labor requirements (and thus employment); (c) CEAI took great pains to avoid understating the impact of the employment multiplier by rounding it up from 2.8 to 3.0. This raised the total employment estimate by 7 percent. In order to reflect the timing of both, average of direct and indirect (the latter being slightly larger) was used as the multiplicand, in the employment multiplier, rather than just direct employment. This raised the total employment by about 1 percent. Comparison with Other Estimates There are very few other employment impact estimates for either the LPW pro- gram or for other programs that are sufficiently similar to be comparable J' Most of the studies that do exist, however, suggest that the employment impact is sub- stantially higher than CEAI's estimates. In general, most of the discrepancies appear attributable to these studies' failure to make adjustments for crowding out, substitution, and price and/or productivity changes. Brief specific comparisons with these studies -- which represent most of the available methodological approaches are presented after the discussion of the adjustments made by CEAI. The Impact of CEAI Adjustments CEAI's adjustments for price and productivity changes, crowding out, and sub- stitution substantially reduced estimated employment. Specifically: CEAI's adjustment for price and productivity changes reduced estimated employment impacts by at least 11 percent of the unadjusted level^ 2 (11,100 person-years of direct employment, or about 33,500 person-years of total employment). • CEAI's adjustment for substitution reduced estimated employ- ment impacts by 20 percent of estimated gross adjusted employ- ment (17,730 person-years of direct employment, or 20,625 person-years of total employment). • CEAI's adjustment for direct crowding-out in the period of the LPW program reduced estimated employment impacts by nearly 9 percent of estimated gross adjusted employment (7,975 person- years of direct employment, or 24,190 person-years of total employment). • CEAI's simulation of an LPW program beginning in the second quarter of 1974 indicated that construction employment would have been nearly 9 percent (7,880 person-years) higher than adjusted gross employment estimated for the actual program, and total employment would have been over 13 percent (36,200 person-years) higher than that estimated for the actual LPW program. !1 A11 BLS studies, upon which CEAI's estimates of direct employment impact are based, are excluded from this section, since any comparison would simply involve CEAI's adjustment procedures, which were discussed in Chapter 2. 12 This figure is based on an alternate estimate direct of employment using labor re- quirements adjusted by the BLS to 1977 for all LPW expenditures. Most other studies use labor requirements adjusted to some earlier year, which makes the discrepancy even larger. Without netting out the substitution and direct crowding-out effects, CEAI esti- mated that direct (construction industry) -employment impacts would be 88,640 person-years, and total employment impacts would be 270,170 person-years. These employment figures are about 40 percent higher than those shown in Table 4.1. The estimated cost per person-year of employment would then be about 29 percent'lower than shown in Table 4.1 ($67,224 per Derson-year for direct employment, and $22,055 per person-year for total employment). Had the problem begun in 1974.2, estimated total employment impacts gross of substitution would have been over 50 percent higher than employment shown in Table 4.1, and the estimated cost of employment would thus have been some 35 percent lower. Job Cost Estimates Most of the available estimates of employment impact of public works expen- ditures summarize their findings in terms of the cost of jobs, or person-years per $100,000 (which is the reciprocal of cost). These results are summarized, by type of employment, in Table 4-3. In order to facilitate the comparison, CEAI's estimates have been adjusted to be gross of substitution and construction crowding out and to be based exclusively on 1977 price and productivity adjustments. The Congressional Budget Office estimated the total employment of injections of public works expenditures, using BL5 data as an input for a macroeconomic simu- lation of the economy J Two estimates are obtained, based on different assumptions about labor intensities of the projects. The resulting impact is estimated to be 6.4 to 8.0 person-years of employment per $100,000 (as compared with CEAI's estimate of 5.55). The estimates are for 1975, however, and CEAI estimates that adjustment to 1977 price and productivity levels would have lowered the results to 5.31 and 6.40. Moreover, the lower figure is based on ''average" labor intensity, while the upper value is set at an extremely high level in order to produce an upper limit. These estimates are thus guite consistent with CEAI's projection of LPW employment generation. T3 The exact references for this and the following studies are given in the notes to Table 4-3. 14 The simulation is not exactly comparable to the LPW program, since a continuous injection is assumed. CEAI has interpreted the longest term (24 months) results as the equivalent of CEAI's estimates. The exact manner of injection of expen- ditures is not adequately documented, so that no methodoloqical comparison is possible. 15 The upper limit is derived under the assumption of expenditure on projects with a labor intensity of 0.6. By comparison, materials costs alone averaged about 55 percent of direct LPW expenditures, and were not less than 42 percent for any construction category. 69 Type of Employment CEAI Direct 1.06 Indirect 1.09 Total 3.23 TABLE 4-3 LABOR COST ESTIMATES (person years per $100,000) CEAI 1977 , Sulvetta,& - Estimate C.B.0. c Thompson Vernez e AFL-CIO 1 " 1.82 - 2.4 3.34 7.0 1.88 - - 1.58 5.55 6.4-8.0 a) Best estimate. Taken from Table 4-1. b) Based on labor requirements adjusted to 1977 by the BLS figures are gross of substitution and crowding out in the construction industry. c) Source: U.S. Congress, Congressional Budget Office, Temporary Measures to Stimulate Employment: An Evaluation of Some Alternatives , Washington, 1975. Estimates are based on 1975 data. d) Source: U.S. Department of Commerce, Economic Development Administration, Program Analysis Division, An Evaluation of the Public Works Impact Plan (PWIP ), by Anthony J. Sulvetta and Norman L. Thompson. PB-263-098. Springfield, VA: N.T.I.S., 1975. Estimates are based on 1972 data. e) Source: George Vernez, et. al_. , Regional Cycles and Employment Effects of Public Works Investments. R-2052-EDA, Santa Monica: Rand, 1977. Estimates are based on 1974 data. f) Source: George Meany, "Statement before the House Committee on Public Works, on H.R. 11 and Related Measures to Increase Authorization for the Local PUblic Works Capital Develop- ment Investment Act of 1976," February 1, 1977. 70 Sulvetta and Thompson base their estimate of direct employment impacts on a sample of 226 PWIP projects. Their estimate of 2.4 person-years per $100,000 ^is based on 1972 data. (For comparison, CEAI's 1977 estimate is 1.82) When adjusted for 1977 price and productivity levels, however, CEAI estimates that the Sulvetta and Thompson value falls to 1.6 person-years per $100,000, which is also consistent with CEAI's own estimate. Vernez et. al . base their direct employment impact estimates on BLS data. That the resulting estimate (3.34 person-years per $100,000) should be so much higher than CEAI's estimate (1.82) is therefore a bit surprising. Over half of Vernez's categories, however, are civil works'* 7 -- mostly water projects that were excluded from the LPW program. These project types have relatively high labor re- quirements. If they are excluded, the average for Vernez's employment impact falls to 2.82 jobs per $100,000. The rest of the discrepancy is explained in terms of changes in productivity and prices between 1974 and 1977. Vernez's indirect employment estimate (1.58 jobs per $100,000) is below the CEAI estimate (1.88). Adjusted to 1977 price and productivity levels, 1 ' 8 CEAI esti- mates Vernez's indirect employment impact value to be 1.32 jobs per $100,000. CEAI believes that much of the discrepancy between this value and Vernez's result is due to the use of a partial input/output table in Vernez' analysis. The partial table includes only 13 two-digit manufacturing industries, one mining industry, and the transportation, wholesale, and retail services sectors. Although the omissions of direct materials requirements may be small, the omission of derived demand for labor is likely to be substantially larger. The AFL-CI0 estimates of 7.0 person-years per $100,000 of direct employment and 3.5 to 7.0 person-years per $100,000 of indirect employment is high by any standard. The estimate is essentially undocumented, although it is reported to have been "based" on the CB0 and BLS studies. Given the available sources, how- ever, such an estimate can only have been made by using quite old data, unadjusted for price or productivity changes, and/or by using values at the top of the esti- mated ranges. Since documentation is completely lacking, adjustment for comparison is impossible. 16 The estimate is originally made in terms of jobs and converted to person-years at the rate of 0.079 Dersnn-vpars nor inh. the rate of 0.079 person-years per job. Vernez assumes equal expenditure on 22 project types, 12 of which are public works. CEAI uses just one of these public works categories, primarily as cate- gory of miscellaneous outdoor projects. 18 Omission of the 12 civil works categories leaves the employment impact value virtually unchanged. Employment Multipliers Estimates of total employment can be compared by analysis of the employment mutipliers. CEAI's Macroeconomic Model simulation of the economy produced an employment multiplier of 2.8, which was judgmentally rounded up to 3.0. Vernez is the only author who explicitly estimates both the direct and total employment impacts necessary to measure an employment multiplier. In this esti- mate an econometric model, which converts expenditures to physical amounts of em- ployment and materials and uses these as inputs, is simulated for the 1970-71 re- cession. An employment multiplier of 3.7 is estimated for sewers and one of 3.9 is estimated for public buildings. -° Aoout nait ot tne cnrrerence Detween tnese estimates and CEAI's LPW estimate appears to be accounted for by differences in the level of crowding out existing in a recession and that existing in the LPW period. CEAI believes that the rest of the discrepancy is attributable to the specifications of Vernez' s model. Since it is not a complete macroeconomic model of the economy, it does not appear t£ capture all of the (negative) feed- backs of expenditure that do exist. Policy Implications The primary policy implication of CEAI's findings is that the LPW program is an expensive means of creating employment. A person-year of employment in the construction industry is estimated to cost nearly $95,000, and overall cost is over $31,000 per person-year. (Of course, to the extent that the LPW program is designed to meet other goals such as addinq to local infrastructure, it is unfair to compare the costs per job directly with the costs of other programs.) Most of the other policy implications pertain to the reasons for this level of cost. Much of the high cost of job-creation is attributable to the LPW's focus on the construction industry. The industry is not particularly labor intensive, and wages tend to be high.^ 1 Both of these factors imply a high cost per person- year of employment. If the goal of a policy is to create construction industry jobs -- as it clearly was -- their high cost must be recognized. 19 The employment multiplier is defined as the total employment impact, divided by the direct employment impact. 20. An estimate cited by Vernez for comparison results in an employment multiplier of 2.8 for water and waste treatment facilities and 3.3 for educational facili- ties. See Roger H. Bezdek and Bruce Hannon, "Job Impact of Alternatives to Corps of Engineers Projects," Engineering Issues , October 1973, pp. 521-531. Bezdek and Hannon computed total job estimates. For comparability, Vernez trans- formed these estimates into employment multipliers. These results tend to con- firm CEAI's aggregate estimate of 3.0 rather than Vernez 's estimates. 21 In part this is due to (and compensates for) the intermittent nature of con- struction employment. Unionism is also a factor in many areas, and the Davis- Bacon and other program-specific requirements are estimated to have increased costs by 10 percent, and possibly more. The LPW program's impact was spread out ever a longer period of time and occurred later -- both absolutely and relative to the business cycle -- than was originally intended. This lowered the employment impact in two ways: First, construction price and productivity increases continually erode the employment generated. 22 Second, crowding out increasingly offset the employment impacts. A shorter, more concentrated program would have had a substantially larger employ- ment impact. A final policy implication of the above comparisons is the importance of using forecasting in program evaluation. At the time of the debate concerning-- and approval of--the LPW program, the available "state of the art" analysis con- sisted of ex post studies of employment impacts. Even the best and latest of these studies used 1974 or 1975 data. Moreover, some of the studies focused ex- plicitly on recessionary conditions. By contrast the peak impact of the LPW program occurred at least three years after the effective dates of all of the studies and indeed three years after the trough of the recession (March, 1975) During this period both cyclical and secular changes in many variables and economic relationships acted to reduce the employment impacts of the LPW program to levels substantially below those originally estimated. If the impact estimates upon which policy makers base their decisions are to be accurate, they must con- sider the expected life of the program and they must include forecasted adjust- ments for events which are likely to occur during this future period. 22 Since the general value of the dollars decreased in value over this period, the comparison of the LPW program with other programs having earlier expenditure pattern is inappropriate unless the necessary adjustments are made. CHAPTER 5: INDUSTRY IMPACTS It is useful to trace the flows of LPW expenditures through the materials producing sectors of our economy for two reasons. First, approximately 60 percent of all construction expenditures are transmitted directly to the industries which produce materials and provide services for the construction industry. Since the materials producing industries are a significant beneficiary of the LPW program, and no analysis of the program would be complete without an examination of these impacts. Second, in order to examine the inter-regional impacts discussed in Chapter 7, one must first know the flows of funds to the various materials producing industries. The construction industry, of course, is itself a significant industry. However, the calculation of the impacts upon the construction industry is quite straightforward and was discussed in earlier chapters. The amount of total funding received by the construction industry is the entire $6 billion of LPW expenditure. The timing of the expenditure pattern is shown in Table 2-2. The amount of construction industry employment generated by the construction activity is shown in Table 4-1 . The Pattern of Industry Impacts Methodology The methodology used to identify the industry impacts is described in Chapter 2. Briefly summarized, input/output categories were aggregated into 57 major material supply industries, which are shown in Table 2-3 and Table 5-3. LPW expenditures were aggregated by type of construction expenditure as shown in Table 5-1. The quarterly LPW expenditure flows were then adjusted to reflect differences between the timing of total construction expenditures and the timing of certain materials purchases.' These phasing adjustments are shown in Table 5-2. The time-phasing was developed from information supplied by contractors and material supply trade organizations. The actual adjustment of expenditures was based upon the average size and duration of the projects and was somewhat judgmental, due to the variety in type and scope of projects funded by the LPW program. The mechanics of the adjustment are illustrated as follows: In an industry, such as lumber, which contributed inputs primarily in the first half of a project, an increasing proportion of expenditures in the latter quarters was removed and reallocated to the earlier quarters up to the point where 50 percent of the total funds for that round had been spent. Table 5-1 CLASSIFICATION OF NEW CONSTRUCTION CATEGORIES FOR MATERIALS REQUIREMENTS STUDY Private Sector 1. Residential Single-Family Dwellings 2. Residential Multi -Family Dwellings 3. Mobile Homes 4. Residential Additions and Alternatives 5. Nonhousekeeping Residential (Hotels, Motels, Dormitories, etc.) 6. Industrial 7. Offices 8. Stores, Restaurants, Garages, Warehouses 9. Religious 10. Education (except Housing) 11. Hospitals 12. Miscellaneous Nonresidential Buildings (including Transportation Terminals) 13. Farm Nonresidential 14. Oil & Gas Exploration and Drilling 15. Railroad 16. Communication 17. Electric Utilities 18. Pipelines and Gas Utilities 19. Miscellaneous Nonbuilding Construction (Bridges, Dams, Playgrounds, Airfields, etc. Public Sector* 20. Highways 21. Military 22. Conservation and Development (Parks, Waterways, etc.) 23. Sewer Systems 24. Water Systems * Note: public education, hospitals, industrial residential and miscellaneous all are assumed to have material requirements identical to the corresponding private sector. TABLE 5-2 TIMING OF MATERIAL SUPPLY DEMAND FOR BUILDINGS Supply Industry 8 Lumber 9 Veneer and Plywood 10 Mill & Wood Products 21 Structural Clay Products 23 Cement, Concrete, Gypsum 24 Other Stone & Clay 31 Plumbing & Heating Equipment 33 Other Structural Metal Construction Phase First 50% of Project First 50% of Project Final 50% of Project Final 50% of Project First 40% of Project 30% to 60% of Project From 35% to 100% of Project From 20% to 50% of Project 76 The adjusted expenditures were then injected into the inverted input/output matrix to obtain industry impacts on a quarterly basis. The results, in current dollars, are shown in Table 5-3. Finally, employment estimated, in person-years, were de- rived using actual and forecasted annual productivity indices.- These estimates are shown in Table 5.4 and are discussed further in Chapter 4. Direct and Indirect Requirements for Construction An input/output table contains coefficients measuring the direct purchases required per dollar of activity for each type of construction. The portion of the Chase/INFORUM Model pertaining to construction requirements is presented in Appen- dix D, Table D-l . However, we must calculate the indirect effects as well as the direct effects, so that we may properly assess theindustrial impact of construction spending. Only then, for example, can we estimate the cumulative effect on the steel industry of contractors' demands for such steel -using products as structural steel, valves, air conditioning units, etc. The analysis of such total economic effects utilizes a matrix closely re- lated to the input/output table. The new matrix* called the Leontief Inverse or the matrix of direct and indirect requirements, quantifies the result of the indefi- nitely long chain of Droduction caused by any change in the economy's final demand. For example, motor vehicles require tires, which in turn create demand for rubber, steel and polyester. The process becomes an infinite chain because the steel indus- try also requires parts for its own hauling equipment, which is made by the same motor vehicle industry we started with.^ In Table D-2 we illustrate the portion of the Leontief Inverse from the Chase/INFORUM Model that deals with construction requirements. That is, these are the direct plus indirect requirements for materials per dollar of construction ac- tivity. At this point, for example, we can see precisely how much the metals are affected by construction activity. Industrial construction, for instance, requires 9.5 QC en cc I— z=> I/O CJ Q Ll_ 2: O §2K§i|§ 5 gid j|p^d5rj=!p^!s!i:i^iiss:. 80 TABLE 5-4 INDIRECT EMPLOYMENT, BY INDUSTRY (person years a ) 1979 TOTAL 1. Agriculture 253.3 497.3 142.5 8.7 901.8 2. Mining 923.0 2280.3 .600.5 37.0 3840.8 3. Food Products 30.4 70.9 18.9 1.2 * 121.3 4. Floor Coverings 5.3 11.0 2.8 .2 19.3 5. Textiles & Apparel 156.4 359.9 94.4 6.0 616.7 6. Lumber 267.0 494.0 108.4 8.7 878.2 7. Wood Products 379.0 932.2 291.5 17.7 1620.4 8. Furniture 171.0 390.4 117.1 7.9 686.4 9. Paper Products 285.3 679.3 182.3 11.7 1158.6 10. Paints 75.2 172.2 45.4 2.8 295.6 11. Chemical Products 229.5 537.3 137.7 8.6 913.1 12. Petroleum Products 77.9 195.6 49.2 2.8 325.5 13. Rubber Products 77.7 218.3 52.5 3.3 351.8 14. Plastic Products 215.9 498.8 130.1 8.5 853.3 15. Leather Products 6.9 16.8 4.6 .3 28.6 16. Glass 113.4 260.2 69.8 4.7 448.1 17. Cement & Clay Products 2291.3 4778.5 1120.0 75.3 8265.1 18. Iron and Steel 1202.4 2599.6 647.5 43.2 4492.7 19. Non-Ferrous Metals 486.9 1116.5 290.9 18.8 1913.1 20. Fabricated Metal 387.6 839.4 217.0 14.6 1458.8 21. Plumbing & Heating Eq. 141.0 335.8 11.1 6.3 494.2 22. Structural Metal Products 1885.9 4221.3 857.2 65.0 7029.4 23. Metal Stampings 243.3 572.9 146.9 9.5 972.6 24. Engines and Turbines 25.2 68.2 15.6 .8 109.9 25. Const, Mine, Mater Hndlg Eq 230.0 525.1 141.4 9.2 905.8 26. Misc. Non-el ec Mach 297.6 682.0 175.8 11.1 1166.5 27. Serv. Indust. Mach. 127.7 285.1 79.8 5.3 497.9 28. Transf & Switch 183.3 414.4 110.4 7.5 715.6 29. Appliances 31.1 69.0 19.3 1.3 120.7 30. Light & Wiring Eq. 323.3 722.9 196.8 13.3 1256.3 31. Communication Eq. 29.6 63.7 19.0 1.1 113.4 32. Elec Mach, NEC 177.6 410.5 110.4 7.0 705.5 33. Transp. Commodities 50.0 112.6 28.2 1.7 192.5 34. Instruments 166.6 354.5 96.0 6.3 623.4 35. Misc. Manufacturing 373.4 836.5 218.3 13.8 1442.0 36. Transp. Servs. 1533.7 3639.8 933.9 56.7 6164.1 37. Wholesale & Retail Trade 5429.1 11848.6 3006.5 193.5 20477.7 38. Business & Legal Services 4284.8 6621.5 1708.7 109.8 12724.8 39. Other Services, NEC 1692.4 3831.2 992.5 62.9 6579.0 TOTALS: 24861.3 52564.1 13191. 91480.4 a)Unadjusted for Substitution and Crowding Out Table 5-5, where they are compared with the shares of expenditure impact. The two major differences are in Services and Metals. The employment share of services is just over 50 percent of all indirect LPW employment—a third larger than Services' expenditure share. The reason for this is relatively low pro- ductivity in the service sector. Metals, by contrast, receives only 7 percent of the employment impact--or 30 percent of its share of expenditure impact. This is due to high productivity in that sector. Cyclical Characteristics of the Materials Supply Industrie s An important policy issue is the extent to which the LPW program relieved the cyclic distress of industries related to construction. It is first necessary, however, to ascertain the degree of cyclic sensitivity of these in- dustries. This involves the timing and volatility of the cyclical movement and the degree of distress suffered by these industries. Timing and Amplitude of Industrial Cycles The timing and amplitude of cycles of individual industries were measured using the Federal Reserve Board's indices of industrial production. Definitions of the construction related industries on which these indices are based are given in Table 5-6. The individual industry indices were compared to the four most recent. NBER reference cycles, which are shown in Figure 5-1 and are defined as follows: 8 Contractionary Period Cycle Peak Trough I. August 1957 April 1958 II. April 1960 February 1961 III. December 1969 November 1970 IV. November 1973 March 1975 The timing of industry cycles is summarized in Table 5-7 in terms of the number of months that the turning points of the industrial production indices lag (lead) the NBER reference cyels. Although there is considerable variation among industries, there are significant differences between the cycles of construction related indus- tries and the movement of overall economic activity. The industrial production in- dices tend to lead the NBER reference cylcles at both the peak and trough, although the trough leads are both smaller and less prevalent. The peak lead is as much as four guarters for many industries in Cycle I. In both peak and trough, however, the turning points of the industrial production indices show a tendency to become more coincident with the NBER reference peaks and troughs in Cycles III and IV then in Cycles I and II. Peak and trough values of the industrial production indices and percent changes during recoveries and recessions are shown in Table 5-8. The results show a consid- erable variety of amplitudes of cyclic movement -- both absolutely and in The cycles are defined as beginning with the recession (i.e., with the peak). Implicitly the cycle continues until the next peak, althouqh that peak is con- sidered part of the next cycle. TABLE 5-5 SHARES OF INDIRECT LPW IMPACT BY INDUSTRY GROUP Industry Group Agriculture Mining (except Stone & Clay) Food Products Textiles Lumber & Wood Products Chemicals (including Paints) Petroleum Products Rubber Products Plastic Products Stone, Clay & Glass Products Metals Fabricated Metal Machinery Other Manufacturing Services Share of Expenditure Impact (%) a 1.0 2.7 0.3 0.6 5.1 2.8 3.5 0.5 1.1 11.8 10.5 13.9 6.5 2.1 37.5 Share of Employment Impact 2.6 0.1 0.7 4.7 1.3 0.4 0.4 0.9 11.1 7.0 10.9 6.1 2.5 50.2 a) Percentages do not sum to 100 due to rounding. 83 3 G) CC en .J a a en £-1 CO 2: a a CO u _j o >i o a z: u u L- CD O X (O \\\\WW\\\l\ \TTTTTTTT \\ \\W\WA \ \ \ \ \WTTTT \\\\\\\\\\\\\\\U\\ TTTVTTV XXX VTVTTTTYft Vf Id Q Q 00 KLNULGNOO 8d TABLE 5-6 CONSTRUCTION RELATED INDUSTRIAL PRODUCTION INDICES INDUSTRY NAME CARPETING LOGGING AND LUMBER LUMBER PRODUCTS HOUSEHOLD FURNITURE BUILDING PAPER AND BOARD PAINTS PETROLEUM REFINING PLASTICS PRODUCTS NEC CEMENT STRUCTURAL CLAY PRODUCTS CONCRETE, MISCELLANEOUS CLAY MANUFACTURING COPPER ALUMINUM NONFERROUS MILL PRODUCTS STEEL MILL PRODUCTS HARDWARE, PLUMBING, STRUCTURE METALS OFFICE, SERVICE AND MISCELLANEOUS CONSTRUCTION AND ALLIED EQUIPMENT SPECIFIC AND GENERAL INDUSTRIAL EQUIPMENT MAJOR ELECTRIC EQUIPMENT AND PARTS HOUSEHOLD APPLIANCES MISCELLANEOUS APPLIANCES EQUIPMENT, INSTRUMENTS AND PARTS SJ_C_ CODE 227 241 , 242 243, 244, 2^9 251 266 285 291, 299 307 224 325 326, 327 3331 3334 335 331 , 332 342, 343, 344 357, 358, 359 353 355, 356 361, 362 363 3634 , 3635, 3636, 3639 381 , 382. 383, 384 CMCMOOCMCMOO-OOOOP - r- -Q vd- o o o r— inCftnCOOlDOJOOlDCOCMOOOlO^MWCOC COWCVIr-On CMCM i— OCM-QCMCMOOi— CM CM ^ LD CM C CU i— fOOr-Oi-r-WNON(V)r-OOMNn(MOiOr- ioojinoOi-LnooomLi-) cy> cm cm cr> r-Mr-cgOlJSJDOOOiOOvlOr-fT^^COOOISSM X a; CI) -o c c CU ^ (1) C 4- o cu <_ 4-> CJ C£ -s LU o CO cri-=d-i — en cm n o jd cm = •> cm oo o o c i — 4-> 3 - CM O i — CTiOOi — O -O CO CM O c — i — i — iDCOUOCOr o i/> o -r- cru 1 c: oa ■<- 3CO"5C5 CU "0 Q.+- 3 s.sa -(-> C S- -C • 1/1 r— ■!- +-> ■»-> cu S- E s- cu s- a» -i- C o CU T- CU r— « CJ -l-> Q. CD-Q CO r— c cu zs u Q. E 14- CU CU -r- on S- CD E 3 »r- 4-> ra E i- c Q. 3 C CU S M- C cu fC O 33 O Z3 a) r— cu +-> o Or— o +-> -a 4- o Cl U-J-JXDQ Q- Q. C o co CJ o < 2: co re o o CO 36 CD O < O 3C CC lo ro co o r^ « LO taCJUOO (71 rH OM£l OlO O CTi C7> CVi CTi CTi CTl jooooooocn omnitHHC -I O >-i t-n O U OlfOr t" MD O O O TNOCTiO^kDlOrvNO 25 i— t— • E CO -r- CO 25 4-> C S- -r- 25 _j (_> t- o) *+- -a 25 25 Q_ (DO "O "O U_ (t> Qi S_ COO. Q. T) rt S--0 E Z D- i— CO 25 00 -co o c ai o ■> c s-^r-i- to p— -i-4 j-!-a)a>-04->o+Js i. COjd to i— c 5- to ( - cr> E zj •!- -r- +■> n3 ! 30=50=3fCa>'— <_ )_l JICQD.D.Q.UWU _q s- o o . ol •— E CU -r- 23 r- 25 CO +J "O E O-r- (_) c i— i— ac t_> i — u +-> "O H- O O- < CXZWIOOW5 E -* 25 fT3 E QJ 88 Takina the cycles as a whole, however, metals in general, and steel mill products in particular, show far the greatest volatility. Cement, lumber, construction equipment, and appliances also have relatively large fluctuations. There is also a general tendency (except for steel mill products and construction equipment for the recession of Cycle IV to be substantially more severe than previous declines in production. The Influence of Construction The ability of a program such as LPW to stimulate a material supply industry depends in large part on the importance of construction as a market for that industry. The primary difficulty is that an industry may have many non-construction uses for its outDut. Metals are a major examDle of this, 10 but even lumber, which is especially stronaly influenced by construction, has the steadying influence of relatively stable demand for paper products. The question of the influence of construction — and the entire issue of volatility—is also complicated by the fact that some demands especially for building materials and equipment, are subject to substantial short-term inventory fluctuations. There are two cases that show the effects of a cycle in construction activity relatively clearly. The first is the recession in Cycle IV, where the downturn in construction coincided fairly exactly with the movement of the general economy. The downturns for construction related industries in Cycle IV are generally more severe than in other cycles. The fall in the industrial production indices for Lumber & Wood Products, Stone, Clay & Glass, and many Machinery categories (including appliances) are 50 to 100 percent larger than occurred in previous recessions. Metals, however, are not especially adversely affected. The second case occurred during 1966-1967, The cycle in construction activity which resulted from the 1966 "credit crunch" was much greater than in other components of final demand. There was no NBER reference cycle durinq this period, and income rose through the construction cycle. Thus the movements in mate- rial supply industrial production indices can be interpreted as reflecting primarily the influence of construction. Table 5-9 (which can be interpreted analogously to Tables 5-7 and 5-8 except that the construction peak and trough are used instead 9 The size of the expansion between Cycle II and Cycle III is primarily due to the great length of this recovery. The metals industries face several volatile markets, of which construction is only one. For the iron and steel industry, for example, both automotive products and machinery are major markets. Both of these downstream markets show substantial volatility, which can cause major swings in the production and sales of steel. Copper producers are in a similar position. Furthermore, since copper is a world-wide commodity, this industry is subject to variation from both the supply and demand side. Aluminum tends to be somewhat more stable, since it is less dependent on the automotive market than steel, and because the single largest market for the aluminum industry-containers and packaging—is relatively stable. 39 of an NBER reference peak and trough) shows these effects. Most of the construction related industries also turn down, and do so in the lagged manner that one would ex- pect. In most cases, the percent declines in industrial production indices, in fact, are remarkably similar to the subsequent recession in Cycle III. Metals, however, are relatively unaffected by the construction cycle. Concrete falls relatively sharply, reflecting an especially great importance of construction to that industry. Appliances, office equipment, and carpets turn down relatively little or not at all, which reflects the fact that the income (which did not decline) is at least as important a factor in demand for these products as is construction activity. The above analysis suggests that metals are affected by construction activity only slightly. Stone, Clay & Glass—especially concrete—are very substantially affected. For the type of construction downturn which occurred in 1966-67 - a collapse in single family home construction - lumber and wood products were also substantially affected. Fabricated Metal Products and Machinery are only spottily affected by construction activity. Measures of Industry Distress A convenient measure of cyclical distress of an industry is size of fluctuations in output—especially the downturns. The difficulty with this measure is that some industries (e.g., lumber or steel) are chronically volatile, and they may have adjusted to this pattern. The fundamental question is whether or not high degrees of cyclical ity are compensated for by higher rates of return. In order to test this possibility, CEAI calculated the average rate of,, return over the past 20 years for a selection of 21 manufacturing industries. Upon comparing the rankings of the mean rates of return with the relative cyclical ity of industries, CEAI found no reason to expect the average of return over the long term to be any higher in cyclical industries than in acyclical industries. It does appear to be the case that the amplitude of movement of the rate of return is larger among cyclical industries than among acyclical in- dustries, but this only confirms the use of fluctuations in output as a measure of cyclical distress. The measure of cyclical distress of an industry that is used in the sub- sequent analysis is the decline in output (shipments) from the previous peak. This The calculations were based on data on return on book value drawn from industry composites constructed by Standard and Poors. Such data are based originally on financial statements of companies, which introduces a wide margin of error into industry rate of return analysis. The difficulty with this type of analysis is that the companies are not confined to one industry, and the stated primary product of a company may not have a dominant influence on the rate of return. 90 can be measured in absolute or percentage terms. The distress of construction related industries at the time of a cyclic trough is shown in the last portion of Table 5-8 and the last column of Table 5-9. The cumulative level of industrial cyclical distress at two-quarter intervals was simulated for the last two recessions under the assumption of the absence of counter-cyclical policy, and is presented in Appendix E. The Size of the LPW Program Relative to the Industry Impact Two comparisons of the stimulus of the LPW program with industry distress were made. First the LPW stimulus for the peak year of the program, 1978, was compared to the total industry distress at the trough of the 1974 recession. '^ Second, the total LPW stimulus was compared with the amount of stimulus that would have been required to overcome 20 percent of industry distress. Table 5-10 presents the estimated value of shipments induced by the LPW program, by industry J 3 in Table 5-11 the value of shipments at, and date of, the peak and trouqh of the last recession are presented. Industry distress is shown in absolute terms and as a percent of the previous peak. The 1978 LPW stimulus is also presented both in absolute form and as a percent of industry distress. Table 5-12 compares the total LPW stimulus—both actual and as a percent of industry distress—with industry distress and with a potential tarqet of 20 percent of industry distress, which is used as a benchmark of "measurable" impact. The peak year impact of the LPW program on materials supply industries was small. As shown in Table 5-11, the impact ranges from 0.02 percent of distress to 10.21 percent of distress, with an average impact of 2.4 14 percent of industry distress. Only five industries receive impacts greater than 5 percent of industry distress. 12 The materials industry stimulus for 1978 is estimated to be 54 percent of the total LPW program materials industry stimulus. A single year's stimulus is used for comparison with distress because all shipments data are on an annual basis. The data in this table are based on Table 5-3. The industry categories have been aggregated to the 39 industries used in Table 5-4. Expenditures have been aggregated from a quarterly basis to an annual basis, and have been transformed from current dollars to constant (1972) dollars. All values in the following tables are also on an annual basis in 1972 dollars. This excludes 29 percent of 1978 LPW impacts, which fell on industries that did not experience measurable distress. The greatest impact is received by Structural Metal Products. Other industries with impacts of over 5 percent are Electrical Machinery, NEC; Plastic Products; Cement & Clay Products; and Construction, Mine & Material Handling Equipment. TABLE 5-10 LPW IMPACTS ON INDUSTRY OUTPUTS (SHIPMENTS) (Thousands of 1972 Dollars) Material Suppl iers 1. Agriculture 2. Mining 3. Food Products 4. Floor Coverings 5. Textiles & Apparel 6. Lumber 7. Wood Products 8. Furniture 9. Paper Products 10. Paints 11. Chemical Products 12. Petroleum Products 13. Rubber Products 14. Plastic Products 15. Leather Products 16. Glass 17. Cement & Clay Products 18. Iron and Steel 19. Non-Ferrous Metals 20. Fabricated Metals 21. Plumbing & Heating Equip. 22. Structural Metal Products 23. Metal Stampings 24. Engines and Turbines 25. Const, Mine, Mater Handlg Eq. 26. Misc. Non-el ec Mach 27. Serv. Indust. Mach. 28. Transf & Switch 29. Appliances 30. Light & Wiring Equip. 31. Communication Equip. 32. Elec Mach, NEC 33. Transp. Commodities 34. Instruments 35. Misc. Manufacturing 36. Transp. Services 37. Wholesale & Retail Trade 38. Business & Legal Services 39. Other Services, NEC TOTALS Program Total : 3,078,824.8^ 1977 1978 1979 8865.3 20886.3 5849.0 382.3 32304.9 80950.3 21317.4 1296.1 2672.4 6391.6 1767.7 115.8 277.9 582.9 151.9 9.3 5410.0 13533.1 3736.1 245.4 13083.7 25195.2 5746.9 471.2 14592.2 34490.0 10640.5 654.9 4616.8 1054.2 3162.2 220.5 10912.5 26663. C 7383.4 474.3 14738.7 11020.1 2996.3 189.8 16215.6 40031.5 10903.3 703.3 17610.8 46360.1 12154.8 699.9 3185.7 8843.4 2233.0 136.8 8637.6 20451.7 5592.6 373.7 235.0 580.9 161.4 10.8 3856.3 8847.5 2444.1 162.7 82488.1 176803.8 43679.7 3012.2 57717.2 129978.5 33668.9 2244.5 36033.6 84855.3 22978.1 1524.6 23641.6 52041.6 13885.9 940.3 4512.6 10745.5 3666.9 214.1 64119.5 156188.8 33432.2 2598.3 9489.0 22343.6 6024.7 388.8 1665.4 4777.6 1172.9 64.3 9200.7 22052.6 6223.4 406.5 10864.1 26120.0 7085.1 461.6 7915.0 18530.7 5349.9 363.2 5499.6 12845.4 3533.8 238.7 1369.7 3171.9 926.1 64.1 9698.2 21687.2 6101.6 413.3 1156.3 2546.8 800.1 46.1 6483.6 15085.2 4138.7 272.2 2628.3 6168.6 1653.8 104.3 5289.1 11521.1 3265.4 218.4 13629.2 31786.6 8512.9 546.8 50229.9 123754.8 33152.3 2069.1 86865.6 201426.4 54117.6 3482.6 68690.3 158916. b 42716.4 2743.8 135389.0 316840.6 84660.9 5410.4 851790.9 1676071.0 516987.9 33975.0 a) Equivalent to $5,714,065 thousand in current dollars during the LPW program. owcnr ,— C\J r— C\J riOLncMinONOO^-co^'ficocoocoor :> rom 'vt-r- uDr— oudodo co lo o ^d- c\j c ■incOW[NlTjr-CMlfl03N5 rcrit-icocoocny30inir)Ln^ooco^(NJCO'sfnir;r-'/)0^wi-Mu3r-cou3MU3 ro as joiominnONNi — m ic in co ' — ro^DUDC^riorocaii — m«3^rofoio , *<3-i — ■)cocccociocoiorocoffo^vo^-owi-row - o-> lo . — PTi--C\jroir)roooooi£)OLn^-, — worn TiDLnu;inco^^Oo3^0cnwDi-cOi-crnDo ■) ITS LO LO LO LO I ) LO LO LO LD ■? - I£) N lO LO CTl W i — COr -OOOlCOi-NOC iriN lo cti r t"r— LOCVJLOCNJr-.CM': _ .jai o^d aid — i — r^Lf. _ _ .■) «D LDlDi — i — ^UJOLnN^tOMMLfinCM^^rcOi — LOCOIMOCOCTiCOO i h^-Ln mi — romocTnoLnLnNLOCVih. cxjn^i — maiNcnocoocOLfii — cocooom j^t-i — coi — en oo rv o aii — in lo m ^- m r- r— c\j .— ld , — ^-NintOLnn^ixtLnoDcMN LOl I— LD LO "=3" .— r-rO(M CM r— I— I ■=*" r— C\J CO I i COLO -coc\jc\jLncocxicrir— LnoimcooNr :> cm lo lo lo ■ — locoooni — r~.cz3 r^r^- _isOi^cvjo">coc\ji- co cm "yDOvHoairocMcocor^io^t-tP. . cm CO >— ^ co ■— c\ic\ja^r^cricx)cocriO=^-cvicoooooi— Or- loo ootNjoror-oor-roOi — Lnir)LOCrn£iiM lo r CMr-r-^COLD s- c o o r— S_ Q- X E O a> 3 o T3 O O 13 Z5 S- r- 3d. ^tuz a)

u CD en (/) u -r- a: a; ai _ _ t 3 -D o3 r- £ C I Z5 2 00 ■. ocu (o ro°3-^ c-ooicd: c s. 10 c 3lou) 2:1— .oacc +-> 03 cu u -i— +-> a>~ ra - -Q O -— C +-) S- Cl-i- O) -I u > c: 1— . ; Q IOOh :llcl wn Jhi-i^l-SmO I" LO LO IN co c - LO LO N COCTl Or r lo 10 n oocn Or J CM CM CM CM CM C 93 TABLE 5- 12 STIMULUS NEEDED TO PRODUCE MEASURABLE CHANGE (MilTions of 197? Dollars) Industry Total LPV I % of 20% of Distress Stimul us Distres z Distress 1. Agriculture 2672 36.0 1.3 534.4 2. Mining 3993 135.9 3.4 798.6 3. Food Products a 10.9 a a 4. Floor Coverings 711 1.0 .1 142.2 5. Textiles & Apparel 5956 22.9 .4 1191.2 6. Lumber 1312 44.5 3.4 262.4 7. Wood Products 1609 60.4 3.7 321.8 8. Furniture 2098 9.0 2.4 419.6 9. Paper Products 1854 45.4 2.4 370.8 10. Paints 435 18.9 4.3 87.0 11. Chemical Products 4719 67.8 1.4 943.8 12. Petroleum Products 3065 76.8 2.5 613.0 13. Rubber Products 1248 14.4 1.2 249.6 14. Plastic Products 293 35.0 11.9 58.6 15. Leather Products 389 1.0 .2 77.8 16. Glass 325 15.3 4.7 65.0 17. Cement & Clay Products 2781 306.0 11.0 556.2 18. Iron and Steel 4314 223.6 5.2 862.8 19. Non-Ferrous Metals 3638 145.4 4.0 727.6 20. Fabricated Metal 1568 90.5 5.8 313.6 21. Plumbing & Heating Eq. 618 19.1 3.1 123.6 22. Structural Metal Products 1530 256.3 16.7 306.0 23. Metal Stampings 1876 38.2 2.0 375.2 24. Engines and Turbines 428 7.7 1.8 85.6 25. Const, Mine, Mater Handlg Eq 436 37.9 8.7 87.2 26. Misc. Non-el ec Mach 5053 44.5 .9 1010.6 27. Serv. Indust. Mach. 838 32.1 3.8 167.6 28. Trans f & Switch 684 22.1 3.2 136.9 29. Appliances 1034 5.5 .5 206.8 30. Light & Wiring Eq. 969 37.9 3.9 193.8 31. Communication Eq. 606 4.5 .7 121.2 32. Elec Mach, NEC 154 26.0 16.9 30.8 33. Trans p. Commodities 25840 10.5 -- 5168.0 34. Instruments 1116 20.3 1.8 223.2 35. Misc. Manufacturing 921 54.5 5.9 184.2 36. Transp. Servs. 3603 209.2 5.8 720.6 37. Wholesale & Retail Trade 4092 345.9 6.7 818.4 38. Business & Legal Services a 273.1 a a 39. Other Services , NEC a 542.3 a a 92,778 3078.8 18,555.7 a) Industry shipment rose throughout the NBER reference cycle In terms of total expenditures, the manufacturing industry most heavily affected was Metals, an industry group which experienced considerable cyclical distress; however, construction expenditures provide only a small portion of the final demand for the metals industry. The Lumber & Wood Products and Stone, Clay & Glass industries suffered considerable cyclical distress and are strongly influenced by construction; but except for Cement & Clay products these indus- tries received little stimulus from LPW expenditures in terms of absolute magni- tude. All other manufacturing industry groups also received only a relatively small share of LPW expenditures. It should also be noted that the peak year impact of the LPW program occurred three year or more after the trough (when distress was greatest) in all but two cases. In these exceptions the impact was 2 percent of distress or less. 16 In Table 5-12, the impact of the active LPW program relative to the distress of each industry is shown. If 20 percent of distress is taken as the minimum amount of relief of distress which could be called significant, it is clear that the LPW program was not a significant factor in relieving distress for any manu- facturing industry. Indeed, only four industries^ 7 receive a total impact of over 10 percent of industry distress. The average industry impact for the entire LPW program is only 3.3 percent of industry distress,^ which is only one-sixth of the rule of thumb for a perceptible industry effect. In current dollars, therefore, the LPW program would have to be $36 billion in order to have an impact that average 20 percent of industry distress, assuming a similar mix of project types. If one's objective was to relieve 20 percent of industry distress in a sinale year, the entire sum would have to spent in that yearly. Leather Products, which received an impact of 0.15 percent of distress, is predicted to recover beginning in 1978. Electrical Machinery, NEC (16.9%) Structural Metal Products (16.7%); Plastic Products (11.9%) and Cement & Clay Products (11.0%). This is calculated as total LPW stimulus divided by total industry distress. Twenty-six percent of LPW stimulus occurred in industries with no measured distress. period of greatest distress. An analysis of the timing of the LPW program is included in Chapter 8. The small size of the impact of the LPW program on materials industries is primarily due to: (a) the small (in macroeconomic terms) size of the program: (b) the lack of any materials industry targeting criterion in project selection^; (c) the heavy dependence of construction materials suppliers on the other non- construction types of final demand. In view of these factors, it is perhaps un- reasonable to expect a program, whose size and composition is designed for the needs of the construction industry, also to be an adequate and effective counter- cyclical program for the range of industries that happen to supply construction materials. In a broader policy context, in fact, the LPW program was not required to have major impacts on these industries, since it was only one of several policy tools used to stimulate economic activity and employment. As such, its impacts beyond the construction industry must be considered in the context of the package of policy tools used. It is only as a contributor to this package -- not as an individual policy -- that the LPW program can be said to have had any significant effect on industries other than construction. 20 Over 25 Dercent of the impact of the LPW proqram occurred in industries with no distress. Only Structural Metals, Cement & Clay Products, Plastics, and some Machinery received a stimulus that was even remotely significant. The "relief" afforded to Cement and Clay Products is now (1978) exacerbate regional shortage in that industry. The relief afforded standard Metals, Plastics and Machinery was too late in the recovery of these industries to be considered timely. 96 CHAPTER 6 IMPACTS UPON LOCAL GOVERNMENT The relationship of the LPW program to local governments is important for two reasons. First, local government actions affected the nature of the pro- gram, the projects chosen, and the amount of substitution which resulted. Second, local governments were major intended beneficiaries of the program. Where local governments built useful projects which would not otherwise have been built, the citizens of the locality were directly aided. The local govern- ment was also aided where LPW projects resulted in municipal facilities which improved the operation, increased the efficiency, or reduced the maintenance costs of local government activities. Finally, to the extent that local governments built projects which would have otherwise been built with local funds, the locality benefited by having their funds available for other pur- poses The Fiscal Situation of Local Governments Secular Trends At the inception of the LPW program, the fiscal situation of many locali- ties was extremely unfavorable by post-war United States standards, due to a combination of the severe 1973-1975 recession and an apparent change in the feelings of Americans toward locality-provided public goods. Prior to the decade of the 1970 ' s there appears to have been a popular consensus that local governments should have the option of raising tax rates to provide additional public services which were deemed necessary. During the post-war period the state and local tax burden, 2 rose steadily as shown in Figure 6-1, doubling between 1954 and 1970. Only in the early 1970 's did public opposition become sufficiently strong to stem increases in the tax burden. At first, when it appeared that local government activities would have to be curtailed due to a lack of local taxing ability, the Federal government used its powers to aid local governments. During the period 1970 through 1973, federal grants-in-aid to state and local governments rose by 66 percent. While there have been pre- vious periods in which grants-in-aid grew dramatically, the growth durina the 1 2 Substitution is often considered an unmitigated problem with government grant programs, while improved locality fiscal position is considered an unmitigated advantage. Actually the problem and the advantage are the two sides of the same coin and a partial conflict exist between the "jobs creation" and "local- ity aid" goals of the LPW program. To the extent that substitution is minimized and the jobs creation impact is maximized, localities will have fewer "freed" funds available for adding to their operating expenses and improving their financial positions. To the extent that localities use LPW funds for projects which have been undertaken anyway, their financial position is improved but the job creation impacts, and particularly the "targeting" elements of the job creation impacts are minimized. (State and Local Tax)/(Personal income - transfer payments - Federal Income Taxes) UJ g S 98 1970-1973 period was from a high base and, when combined with data on the locality tax burden, provides an unambiguous indication that Americans believed that localities were in dire need of outside aid. However, the strength of the popular antagonism to increased state and local government expenditures which emerged in the early 1970' s appears to be growing contin- ually. The success of Proposition 13 in California and the widespread support of similar measures in other states indicates the magnitude of the shifts in perception of need for local governments to increase their revenues. Many reasons for the "taxpayer revolt" have been suggested, including the sharp increase in the burden felt by taxpayers during the post-war period, the slowdown in growth of real income during the 1970's, a growing disenchantment with many of the activities regularly undertaken by local government, and a feeling that the need for some local government services has lessened due to demographic and other changes. Yet whatever the causes, the result is clear -- a change in taxpayers' preferences away from increased local government activity. This change in the secular trend has made' it far more difficult for localities to manage their fiscal affairs in the 1970's than the 1960 ' s since the option of in- creasing tax rates to finance desired expenditures is far more circumscribed. Locality construction expenditures have been particularly affected by the reduced preference for state and local services. As shown in Figure 6-2, con- struction expenditures are the only major component of state and local expendi- tures to have declined in relative importance. Interestingly, wage and salary disbursement which are often singled out as both a major part of the increased state and local burden and a major area in which cutbacks can occur have been a remarkably steady percentage of state and local expenditures throughout the post- war period. Part of this decline in construction can be traced to a dramatic reduction in expenditures for construction of new educational facilities. How- ever, as shown in Figure 6-3, many other types of state and local construction expenditures, including that for highways, housing, and miscellaneous buildings, have also been reduced dramatically. The extent to which these reductions in construction expenditures are due to reductions in the perceived need for (and/ or increases in the perceived disadvantage of) these types of facilities and the extent to which these reductions were due to state and local budgetary restric- tions is, of course, unknown. In view of the large backlog of unfunded projects found by CEAI surveyors in virtually all localities, both factors are certain to be at least partially responsible. Cyclic Position In addition to the secular problem faced by local governments, these govern- ments faced substantial additional hardship as a result of the recession of 1973- 1975. The impact of recessions upon state and local expenditures is generally quite modest. 3 Until the 1973-1975 recession there had been considerable discus- sion in economic writings as to whether local governments were procyclical (i.e. 3 Prior to 1970, the trend toward increased state and local government activity was so great that some commentators (e.g. Rafuse, Robert W., "Cyclical Behavior of State and Local Finances" in Essays in Fiscal Federalism ed. Robert Musgrave, The Brookings Institution, Washington, D.C., 1965) concluded that the trend has, and would continue to, overwhelm any cyclical impacts. 99 100 UJ >- 1 % < a Sri 331 2' St; §1 101 "local governments, aiven a modest reduction in revenues and perhaps even an in- creased ability to float bonds, manaqed to maintain and/or increase expenditures -- thus acting to aid the economy). 4 In view of the lack of consistency in conclusions on this issue, and the lack of clear procedure for dating state and local government cycles, CEAI has conducted an additional assessment of the cyclical response of local governments. In the process it was found that the nature of the response of state and local government to the economic cycle is far more complex and difficult to determine than has been suggested by most commentators. For example a simple examination of state and local expenditures is not sufficient to examine the cyclicality of local government expenditures. The trend in expenditures for the last few decades has been rising so steadily that prior to 1970, cycles as such were vir- tually nonexistent. Also, state and local expenditures and taxes are not indepen- dent; even in constant dollars the two variables have a correlation coefficient of almost 0.99. The obvious solution to this problem, combining the two concepts and examining surpluses or deficits, produces the following difficulty of inter- pretation: surpluses may respond either to the general economy or to local government policy, and the two responses will have the opposite significance. For example, if surpluses begin to fall, it could signify expansionary spending from a comfortably sated treasury, or revenue losses resulting from the onset of a recession. Similarly an upturn in surpluses could reflect a recovering economy or a local belt-tightening campaign -- perhaps required by increasing financial pressures. In both examples the first response reflects expansionary conditions, while the second reflects contradictory conditions. Several other potential measures of fiscal tightness exist. One of these is the interest rate at which local governments borrow. Sharp rises in the interest rates paid by local governments relative to other interest rates can indicate an increase in the perceived degree of "risk" in lending to local govern- ments. Construction expenditures form final type of evidence on cyclic positions of local governments. They are durable and constitute an interruptable expenditure that is especially susceptible to fiscal distress. A representative locality fiscal cycle can be characterized as follows. The onset of fiscal distress (downturn from peak) is marked by rising locality interest rates and falling revenues and surpluses (although surpluses may already be falling). During the period of distress, interest rates will be high, and localities will re- duce expenditures -- particularly construction. As budgets are tightened surpluses will rise again. This will occur during the distress if it is prolonged, but will coincide with the recovery if the distress period is brief. The beginning of the locality recovery (trough and upturn) is marked by declining locality interest rates, recovery revenues, and rising surpluses. Once the recovery is firmly estab- lished, expenditure (including construction) will rise again, and surpluses will be adjusted downwards. This cyclic pattern of movement for these variables is not always consistent. This pattern is useful, however, for identifying and interpret- ing cycles in state and local government fiscal conditions. 4 Virtually all of the literature deals with the issue of (and uses the terms "pro- cyclical" and "counter-cyclical" from the perspective of) the effect of state and local expenditures on the economy, rather than their response to the economy. Thus there has been very little analysis of the actual cyclic position of state and local governments. Figure 6-4 and Table 6-1 show revenues for state and local governments in current dollars (REVCR) and in constant dollars (REV72). Expenditures are also shown in current dollars (XPENDCR) and in constant dollars (XPEND72). These variables all have a strong upward trend, and -- as noted earlier -- expendi- tures and revenues move closely together. Two measures of state and local government surpluses are shown in Figure 6-5 and Table 6-2. One (SURPLUS) in- cludes state and local contributions for social insurance, while the other - (NTSURPLUS) is net of these contributions. Several ratios measuring relative state and local interest rates are shown in Figure 6-6 and Table 6-3 including: state and local interest rates relative to the rates on U.S. bond issues (Sl.US): state and local interest rates relative to a composite corporate bond rate (SLCORP); municipal bond interest rates relative to Moody's AAA corporate bond rates (MUNC3A) ; and state and local government BAA bond rates relative to state and local AAA bond rates (SLBAA3A). After comparing the Dath of these variables to the path of the economy and the timing of the national business cycle, as estimated by the National Bureau of Economic Research (NBER), CEAI found that locality fiscal cycles lag the NBER cycles. A comparison of CEAI ' s estimates of local government cycles to the NBER reference cycles is shown on Table 6-4. Prior to the most recent cycle, the locality cycle lags appear surprisingly short, relative to the lag that might be suggested by identification and budgetary response lags. In these earlier cycles the locality cycle peak does not consistently lag the national peak, and the locality trough appears to occur approximately one quarter after the trough in the general economy. This result, however, may be an artifact of a strong upward trend, which results in a shorter recession with later peaks and earlier troughs than would occur if the trend were either non-existent or downward. As shown in Table 6-4, the 1973-1975 recession (after the trend in local government activity was reversed) produced a far lengthier period of locality distress. While the locality peak did occur three quarters after the NBER reference cycle peak in this case, the final trough of the cycle for local governments did not occur until the fourth quarter of 1976 -- almost two years after the trough in the cycle for the general economy. The reason for the extreme severity of the 1973-1975 cycle on local govern- ment, are not hard to discern. As noted above, the trend of increasing local revenues, tax burden, grants-in-aid, and expenditures had been dampened and in some cases even reversed. Also the 1973-1975 recession was substantially more severe than previous recessions. Finally, the well publicized financial diffi- culties of some of the nation's largest cities undoubtedly contributed to a grow- ing feeling on the part of both local government officials and citizens that fis- cal restraints on the part of local governments were more necessary tha before. The LPW program can be seen as a partial response to the unusual distress faced by state and local governments and the relatively large reductions in construction that appeared to be resulting from this distress. 103 2 "*- < i P UJ III on 7? fi t!> rv: B a! O O SNomia TABLE 6- STATE AND LOCAL REVENUE AND EXPENDITURE, BY QUARTER, (BILLIONS OF DOLLARS) REVENUE, REVENUE. EXPENDITURES EXPEND, CURRENT $ 1972 $ CURRENT S 1972 $ YEAR (RE VCR) 76.6 :REV72) 112.7 (XPENDCR) (XPEND 72) 1965.1 71.7 111.3 1965.2 74.2 114.5 73.8 113.9 1965.3 75.9 115.9 76.7 117.1 1965.4 77.7 117.9 78.3 118.8 1966.1 81.0 121.3 80.5 120.5 1966.2 83.8 123.4 82.9 122.1 1966.3 86.3 125.1 85.2 123.5 „ 1966.4 88.2 126.2 88.7 126.9 1967.1 90.1 126.7 91.5 128.7 1967.2 91.0 126.4 93.5 129.9 1967.3 94,7 129.7 95.3 130.5 1967.4 98.6 133.2 98.3 132.8 1968.1 102.4 136.0 102.6 136.3 1968.2 106.1 138.9 105.8 138.5 1968.3 108.8 140.8 107.9 139.6 1968.4 111.6 142.0 111.4 141.7 1969.1 114.3 142.9 113.6 142.0 1969.2 117.6 144.6 116.6 143.4 1969.3 121.4 147.3 118.6 143.9 1969.4 125.4 149.6 121.4 144.9 1970.1 129.9 151.6 125.6 146.6 1970.2 133.4 152.1 129.5 147.7 1970.3 137.0 153.8 134.6 151.1 1970.4 139.5 153.8 139.0 153.3 1971.1 145.5 157.1 143.9 155.4 1971.2 151.1 160.6 147.9 157.2 1971.3 154.4 162.0 150.1 157.5 1971.4 159.4 165.9 153.8 160.0 1972.1 165.5 169.2 158.9 162.5 1972.2 177.2 178.8 160.8 162. 3 1972.3 175.5 173.9 165.3 163.8 1972.4 191.5 187.4 169.7 166.0 1973.1 190.7 182.3 174.6 166.9 1973.2 192.0 180.3 178.2 167.3 1973.3 194.0 179.5 182.3 168.6 1973.4 197.2 179.3 186.9 169.9 1974.1 201.8 178.3 192.3 169.9 1974.2 208.1 178.6 199.3 171.1 1974.3 214.2 178.1 206.5 171.7 1974.4 217.3 176.1 213.1 172.7 1975.1 223.7 177.1 220.0 174.2 1975.2 231.8 180.2 227.3 176.8 1975.3 240.8 184.0 234.2 178.9 1975.4 246.4 185.7 237.5 179.0 1976.1 246.4 185.7 237.5 179.0 1976.2 253.8 188.1 240.5 178.3 1976.3 258.4 188.9 245.5 179.5 1976.4 269.0 194.1 247.9 178.9 1977.1 277.5 197.2 251.1 178.5 1977.2 281.0 196.0 253.7 176.9 1977.3 288.1 197.1 262.6 179.6 1977.4 301.6 203.6 268.7 181.4 105 1 oc UJ a: a i i i i i i i i I i i i i i i i i i i i i i i i » i i i *i i i i i i i i i i i i » i i i ;- G9 T i 8, 8 5 8 3 sNoinra 106 Year 1965.1 1965.2 1965.3 1965.4 1966.1 1966.2 1966.3 1966.4 1967.1 1967.2 1967.3 1967.4 1968.1 1968.2 1968.3 1968.4 1969.1 1969.2 1969.3 1969.4 1970.1 1970.2 1970.3 1970.4 1971.1 1971.2 1971.3 1971.4 1972.1 1972.2 1972.3 1972.4 1973.1 1973.2 1973.3 1973.4 1974.1 1974.2 1974.3 1974.4 1975.1 1975.2 1975.3 1975.4 1976.1 1976.2 1976.3 1976.4 1977.1 1977.2 1977.3 1977.4 TABLE 6-2 AND LOCAL GOVERNMENT SURPLUS (1972 DOLLARS) (BILLIONS OF DOLLARS) Surplus Including Social Surplus Net of Social Insurance Contributions Insurance Contributions 1.4 -3.7 0.6 -4.6 -1.2 -6.4 -0.9 -6.2 0.7 -4.8 1.3 -4.3 1.6 -4.3 -0.7 -6.9 -2.0 -8.2 -3.5 -9.7 -0.8 -7.7 0,4 -6.8 -0.3 -7.4 0.5 -6.8 1.2 -5.6 0.3 -6.4 0.9 -6.0 1.2 -5.8 3.4 -3.9 4.8 -2.7 5.0 -2.6 4.4 -3.2 2.7 -5.1 0.6 -7.1 1.7 -5.4 3.4 -4.3 4.4 -4.2 5.8 -2.3 6.7 -1.3 16.4 8.4 10.2 2.1 21.3 13.1 15.4 7,3 13.1 4.8 10.8 2.5 9.4 1.0 8.4 -0.3 7.6 -1.3 6.4 -2.4 3.4 -5.5 2.9 -6.0 3.5 -5.6 5.0 -4.4 6.7 -3.2 9.9 -0.3 9.4 -1.1 15.2 4.5 18.8 8.0 19.0 8.3 17.4 6.8 22.2 11.7 20.7 10.2 v> ZOT Old igg U UD •!- s- r~- -o co -(-> •i- to u XJUJ 3 C S- 13 -M C s o a. o _j vo 1/1 r^ CO CXi S- i— 13 +-> -o «£> •i- a) CO ro ro C C C\J CM o >- <_) u_ CT; C •i- c ■o o O 00 3 I— U_ r— CO 3" -Q Q- ^O- _l S- o .J2 > Q +J >> -t-> o fO >> C o -C o o _l c_> CJ oo 114 i-i n. cr CM oo lO —1 CD CO s- Ol CU O) s- 3 -M "O c O) Q., — - x e UJ o CO <£> c: i— O i— 00 lo o +-> s: LO o n -l^- <- — - -t- 1 c o o cu s- 3 — 4-> C "D -r- o C r— CU r— C\J CO Q--r- cc x s: «=<=► 3-— - C " J ■o CU a. E io ra cu +j 4- -r- LO CO CO O i— CNJ '*- o cu o J2 r- E 3 ■z: ^ u s- o .£ >^ o +j >> o >) c o =5 o o o - 1 C_> o oo 115 To examine LPW funding by class of government, CEAI updated the 1972 Census of governments data to a 1976 equivalent. LPW funding by class of locality rela- tive to 1976 construction expenditures are shown in Table 6.5. 9 LPW funding was equal to approximately 40 percent of total city construction spending, 32 percent of total county construction spending, and 25 percent of total school district construction spending. Unpublished Census Department construction data for 1976 were found to exist for 36 of the 50 localities surveyed, of which 25 were cities, 3 were counties and 8 were school districts. By dividing the LPW grants received by these locali- ties by the capital expenditures of these localities one can obtain funding esti- mates similar to those in the last column of Table 6-5.. These results are reported in Table 6-6. The data suggest that this subsample, although not an exact representation of the population data, is at least representative for cities. The limited number of observations suggests that the data for counties should be treated as indicative only. The school district data were skewed by huge non-LPW expenditures in Chicago. Without Chicago the LPW funding for school districts would have been 60 percent of construction spending. The survey data collected by CEAI indicate that localities and particularly cities often base their planning on "capital budgets" which are smaller than one would expect based upon the NIPA data. In fact, 10 percent of the localities surveyed reported that, as a general rule, they undertake no locally financed capital projects at all! The main explanation appears to be that localities often exclude "special" and federally funded projects from their anticipated capital budget. Since projects built with federal or state categorial grant funds are uncertain as to timing, and are, in any case, outside the usual priorities- setting procedure, most localities view grant funds differently from locally raised funds. In the largest jurisdictions, the receipt of federal and state funds is presumed and incorporated into the budget. However, even in these cases the "loss" of the funds is more likely to result in project delay than the trans- fer of funds from other areas. In the medium-sized and small localities, grants- in-aid for construction often are either in a separate budget or are not in the budget documents at all. LPW Expenditures and Discretionary Capital Expenditures Defining the "discretionary" portion of a capital budget is almost impossible. While in most localities, 100 percent of all capital budget funds were reported to be "earmarked", top local officials almost always retain sufficient authority to shift funds and to delay or speed up capital projects as necessary. "Need" cannot be used to distinguish between discretionary and non-discretionary projects, since in many cases, CEAI surveyors found many "critical necessary" bridges, sewage treatment plants, and other facilities which -- due to contractor difficulties, environmental restrictions, a lack of funds, or other causes -- were delayed for many years and in some cases indefinitely. 8 1972 expenditures were multiplied by the ratio of total local capital spending in 1976 to total local capital SDendinq in 1972. This process adjusts for both inflation and real growth (decline) but assumes that the relative importance of construction for the alternative classes of local governments did not change. 9 LPW funding as a percent of capital outlays (which include non-construction expenditures) was 5-8 percentage points lower for each class. 116 In an attempt to sort cut these issues, it was assumed that as a first approxi- mation that expenditures not financed from local funds were not "discretionary". Although localities do have the option of rejecting federal or state construction funds, it was assumed that, except in special circumstances, most local officials would not refuse outside funding. The most recent data available from federal sources describing the percentage of locally financed capital expenditures is for 1973. ^° At that time, federal aid comprised 25 percent of total state and local gross fixed capital formation. Gramlich 11 has estimated that this percentage had risen to 29 percent by 1976. No further breakdown by "locality class from either source is available, in 1976, however, local receipts of crants-in-aid (in- cluding state grants) were 48 percent of total expenditures, while state receipts were only 29 percent of expenditures. Thus locality reliance upon arants-in-aid for construction is likely to be significantly higher than state reliance. As part of our survey effort, CEAI gathered data on the financing of capi- tal expenditures. Data are available for 22 cities, 4 counties and 8 school districts. Based upon these data, Table 6-7 shows a comparison of LPW expendi- tures to locally financed^ 2 capital expenditures, The major difference between LPW expenditures as a percentage of total construction expenditure (Table 6-6) and LPW expenditures as a percentaae of locally financed capital budgets occurs for cities, which rely relatively upon grants-in-aid. For the 22 sampled cities for which data are available, LPW expenditure equaled almost 60 percent of the locally financed capital budget. This suqqests that the LPW proaram provides very important aid to the construction budgets of cities. Based upon the yearly variation in expenditures and the expenditures in localities where few locally financed projects are ever undertaken, CEAI has made a crude, but conservative estimate that 25 percent of all locally financed expenditures are "discretionary". Based upon this estimate, the relation of LPW ex- penditures to discretionary capital expenditures can be calculated on a locality perception basis and on NIPA basis. These results are shown in Table 6-8 and 6-9, respectively. As is evident from Table 6.8 and 6.9, LPW funding formed quite a large percentage of discretionary construction expenditures for recipient localities. While the limited data available and the difficulty of defining "dis- cretionary" make precise estimates impossible, it does seem that LPW expenditures were equivalent to about three quarters of the 1976 discretionary construction expenditures for recipient cities and approximately two-fifths of the 1976 dis- cretionary construction expenditures for counties and school districts. 10 11 Paul Schneiderman, "State & Local Government Gross Fixed Capital Formation: 1958-73; Survey of Current Business , October 1975. Edward Gramlich, "State and Local Budget. The Day After It Rained: Why is the Surplus So High" Broo kings Paper on Economic Activity 1978 Debt, local taxes and fees. O =3 1-h OCQWW _J I— CH O I^J O CD cc o 2 IE Q 1 UJOwW —I o o r^ a-* LO LO CM O O ID ,_! CO vn CM CM LO 0O vo en u +* a o U ~! -r- mO^ O !— T- ■o -M O oo +J CL-— - UJ OU_J i— cc - 3 C ZD \— f— +-> O I— •— « oo ■O 1— Q oo c£5 cr> u s_ + J CD CD 3 CJ> «/, 120 The proportion of LPW recipients who received LPW awards of more than 100 percent of their discretionary capital budgets are of particular interest. Of the 24 cities for which sufficient data are available to make this determination, 19 -- or 71 percent — received LPW awards which were greater than 100 percent of a single year's discretionary capital budgets. ' 3 Half of the counties and 50 per- cent of the school districts also received LPW grants which were in excess of 100 percent of their discretionary capital budgets. Comparison of the Type s of LPW Projects to Nor mal Capital Expenditures Work type codes supplied by EDA were used to aggregate the LPW projects into budget expenditure categories which are comparable with the Census Bureau's local government construction data. A comparison of the distribution of LPW awards by type of project and the Census data for all localities' expenditures is shown in Table 6-10. Clearly, localities were not awarded, or did not choose, projects representative of their normal capital investment patterns. The larg- est budget shares normally go to Education, Sewage, Local Utilities, and High- ways. Education's share was a third smaller among LPW projects than in the nor- mal capital budget, which represents a sharp acceleration of the prevalent down- ward trend in these types of expenditures. The share of local Utilities is over 25 percent smaller in LPW awards. Sewage's share of LPW funds is also over a third lower than in the capital budget, despite its large upward trend in local- ity expenditures. Highways, the only homogeneous major budget category to re- ceive a larger than normal budget share of funds, has a share of LPW funds well over 50 percent higher than normal. The 'Other" category, including items such as locality office buildings received a share of LPW funds nearly 2.5 times its normal budget share. Many of the same relationships are found upon examination of the comparable data for all municipalities, as shown in Table 6-11 Since the historical con- struction expenditure data are collected on a slightly different basis, it is possible to derive an important inference concerning the large relative expendi- ture in the "Other" category noted earlier. Data for municipalities, however, have a separate category for "General Public Buildings". For this category, the LPW share of funds is over four times as laroe as the normal share of the budget. This would suggest that much of the additional "Other" spending for all locali- ties is in the area of general public buildings. As was the case with all localities, the municipal budget categories that are "low- funded" by LPW (i.e., receive a lower than average share of LPW funds) are basic nuts-and-bolts services and/or categories that have relatively strong sources of finance. All, except Housina & Urban Renewal, have substantial reve- nues in the form of fees or more-or-less earmarked taxes. Housing & Urban Renew- al and Transport & Transit also receive major earmarked Federal funding. In addition, Education, Sewage, and Utilities are all such basic services that they are politically guite visible. It appears, then, that the categories deemphasized in terms of LPW funding are those that either are sufficiently important in their own right to warrant relatively reliable funding from other sources, or have major amounts of other non-local funding. T3 See Footnote 7. TABLE 6-10 COMPARISON OF CENSUS TRENDS AND LPW DISTRIBUTION BY BUDGET SHARE, LOCALITIES Three- Year 1973/74 Census Average - 1975/76 Budget Share Average Annual Change LPW Budget Share Education 23.60 - .55 15.7 Highways 12.17 - .10 19.1 Health & Hospitals 3.40 .34 3.4 Sewage 15.37 1.15 9.7 Local Parks & Recreation 4.07 - 12.6 Natural Resources 1.33 - .10 0.9 Housing & Urban Renewal 4.73 -1.10 1.0 Transport & Transit 7.43 - .70 1.5 Local Utilities 14.97 .15 11.0 Water Supply 8.63 .10 7.5 Electricity 6.17 .10 0.3 Gas SUpply N.E.C. .20 - - All Other 12.83 .85 30.3 122 TABLE 6-11 COMPARISON OF CENSUS TREND AND LPW DISTRIBUTION BY BUDGET SHARE MUNICIPALITY Three-Year Census Average 1973/74 - 1975/75 Budget Share Average Annual Change LPW Budget Share Education 8.63 -1.27 4.18 Highways 18.20 .76 21.36 (Own) Hospitals 1.88 -.06 1.81 Sewerage 22.75 2.31 16.35 Parks & Recreation 8.62 .02 14.62 Housing & Urban Renewal 9.63 -2.91 1.17 General Public Buildings 3.73 .31 15.25 Other 26.58 .86 25.30 Of the categories that are "high-funded" by LPW, two contrast sharply with the "low-funded" categories. Local Parks & Recreation facilities which are be- coming increasingly more important as urban areas grow in size but have little or*no built-in revenue sources. The high funding in the "Other" category for all localities appears to be in areas such as General Public Buildings which are use- ful but unable to generate widespread political or financial supoort. The category of Highways appears to contradict the pattern that "high-funded" categories tend to be general use categories without major constituencies or sources of funding. However, upon. closer inspection it was found that 40 percent of the LPW funds devoted to Highways are to "repair, rehabilitate, renovate, /and/ maintain" them. Another 13 percent is for a combination of this work type and a work type called "addition, extension, extend, expand". (For municipalities these percentages are slightly higher.) Thus about half of the LPW funds for High- ways, an exceptionally large fraction, are being used for upkeep. This corresponds with recent reports that road maintenance has'fallen seriously below normal "wear and tear" and suggests an additional characteristic that would lead to high LPW funding: deferabil ity. Upon inspection, this attribute certainly applies to ex- penditures for public buildings, and probably to Parks & Recreation facilities as well . The share of funds for each budget category awarded in Round I are compared to the share awarded in Round II of the LPW program in Table 6.12. This may be interesting since -- although the population of potential projects was determined by the applications -- the selection of Round I projects was made by the Economic Development Administration, while the Round II projects were selected entirely by the local jurisdictions. About half the categories receive similar shares in both Rounds. The shares of Highways and of Parks & Recreation, however, are sub- stantially higher in Round II than in Round I. By contrast, the share of Educa- tion is substantially lower, and the shares of Sewage, Utilities, and Other are slightly lower in Round II than in Round I. Except for the small decline in "Other", all the deviations of Round II from Round I are similar in direction to the deviations of total LPW shares from normal budget shares described above. Thus comparisons of LPW expenditures by type to normal expenditures, and the com- parison of Round I and Round II expenditures support the hypothesis that LPW funds were used disproportionately for types of projects which were (a) without alterna- tive major earmarked funding; (b) relatively easily deferrable and probably had been deferred on occasion; and (c) not visibly important enough to sustain active political support. This hypothesis, in turn, implies that the LPW projects were useful but not so critical that most of the projects would otherwise have been funded locally. This implication plays an important role in explaining the rela- tively low level of substitution found by CEAI.14 14 The issue of substitution is discussed in Chapter 3. 124 TABLE 6-12 COMPARISON OF ROUND I AND ROUND II BUDGET SHARES, ALL LOCALITIES Round I Education 19.1 Highways 11.0 Health & Hospitals 3.0 Sewerage 10.5 Local Parks & Recreation 10.3 Natural Resources 1.0 Housing & Urban Renewal 0.7 Transport & Transit 1.4 Utilities (All) 11.9 Water 8.5 Electric 0.1 Other 31.1 Round II 14.0 15.6 3.5 9.3 13.7 0.8 1.1 1.6 10.6 6.2 0.4 29.8 125 Impact of the LPW Program by Type of Locality The priorities for allocating LPW funds among different classes of govern- ment are obscure. '^ Nevertheless a clear reason for channeling the assistance to the construction industry through state and local governments is the need, or fiscal distress, of those jurisdictions. In analyzing the allocation of LPW funds among different types of jurisdictions, we will therefore assess the im- pact relative to the needs of different types of localities. One can distinguish between the needs of state governments and local govern- ments fairly easily. States have a degree of sovereignty which localities do not possess. While there were major fears of some local government bankruptcies the wake of the 1974 recession, few persons thought that any states (other than perhaps New York) were in serious jeopardy. Also, revealed preferences, as demon- strated by a huge increase in state grants-in-aid to localities, indicate that state legislators believe localities to have more pressing needs than the states. During the period 1966 to 1976, state grants-in-aid increased 224 percent approximately 20 percent per year. During the period 1973 through 1976 these grants increased 34 percent, or approximately 10 percent per year, despite the much higher base. The relative funding of the LPW program was clearly consis- tent with this difference in relative needs. States received only $0.5 billion of the $6.0 billion which was distributed to state and local governments as a whole. Assessing the relative needs of the various types of local governments is far more difficult. A vast amount of diversity exists between the powers, re- sponsibilities, and needs within each of the classes of local government. This diversity primarily exists between -- but also often even within -- the states. Also, it should be recognized that the states have the ability to shift powers, responsibilities, and funds between the various subdivisions in its domain. Existing data which describe local government finances provide some -- although hardly an overwhelming -- indication that municipalities were more in need of counter-cyclic assistance than other classes of local government during the period of, and immediately before, the LPW program. From the period 1972/73 to 1975/76, the outstanding debt of counties grew 31 percent while the outstanding debt of In Section 108(b) of the Local Public Works Capital Development and Investment Act of 1976 "priority and preference" is given "to public works projects of local governments." In the Public Works Employment Act of 1977 this is amended to give the same "priority and preference to any public works project reguested by a State or by a special purpose unit of local government. .. (or) by a school district. " 126 municipalities and townships grew only 23 percent. This would seem to indicate that counties were more expansion oriented than municipalities and townships.^ This tends to be corroborated by data collected from the outstanding debt and interest payments of localities (Table 6-13) which show that the interest rates paid by cities grew more than those paid by counties. This also indicates that the 1974-75 recession may have had a more serious impact upon cities than coun- ties. Nevertheless, in terms of absolute levels, the average interest rates paid by counties remained above the average rate paid by municipalities. Several other bits of information can be gleaned from Table 6-13. First, townships, which appeared to pay an interest rate penalty relative to munici- palities in 1972/73, seemed to face less cyclic pressure, and in fact achieved parity with municipalities by 1975/76. Second, special districts seemed to face the least cyclical pressure, as measured by ''interest rates. Third, school dis- tricts seemed to face considerable pressure, although less than that of munici- palities, despite a limited need for expansion. There is doubt, however, about whether locality type is the relevant dis- tinction for identifying relative needs. An analysis of local government interest rates by rating class indicates that a fairly wide (25-30 percent) disparity in interest rates" between AAA and Baa rated bonds existed until the early 1960 ' s when it dropped in half and remained at the lower level throuah 1974. After 1974 _the differential climbed sharDly back to the level of the early 1960's before receed- ing again in 1977. Apparently in 1974 investors began to perceive the relative risk of investing in lower rated local government securities as having increased substantially. To the extent that this" change in perceptions occurred, it would significantly increase the fiscal distress of localities which had allowed their bond ratinas to deteriorate. This suggests that the increased cyclic fiscal distress, rather than being a locality class phenomenon, may have been related to the past fiscal practices of individual localities, regardless of their class of government. At first one might assume that data on the relative surplus positions of the classes of local governments would provide considerable information concern- ing cyclical distress. However the ambiquity of interpretation noted earlier when discussing local governments as a whole, as well as possible aggregation problems, makes these data difficult to interpret. During the period 1972/73 to 1975/76, the net surplus position of counties fell from $3.8 billion to $1.5 billion. At the same time the surplus position of municipalities increased from a deficit of $0.5 billion to a surplus of $2.5 billion. The interest rate data presented above would seem to indicate that many municipalities felt a need to improve their fiscal situation. However, the level of the surpluses would also seem to indicate that the cyclic fiscal distress of some municipalities is quite limited. The position of school dis- tricts, perhaps as a result of strict budget requirements, consistently showed a modest surplus. The outstanding debt of school districts grew only 1 percent, presumably a reflection of the reduced demographic pressure for school expansions. 127 r— <0 Another indicator of cyclic distress is the growth in public welfare expen- ditures. For municipalities public welfare expenditures grew 32 percent while for counties, the growth was a more modest 12 percent. Nevertheless, public welfare expenses still remain more than twice as large a percentage of total expenditures for counties as for municipalities. Data on total expenditures can be useful for comparative purposes. First, the level of total expenditures provides a measure of the relative responsibili- ties of the classes of locality. Second, the change in total expenditures may provide a measure of the growth of the classes of locality. The data on locality expenditures are presented in Table 6-14. In terms of size of budget, cities would appear to be slightly smaller than school districts but approximately 50 percent larger than counties. The growth rates of the three major classes of governments are not very dissimilar, although the relatively low growth in the expenditures of municipalities may be indicative of relatively great fiscal dis- tress. The LPW funding levels by class of governments are shown in Table 6-15. Cities received more than 4 times as much money as either counties or school dis- tricts, despite the fact that expenditures by cities are only 50 percent larger than counties' expenditures and are slightly smaller than school districts' ex- penditures. In the case of school districts, this disproportionate funding may be appropriate. School district expenditures have grown rapidly despite a sharp decline in the need for new school facilities. Furthermore, as a special purpose district, the level of school district funding would seem to be a local political decision amendable to change by the voters, if desired. In fact, one could argue that the level of LPW funding for school districts, which was greater than the level of funding for counties, was relatively high. In CEAI's judgment, the information available on relative distress by class of government do indicate that counties may have received insufficient funding relative to cities. Per dollar of locality expenditure, cities received 3 times as much aid as counties. The evidence on relative locality distress does not provide a clear indication that cities as a class (as opposed to cer- tain individual cities) faced pressures that were several times as great as those faced by counties. Perhaps the clearest conclusion from this analysis is that targeting counter- cyclic funds by class of government will always be frauoht with difficulties. The degree of cyclic pressure faced by localities must inevitably be more locality specific than locality class specific. For this reason, a targeting mechanism, if one is used, should be directed to individual localities, rather than locality classes. 129 oj ro oo LD 130 13] CHAPTER 7: INTER-REGIONAL IMPACTS The geographical distribution of the impacts of government expenditure pro- grams have important policy ramifications, involving issues such as regional economic development, equity, and the reduction of differentials in the level of unemployment. These issues are particularly important for a "targetable" program such as the LPW program. In order to address these issues, CEAI analyzed the im- pacts of the LPW program on a regional level. Tnis analysis included an estima- tion of the impacts on both the state and the multi-state regional level sJ Some insights gained about local (sub-state) impacts and flows are described, but the primary means of evaluating the geographic impacts at that level is reserved for the separate microeconomic evaluation of the LPW program being undertaken by EDA. Because the incidence of LPW expenditures are known, and are subject to some degree of legislative control, our analysis has focused on inter-regional flows of expenditures. CEAI first divided the impact of LPW expenditures between con- struction value added and the materials requirements for each type of construction. The flows of materials purchases across state lines were then traced using the Commodity Transportation Survey results. These results were then aggregated into inter-state and inter-regional flows of expenditures, /* more detailed discussion of the methodology used is presented in Chapter 2. In this chapter the results are first presented and then analyzed for causes and distributional effects. The chap- ter concludes with a brief discussion of policy implications. CEAI Fi ndings Expenditure Flow and Impact Estimates The state impacts of LPW expenditures, and the inter-state expenditure flows resulting from the LPW program, are presented in Table 7-1. LPW expenditures in the 48 contiguous states are shown in the first column. 3 In the second column, CEAI's estimates of inter-state flows of expenditures are summarized as net inflows There were two basic reasons for carrying out the analysis at the state level and then aggregating to regions. First, part of the analysis consisted of developing new regional definitions on the basis of employment characteristics. (See Figure 2-3 for CEAI's definition of regions.) Since these regions differ from other definitions of regions, the presentation of regional results only would have lead to difficulties of interpretation and comparison with other work. Second, the data the CEAI judged to be most useful were organized on a state-by-state basis. Since CEAI found that labor, contractors, and materials distributors come largely from within the state -- and virtually entirely from within the region -- of the project, the flows of materials purchases by distributors from manufacturers are estimated to constitute all of the significant inter-state flows of expenditure. These figures are taken from LPW awards data. 132 TABLE 7-1 LPW EXPENDITURE, IMPACT, AND FLOWS OF EXPENDITURE BY STATE ($1 ,000) Inc idence of Net Inflow of Net Impact State LPW Expenditure Materials Expenditure of LPW Expendi tures Alabama 53,439 13,201 66,641 Arizona 81 ,038 -15,910 65,128 Arkansas 40,330 -7,635 32,695 Cal i form' a 724,583 -22,543 702,040 Colorado 43,120 -9,671 33,449 Connecticut 128,984 -2,253 126,731 Delaware 39,930 -9,180 30,750 Florida 296,287 -67,467 228,820 Georgia 101 ,909 -13,172 88,737 Idaho 40,779 -10,399 30,380 111 inois 150,983 112,310 263,293 Indiana 62,216 44,866 107,082 Iowa 41,716 -4,722 36,994 Kansas 40,452 -7,312 33,140 Kentucky 46,428 -4,813 41,616 Louisiana 60,550 -3,277 57,273 Maine 40,272 -8,895 31,378 Maryland 99,590 -20,533 79,057 Massachusetts 186,172 -31,090 155,082 Michigan 371,031 -40,390 330,642 Minnesota 47,295 -8,272 39,022 Mississippi 41,554 3,632 45,186 Missouri 52,087 5,098 57,184 Montana 42,976 -8,689 34,287 Nebraska 40,733 -8,161 32,572 Nevada 41,607 -10,374 31,233 New Hampshire 40,103 -8,849 31 ,254 New Jersey 312,953 8,926 321 ,879 New Mexico 62,766 -14,562 48,203 New York 727,375 -63,681 663,695 North Carolina 73,438 -14,978 58,460 North Dakota 39,641 -8,352 31,289 Ohio 211,653 139,707 351,361 Oklahoma 67,311 -14,611 52,700 Oregon 85,987 7,614 93,600 Pennsylvania 268,827 126,958 395,785 Rhode Island 49,003 7,324 56,327 South Carolina 43,978 -8,435 35,543 South Dakota 44,537 -10,051 34,486 Tennessee 54,499 -7,352 47,147 Texas 146,290 31,082 177,372 Utah 41,524 -9,772 31,752 Vermont 41,289 -10,358 30,931 Virginia 62,791 -7,783 55,077 Washington 118,020 -2,794 115,226 West Virginia 41,501 -8,427 33,074 Wisconsin 56,551 13,705 70,256 Wyoming 38,794 -9,664 29,129 133 of expenditures for each state. (An alternative interpretation of this column is net exports of materials.) The impact of LPW expenditures plus net inflows of materials purchases is shown for each state in the third column. 4 Table 7.2 presents the impacts of LPW expenditures, and expenditure flows resulting from the LPW program on a regional basis. The interpretation of the first three columns is the same as for Table 7-1. The last column shows the percentage increase in impact estimated to occur as a result of inter-regional flows of material expenditures. Interpretation of Expenditure Flows The states with the largest net inflows of materials expenditure are Ohio, Pennsylvania, and Illinois, followed by Indiana, Texas, Wisconsin, and Alabama. Indeed, the entire industrial belt from New Jersey to Wisconsin enjoys net in- flows of materials purchases. Rural states have the reverse pattern. The entire Midwest Region -- and most of the plains and mountain states -- have net outflows of materials expenditure. So do the states in the South (except for Alabama and Mississippi) and the northern tier of New England states. The degree of indus- trialization therefore appears to play a major role in the inter-state and inter- regional flows of materials expenditures, with industrialized areas experiencing net inflows of expenditure while rural areas experience net outflows of expenditure. There are some apparent exceptions to this pattern. The North Atlantic region -- and New York, Massachusetts, and Connecticut in particular -- are net importers of LPW materials, despite being industrial states. The same is true of -California. 6 in order to explore these cases more fully, imports and exports were calculated as a percentage of LPW obligations for these states. The states' shares of LPW funds were also compared with the states' shares of shipments of manufactures. For comparison, the same calculations were made for Illinois, Ohio, and Pennsylvania, the largest net exporting states. The results are shown in Table 7-3. A comparison of the industrialized net materials importers (California, Connecticut, Massachusetts, and New York) with the control group of net materials exporters (Illinois, Ohio, and Pennsylvania) shows that all of these states have 5 not precisely correct since the two expenditures are of different types. However, CEAI believes that this procedure provides a useful shorthand summary of the regional impacts. There is some possibility that these results were exaggerated by sample bias in the CTS data. Shipments were sampled by state of origin. In general the non- industrial states were less adequately sampled than the major industrial ones, and in some cases little or no usable data was available. Although the effects are relatively small, this tends to overstate materials exports of industrial states and understated materials exports of rural states. Michigan also had substantial net imports, despite being industrialized. In this case, however, the result is less anomalous because the state's major in- dustry -- automotive products -- is not strongly impacted by construction expen- ditures. 134 TABLE 7-2 LPW EXPENDITURE, IMPACT, AND FLOWS OF EXPENDITURE BY REGION Region North Atlantic b Great Lakes Mid-South Atlantic^ South Atlantic^ Midwest Southwest 9 North Pacific 11 Incidence of LPW Expenditure ($1000) Net Inflow of Materials Expenditure ($1000) -27,842 Net Impact of LPW Expenditure ($1000) 2,010,947 Percent Increase in Impact Due To Expenditue Flows 2,038,789 - 1.4 1,039,960 257,489 1 ,297,449 24.7 212,245 -38,401 173,845 -18.1 553,738 -67,082 486,656 -12.1 288,801 -62,983 225,817 -21.8 1,209,581 -39,305 1 ,170,276 - 3.2 201,775 -21,882 179,893 -10.8 a) (100)(Net Inflow)/( Incidence) b) Maine, New Hampshire, Vermont, Massachusetts, Connecticut, Rhode Island, New York, New Jersey, Pennsylvania, Delaware, Maryland, West Virginia, Virginia c) Kentucky, Ohio, Indiana, Michigan, Illinois, Wisconsin, Minnesota, Iowa, Missouri d) North Carolina, South Carolina, Tennessee, Arkansas e) Georgia, Florida, Alabama, Mississippi, Louisiana f) North Dakota, South Dakota, Nebraska, Kansas, Wyoming, Colorado, Utah g) Oklahoma, Texas, New Mexico, Arizona, Nevada, California, Oregon h) Montana, Idaho, Washington ■o _ o CJJ O CD c i_ ou CD 73 i-v. H a-. Q. cr CO 3 O >- •— 136 California has the lowest propensity to import, probably because of its large size and relative isolation. Connecticut and Massachusetts have the highest pro- pensities to import, which one would expect from their small size. Otherwise these states have \/ery similar import patterns. By contrast, a comparison of - ex- ports of materials with LPW obligations shows a striking difference between these two groups of states. The net importers of materials have at most only one-third as many materials exports per dollar of LPW funds as the net exporting states. Since these importing states are all relatively industrialized, the likely cause for this pattern is their receipt of a relatively high level of LPW obligations. Indeed, two of the net materials importing states (California and New York) have large absolute shares of LPW funds (over 12 percent each). All four materials importing states have shares of LPW funds that are large relative to their ship- ments of manufactures. Thus it appears that where highly industralized states are not importers of materials, the explanation for this phenomenon is large LPW grants rather than a high propensity to import. Several conclusions concerning the propensity to export materials can be stated. In general, a state or region tends to gain more from the flows of mate- rials purchases resulting from LPW expenditures: • The more industralized its economy is. The lower its own share of LPW awards is. • The more central and proximate to markets its location is. All of these factors were operative in the Great Lakes region, the only region to enjoy a positive net inflow of materials expenditures. Net Regional Distributional Effects of LPW Expenditures To gain a complete picture of the net distributional effects of LPW expendi- tures, the incidence and impacts of those expenditures must be compared with a source of funding. A region has gained (lost) from the program as a whole only if it has received a higher (lower) share than it contributed to the funding. To esti- mate this effect personal income taxes^ were used as a measure of contributions to funding of LPW expenditures. Each region's share of LPW awards, net LPW impact, and personal income taxes are shown in Table 7-4. Ratios of these variables, 7 The measure used to represent propensity to import materials is the ratio of the value of materials imports stemming from LPW expenditures to the value of LPW awards. 9 Data for corporate profit taxes are not available by state (or region). The basic underlying problem is the difficulty of making meaninaful assignments of the profits of a multi -state firm to a given state. 137 TABLE 7-4 SHARES OF AWARDS, NET IMPACT AND FINANCING OF LPW, BY REGION Share of LPW Share of LPW Impact, Including Share of Personal 3 Region Awards Net Materials Flows Income Taxes North Atlantic 36.30 36.27 31.72 Great Lakes 18.89 23.40 27.69 Mid-South Atlantic 3.86 3.14 5.19 South Atlantic 10.06 8.78 8.98 Midwest 5.26 4.07 4.13 Southwest 21.97 £l . 1 1 19.96 North Pacific 3.67 3.24 2.33 a) Source: U.S. Internal Revenue Service, Stat i stics of Income, Individual Tax Returns , 1974. 138 The Great Lakes Region and the Mid-South Atlantic Region both received substantially less than their tax "share" of LPW awards. Because of its heavy exports of materials, the Great Lakes Region made up, in terms of net impact of LPW expenditures, about half of their awards shortfall. The Mid-South Atlantic Region, however, lost still further ground as a result of materials expenditure flows. All other regions received above their tax "shares" of LPW awards. The Midwest Region and the North Pacific Region received the greatest ratio of LPW awards to contributions. This is primarily the result of a general pattern of small rural states' receiving proportionately large (although absolutely small) amounts of LPW funds. All of these regions except the North Atlantic Region incurred net losses as a result of materials expendi- ture flows. The South Atlantic Region and the Midwest Region had their net LPW impacts reduced to slightly below their tax "shares." When compared to personal income taxes, LPW expenditures do not appear to have a major regional redistributive effect. One distinction that is ocassion- ally hotly contested is that between the Northeast (North Atlantic) Region) and the "Sun Belt" (South Atlantic and Southwest Regions). These regions benefit to an equal degree in terms of LPW awards, although the sun belt's share of net impact is somewhat lower. The effect on the regional distribution of im- pacts of materials expenditure flows is not large J ° Its effect was fairly con- sistent to reduce the discrepancies in shares of benefits. The reduction in the discrepancy is particularly noticeable for the Great Lakes Reaion. Only for the two The ratio of Share of Awards to Share of Taxes and the ratio of Share of Net Impact to Share of Taxes both measure the degree of regional gain (loss) from the program. Values in excess of unity show a net gain; the larger the value, the larger is the gain. The ratio of Net Impact to Awards measures the degree of gain from the flows of materials expenditures. Again, a value greater than unity shows a net gain and the larger the value, the larger the gain. The largest net regional outflow of materials expenditures is 23 percent of LPW awards. The average net outflow, weighted by LPW awards (excluding the Great Lakes Region), is about 5 percent of LPW awards. Regional shares of LPW awards, net impact of LPW, and personal taxes are extremely highly correlated, and state shares are almost as highly correlated: Correlation Coefficients Between Awards & Awards & Taxes & Net Impact Taxes Net Impact By Region .988 .938 .978 By State .969 .900 .909 139 TABLE 7-5 COMPARISON OF AWARDS, NET IMPACT, AND FINANCING OF LPW, BY REGION REGION SHARE OF AWARDS SHARE OF TAXES 1.14 SHARE OF SHARE NET LPW IMPACT OF TAXES NET IMPACT AWARDS North Atlantic 1.14 1.00 Great Lakes .68 .85 1.24 Mid-South Atlantic .74 .60 .81 South Atlantic 1.12 .98 .87 Midwest 1.27 .97 .77 Southwest 1.10 1.05 .96 North Pacific 1.57 1.39 .89 140 fer from the regional share of personal income taxes by more than 15 percent. Local Targeting of LPW Impacts CEAI's research efforts focused on the issue of inter-regional effects', not local flows of expenditure. The field experience that was gained nevertheless produced some observations about the efficacy of targeting expenditures on small pockets of unemployment. In general it does not appear feasible to target on a geo- graphical area smaller than a city or county. The major reason for this conclusion is that LPW funds flowed through previously established local channels and through the market mechanism. Projects were bid on competitively and contractors ob- tained labor by their own means. If a contractor was unionized, the union hiring hall was used. Contractors tended to have preferred materials distributors. All these factors tend to diffuse the geographic impacts of LPW expenditures. Even when all inputs are procured locally, this typically means at least a 10 to 20 mile radius. Targeting impacts within a smaller area does not appear possible with a construction grants program unless many additional requirements are to be imposed. Policy Implications Inter-regional flows of construction and materials expenditures do exist with a program such as the LPW program. While these flows of expenditures are confined almost exclusively to materials expenditures, they are fairly substantial. Materials expenditures tend to flow from relatively non-industrail ized regions to industral ized regions. For future programs regional exports and imports of mate- rials should be considered in estimating impacts. If this is done, the implica- tion of the preceding analysis is that non-industral ized areas should be given a larger share of funds (perhaps 20-25 percent larger) than their intended ultimate share of program impact. In practice, this was done in some cases of the LPW fund- ing. Expenditure flows also tend to reduce the efficacy of targeting awards on a geographical area. The smaller the area is, the greater will be the leakage. On a regional level this leakage appears to have reduced, by about half, the degree to which the regional share of LPW grants differed from the shares of personal income taxes. At the state level, the loss of specifity in targeting is somewhat larger, although the exact amount will vary with the degree of industral ization and the geographic position of the state with respect to markets. For geographical units smaller than a city or county, the loss of the targeting effect appears to be nearly complete. 11 The Mid-South Atlantic and North Pacific Regions contain only seven states, and they received only 6.4 percent of LPW awards. Only in the case of the Mid-South Atlantic Region did inter-regional material flows increase the discrepancy between LPW awards and taxes. CHAPTER 3: OTHER POLICY ISSUES In carrying out its tasks under this project CEAI gathered information re- lating to the success of LPW in meeting policy goals other than construction- related employment creation, industry stimulus, local government assistance and geographic targeting. These other goals include counter-cyclical timing, main- tenance of construction industry wages at satisfactory levels, minority assis- tance, the creation of social infrastructure and the creation of continuing employment as a result of the increased social infrastructure. The effectiveness of the LPW program as counter-cyclical policy is a significant issue for the macroeconomic evaluation, and it was a major focus of CEAI ' s analysis. Another policy issue of considerable importance is the extent to which the LPW acted to increase the rate of inflation. The other issues, however, are properly topics to be included in the microeconomic evaluation, which is being conducted by EDA. CEAI made no systematic analysis of these issues, and the observations on these latter topics should be thought of as merely indicative. The LPW Program as Counter-Cyclical Policy LPW Program Timing The LPW program was designed to be a brief, sharp injection of funds into the construction industry. The legislation set a maximum of two quarters for implementation J and expenditure was to be completed in another five quarters. Round I funds ($2 billion) were appropriated on October 7, 1975. LPW expenditures effectively started in early 1977. Approximately 20 percent of the total $6 bil- lion in expenditures that was eventually authorized was actually spent in that year. As Round II projects commenced and many of the projects reached their most intensive stage of construction, LPW expenditures rose to a peak in mid 1978. Approximately 61 percent of all LPW expenditures are estimated to take place in 1978. During 1979, LPW expenditures are estimated to drop off steadily, with about 18 percent of expenditures estimated to take place. Only a small trickle of LPW expenditures -- about 1 percent -- is estimated to occur in 1980. The expendi- tures are shown by round in Table 2-2. The purpose of this section is to compare these expenditures with the cyclical character of the economy. This time period does not make specific allowances for applications. It was assumed that LPW projects would be readily available, so that minimal time would be required in making applications. Expenditures did, in fact, begin to be made within two quarters of the appropriation of funds. 142 Table 8-1 shows the timing of major LPW events in relation to the most recent NBER reference cycle. Time is measured in quarters, beginning with the NBER reference peak in 1973.4. The NBER reference trough occurs in the fifth quarter after this peak. The average length of the previous four expansions is used as a proxy for the next peak. z LPW events include the appropriation of each round, and the initiation, peaks, and termination of expenditure flows. The LPW proqram chronology reveals the following features: • LPW expenditures began three years after the NBER reference peak and nearly two years after the NBER reference trough. t LPW expenditures reached a peak roughly one year before economic activity is likely to peak and turn downward. • LPW expenditures will decline steadily during the period in which an economic downturn may be expected to occur. 3 Thus, rather than being counter-cyclical, the expenditure pattern of LPW program enacted in 1976 and 1977 is pro-cyclical. From the NBER reference peak to the estimated completion of LPW expenditures is a period of nearly seven years. In Table 8-2 the time requirements of the LPW program are described. Most of the time preceding the impact of LPW expendi- tures was consumed by various lags. The "identification lag" alone was two years. This lag has several components. First the fact of a recession must be confirmed, which may take several months. In 1973-74 the identification and confirmation of the recession was exceptionally complex. Next the industries which were particu- larly severely affected (e.g. construction) must be identified, and programs must be devised. Finally, sufficient political support must be generated to enact necessary legislation. All these delays are typical of discretionary counter- cyclical fiscal policy, although they may have been somewhat more prolonged than usual in the case of the LPW program. Delays, however, did not stop with the passage of the original local public works legislation. First, President Ford's veto of HR 5247 delayed the pro- gram another two quarters. Then three more quarters passed before the legislative support strengthened to the point of passing Round II. Of the thirteen quarters which elapsed between the economic downturn and appropriation of Round II, probably half of the delay was attributable solely to difficulties in making the final policy decision expeditiously. This means of estimating the next peak in economic activity is admittedly a very rough procedure, especially since the 1961-69 expansion was twice as long as the average. The resulting estimate nevertheless had great plausibility for two reasons. First, this is an estimate that is easily understood and was readily available during the debate to approve the LPW program. Second, many current economic forecasts -- including CEAI's own -- predict a strong possibility of an economic downturn in 1979. This timing coincides with the much cruder estimate of the next economic peak. CEAI's forecast, among others, is relatively optimistic for the 1980s. 143 TABLE 8-1 LPW CHRONOLOGY Quarters from Year 1 973 Peak Event 1974 1 2 3 4 1975 5 NBER Reference Trough 6 7 8 1976 9 10 11 Round I Appropriated 12 1977 13 Round I Expenditures Begin 14 Round II Appropriated 15 Round I Expenditures at Maximum 16 Round II Expenditures Begin 1978 17 18 19 20 1979 21 Average Length of Previous Four Expansionary Cycles 22 23 24 Round I Expenditures Estimated to End 1980 25 26 Overall LPW Expenditures at a Maximum; Round II Expenditures at a Maximum 27 Round II Expenditures Estimated to End 28 TABLE 8-2 TIME REQUIREMENTS OF LPW PROGRAM Program Lags Identification Lag 3 Presidential Veto° Round II Decision Time Over-run^ Total Lags 8 Quarters 2 Quarters 3 Quarters 7 Quarter^ 20 Quarters Program Duration Implementation of Program e Expansion to Maximum Expenditure Winding Down to Completion^ Total Program Duration 2 Quarters 2 Quarters 3 Quarters 7 Quarters 'Elapsed time from NBER reference peak to passage of HR 5247. 'Elapsed time from Senate sustaining of veto of HR 5247 to House override of veto of PL 94-369. c) 'Elapsed time from completion of Round I project selection to completion of Round II project selection. 'Elapsed time from legislative completion date (December 1978) to CEAI's estimated completion date (September 1980). e) 'This includes a 30-day period for promulgation of rules, a 60-day period for ruling on applications, and a 90-day period to get labor on site. 'The completion date is based on elapsed time between Round II Appropriations and the Round II stated completion date of December 31, 1978. The final factor in the long life of the LPW program is the fact that dura- tion of the expenditures is estimated to be 7 quarters longer than was forseen in the legislation. Perhaps one quarter of this is attributable to the fact that the 90-day requirement for on-site labor was met in form only for many projects. The size of projects is another major factor, even though the funded projects were, on average, quite small. The average sized LPW project required more time to com- plete than was projected in the appropriating legislation. A possible failure to recognize the degree to which construction slows during the winter months may also explain some of the delay. Finally, previous delays may have played a part in the time over-run, in that, as the economy reached a more exuberent phase of the cycle the pace of all construction tends to slow. The LPW legislation designed a concise program of expenditure. Yet three fac- tors -- the process of identifying the problem, the growth of political pressure for this type of program, and the excess of actual construction time over that envisioned in the legislation -- each added to the elapsed time an amount roughly equal to the conceived duration of the program. These delays are primarily respon- sible for the pro-cyclical nature of the expenditure. At least three years of the delays -- the presidential veto, the elapsed time between Round I and Round II, and the time over-runs -- appear to be singularly related to the LPW program itself. LPW Expenditure Timing And Employment When compared to the movement of the unemployment rate, LPW expenditures are, if anything, even less counter-cyclical than they appear in relation to the NBER reference cycles. Unemployment peaked at just over 9 percent in May 1975 (see Figure 4-1). It fell below 8 percent in January 1976 -- shortly before President Ford vetoed HR 5247 -- and it first reached 7.5 percent in May 1976 — two months before President Ford's second veto was overriden. The 7 percent floor on the un- employment rate was finally broken in August 1977 - shortly before Round II expen- ditures began. Unemployment reached 6 percent in April 1978 -- shortly before the peak of LPW expenditures -- and it has remained at that plauteau since. CEAI forecasts, however, that the unemployment rate will begin to rise -- coincident with the tapering off of LPW expenditures -- during 1979. The timing Dattern with respect to construction unemployment is much the same. The 1975 unemployment peak, however, was much higher (21 percent), the decline was much sharper (to 9.2 percent), and the construction unem- ployment rate has already risen by 1.5 percent since the peak period of LPW Three quarters of this time is an extremely small trickle of expenditure estimated to occur in 1980. In a sense, this overstates the time over-run, because the expenditure is so small. Nevertheless, the period involved is already past the likely date of a possible economic downturn. Thus shorten- ing the estimated time over-run to four quarters does not affect the nature of the relationship between the timing of the business cycle and the timing of LPW expenditure. Estimated total employment generated by the LPW program was -- at its maximum -- only about 0.1 percent of the labor force. Thus the pattern of changes in the unemployment rate was not significantly influenced by the LPW expenditures themselves. However, to the extent that the LPW program was part of a wider fiscal stimulus effort, it is fair to conclude that the impact of the total pro- gram helped to make one aspect of it -- LPW expenditures -- more pro-cyclical. 146 6 expenditures. Thus, for both the economy as a whole and for the construction in- dustry, LPW expenditures began well after the peak of unemployment and thereafter move inversely with the unemployment rate. As noted in Chapter 2, the delayed timing of LPW expenditures also contrib- uted to the dilution of the impact of LPW expenditures. Had the program taken place three to four years earlier, employment would have been over a third higher: Crowding out would have been reduced by nearly 25 percent, and at least 15 percent more employment would have been generated per dollar because of greater purchasing power and lower labor productivity. LPW Expenditure Timing and Industry Impacts The LPW program did not have a significant counter-cyclical impact on material supply industries. As was noted in Chapter 5, ; the impact on any single industry was "too little" to provide major relief for the cyclic distress of that industry and "too late" to benefit industries at the time of their cyclic distress. The date of the peak and trough of constant-dollar value of shipments is shown for each impacted materials industry in Table 8-3. This table also shows two measures of industry recovery -- the year in which shipments surpassed the previous peak, and the 1978 shipments as a percentage of the previous peak's shipments. Of the 36 industries impacted by LPW expenditures, 34 had reached their trough in 1975--over a year before LPW expenditures began. Nearly half (15) had regained their previous peak by 1976 -- before LPW expenditues began, and 75 percent had achieved that level by 1977 --before LPW expenditures peaked. In 1978, the year of peak LPW expenditures, shipments for these industries averaged 6.7 percent above the level of the previous peak.? In that year only 6 industries had not yet reached the previous peak, and they were not greatly aided by the LPW program. ° In general, LPW expenditures did no more than add very slightly to demand for these industries (for only one quarter) employment attributed to the LPW program is only about 1 percent of construction employment. Movement of the construction industry unemployment rate is therefore not substantially influenced by LPW expen- ditures themselves. The contribution of small. In only a yery few industries (Cement & Clay Products, Plumbing & Heating Equipment, Structural Metal Products, and Light & Wiring Equipment) was the impact over about 0.3 percent of production. The largest degree of impact was about 1.2 percent of shipments. For these industries, the maximum (1978) impact of LPW expenditures averaged 0.37 percent of shipments. Only in one case (Cement & Clay Products, with an LPW impact of 1.2 percent of shipments) was the impact greater than 0.5 percent of shipments. 147 C (UD- O) E E a. v) H O O ^H O r hiJi O OOCOCi ro r 30000c--IC\IOOOl-H<-lOr j CO ro o r-- oo >= <^Nmir)uiLnLntr!ir)uiifi^^cOir)Lnir)ir)iflir)ir;iOLntr)ifiir/Lntr)i/)ir)LnLnLf)'; CT^O^CT^C^CTia^CT^C7^CriCT^C7^C7^CriCnCT^C^CT>CT^Cr^ O) •<- t- 3CJ1 I— S-CL+J 00 <_> r u c s- •■- a> t- s- +j -i- i > -Q "O C QJ ( 3r-+)+JI+J 0)3 3 -a o +-> 0S-4J i S- O -r- +J O) =5 O 13 i -na.UD.aa. h- - QJ +-> c c -z. -a (J C J n- •r- O O -a c lu to s S_ -r- - E > C -r- 1 =5 oo 00 -r- 4-> -C E +-> Z5 5- IBS C-O O) 3 r0 ■-> O E C a.' O C oS U U lOU oo "2: i— i Co3-r-^ E 21 3 * E -M • -co •i- 4-> rs • oo S- . <- jz E u c +-> u t D1C W S- IC V. lu (_> s: oo i— uj |— »-■ 21 1— 2 after the recovery had been accomplished. Potential Counter-Cyclical Impact A large portion of the delay in implementing the LPW program is due to the political decision-making process rather than lags inherent in the program it* self. For example the actual appropriation of Round I funds, did not occur un- til three years after the onset of the recession. For the purposes of policy analysis, it is useful to estimate the impact of an LPW type program which started as soon after the recession as possible. It is, of course, impossible to eliminate the identification lag entirely, but is is reasonable to ask what would have happened if the LPW program had started 90 days after the earliest identification of the 1973-1975 recession. Determination of the date of earliest possible identification of the reces- sion is never easy, but it is particularly difficult for the 1973-1974 recession due to the great uncertainty among economists and businessmen as a result of the oil embargo. Nevertheless, with the quadrupling of oil prices in January, 1974, it should have been clear that a problem existed, even though the importance of the oil price increase was not generally recognized, and the coming agricultural crisis was, of course, unforeseen. 10 With 1974.1 considered as the date of earli- est possible identification of the downturn, simulation of an LPW type program 11 starting 90 days after the earliest possible identification thus begins in 1974.2. For comparison, a simulation was also made of the LPW proqram beginning in 1975.1 -- 90 days after most commentators conclusively identified a serious downturn in 1974.4. The impact of LPW expenditures on the materials supply industries may have been even later than the expenditure peak in 1978.2 suggests. CEAI did not include estimates of the lag structure of the ripple effect of derived demand as it spread through the economy. CEAI had been warning of a peak in 1973 and a recession in 1974 for several years, and in 1974 warned that "not only is the recession here, but it will stay with us for the rest of the year." By contrast Fortune's respected "Business Roundup" argued in July, 1974 that "By Comparison with the past eighteen months, the next eighteen will seem almost sunny... the growth of real GNP should continue as a 3 percent rate the rest of this year and move up toward 4 percent during 1975." As late as September, 1974, "Business Roundup" suggested that there was only a risk--" by no means probable--that the U.S. will sink into a severe recession." In the simulation the actual timing of the LPW expenditures was used, i.e. a $2 billion Round I was followed by a $4 billion Round II 3 quarters later. CEAI believes that this procedure makes adequate allowance for any uncertain- ties about the seriousness of the recession prior to the second half of 1974. 149 Two versions of the 1974.2 simulation were run. The first injected $6 bil- lion of LPW expenditure in current dollars; the second used the deflated real value equivalent of the LPW expenditures. Employment impacts for the construc- tion industry are summarized in Table 8-4, and total employment impacts are sum- marized in Table 8-5. ' 2 Employment in the current dollar simulation is over 20 percent higher than for the Deflated 1974.2 simulation. The difference between this result and the Deflated 1974.2 result (described below) reflects the decline in purchasing power of $6 billion in the intervening 11 quarters. Construction employment for the Deflated 1974.2 simulation is nearly 9 per- cent higher than that which occurred in the actual LPW program. Total employment is 13 percent greater than that which occurred in the actual LPW program. This implies increases of approximately 16 percent in the generation of non-construc- tion employment. During the first four quarters of the Deflated 1974.2 simulation, real GNP was projected to increase by 0.8 bill ion, '3 which is 15 percent more than the increase attributed to the actual LPW program. In the next four quarters the simulated increase in GNP was 2.4 billion, 10 percent greater than the effect of the actual LPW program. The total increase in real GNP as a result of the Deflated 1974.2 simulation was $3.9 billion, which is 11 percent more than the 3.5 billion increase which is estimated to have resulted from the actual LPW program. 14 Other differences between the Deflated 1974.2 simulation and the actual LPW program also exist. While the actual LPW program was estimated to have a very slight inflationary impact beginning in the third year, the Deflated 1974.2 simula- tion, if anything, tends to lower prices by a miniscule amount by reducing the de- cline in labor productivity. Similarly, although impacts on three month Treasury Bill rates are virtually identical for the two simulations, the Deflated 1974.2 simulation has a marginally, but pervasively, smaller effect on long term interest rates than does the actual LPW proaram. 12 The data in Tables 8-4 and 8-5 extend for 11 quarters only. The twelfth quarter of the 1974.2 simulations is, of course, 1977.1, which is the first quarter of actual LPW expenditure. Since the baseline for 1977.1 thus includes the actual LPW program, it was necessary to subtract actual 1977.1 expenditures from the hypothetical twelfth period expenditures of the 1974.2 simulation. Thus the twelfth period of the 1974.2 simulation cannot be compared to the twelfth period of the actual LPW program simulation. This value and subsequent values of GNP change are measured in constant 1972 dol lars . The 1974.2 simulation in current dollars produced increases in real GNP that were 30 percent higher than those resulting from the actual LPW program. TABLE 8-4 Program Quarter TOTAL (first 11 quarters) CHANGE IN CONSTRUCTION INDUSTRY EMPLOYMENT (Persons Employed) Actual LPW Program 300 1,200 1,800 3,100 4,700 7,100 6,900 6,800 6,000 4,600 2,800 45,300 $6 Billion 1974.2 Program Deflated 1$74.2 Program 300 300 1,500 1,300 2,300 2,000 3,800 3,300 5,500 4,800 8,800 7,500 8,800 7,500 9,000 7,500 8,000 6,800 6,800 5,300 4,500 3,300 59,300 49,600 TABLE 8-5 CHANGE IN TOTAL EMPLOYMENT (Persons Employed) Program Quarter Actual LPW Program 300 $6 Billion 1974.2 Program 300 Deflated 1974.2 Program 1 300 2 2,000 4,000 4,000 3 4,100 5,800 5,800 4 8,100 11,500 9,500 5 12,600 15,300 13,300 6 17,100 23,000 19,300 7 19,400 25,300 21,500 8 19,500 25,500 21,500 9 17,400 23,800 17,800 10 13,100 20,000 14,000 11 8,800 14,000 8,000 TOTAL 122,400 168,500 135,000 (first 11 quarters) 152 The major reason that the estimated impact of the Deflated 1974.2 simula- tion is greater than that of the actual LPW program is that the earlier simula- tion occurs at a more propitious time of the business cycle. Because the expen- diture started during the recession and wound down when the economy recovered, crowding out and other adverse effects such as inflation and interest rate impacts were minimized or prevented entirely. Furthermore, under the 1974.2 simulation employment and GNP impacts occurred when they were needed most. This simulation produced truly counter-cyclical results. The 1975.1 simulation was run with a deflated expenditure equivalent to the actual $6 billion LPW expenditure. In this simulation, however, it was assumed that the total expenditure was made at one time rather than in two rounds. In terms of total impact upon real GNP and employment, the results of this simula- tion were quite similar to the Deflated 1974.2 simulation. Thus it appears that the benefits of counter-cyclical ity could have been achieved if the program had been initiated in early 1975, when the recession was universally acknowledged. This result is hardly surprising. The 1974 recession was particulary long and unusually severe. Virtually all of the direct stimulus of the single phase 1975.1 program occurs in 1975 and 1976, well before the factor markets began to show the first evidences of tight conditions. Policy Implications of LPW Program Timing The timing of the LPW program was such that it was not an effective counter- cyclical policy tool. The program began after the recovery was well under way, achieved its peak impact at a time of relatively exhuberant economic activity, and will find down during a time when there is danqer of an economic downturn. Time is critical for counter-cyclical expenditures. Yet, prior to its en- actment, debate over LPW consumed well over a year, and perhaps as much as two. This delay was critical in terms of the counter-cyclical efficacy of the program. It would seem to further the goals of discretionary counter-cyclical expenditure programs if their proponets either carried the day immediately and overwhelmingly or lost the fight entirely. Unfortunately the political process, which requires that the problem becomes generally obvious, seriously complicates all federal efforts for effective counter-cyclical policy. Nevertheless, the CEAI simulations do indicate that if action is taken promptly its effectiveness could have been en- hanced. A list of the post war recessions and an indication of their duration is given in Table 8-6. Although the 1973-1975 recession was the longest of the post war recessions, the difference in duration is surprisingly small. Since the earlier reces- sion were not obfuscated by the oil embargo it is likely that the shorter duration of the The purpose of these simulations is to assess the effect of the LPW program in the absence of delays due to political decision-making. The two-round program simu- lated in 1974.2 seems appropriately to reflect some degree of uncertainty that existed during 1974 (see footnote 10). By the end of 1974, however, the extent of the recession and the need for the LPW program were clearly known. Thus abstrac- tion from political decision-making implies implementing the entire $6 billion program in 1975.1 . 153 TABLE 8-6 DURATION OF POST-WAR RECESSIONS AND RECOVERIES TO PEAK Cycle Durati on 9 July 1953 - March 1955 21 months February 1957 - February 1959 25 months January 1960 - November 1961 23 months October 1966 - October 1967 (growth recession ') 12 months September 1969 - January 1972 29 months November 1973 - May 1976 a ^ 31 months reattainment of the peak level of production. ^A local p May 1976. recessions would be offset by an earlier identification. Thus the possibility of an effective counter-cyclical works program exists, although unless the political action is exceptionally prompt this possibility is relatively small. This does not necessarily support a standby public works program, since other issues must be considered, including: (a) the possibility of increased "project delay" sub- stitution; (b) the planned size of the program; (c) the possibility of a reces- sion which impacts construction mildly; (d) the size and type of projects and (e) the possibility of false indications of a recession triggering the program must first be considered. The design and implementation of the LPW program also contributed the delays. The time limits on steps of implementation appear to have worked reasonably well in terms of relapsed time, although there was considerable slippage in getting labor on site. The multiplicity of goals and selection criteria, however, were complex and led to results that were not wholely satisfactory, adding to the pressures for Round II. The greatest timing "shortcoming of the actual program was that effective control over the termination of expenditures was lost. Thus the program is continuing well beyond the expected period, and the impact is correspondingly delayed. Ideally the program would wind down in the strongest part of the recovery. Davis-Bacon Reguirements The Davis-Bacon reguirement is designed to remove the incentive that con- tractors might otherwise have to "import" low cost labor from other regions. The reguirement does, of course, raise the costs of LPW projects. CEAI included guestions on this effect in its survey of LPW contractors. The results are shown in Table 8-7. Based on CEAI's sample Davis-Bacon reguirements raised wages for slightly under half the LPW contractors. The tendency for Davis-Bacon reguire- ments to raise wages was markedly higher among non-union contractors than union contractors. The survey also produced some evidence that Davis-Bacon reguirements (and/ or high wages) affect the skill mix of labor. In response to a guestion on the effect of Davis-Bacon requirements on hiring practices, two union contractors responded that, since they had to pay hiaher wages anyway, they took care to employ highly skilled workers. Three non-union contractors noted (with pride) that they had skilled employees and paid them well. This tends to suggest that Davis-Bacon reguirements raise the skill mix of LPW work forces. 155 TABLE 8-7 SAMPLE OF DAVIS-BACON EFFECTS a) Union Status of Respondent Effect on Wages Unionized .Non-un ionized Total Raise Wages 3 8 11 No Effect 9 7 16 Total 12 15 27 Higher waqes reduce the employment that can be aenerated per dollar. Skilled labor reduces to quantity of labor required for a given project. Both effects tend to lessen the employment impact of the LPW program. CEAI's employ- ment estimates, which used average labor requirements, took no account of Davis- Bacon requirements. Since these requirements, if they have any effect, appear to reduce the employment impact, this effect re-enforces CEAI's judgment that our employment estimates have not tended to understate the LPW program's employ- ment impact. The Value of LPW Projects Ideally, the LPW program would be composed of those projects for which localities thought the need was pressing but not sufficiently pressing to be worth financing locally. If higher priority projects were selected, the problem of substitution would arise; if lower priority, projects were chosen, the social welfare would not be maximized. Attempts to explore the question of the usefulness of public works projects often lead to an endless discussion as to whether or not there is a shortage of public infrastructure. A person's opinion on whether or not there is a shortage of public infrastructure clearly depends upon the person's own particular prefer- ence function. Until recently it was often argued that the lack of a market sys- tem for public goods and/or competition between localities to attract businesses produced a less than optimal level of public infrastructure J 6 Indeed, the appar- ent lack of local government resources for "essential" projects forms a major link in the justification of the program as expressed in Section 201 of the 1976 act. However, recent political actions, including the success of Proposition 13 in California would seem to indicate that the view that public services are under- provided is by no means unanimous. One can avoid becoming embroiled in this argument by noting that the LPW program could easily have increased the country's welfare even if there were no projects whose value exceeded their cost. The reason for this apparent paradox is simply that the benefit of the achievement of other social goals including employment creation must be added to the benefit of the project, be- fore comparing the project's benefit to its cost. If this procedure were not followed, one might take the absurd position of rejecting useful construction projects because their direct benefits were slightly lower than their costs while deciding to finance totally unproductive employment programs to aid the unemployed. CEAI's survey indicates clearly that in most cases LPW funds were used for projects which were generally desirable but of less than the highest priority. This finding supports our conclusion that substitution was only 20 percent of total fundinq as well as the view that LPW projects were not "make work" opera- tions. Many cases existed in which it seemed unlikely that local funds could be raised for the purpose. Many of the projects (such as new municipal buildinqs) C.F. John Kenneth Galbraith, The Affluent Society (Hauqhton, Mifflin, Bost, 1958) 157 involved issues of direct concern to locality employees while many others (such as construction activities designed to increase the efficiency of local govern- ment operations or to provide for recreational services) were without broad ac- tive constiuencies. The types or projects included renovations of existing buildinqs to reduce maintenance expenses, the establishment of retail centers, road construction and repair, the construction of municipal facilities of all types, and recreational facilities. In virtually ail cases in which substitution did not appear to be a factor, the projects appeared useful but it is unknown whether or not the projects were cost effective. Indeed, CEAI found little evi- dence of the use of cost/benefit analysis for projects of the rather little size of the LPW projects on the local level. The Creation of Continuing Employment During the debate on whether or not to establish the LPW program, program proponents argued that public works programs cause continuing employment gains.17 For example, after the construction of a sewerage treatment plant workers are needed to run and maintain the plant. The fallacy of this argument is that, even if additional workers are hired, the increase in employment may be offset by cut- backs elsewhere. If the increase in employment is not offset, taxes may be in- creased and private demand reduced. Although this issue is clearly beyond the main focus of this project, CEAI surveyors did ask localities if they expected the projects to result in continuing operational, maintenance, or program expenses (savings). In general, this appeared to be an issue to which localities had devoted relatively little thought perhaps because for the types of projects financed by LPW the net changes were expected to be small. From the responses which were received, it appears that for approxi- mately one-third of the projects a small increase in maintenance costs were ex- pected. For another one-third only insignificant changes were expected. For approximately one-quarter of the projects a small decrease in maintenance and opera- tion costs was expected. In most cases where the changes were judged small, no continuing changes in employment were expected. However occasionally the addition (elimination) of a maintenance position or the addition of a program director or elimination of an elevator operator position was expected. For the remaining proj- ects (less than one-tenth of all projects) significant increases in program costs were expected as a result of the LPW projects. In these cases, substantial employ- ment gains were also expected. For example in Passaic County, New Jersey, a major hospital renovation was undertaken which permitted an expansion of staff. However in most cases in which major increases in operation and maintenance costs were ex- pected, the locality officials reported that the project (or a less expensive sub- stitute) would have been undertaken even without the LPW program. In addition to the cases of complete substitution there were many cases in which a less expensive substitute would have been built in the fairly near future. Thus, there were few cases in which continuing employment gains could be attributed to the LPW program, even without considering the possibility of offsetting reductions elsewhere in locality budgets. 17 Congressional Record", July 21, 1976 P. SI 21 21 -SI 21 29. 158 Inflationary Impac t Fears that the LPW program would be inflationary ' 8 appear to have been exagger- ated. The net impact of the LPW program on the consumer price index was essentially zero. Nonresidential construction prices did rise somewhat. However, even here the net rise in prices was only about 0.1 percent. Furthermore, the impropitious timing of the LPW project contributed to its inflationary impact. Had the LPW program com- menced 90 days after the earliest identification of the downturn, rather than when it actually occurred, the program would have had no inflationary impact in terms of consumer, producer or construction prices. The increase in inflation resulting from the higher demand would be completely offset by the reduction in prices resulting from a less severe recession. 18 "Congressional Record", July 21, 1976 P. SI 2128-S12129. 159 CHAPTER 9 DATA DEFICIENCIES AND SUGGESTIONS FOR FUTURE RESEARCH General Data Deficiencies The Impacts of Federal Grants-in-A id A paucity of data exists describing the impact on local economic activity of federal grants-in-aid to state and local governments. While several studies designed to trace flows of specific types of federal expenditures through the economy have been commissioned, no comprehensive data exist indicating the ex- tent to which particular types of state and local expenditures are financed through federal funds. Therefore, the extent to which local construction ac- tivity is influenced by federal (not to mention state) closed end matching grant programs is unknown. Specific areas in which data would be useful in- clude: (a) quarterly estimates of federal grants by type of federal program, (b) estimates of the proportion of local government spending which is financed through matching and closed-end grants -- by type of grant and cumulatively, (c) estimates of the amount of state and local construction expenditures which are financed through non-categorical funds, such as revenue sharing and (d) ideally, estimates of federal (and state) contributions to local government expenditures by type of locality. Local Economic Conditions Perhaps the most shocking gap in our understanding of the U.S. economy is in the area of "small area" economics. Of course, the vast number of small areas makes data collection difficult and expensive. However, at some point a decision must be made as to whether or not the detailed "targeting" of federal pro- grams is worth this cost. If such targeting is deemed appropriate, it is neces- sary to have accurate federal estimates of local economic activity, income, over- all employment, unemployment by type, and locality fiscal health (measured, for example, in terms of the ability to float bonds at the best available rates). This requires the systematic establishment of a mutually exclusive and exhaustive categorization of all the areas of the country based upon standardized criteria, including natural economic boundaries and political boundaries and jurisdictions. The Congress and the Executive Branch should agree upon the boundaries and data needs so that the data can best fulfill the needs of both branches of the govern- ment. The present maze of differences -- both in geographical coverage and po- litical authority -- between the small area categories is not used for federal qrant targeting. Thus, even the limited data available cannot be used effectively. 160 C onstruction Data In 1978, construction will represent almost 9 percent of Gross National Product. Nevertheless, the data available on construction activity are among the poorest for any segment of our economy. Data on construction labor are particularly poor since the data are presently gathered on an establishment basis. Workers in particular skill categories are classified on the basis of the type of firm that employs them, rather than on the basis of their skill category. This leads to spurious reported changes in employment by skill category as workers change employers. Furthermore, since many firms consider themselves "general contractors", there is, in many cases, absolutely no way to estimate changes in employment by skill categories. Another major problem with construction data is the woeful inadequacy of construction price deflators. The primary difficulty with these data is that the indices are based primarily on input costs and take no account of changes in productivity nor of the varying amounts of value added in construction of different types. Another major deficiency in the construction price data is that virtually all of the Commerce Department's composite construction indices depend upon private sector sources. These private sector sources often do not put the same high level of effort into data collection that we have come to expect from U.S. government agencies. The data gathered are often unrepresenta- tive of types of construction. Furthermore, the construction index data, in- stead of viewing buildings in terms of their generic purpose, concentrate upon the types of construction materials which are used to construct the building. Another major weakness in construction data is that the detailed data on the stock of, and new additions to, commercial and industrial buildings are collected by a private company, F.W. Dodge, rather than by the federal govern- ment. Since much of these data are proprietary, access to the data is severely restricted through the collection of fees which are far greater than the costs of distributing the data. Estimates of the amount of employment generated by different types of con- struction activity are collected by the Bureau of Labor Statistics (BLS). How- ever, the collection of these data is extremely sporadic. Part of the problem is incomplete coverage. Only a dozen construction categories are available in updated form. Among the major omissions are commercial and industrial buildings and renovation and repair activities. Other problems arise with the vintage of the data. Although BLS updates these data for productivity and price changes, the only primary information on the amount of employment generated by specific types of construction activity are, in many cases, 10 to 15 years old. Estimates of the use of materials in construction (and thus of indirect employment) must be based upon a 1967 input/output table. This input/output table is, of course, now 10 years old and cannot be expected to reflect present day materials use -- in either the manufacturing or construction industries -- with complete accuracy. 161 Some private sources, including CEAI, have endeavored to estimate the change in the input/output coefficients which have occurred during the Dast 10 years. These estimates, however, are based upon fragmentary data and do not substitute for a more timely, exhaustive study of the inter-industry relationships in our economy . LPW - Specific Data Deficiencies Much of our analysis of the LPW program was hindered by a lack of informa- tion concerning the actual flows of LPW funds. The two major aspects of this problem were that (a) the LPW program is still continuing, and thus much infor- mation concerning the flows of funds is at present unknowable; and (b) much of the data concerning LPW project applications, their disposition, and LPW dis- bursements were either unavailable, contained errors, or lagged actual events by several months. The explanation for these problems clearly lies in the vast amount of data which must be gathered and .processed in a program designed to provide $6 billion in relatively small grants. It must be recognized that suf- ficient resources simply were not available to check and clean the data as much as might be desirable from the point of view of a macroeconomic analysis. Suggestions for Fut u re Work The Congress wisely decreed that a study of the economic impacts of the LPW program be undertaken so that, should the economy require another dose of counter- cyclical medicine in the future, an analysis of the value of public works funding would be available. To a large extent this evaluation answers many of the crucial issues regarding the macroeconomic impacts of LPW programs. Additional work could certainly be undertaken in terms of: (a) additional analysis of the timing of the spending impacts; (b) additional analysis of the exact employment creation impacts; (c) an expansion of the survey sample to include more localities and more projects; and (d) a more detailed investigation in some of the areas only briefly touched upon by the study. Moreover, an analysis conducted in 1980, after the program is complete, would be able to add precision to many areas in which CEAI was forced to provide estimates. 162 Perhaps the most important aspect of evaluation of the LPW program which remains to be undertaken is to compare a.nd contrast the various types of counter- cyclical programs which were adopted in the wake of the 1973-1975 recession. This type of effort would integrate the results of studies of individual programs that have been undertaken by various arms of the federal government since-1973. Such an integration would be extremely useful in that it would identify the ex- tent to which these programs are supplementary, complementary, overlapping, and inconsistent. It would also highlight the relative strengths of the different program designs in attaining individual policy goals. This work would greatly facilitate efforts to prepare comprehensive contingency plans for recession of alternative types and magnitudes. Another issue which needs to be explored in much greater detail is that of project delay substitution, i.e. the extent to which local governments delay fundinq various activities because they believe that federal funds will be avail- able in the near, or perhaps not so near, future. Delays could occur in response to anticipation of a specific program -- such as counter-cyclical public works -- or simply in the general hope of future funding. They can occur whether or not the project involved actually receives federal funding. Localities that signifi- cantly change their spending patterns on this basis are practicing an insidious form of grantsmanship, since locality effort for worthwhile projects is reduced as the localities await federal funds. If a counter-cyclic program is expected in the near future, the impact of the delay may peak just when the need for addi- tional locality expenditures is the greatest. The potential impact of project delay substitution affects not only the LPW program, but also many tyoes of federal aid and even tax policy. (In fact, general purpose construction arant programs may cause significantly less amounts of project substitution than an anticipated special purpose construction program.) The Local Public Works program is a fairly unique type of countercyclical grant program. An inherent problem in its analysis is that only one observation of the impact of this type of program exists. Inevitably, "unusual" circumstan- ces -- which were particularly severe in the wake of the 1973-1975 recession -- have acted to make it difficult to analyze the ceterus parabus impacts of the programs. Given the severity of the recession upon local governments, the change in the secular trend away from local government spending in general (and construc- tion activity in particular), the crisis in some of the largest cities of our country, and the usually large amount of inflation experienced during this period, these results must inevitably be more tentative than if additional observations of an LPW-type program were available. It is almost axiomatic in economics that ob- servations of a similar program -- either in the U.S. or elsewhere -- and an analy- sis contrasting the LPW program with these other programs, would provide additional information upon, and clarification of, the economic impacts. 163 Despite the inevitable data gaps and limit to the exploration of specific areas, CEAI believes that this study has answered many of the broad questions dealing with: (a) the speed at which countercyclical construction grant funds can be utilized; (b) the magnitude and types of the employment impacts; (c) the amount of substitution which can be expected; (d) the level and direction of inter-regional transfers; and (e) the feasibility and usefulness of targeting construction grant funds. Although further work would, of course, be useful, CEAI believes that in most cases, the answers to these questions contained in this study are within the range of permission necessary to provide a useful guide to the Congress and the Executive Branch in formulating future policy. 164 PROJECT TYPE AND CONSTRUCTION CATEGORY CLASSIFICATIONS 165 TABLE A-l DEFINITIONS OF ACTIVITY CODES AND STRUCTURE CODES* Type of Activity 1 CONSTRUCT, BUILD, CREATE, INSTALL (NEW CONSTRUCTION) 2 ADDITION, EXTENSION, EXTEND, EXPAND (NEW CONSTRUCTION) 3 REPAIR, REHABILITATE, RENOVATE 4 DEMOLISH, REMOVE (BUILDING OR FACILITY) 5 COMBINATION OF CODES 1 AND 4 (SECOND AND THIRD DIGITS) WILL REFER TO THE TYPE OF STRUCTURE BEING BUILT 6 COMBINATION OF CODES 2 AND 3 9 MISCELLANEOUS CODES Type of Structure or Facility 01 POLICE STATIONS 02 FIRE AND/OR RESCUE STATIONS 03 JAILS, PRISONS, DETENTION FACILITIES 04 MUNICIPAL OFFICE BUILDINGS, INCLUDING TOWN HALLS, COUNTY COURTHOUSES, MULTIFUNCTION MUNICIPAL OFFICE BUILDINGS 05 HOSPITALS, CLINICS, NURSING HOMES, HEALTH CENTERS, SANITARIA 06 ARENAS, STADIUMS, BLEACHERS, PAVILIONS 07 AUDITORIUMS, THEATERS 08 GYMNASIUM, SWIMMING POOL, OTHER RECREATIONAL BUILDINGS AND STRUCTURES 09 COMMUNITY CENTER, SOCIAL SERVICE CENTER 10 SCHOOL AND OTHER LEARNING OR TRAINING FACILITY 11 LIBRARY 12 MUSEUM, CULTURAL CENTER, SCIENCE CENTER, ART GALLERY 13 TERMINAL BUILDINGS (AIR, WATER, RAIL) 14 GARAGES, PARKING STRUCTURE (NOT PARKING LOT, SEE #48) 15 FACTORIES, CANNERIES, PROCESSING PLANT 16 SHELL INDUSTRIAL BUILDINGS, WAREHOUSES, MARKETING FACILITY 17 PORT FACILITIES, HARBOR DEVELOPMENT 18 AQUACULTURE FACILITY, FISH HATCHERY 19 ELECTRIC POWER PLANT, GENERATING FACILITY 20 DWELLING UNITS, HOUSES, APARTMENTS 21 HISTORIC BUILDINGS, REGARDLESS OF TYPE 29 MISCELLANEOUS STRUCTURES AND BUILDINGS NOT ELSEWHERE CLASSIFIED 30 DAMS, LEVEES, DIKES, FLOOD CONTROL STRUCTURES 31 WATER LINES *An EDA Work-Type Code consists of three digits. The first is an activity code. The last two are a structure code. 32 WATER SYSTEMS (LINES PLUS WELL, INTAKE, OR RESERVOIR AND/OR TREATMENT FACILITY, PUMPING STATION, ETC.) 33 WATER SOURCE DEVELOPMENT (RESERVOIR, WELL, INTAKE STRUCTURE) 34 WATER TREATMENT FACILITY (POTABLE) 35 SEWER LINES, MAIN, TRUNKS 36 SEWER SYSTEM (LINES PLUS OUTFALL, PUMPING STATION(S) AND/OR TREATMENT FACILITY) 37 SEWAGE TREATMENT PLANT, WASTEWATER TREATMENT FACILITY 38 COMBINED WATER/SEWER PROJECT 39 STORM DRAINS, EXCLUDING DRAINAGE DITCHES 40 DRAINAGE DITCHES, CANALS, AQUADUCTS 41 STREETS, ROADS, HIGHWAYS (MAY INCLUDE SIDEWALKS, CURBS, AND/OR GUTTERS) 42 SIDEWALKS, CURBS, AND/OR GUTTERS 43 COMBINED WATER/SEWER AND/OR HIGHWAY/STREET/ROAD AND/OR SIDEWALKS, CURBS, GUTTERS 44 SITE DEVELOPMENT (GRADING, CLEARING, ETC., PLUS INSTALLATION OF MULTIPLE UTILITIES) 45 FENCES 46 BRIDGES 47 RAILROADS, TROLLEYS, STREETCARS (INSTALLATION OF LINE, RAILS, ETC.) 48 PARKING LOTS 59 MULTIPLE UTILITY-TYPE FACILITIES, OR UTILITY- TYPE PROJECTS NOT ELSEWHERE CLASSIFIED Miscellaneous Codes 989 RIGHT-OF-WAY CLEARANCE, GRUBBING, CLEARING, LANDSCAPING, PLANTING, RIVER, STREAM, CHANNEL, SHORELINE, OR DRAINAGE DITCH CLEARANCE 990 PARK DEVELOPMENT (CITY PARK, COUNTY PARK, ETC.) (PARKS OR RECREATIONAL FACILITIES INVOLVING MAJOR STRUCTURES SHOULD BE CODED UNDER #8) 991 INDIAN ACTION TEAMS 999 MISCELLANEOUS PROJECTS NOT ELSEWHERE CLASSIFIED 167 TABLE A-2 CLASSIFICATION OF WORK-TYPE CODES INTO BLS CONSTRUCTION CATEGORIES Construction Category Private Multi -Family Housing Private Single-Family Housing General Hospitals Elementary and Secondary Schools Federal ly-Aided Highways Sewer Lines Sewer Plants Civil Land Works Public Housing Federal Office Buildings Utilities Structure Type Codes 20 b , 59-99 b , 6, 7, 8, 9, 10, 11, 50-99 u 41, 42, 43 b , 45, 46, 47, 48, 59-99 b 31, 32, 33, 35, 36, 38, 39, 43 b , 59-99 b 34, 37 17, 18, 30, 40, 44, 59-99 b , d 1, 2, 3, 4, 12, 29, 59-99 u 13, 14, 15, 16, 19, 59-99 b a) Last two digits of work-type code refer to Table A-l b) Structure type has been allocated to more than 1-0 category on a project-by- project basis. c) Includes all projects with work-type codes of 221 or 621. d) Includes all projects with activity codes of 4. e) Includes all projects with work-type codes of 121, 321, or 521. 168 TABLE A- 3 ADJUSTMENTS OF BLS JOBS PER MILLION $ ESTIMATE (Construction Industry Jobs per $ Million of Contract Cost) Construction 1977 Category Estimate Percentage Change 1976-1977 1978 Estimate 1979 Estimate 1980 Estimate Private Multi- Family Housing 16.5 -13.2 14.3 12.4 10.8 Private Single- Family Housing 14.8 -11.9 13.0 11.5 10.1 General Hospitals 19.3 - 7.2 17.9 16.6 15.4 Elementary & Secondary Schools 16.0 -23.1 12.3 9.5 7.3 Federally-Aided Highways 18.0 -10.0 16.2 14.6 13.1 Sewer Lines 13.5 - 8.2 12.4 11.4 10.4 Civil Land Works 14.1 - 7.8 13.0 12.0 11.1 Public Housing 20.4 -12.8 17.8 15.5 13.5 Federal Office Buildings 17.5 - 7.9 16.1 14.8 13.7 Utilities 11.9 - 8.0 10.9 10.1 9.3 169 TABLE A-4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 CLASSIFICATION OF WORK-TYPE CODES INTO INFORUM CATEGORIES INFORUM Industry Corresponding Structure Codes Single-Family Dwellings 20* Multi- Family Dwellings 20*, 50-99* Mobile Homes 20*, 50-99* Residential Additions 20*, 21 (With Activity Code of 2 or 6) Hotels, Motels Industrial 15, 50-99* Office 4, 50-99* Stores, Garages 14, 16, 50-99* Religious Educational 9, 10, 11, 50-99* Hospitals 5 Other Buildings 1, 2, 3, 6, 7, 8, 12, 13, 21, 29, 50-99* Farm Construction Oil , Gas Railroad 47, 50-99* Telephone, Telegraph Electric Utilities 19, 50-99* Pipeline and Gas Utilities 50-99* Misc Non-Buildings 50-99* Highways 41, 42, 43*, 45, 46, 48, 50-99* Military Conservation and Development 17, 18, 30, 40, 44, 50-99* Sewer Systems 35, 36, 37, 38*, 39, 43*, 50-99* Water Systems 31, 32, 33, 34, 38*, 43*, 50-99* Residential Renovation/Alteration 20*, 50-99* Combined Misc Non-Build/Highways 50-99* Combined Highways/Sewers 43* Combined Highways/Water 43* Combined Sewers/Water 38*, 43* Combined Highways/Sewers/Water 43* Combined Other Build/Misc Non-Build 50-99* Combined Other Buildings/Highways 50-99* Demolition (Activity Code of 4) ♦Indicates Structure Code has been split by hand to several Input-Output Code categories : —Structure Code 20 to Input-Output Codes 1, 2, 3, 4, and 25 --Structure Code 38 to Input-Output Codes 23, 24, and 29 --Structure Code 43 to Input-Output Codes 20, 23, 24, 27, 28, 29 --Structure Codes 50 through 99 tc Input-Output Codes 2, 3, 6, 7, 8, 10, 12, 15, 17, 18, 19, 20, 22, 23, 24, 25, 26, 31, 32 170 APPENDIX SELECTED SURVEY INFORMATION 171 TABLE B - 1 SELECTED SURVEY QUESTIONS Project-Specific Questions Pertaining to Substitution What were your plans with respect to the (name) project if the LPW application had been rejected? If you would have undertaken the (name) project anyway, when would the project have been started? (This year, next year, etc.) When would it have been finished? Would it have been fully funded? How would you have funded the (name) project? (Tax increase, bond issue, transfer of funds from another capital project or other spending categories, capital or operating budget?) Why was application made for LPW funds in the case of the (name) project? Have you ever tried to fund the (name) project before? By what method? (bonding, etc. ) Questions on Rejected Projects What are your plans with respect to this project now that your application has been rejected? If you plan to undertake this project, how will you fund it? (Capital or operating budget) Would it have been fully funded? If you plan to undertake this project, when do you plan to begin (have you begun) work? Why was the application for LPW funds made in the case of this particular project? Have you ever tried to fund this project before? By what method (bonding, etc.)? Questions on Local Government Participation How much money are you contributing to the (nam e) project? How much money is your parent government contributing to the [name) project? How would these funds have otherwise been spent? What expenses (operational, maintenance, or program - positive or negative) do you expect will be required as a result of the (name) project? Questions on Bonds When was the last bond issue in your jurisdiction? What type of bond issue was this? What was the fate of the last bond issue submitted to the voters? Passed; Failed; Year? If you needed money for a capital project, how do you think the community would feel about raising new taxes, floating a bond issue, etc? Are there any institu- tional reasons for not raising taxes or floating a bond? (Specify) 173 TABLE B-2 DJ s tributions of Population and Survey Sampl e Number in Percent of Percent of , Universe -' Criterion Range or Category North Atlantic Sample 19 Sample Region 38% 37:13 Great Lakes 10 20 19.5 Middle South Atlantic 2 4 3.7 South Atlantic 5 10 10.2 Midwest 2 4 4.9 North Pacific 2 4 3.2 Southwest 10 20 21.3 Total 50 100% 100.0? Jurisdiction Type County 6 12-2 15.1% City 36 72 69.0 School District P, 16 15.8 Total 50 1002 100.02 Population 0-50,000 26 522 52.1$ 11.4 W 20<2 b/c/ 50,000-100,000 7 14 100,000-500,000 10 20 Over 500,000 7 14 Total 50 1 00°' 100.0% LPW Funds $0-200,000 11 222 37. M*! 31.8 ^ 16.9 % 13.5 4/ $200,000-5500,000 10 20 $500, 000-$l ,000,000 12 24 Over $1,000,000 17 34 Total 50 loo'r 100.02 SMSA Yes 35 702 e/ No 15 30 e/ Total 50 100% 174 a) Fraction of all LPW funds (obligations) received by jurisdictions in this category, except as otherwise noted. b) Based on jurisdictions in the awards data base for wihch population data exist. c) Inciudes New York City. If this one jurisdiction's share of LPW funds is removed from the population class, the percent of the universe is approximately 14%. d) Percent of jurisdictions, unweighted by LPW obligations. These four obligation classes were assigned weights of 1, 1, 2, and 4, respectively. This brings the adjusted percentages of the classes to 23.6, 20.2, 21.5, and 34.3, respectively. e) The proportions of two-thirds SMSA and one-third non-SMSA, was requested in order to over-sample urban areas. Components may not add to totals due to rounding. 175 APPENDIX C SUBSTITUTION CROSS-TAB RESULTS 176 ° o| in 5 5| 4* O Ol LD ej cr> -J «* 177 i— CM i— r—\ C\J Ol r- LT) , v c: fO o <_> ■'- , 1 1 - - w Tl o +J fO O 00 Irt o 180 c o > CO 0\ «* <— o °182 LO r— SI _j c *e Z3 r— CTjQ •i- o "O O <— O r- O r- 18^ o mj !2l APPENDIX D: CHASE INFORUM RESULTS 186 SUBMddflS "IVIM1W 187 SbBIlddflS 1VIM31VW saanddns iviaaivw 189 s«3iiddns ivibBivw 190 SaSIlddflS lVIM3iVW 191 S«3HddnS 1VIM31VW 192 APPENDIX E: SIMULATION OF TWO "NORMAL" RECOVERY PATTERNS 193 Examination of the "normal" impact of business cylces on construction related industries is complicated by the existence of federal country-cyclical policy. Moreover, every cycle will contain some random evnets, which cause the recovery to differ from a "normal" cycle. Thus CEAI has simulated two recessions, usinq its Macroeconomic Model and INPRO Model, to show the effects of a moderate and a severe business cycle on these industries. These two simulations are based on the actual historical cycles of 1970 and 1974, respectively, except that no federal counter-cyclic fiscal and monetary policies were assumed to have occurred, and no changes to the behavioral relationships in the model were made to reflect differences between actual events and the models' simulation of the time periods. The two simulations began in 1961.1 and 1974.1, respectively. The changes in actual federal policy that were entered into the models are summarized in Table E-l . The impact of these changes was to increase the severity of the recessions and slow the path of the recoveries. The simulated recovery patterns for the economy in both simulations are described in Table E-2. The simulated changes in the industrial production indices of selected con- struction related industries are summarized in Table E-3 and Table E-4. 194 TABLE E-l CHANGES FROM ACTUAL HISTORY Policy Simulations Revenues 1970 1974 The Phase-out of the income tax surcharge in 1970 was revised as were the Phase II (1971) personal tax cuts. The investment Tax Credit was not re- instituted. No changes in 1974. In 1975 and 1976 the tax reduction act passed in March was eliminated. Expenditures No changes were made as substantial cuts in pur- chases of goods and ser- vices were offset by above average growth in transfer payments. A large increase in social security benefits seemed to be related more to in- flation than the slow real growth. In 1975 and 1976 trans- fer payments to persons were reduced by approx- imately $4.5 billion. Grants-in-aid were re- duced by an amount which grew steadily from $5 billion to $18 billion, with 50% of the change being reflected in state and local expenditures. In 1976, Federal direct purchases were also re- duced by $3 billion. Monetary Policy Assumed steady 4% growth in the non-borrowed mone- tary base less currency. Assumed steady 7.5% growth in the non-borrowed mone- tary base less currency. - 01 ^s &s r-. U0 UD 00 00 CO LD 00 «* CO co lo CT> ' «3"' C\j .-H CM St O 1—1 £ CD +-> s-s &s UD O &S &S =3 "vT O CT> r-H «3" CI rH O ld r-~ rH ' ^}-' CO t— 1 t— 1 LD Cr CO S- 0J CD VD tO &S s-s Q. cr> en «=t <-< CNJ CO O CD 3 CO* CO CM .-< CO* «3" CO CM CO O cy c CD IX) 01 E cr 1 1— 1 O ■st |J- OJ E S- 0) 00 &s &? O <~D S-S S^ =3 cr ^r So LD 1^. CM CX> 1—1 >=t l" r-I 1" CVJ co" r-J ^ CM +-> c CD E +J CD > C c p _ c Q. 2: +->^ c: c CD c c CD C C C C c 00 c c rs O O O O CD S- TD "O fC O fd +-> +-> a: •4-> +-> S- to rd fO ro O- CD O 3 3 3 3 E Z3 13 3 3 S- Q 13 3 E E E E CD E E =) E E E E E C +-> CD L0 00 >> >i to 00 c S- f— c O ^ fO -Q +-> fC ^r f0 O St O <3- O -=3- O «3- r^ r»» r^. r^ Q. r^ r^ r^ 1^. -O to r^ r^. Z5 O ct> en CT> CT> E cr> en cr> cr\ CD C en cr> -O h- c CD CL X O c C •r- C_> > <_> ID O U_ 196 r^ lo o CMvooOO^unoooroLn rocMOOfor^^or—cnoo co ro ro co CO o O ^- O CM UD hs co o en lt> r-- o Ifi W kD lO r-^ r» co en cm en "^J- en r». ro lo ^h en oo o co co ud ld i— i Ln *d" en CM CM 00 00 t— I CO CM CM 1^ CM CM <£> CO O <^3 "* CO CO N ID un cm cm cm o en r-~ r^. m rH o o ^- <3- o Q. i— OJ CD CD CD CD T3 4-> 4-> a jd n vi i— c i/> _i zn ca a. a_ o -i- CD CD -i- CD i— W U U < CO O 'i- 197 lO CO N Kl (7i n ' cti r-. lo oo o~> lo ^j-ojLn^HovDeor-- 1 -^ cr> cr> 01 w in w oo ^- o-i o o oo IX) CT> LO .— I o^u->c\jor-^c\ji-nc\j'^" o r^. o o cm o <£> r-» CM CD VO CD 'X) CM CO O O^t-C^r--.iX)00^OCM00CTi , -D fr 5 .-n en O O O r^ oo o oo .-I «3" cn en g r— cu -a cu cu a> aj -a CL JD -Q 00 i— <_> _i __i « r- (U +> a. a_ o oo ct z 00 jo ft U. S. GOVERNMENT PRINTING OFFICE : 1981 340-997/360 PENN STATE UNIVERSITY LIBRARIES A00007JH4S1T4