College and Research Libraries Research Notes The Economics of Economics Journals: A Statistical Analysis of Pricing Practices by Publishers H. Craig Petersen This study uses multiple regression analysis to investigate price determinants of the top-ranked economics journals. Holding other factors constant, the study found that the prices charged to libraries in the United States are significantly higher for journals from for-profit publishers and for those originating in Europe. The estimated price differential for European journals is too large to be attributed entirely to the extra cost of shipping the periodicals to the United States. Another finding was that there is a positive and significant correlation between a journal's impact (measured by frequency of citations) and its price. The results suggest that journal prices are not always cost based. One implica- tion is that, as the main buyers in the market, libraries should not passively acquiesce to all price increases. Rather, working through their associations, libraries should require publishers of journals whose prices appear to be exces- sive to justify their pricing policies. • ide variation occurs in the prices of economics journals. Often, these differences cannot be ex- plained by publishing costs alone. For example, the 1990 subscription price to libraries in the United States for the Journal of Econometrics was $575, while the rate for the Journal of Economic Issues was only $40. But the two journals have about the same number of pages per year and are generally similar in appearance. Clearly, factors other than cost must determine the relative prices of such publications. This paper reports the results of a statistical analysis of pricing practices by the publishers of economics journals. It is similar to a previous study by the author, but incorporates more sophisticated methodology, a better-defined sample of journals, and additional explanatory variables.1 The objective of the analysis is to determine the noncost factors that af- fect prices of the leading economics jour- nals. Specific hypotheses to be tested are that (1) prices of periodicals from for- profit publishers are higher than those of other publishers, (2) European journals have higher prices than those of publications origi- nating in the United States, and (3) prices of more prestigious journals are higher. H. Craig Petersen is a Professor in the DeJXlrhnent of Economics at Utah State Universih;, Logan, Utah 84322. A Journal of Economic Literature article by S. J. Liebowitz and J.P. Palmer eval- uated the relative impact of economics journals by using the number of times articles from a particular journal were cited in other journals. These counts were adjusted by the size of each pub- lication and were used to formulate a ranking of economics journals.2 There- sults reported here are based on that rank- ing. Starting with the top 100 journals as determined by Liebowitz and Palmer, nine journals were deleted (e.g., Yale Law Re- view) from this study because they orig- inated in another discipline. Another 10 had to be eliminated because of missing data. The remaining 81 journals from the Liebowitz and Palmer ranking consti- tute the sample used here to investigate the determinants of journal prices. The price data are 1990 one-year sub- scription prices charged to libraries in the United States. 3 Table 1 shows rela- tionships between mean prices and jour- nal characteristics. Note that prices are higher for journals from the for-profit publishers and those originating in Great Britain and Europe. Some plausible rea- sons exist for these differentials. The com- Economics of Economics Journals 177 mercia! publishers can be assumed to be maximizing profit, while institutional and societal publishers tend to have other goals, such as disseminating knowledge. Thus, it is not surprising that commercial suppliers, facing similar costs, would charge higher prices. Higher prices in the United States for journals published abroad could reflect several factors. The most likely is additional distribution costs of shipping journals to the United States. Another explanation for the observed price differentials is that the demand for some journals is more inelastic because those periodicals are considered to be more useful or more prestigious. If such journals are not randomly distributed by nation of origin or type of publisher, this factor may explain the price differences shown in table 1. Note also the circulation data in table 1. Four journals have less than 1,000 sub- scribers. At the other extreme is the American Economic Review with a circula- tion of 26,000. The table suggests that prices are inversely related to journal circulation. This is consistent with the existence of economies of scale in pub- lishing and distribution. TABLEt PRICE CHARACTERSTICS OF ECONOMICS JOURNALS Mean One-Year Characteristic No. 1990 Library Price Total sample 81 $107.00 Type of publisher Commercial 25 217.40 Academic society or other 56 57.71 Nonprofit institution Country of origin United States or Canada 50 62.38 Great Britain 16 113.94 Europe 10 343.80 Other countries 5 57.40 Circulation (number of copies) 1,000 or less 4 203.50 1,001-2,000 30 i48.63 2,001-5,000 35 73.63 More than 5 000 12 68.08 178 College & Research Libraries March 1992 PRICE= ao + atFREQ + a2LENGTH + a3CIRC + a4ADV + asiMPACT + a6PROFIT + a7GB + asEUR + a9 OTHER + u with variables defined as follows: PRICE = 1990 one-year subscription price to libraries. FREQ = Number of issues per year. LENGTH = Number of pages per issue. CIRC = Circulation. Number of copies per issue. ADV =Advertising found in the journal. Yes= I, No=O. IMPACT= Impact or usefulness of journal. Citations per character as calculated by Liebowitz and Palmer. PROFIT= Journal published by a for-profit firm . Yes= I, No=O. GB =Journal published in Great Britain. Yes= I, No=O. EUR =Journal published in Europe. Yes= I, No=O. OTHER = Journal not published in United States, Canada, Great Britain, or Europe. Yes=l, No=O. u= Randomly distributed error term. FIGURE1 A general problem with the data in table 1 is that they mask interactions be- tween the factors that affect journal prices. For example, although the com- mercial publications have higher prices, the data do not prove differential pricing by publisher type. It could be that the journals from for-profit publishers are longer, more expensive to produce, or are perceived as more prestigious than those from societies and other nonprofit institutions. Sorting out the effects of individual determinants of journal prices requires statistical analysis. Of the many available techniques, the investigators chose multiple regression because, in addition to separating the effects of different vari- ables, it also provides quantitative esti- mates of the magnitudes of those effects. The regression model used is described in the next section. MODEL The factors that affect journal pricing can be divided into four categories- cost, systematic noncost, demand, and random. Among the most important of the cost factors are number of pages per issue, number of issues per year, circula- tion, whether advertising is accepted (advertising generates revenue that can offset production costs), and distribu- tion costs. Possible systematic noncost factors already noted are type of pub- lisher and nation of origin. Demand-re- lated factors affecting journal prices include the prestige or perceived value of the publication. Any remaining varia- tion in prices is assumed to reflect ran- dom influences. This study estimated that journals published in Europe cost about $168 more than those originating from the United States or Canada. Based on the previous discussion, the following equation is suggested to ex- plain variations in the price charged to libraries for economics journals. This model should be viewed as a reduced form equation THAT incorporates the net effects of cost, systematic noncost, and demand-related influences on prices. The coefficient of PROFIT estimates price differentials for journals provided by comrilercial publishers relative to those from societies and other nonprofit institutions. The coefficients of GB, EUR, and OTHER estimate prices of journals published in Great Britain, Europe, and other countries in comparison to prices for journals from the United States and Canada. Economics of Economics Journals 179 PRICE=-19.6644+ 17.7148 FREQ + 0.1154 LENGTH- 0.0051 CIRC (3.92) .... (0.89) (-2.35) •• -11.4022 ADV + 54.3712 PROFIT+ 0.6562 IMPACT (1.68). (-0.71) (2.63) ••• + 23.5423 GB + (1.19) 167.6590 EUR + (5.26)··· 5.3231 OTHER (0.17) R = 0.739 n= 81 • Significant at 0.05: one-tail test. •• Significant at 0.05. ••• Significant at 0.01. FIGURE2 The coefficient of CIRC captures scale economy effects. An inverse relationship is predicted. That is, journals with larger circu- lations are likely to have lower production costs and, consequently, lower prices. 4 With respect to the impact variable, journals perceived to be more useful should be able to charge higher prices. The number of Social Science Citation Index listings calculated by Liebowitz and Palmer is used as the index of impact or usefulness. Hence, it is expected that the estimated coefficient ofiMPACfwould be a positive number.5 REGRESSION RESULTS Coefficients of the regression equation for library prices were estimated using ordinary least squares and data from the 81 journals. Results are provided below with t-statistics in parentheses. The statis- tically significant coefficients are desig- nated with asterisks. A one-tail test was used in one case because the sign of the coefficient could be predicted from economic theory. Approximately 74 per- cent of total variation in journal prices is explained by the independent variables of the model. The coefficients of the cost variables are consistent with prior expectations. Num- ber of issues each year is significantly and positively related to price. Prices increase as the number of pages per issue increase. Prices and the presence of advertis- ing are inversely correlated. However, the coefficients of LENGTH and ADV (adver- tising) are not statistically significant. Note that the regression analysis indi- cates a statistically significant inverse re- lationship between price and journal circulation. This negative coefficient re- flects economies of scale in journal pub- lishing. The size of the coefficient implies that for each 1,000 increase in circulation, the price declines by approximately $5. The demand-related effect of journal usefulness on prices was estimated by the coefficient of IMPACT. The coeffi- cient was positive (as predicted) and statis- tically significant based on a one-tail test. That is, more frequently cited journals were found to have higher prices. Journals from forprofit publishers are significantly more expensive. With respect to systematic noncost fac- tors, those journals published in Great Britain, Europe, and other (non-U.S.) na- tions were determined to command higher prices than those originating in the United States and Canada. However, the coefficient is only statistically signif- icant for European journals. For those observations, the differential is quite large. Holding other factors constant, this study estimated that journals published in Europe cost about $168 more than those originat- ing from the United States or Canada. Some differential would be expected because of distribution costs for European journals shipped to the United States. The real question is whether the differentials are greater than could be attributed to these costs. The estimated coefficient for Great Britain provides a frame of refer- ence. The mean price of British journals is about $24 more than that of U.S. jour- nals. Assume that this value approxi- 180 College & Research Libraries mates the additional costs of shipping journals across the Atlantic Ocean. Note that the price premium charged for European journals is about seven times that amount. Some of the additional differential could reflect higher production costs in Europe, but it is unlikely that costs would be seven times greater than those in Great Britain. Apparently, European publishers have a different pricing policy than those of other journal suppliers. For type of publisher, the coefficient of PROFIT is positive and statistically sig- nificant. With other factors held constant, journals from the for-profit publishers are estimated to cost about $53 more than those from academic societies and other nonprofit institutions. This conclusion is consistent with prior expectations. CONCLUSIONS AND IMPLICATIONS The results of this study provide some preliminary findings on pricing of economics journals. Noncost factors can have a significant impact on prices. In particular, statistically significant price differentials exist for journals from for- profit publishers and for those originat- ing in Europe. The European premium is too large to be attributed solely to addi- tional costs of production and distribu- tion. More frequently cited economics journals command higher prices. Rapidly increasing journal prices and budgetary restrictions have caused librar- ies to severely limit purchases of new jour- nals and to cancel subscriptions for journals already in the library's collec- tion. As this practice continues, it will reduce access of faculty and graduate students to new ideas in their fields. Over time, the result will be diminished research and teaching productivity. The finding of this research, that non- cost factors are significant determinants of journal prices, suggests that rapid price increases are not inevitable, and that the groups affected should not pas- March 1992 sively acquiesce. The question is: Which groups have the incentives and in- fluence to facilitate change? One possi- bility is the scholars who use journals for their research and submit articles for publication. If prices of certain journals become too high, scholars could use their professional associations to estab- lish other, less expensive publications. The problem is that scholars have little incentive to do so. They have access to library subscriptions of journals at no cost and can usually purchase in- dividual subscriptions at rates signifi- cantly less than the library price. Cancellation of library subscriptions may not only deprive the scholar and the graduate student of access to library sub- scriptions but also to personal subscrip- tions which are sometimes contingent on an institutional subscription. Also, scholars are evaluated not only on how frequently they publish but also on the prestige of the journals in which their work appears. Articles in new, little-known journals do not contribute as much to professional advancement as do articles in the top journals. Con- sequently, it is unlikely that scholars will abandon such journals for less expen- sive, but less prestigious substitutes. Societies can establish new journals and invest them with excellent editors and boards, but the process is a lengthy one. Another group that could influence journal pricing is library associations. Because their budgets are affected, li- braries have a strong incentive to oppose price increases. Because they represent the largest purchasers in the market, these associations may be able to exert their economic influence in dealing with journal suppliers. At the very least, pub- lishers whose prices are significantly higher than charges for comparable pub- lications should be asked to justify their pricing practices by providing detailed information on costs. REFERENCES AND NOTES 1. See H. C. Petersen, "Variations in Journal Prices: A Statistical Analysis," The Serials Librarian 17:1-10 (1989). 2. SeeS. J. Liebowitz and J.P. Palmer, "Assessing the Relative Impacts of Economics Economics of Economics Journals 181 Journals," The Journal of Economic Literature 22:77-88 (Mar. 1984). Data used in this analysis are taken from table 2, col. 2., p.85. The rankings are based on citations to articles in a particular journal by articles from other journals. These data were obtained from E. Garfield, ed., Social Science Citation Index (Philadelphia: Institute for Scientific Information). Data were adjusted for journal length by dividing the total number of citations by number of characters in the journal. 3. Library prices as reported in Faxon Inc., Faxon Librarian's Guide to Serials:1990 (West- wood, Mass.: Faxon, 1990). 4. The inclusion of circulation data is a unique contribution of this study. This information is available for most U.S. journals, but can be very difficult to obtain for foreign journals, especially those from commercial publishers. 5. The citation data were normalized by assigning the journal with the most citations per character (the Journal of Political Economy) a value of 100. For all other publications, the data represent the percent of citations per character compared to the most frequently cited journal. OCLC/AMIGOS Collection Analysis Systems Make a wise investment. Choose from three options to analyze your library's data: Collection Analysis CD compares quantitative data BCL3 Tape Match measures against a standard Tape Analysis fits individual specifications Available exclusively from AMIGOS Bibliographic Council, Inc. 12200 Park Central Drive, Suite 500 Dallas, Texas 75251 214/851-8000 or 800/843-8482