I a 5-1414 '45 7 September 1984 Economic Impacts of Changes in United States Rice Price Variability on Market Efficiency, Q Marketing Margin and Producer Viability The Texas Agricultural Experiment Station Neville P. Clarke, Director The Texas A&M University System College Station, Texas [Blank Page in Orifinai t?‘ Economic Impacts of Changes in United States Rice Price Variability 0n Market Efficiency, Marketing Margin, and Producer Viability W. R. GRANT, agricultural economist National Economics Division, Economic Research Service U.S. Department of Agriculture I. W. RICHARDSON, associate professor The Texas Agricultural Experiment Station (Department of Agricultural Economics) B. W. BRORSEN, assistant professor Department of Agricultural Economics Purdue University M. E. RISTER, assistant professor The Texas Agricultural Experiment Station (Department of Agricultural Economics) M‘ W‘ This research was funded jointly by the United States Department of Agriculture, the Texas Agricultural Experiment Station (Projects 6337, 3694, and 6507), the Texas Rice Research Foundation (Econo-Rice Project), and the Department of Agricultural Economics, Texas A&M University. CONTENTS Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Market Efficiency Methods . . . . . . . . . . . . . . . . . . . . . 4 Market Efficiency Results . . . . . . . . . . . . . . . . . . . . . . 4 Impact of Price Variability 0n Marketing Margins Methods . . . . . . . . . . . . . . . . . . 4 Impact of Price Variability on Marketing Margins Results . . . . . . . . . . . . . . . . . . . 5 Producer’s Viability Methods . . . . . . . . . . . . . . . . . . . . 6 Producer’s Viability Results . . . . . . . . . . . . . . . . . . . . . 8 Base Comparison of Policies . . . . . . . . . . . . . . . . . . 8 Comparison of Lower Variable Costs . . . . . . . . . . . l0 Changing Market Alternatives . . . . . . . . . . . . . . . . . ll Combination of Marketing Alternatives/ Cost of Production Changes . . . . . . . . . . . . . . . . . . . l2 Reverting to l960’s Policy . . . . . . . . . . . . . . . . . . . . . l2 Limitations and Conclusions . . . . . . . . . . . . . . . . . . . . . l2 Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 SUMMARY This study investigates the impacts on the Texas rice industry of the shift from stable economic conditions and a government policy of supply control in the 1960s to the more variable economic conditions and a market- oriented farm policy of the 1970s. Markets were found to adjust quicker to changes in information under the market-oriented policies of the 19703. The increase in risk due to increased price varzaaility significantly in- creases marketing margins. The shift to the conditions of the l970’s decreases a Texas rice producers chance of survival. Keywords: Market efficiency, marketing margins, producer viability, producer survival, producer success, policy changes, rice. 2 INTRODUCTION p Agricultural legislation applicable t0 rice dates from f the early 1930’s with the enactment of the original Agricultural Adjustment Act (Public Law 10, 73rd Con- gress) of 1933 (Holder and Grant 1979). The basic ag- ricultural legislation currently aflaecting the industry had its origin in the Agricultural Adjustment Act of 1938. The 1938 Act endeavored to stabilize supplies and prices through acreage adjustments, government loans, and marketing quotas. The first loan activities or government purchases occurred in 1948. Acreage allotments and marketing quotas were instituted in 1955. The industry operated under a price support-acreage allotment- marketing quota program through 1975. With the pas- sage by Congress of the Rice Production Act of 1975, the rice supply no longer was controlled through marketing quotas and allotments. Instead, the rice program changed to a market-oriented program with deficiency payments based on the diiIerence between a weighted August-December farm price and the target price for the 1976 and subsequent crops. U.S. government programs stabilized domestic prices from 1955 through 1971 (Fig. 1). The world rice situation from 1972 to 1974 was characterized by re- duced exportable supplies and increased import de- mand; United States prices climbed, in 1973, to the highest price ever recorded. The sharp rise in prices triggered a suspension of domestic marketing quotas for DOL/CWT 32.5 — 30.0 - 27.5 —l 25.0 — 22.5 - 20.0 - 17.5 - THO-DU 15.0 — 12.5 - Texas Mlll A‘ 1 \ , -__/ ___- \ 5 0 -"' ------ ---_v"__\' ' _ ~ - - - I - ' ' _ ~~ . “ F‘*-"’\€___ the 1974 and 1975 crops and opened the way for expan- sion of U.S. rice acreage. The shift to target price programs in 1976, however, placed emphasis on defi- ciency payments as a means of support to producers. Had the farm program in eiIect during the 1960s con- tinued to the present, U.S. rice prices would have restabilized after the 1972-74 rise in world prices. Prices supported at 65% of parity would have been above the farm price every year since 1975 (except for 1977). Instead, farm prices have remained unstable, generally below the relative level of support of the 1960’s. This research evaluates the effect of the shift in farm policy and economic environment on: (a) market eiIicien- cy, (b) marketing margins, and (c) producer viability. It investigates the short-run interactions among rice prices by analyzing dynamic adjustments in the markets under two policy situations—in 1960-71 for high price sup- ports, and 1974-81 for low price supports. The margin between farm and mill prices and the variability of farm and mill prices are examined for each period. 1 The Texas rice producefs survivability is then evaluated for the two policy periods using the Firm Level Income Tax and Farm Policy Simulator (F LIPSIM V) (Richardson and Nixon 1982). 1A continuous retail rice price series for both periods is not available. Thus, margins between mill and retail prices are not analyzed. YEAR Figure 1. Season average rice prices, 1955-82. Data Source: USDA-ERS, Rice Outlook and Situation MARKET EFFICIENCY METHODS Sporleder and Chavas (1979) define price efficiency in terms of how quickly and accurately changing supply and demand conditions are reflected in prices, i.e., the shorter the adjustment period, the more efficient the market. Fama (1970) suggests tests of market efficiency can be carried out based on a division of all information into three sets: (l) a strong form test which encompasses all information, including that possessed by insiders; (2) a semistrong form test which includes all publicly availa- ble information; and (3) a weak form test which includes only the information available in historical prices. This study is concerned with market efficiency only in the weak form sense. Prices in different locations may not be competitive- ly determined. If this is the case, prices may be deter- mined by a dominant or barometric firm. In such a price leadership model, prices in other locations change only after prices have changed in the dominating location, resulting in a lag in price adjustments. The markets would, therefore, be inefficient in the weak form sense. The data used in this analysis consist of weekly milled rice prices, quoted on Monday for each week, in Arkansas, California, Louisiana, and Texas covering the periods from October 1966 to September 1971, and October 1974 to September 1981 (USDA, Rice Market News). Ranges of these weekly milled rice prices typi- cally do not exceed 5O cents per hundredweight. The transition period between October 1971 and September 1974 was omitted so the structural changes could fully impact on the markets. The methods used for the market efficiency tests are discussed in the Appendix. The long-run multipliers and their standard errors (Schmidt 1973; Dhrymes 1973) were estimated to test market efficiency? As a measure of the speed of adjust- ment, the number of time periods needed for the inter- mediate-run multipliers to stabilize within 5% of the long-run multipliers were calculated. MARKET EFFICIENCY RESULTS During the early period, Louisiana prices had a significant impact on prices in all four locations; howev- er, no milled rice price in other locations had a signifi- cant impact on Louisiana prices (Table 1). This indicates that Louisiana mills were price leaders during 1966-71. Louisiana was the major producer during this period and was also a major shipping point for PL480 rice. Thus, its dominant position is not surprising. The analysis indi- cates that only the long-run multipliers measuring the impacts of Louisiana prices on the other three markets are significant (Table 2). Cross-price adjustments were relatively slow, indicating all markets were inefficient in reflecting Louisiana prices. When the cross-price multi- plier was significant, it took from 16 to 23 weeks for the full impact to occur. All markets were inefficient in the later period in the sense that price adjustments were not instantaneous 2Long-run multipliers measure the cumulative impact of one variable upon another variable. A standard error is a measurement of the dispersion (spread) of values. 4 TABLE 1. ESTIMATION OF INSTANTANEOUS PRICE ADJUSTMENTS" Independent Variables Price Series Louisiana California Arkansas Texas 1966-71 Louisiana 3.23* .32 .55 1.02 (.002) (.96) (.82) (0.82) California 698* .86 f,3.00* .96 (.0001) (.55) '3 (.004) (0.47) Arkansas 13.50* 2.89* 1.93 2.43* (.0001) (.005) (.06) (.02) Texas 3.54* 1.32 .69 2.62* (.0008) (.24) ( 70) (.01) 1974-81 Louisiana .65 2.40 3.70* 2.44 (.59) (.07) (.01) (.06) California 1.37 .40 5.42* .19 (.25) (.76) (.001) (.90) Arkansas 3.45* .36 3.24* .72 (.02) (.78) (.02) (.54) Texas 3.15* 9.73* 8.91* 13.76* (.02) (.0001) (.0001) (.0001) “Probabilities of a greater F-value are in brackets. The null hypothesis is that the coefficients for all lagged values for a given variable in the equation for the given price series are zero. *= significantly different from zero at the 5% level. (Table 1). With the markets now volatile in price fluctua- tion, however, the adjustment period for new informa- tion in one market to transfer to another ranged from 11 to 13 weeks, much shorter than with the earlier period (Table 2). The changes occurring in government pro- grams in the 1970’s also resulted in major shifts in regional production patterns (Grant 1982). These re- gional shifts plus the effects of price destabilization impacted the market pricing structure. Arkansas became the leading rice-producing state. Mill prices in Arkansas moved into a dominant price leadership role alongside Louisiana. The impact of Louisiana mill prices diminish- ed, but they still had a long-run impact on mill prices in all other states (Table 2). Changes in Texas mill prices lagged behind all other mill prices. IMPACT OF PRICE VARIABILITY ON MARKETING MARGINS METHODS If mills are risk averse, increases in price variability should result in increased margins. The price support- acreage control program during 1960-71 stabilized U.S. prices to the extent that the rice industry experienced very little risk from price variability (Fig. 1). Taxpayers assumed this risk through the costs of the government program. As prices rose sharply in 1973, however, price variability increased. The suspension of marketing quotas in 1974, the change to a target price program in s; ill 1976, and higher rice exports in the 1970’s all forcej, both millers and producers to contend with increase price variability (Grant, Holder, and Ericksen 1980). A discussion of the theory and methods used in analyzing TABLE 2. LONG-RUN MULTIPLIERS OF MILLED RICE PRICES“ “Price Multipliersb Adjustment Period‘ (Weeks) Series LA CA AR TX LA CA AR TX 1966-71 Louisiana 1.24 .28 .28 — .31 13 22 19 16 (.18) (.27) (.29) (.29) California .47* .95 .28 — .11 23 12 22 26 (.15) (.19) (.20) (.20) Arkansas .66* — .22 .82 — .09 2O 3O 19 28 (.15) (.19) (.20) (.20) Texas .53* — .16 .23 .45 22 25 2O 12 (.20) (.26) (.30) (.30) 1974-81 m Louisiana 1.22 .40* .78* .19 9 12 13 12 (.12) (.13) (.13) (.10) California .40* 1.10 .70* .07 13 8 13 16 (.12) (.13) (.13) (.09) Arkansas .49* .31* 1.65* .18 13 13 11 12 (.13) (.15) (.15) (.10) Texas .48* .57* 1.03* .73* 13 13 13 6 (.13) (.15) (.15) (.10) “Long-run multipliers are dynamic multipliers that measure the total impact of one variable upon another variable. “Standard errors in parentheses. * = significantly different from 1 (0) at the 5% level for own (cross) multipliers. ‘Weeks required for the intermediate-run multipliers to stabilize within 5% of the long-run multiplier. the impact of increased price variability on marketing margins is given in the Appendix. Unweighted season average prices and monthly prices received by Texas farmers (USDA, Agricultural Prices) and by Texas mills (USDA, Rice Market News) for 1960 to 1981 were used in evaluating price variabili- ty. A continuous monthly series is not available for retail prices. The missing data in the monthly prices received by Texas farmers (September 1976 to July 1979) were estimated by regressing the Texas price on the monthly prices received by U.S. farmers (January 1960 to july 1982) Identification of the shortening of the adjustment period during the late 1970’s is significant. The shorter adjustment period indicates the market-oriented pro- grams in effect during the late 1970’s result in less inefficient rice markets than the stable price support programs in the 1960s Markets were inefficient in the weak form sense during both periods, however, since both had multipliers significant at the 5% level. Annual Texas mill prices (adjusted to a rough rice equivalence by a factor of 0.71) (USDA, Rice Situation and Outlook) and annual Texas farm prices were used to calculate the margins. The supply of marketing services was estimated by regressing the margins against quantity and the shiftersof the supply of marketing services. llAnnual quantity milled in Texas was used as the quantity moving through the marketing channel (USDA, Rice Situation and Outlook). This quantity is assumed to be , xogenously determined. The coefficient of variation “(standard error expressed as a percent of the mean, or average) of monthly Texas mill prices within each mar- keting year was used to represent price variability. Mills usually maintain short-term inventories (one to two months) and should, therefore, be influenced by short- term price variation. Milling costs were used to repre- sent the other shifters of the supply of marketing ser- vices. Data on milling costs were available only for six years during 1960-81 (Holder and Grant 1979). The available data were regressed against the statistic “Total Marketing Bill” (USDA, Agricultural Statistics) and mill costs were then estimated from this equation. In 1975, the elasticity3 of rice production with re- spect to farm price was 0.35 (i.e., a 1% change in farm price shifts production 0.35% in the same direction) and the elasticity of demand with respect to mill price was -0.83 (i.e., a 1% change in mill price shifts demand 0.83% in the opposite direction) (Grant and Leath 1979). These are aggregate demand and supply elasticitiesfor the United States. They should also represent demand and supply in Texas since rice prices in different loca- tions of the United States follow each other very closely (Brorsen 1983). Given these elasticities and the es- timated supply of marketing services, the portion of the increased margin that would be shifted to the producer can be calculated using the method outlined in the Appendix. IMPACT OF PRICE VARIABILITY ON MARKETING MABGINS RESULTS The shift in farm policy emphasis, from high sup- port prices and effective supply controls to the free market, coupled with the growing interdependence of sElasticity is the ratio of the percent change in one variable to the percent change in another variable. TABLE 3. MONTHLY RICE PRICE MEANS, STANDARD DEVIATIONS, AND COEFFICIENTS OF VARIATION FOR THE TWO POLICY PERIODS 1960-71 1974-81 Coefficienta Coefficient \y Standard of Standard of Price Mean Deviation Variation Mean Deviation Variation ¢/cwt ¢/cwt % ¢/cwt ¢/cwt % Farm 505.1 22.6 4.48 953.4 189.3 19.86 Mill: Arkansas 987.8 29.7 3.01 2061.8 369.9 17.94 California 1004.3 61.0 6.08 2133.9 447.2 20.96 Louisiana 853.6 47.8 5.60 1882.4 422.0 22.42 Texas 998.6 28.0 2.80 2062.6 373.6 18.11 the U.S. rice industry with world markets, increased rice price variability in the 1970’s. The coefficient of variation for Texas farm prices changed from 4.99 with price-stabilizing programs during 1960 to 1971 to 21.17 with the market-oriented policy of 1974 to 1981 (Table 3) (the greater the coefficient of variation, the greater price-variability exists). Similar magnitudes of increased mill price variability are reflected by the coefficient of variation measures associated with the respective states’ mill prices (Table 3). Interestingly, the medium-grain mill prices (California and Louisiana) were more variable than the long-grain mill prices (Arkansas and Texas) during both policy periods. In accordance with Gardnefs model, an attempt was made to associate the observed widening of the Texas mill-farm gate marketing margin during the 1970’s with the respective factors of importance, i.e., quantity of rice milled, milling cost, and a measure of the in- creased rice price variability. The estimated inverse supply of marketing services was MAR = -0.116 + 0.0490 QM + (1) (.12) (.65) 0.132 VARTX+ 1.193 MILLC (2.18) (.85) where MAR (i.e., marketing margin) is the Texas mill price adjusted to a rough rice equivalent minus the Texas farm price (dollars per hundredweight [cwt]), QM is the annual quantity of rice milled in Texas (million cwt of rough rice), VARTX is the annual coefficient of varia- tion of monthly milled prices in Texas, and MILLC is an estimate of annual milling costs (dollars per cwt of rough rice). R2, coefficient of determination, equals 0.629. The t statistics are in parentheses under their respective regression parameter estimate. The variable repre- senting price variability over the data period, VARTX, is significant and positive indicating mills are risk averse. Meyer (1983), in an analysis of mill buying response in bid/acceptance markets, also found rice mills were risk averse, i.e. , they reduced their bids for rough rice when faced with a higher level of price volatility. The other variables are relatively insignificant/f These results indicate the widening in the farm-mill price margin is significantly associated with the increase 6 in price variability accompanying the market-oriented*b' farm policy emphasis of the 1970’s. The increased varia- bility in Texas mill prices (average annual coefficient of variation shifting from 1.57 during 1960 to 1971 to 8.91 during 1974 to 1981) implies an increase in mill-farm margin of $097.5 The average farm price during 1974- 1981 was $9.81/cwt, while the average Texas mill price for the same period was $20.51/cwt. Using these price levels and the earlier discussed elasticities (production at 0.35 and demand at —O.83) in Appendix Equations 12 and 13 indicates the impact of such increased price variability is to increase retail prices by $.45/cwt and to decrease farm prices by $.52/cwt. Attention needs to be focused on production costs, marketing margins, and alternative land tenure arrange- ments to address the issue of which parties (i.e., owner/ operator producers, tenants, and/or landowners) are be- ing adversely affected by increased rice price variability. The shift in farm policy emphasis increased farm price variability and may significantly reduce producers’ chances of survival. PRODUCERS VIABILITY METHODS The Firm Level Income Tax and Farm Policy Simulator (FLIPSIM V) was used to analyze a typical rice farm on the west side of the Texas Rice Belt.6 The computer model simulates the annual production, farm policy, marketing, financial management, and income tax aspects of a typical farm over a 10-year planning horizon. The model recursively simulates the farm by using the ending financial position for one year as the beginning financial position for the next year. The model repeats the 10-year planning horizon for 50 iterations using random crop prices and yields drawn from a multivariate probability distribution. By changing the “The insignificance of MILLC may be due to a data quality problem. The “Total Marketing Bill” statistic, used to estimate the missing milling cost data, poses a limitation on the study. Unfortunately, a Y continuous milling cost statistic is not available. sEstimating the margin equation with price/cost variables deflated by the Consumer Price Index results in an even higher impact 0n mill- \ farm margins ($1.11 when reinflated to the average price level of the‘ H 1976-82 period). 6A typical farm was defined for Jackson, Matagorda, and Wharton Counties. "N TABLE 4. PROBABILITY DISTRIBUTIONS OF RICE YIELDS AND PRICES FOR TEXAS GULF COAST RICE PRODUCERS O Rice Yields Rice Prices First Second 1960-71 1974-81 C C Item mp mp luiy January July January Cwt Dol/cwt Mean 45.82 11.36 5.09 4.77 9.73 9.09 Ranked Deviation From the Mean 1 ——4.89 -11.36 -0.74 -0.75 -3.21 —3.34 2 —3.82 — 4.56 -0.41 -0.29 -3.04 -3.11 3 —3.12 — 2.47 —0.21 -0.10 -2.17 —2.68 4 —2.00 0.39 -0.11 -0.02 -1.04 2.00 5 —0.87 0.85 —0.09 0.00 '—0.87 -1.64 6 0.83 2.96 0.10 0.09 0.10 0.26 7 2.37 3.52 0.16 0.09 0.73 0.99 8 3.02 4.41 0.20 0.10 1.18 1.66 9 4.92 5.32 0.31 0.20 2.81 3.49 10 6.50 5.71 0.40 0.37 3.88 4.77 Correlation Coefficient 0.44 0.54 0.91 farm policy parameters, one can quantify the probable impacts of alternative farm policies on a farm’s viability.7 An earlier version of F LIPSIM is described and documented by Richardson and Nixon (1982). The ver- sion of FLIPSIM used for this study has been revised t0 include the provisions of both the 1981 and 1982 income tax policies and the 1981 farm program. For this analysis, a typical Western Texas Rice Belt rice farm was simula- ted recursively over a 10-year planning horizon and the horizon was replicated 50 times using different rice prices and yields randomly drawn from empirical dis- tributionss The model was used to simulate the typical farm under two scenarios: (1) the 1960’s rice program (1960-71), and (2) the 1970’s rice program (1974-82). The same assumptions regarding machinery depreciation (cost recovery), family size, family living expenses, in- come tax and social security schedules, machinery re- placement, interest rates, growth, etc. were used for both scenarios. Information necessary to model a typical Western Texas Rice Belt farm in F LIPSIM is provided by Gerlow (1982) and by Richardson and Bailey (1982). The typical farm selected for this study has 1,700 acres of land. Rice is planted on the same cropland every other year and idle cropland is cash leased for grazing. The operator has an initial debt to asset ratio of 40%. Three tenure 7Viability in this case refers to the probability the farm will be an economic success and the farm will be able to survive for 10 years. Probability of success is measured as the probability the farm will generate sufficient income and retained earnings to have a positive after-tax net present value of net family withdrawals and change in net 1p worth. Assuming a real after-tax discount rate equal to 4%, the probability of success indicates the chance a farm will provide a 4% or greater real return t0 initial equity. Survival in this case is defined as the farm maintaining its leverage ratio at levels acceptable to local gfmancial institutions, i.e., remaining solvent. Empirical probability distribution: A statistical density function in which the probability of observing a particular range of values is the same as the actual occurrence of observations in that range in empiri- cal data. situations (full owner, part owner, and tenant operator) are analyzed to determine the impact of the two policy scenarios on different tenure arrangements. The part owner owns 412 acres of cropland and leases the remain- der on a share lease. Leased land is rented on a crop- share basis with the landowner receiving 10% of the crop and paying his share of grain drying costs and sales commission (Gerlow 1982). The cash production costs for rice in the study area are summarized in Appendix Table 1. Additional information to describe the farm and as- sociated production costs is available in Richardson and Bailey (1982). The simulation model was run assuming all costs, mean prices, and policy parameters are held constant throughout the planning horizon. Long-term interest rates were 10% and intermediate interest rates were 12% over the entire planning horizon. Land values were held constant at their current levels over the planning horizon. Bivariate probability distributions for rice yields (first crop and second crop) were developed from pro- ducer yields in the study area. Actual farm yields for 5 years (1977-1981) were used to develop empirical proba- bility distributions for first and second crop rice yields. The empirical probability distributions for rice yields are summarized in Table 4 in terms of the means and ranked deviations from the means. Yield for the second crop is correlated (p = 0.44) to yield for the first crop in the simulation model using the procedure suggested by Richardson and Condra (1981, p. 433). The yield dis- tribution reported in Table 4 was used for both policies analyzed. Many marketing strategies exist for rice producers in the Texas Rice Belt. A typical strategy is to sell after harvest in July (first crop) and in January (second crop) (Gerlow 1982). To simulate this practice, empirical bivariate probability distributions for July and January rice prices were developed for both policy periods ana- lyzed, 1960-71 and 1974-81 (Table 4). Prices in January during these periods are highly correlated (p = 0.85 and 7 p = 0.87, respectively) to ]uly prices. The typical rice farm was simulated for both the 1960’s farm policy and the 1970’s farm policy. Under the 1960’s policy, the farm has a 688-acre rice allotment and it is assumed the farm cannot plant rice in excess of this allotment, i.e., an effective marketing quota based on acreage. Grain sorghum is assumed to be planted on the cropland without a rice allotment (175 acres) under the 1960’s farm policy. The acreage allotment under the 1970’s farm policy is for 748 acres of rice and determines only the portion of the crop eligible for price supports and deficiency payments. These allotments were es- timated based on the Texas allotments for rice between 1960 and 1980, the acres of rice planted in Texas be- tween 1960 and 1980, and the typical farm’s cropland acreage‘) The average nominal loan rate between 1960 and 1971 was $4.68/cwt (91.9% of the average price in ]uly). Over the 1976-81 period, the average nominal loan rate was $6.98/cwt and the average nominal target price was $9.30/cwt (71.7% and 95.6% of the average price in ]uly, respectively). To compare the two rice policies, both the price distribution and the average loan rate for the policy in the 1960’s were scaled up to levels comparable to the 1974-81 rice program. The empirical price distribution for the 1960-71 period (Table 4) was scaled up to the same mean as the 1970’s policy ($9.73 for ]uly and $9.09 for January) plus the marketing margin adjustment for price risk determined above. This adjustment for mar- keting margin was made because returning to the 1960’s policy would reduce both the price variability and the marketing margin, thus increasing the mean price re- ceived by Texas rice producers. Given this adjustment for price, the loan rate for the 1960’s rice policy was increased to $9.95/cwt, or 91.9% of the adjusted mean price ($10.83/cwt.). The average rice price for 1974-81 (Table 4) and the average loan rate and target price for 1976-81 were used directly in the simulation model for the new policy. 10 All mean prices (january and ]uly) and policy variables (loan and target prices) were held con- stant over the 10 years simulated for both farm policies. PRODUCEPKS VIABILITY RESULTS Three typical rice farms on the western side of the Texas Rice Belt were simulated under both the 1960’s rice policy and the 1970’s policy. It was assumed in the base analysis the operator does not change his marketing practice from the old policy to the new policy despite increases in price risk and marketing margins. Addition- QUnder the old rice program, planted acreage of rice in Texas was 99.4% of the Texas rice allotment (Grant, Holder, and Ericksen 1980). Under the new rice program, Texas producers have overplanted their allotment by 14% on the average. Given the farm produces 850 acres of rice under the new policy, his allotment is 748 acres. Deflating the 748-acre base under the new program by the ratio between the average rice allotment for Texas under the new program (500,000 acres) and the old program (460,300 acres) yields the farm’s rice allotment under the old policy, or 688 acres. mThe national loan rates were converted to a long-grain loan rate consistent with the actual loan rate for Texas rice. A comparable adjustment was made for the rice target price. 8 al analyses under the new policy assuming 10% and 15% reductions in variable costs, both with and without a 50% decrease in the marketing margin, were also simu- lated. Base Comparison of Policies The base results indicate the lower price variability and smaller marketing margins associated with the rice policy during the 1960’s generally resulted in greater producer viability (success and survival)hthan under the 1970’s rice policy (Table 5). For a tenant rice producer with 1,700 acres of cropland, the 1970’s rice policy provides only a 70% chance of being an economic suc- cess (providing a 4% or greater real return to initial equity) while the 1960’s policy provided a 100% chance. A part owner has an 88% chance of being an economic .1; success under the new policy versus a 98% chance under the 1960’s policy. Due to the high level of debt on cropland ($600,000), and the associated required princi- pal and interest payments, the full owner has a low probability of success under both farm policies. The probability that tenant farm operators will maintain their financial leverage ratios below maximum levels set by their creditors (i. e., that they will survive) is reduced from 86% under the 1960’s policy to 70% under the 1970’s policy. Similarly, the probability of survival decreased from 98% to 88% for the part-owner operator. The probability of survival was 98% and 100% for the full owner under the 1970’s policy and the 1960’s policy, respectively, due to the high initial net worth of the operator (60% equity in 1,700 acres of cropland). The part owner’s equity in 412 acres of cropland similarly contributed to the higher probability of survival relative to the tenant producer. Average after-tax net present value is approximately $55,000 less for tenant rice farmers under the 1970’s rice policy than the 1960’s policy (Table 5).n For part own- ers, however, average after-tax net present value is greater by $64,000 under the 1970’s rice policy. Similar- ly, this value is greater for full owners under the 1970’s rice program. These mixed results are also observed for the three farm operators’ average ending net worths. The 1970’s rice policy is associated with greater average after-tax net present values for part and full owners due to these operators receiving all or most of the benefits from deficiency payments, while the tenant shares this program benefit with his landowner. The 1970’s rice policy resulted in greater absolute and relative variance in after-tax net present value than Texas rice producers had experienced under the 1960’s policy (Table 5). The relative variance (standard devia- tion) in after-tax net present value for part owners more than doubled as a result of the policy change. Similar results are observed for the other tenure arrangements. Examining the extremes of the after-tax net present q! value and ending net worth distributions reveals these distributions are skewed more to the right for the 1970s “After-tax net present value is the discounted stream of family with- drawals and changes in net worth for the farm operation over the 10- year planning horizon. fl 8 5 rice policy than for the 1960’s rice policy (Figs. 2, 3, and é 4). In the case of a part owner, the minimum after-tax net present value is $80,000 less for the 1970’s policy while the maximum is about $138,000 greater (Table 5). Similar results are also observed for the tenant and the full owner. These distributions were shifted to the right due to the benefits of the 1970’s rice policy (deficiency payments and price supports) and the increased price variability associated with this policy period. The 1970’s program’s benefits provided downside income and price protection from the increased price variability while the increased price variability also provided an opportunity for high prices and returns. The reduced probabilities of success and survival for tenants and part owners suggest, the same interest costs, credit availability rules, and income tax schedules, the 1970’s rice policy is associated with higher average ending leverage (debt/equity) ratios for these farm operator types (Table 5). The average ending leverage ratio for tenant operators increased 118% due to the policy change; the increase was 58% for the part owner and minimal for full owners. The simulation results indicate the 1970s rice poli- cy has different effects on structure than the 1960’s policy. The 1970’s policy reduces the chances of a tenant operator surviving more than it reduces the chances of survival for full owners and part owners. Since 57% of the rice farmers in the Texas Gulf Coast were tenant operators in 1979 (Mullins, Grant, and Krenz 1981), the however, the program benefits were not sufficient to 1970’s rice policy is likely contributing to a structural fully compensate these producers for the increased price change among rice producers in Texas. Mullins et al. variability and the marketing margin change. also indicated approximately 47% of all rice farmers in The financial well-being of part owner and tenant the United States were tenant operators in 1979 so the rice producers in the Texas Rice Belt has been worsened 1970’s rice policy is likely causing similar changes in the by changes in the national farm program for rice. Given structure of U.S. rice production. TABLE 5. COMPARISON OF THE 1960'S RICE POLICY TO THE 1970'S RlCE POLICY FOR RICE FARMERS (FULL OWNERS, PART OWNERS, AND TENANT OPERATORS 1N THE WESTERN TEXAS RICE BELT Full Owner Part Owner Tenant 1970'S 1960'S 1970'S 1960’S 1970'S 1960'S Item Policy Policy Policy Policy Policy Policy After-Tax Net Present Value ($1 ,000)a Mean -10.61 —92.34 367.19 303.57 551.84 606.47 Std. Dev. 197.87 99.51 207.79 84.74 363.60 230.58 Minimum —537.69 —312.53 -87.06 —7.61 —36.46 18.63 Maximum 493.13 89.29 692.17 454.68 1,030.45 822.16 Present Value of Ending Net Worth in Year 10 ($1,000) Mean 961.62 878.45 578.18 498.98 551.26 578.62 Std. Dev. 194.19 99.51 173.55 77.51 287.20 172.10 Minimum 504.81 658.27 249.31 297.57 89.86 144.95 Maximum 1,463.93 1,060.09 885.33 647.84 978.04 769.75 Leverage Ratio in Year 1O (fraction) Mean 0.64 0.63 0.57 0.36 1.07 0.49 Std. Dev. 0.35 0.19 0.68 0.28 1.45 0.79 Minimum 0.29 0.34 0.14 0.16 0.04 0.05 Maximum 1.88 1.05 2.74 2.03 4.00 2.66 Probability of Successb » 0.56 0.22 0.88 0.98 0.70 1.00 " Probability Oi Survival‘ 0.98 1.00 0.88 0.98 0.70 0.86 “Net present value is the present value of net annual family withdrawals plus the present value of change in net worth over the 10-year planning horizon. After-tax net 4"\ present value is largest for the tenant and smallest for the full owner due to the amount of initial equity each has invested, the amount of net gains each has from leasing idle land for pasture (none for the tenant), and the amount of retained earnings for each farm. Annual interest and principal payments on cropland for the full owner exceed the annual crop share rental cost of tenants who have greater annual retained earnings. “Probability 0f success is the probability that net present value will be greater than or equal to zero, assuming a discount rate of 4%. “Probability of survival is the probability that the farm will maintain its leverage ratios at less than maximum levels established for local financial institutions. 40- 30- 25- PGICGIII 20- 15- 10—' I I I I I -550 -350 -150 50 250 450 650 850 1050 Thousand Dollars Figure 2. Probability distribution of after-tax net present value, tenant operator. “i 45- 35— 30- 25- Percent 20- 0_ . _ ~ , ‘\___gl,--- u.“ I I I I I -550 -350 -150 50 250 450 650 850 1050 Thousand Dollars Figure 4. Probability distribution of after-tax net present value, part owner-operator. 10 s04 45— 40- Percent I T v 1 1 -550 -350 -150 50 250 450 650 B50 1050 Thousand Dollars Figure 3. Probability distribution of after-tax net present value, owner- operator. Comparison of Lower Variable Costs The simulation results indicate that reducing vari- able costs while maintaining yield under the 1970’s policy improves producers’ viability (Tables 6, 7, and 8). A 15% decrease in variable costs increases the probabili- ty of success for a rice producer with full ownership from 56% at the base level of costs to 96%. The probability of success for part owners was increased from 88% to 100% when variable costs are reduced 15%. The decrease in variable costs has only a minor impact on the success and survival probabilities for tenant producers. The 15% reduction in variable costs increases the likelihood of tenant producers remaining solvent over a 10-year period from 70% to 72% and does not increase their probability of success. Average after-tax net present value improves con- siderably for all farm types when variable costs are reduced (Tables 6, 7, and 8). A reduction of variable costs by 10% increases average after-tax net present value relatively more for full owners than for part owners and tenant producers. Average after-tax net present value for part owners increases 159% over the base analysis if variable production costs are reduced 10%. Higher ending net worths also result for all farm tenure arrangements when variable costs are reduced. A review of the results of field record-keeping sys- tems in Wharton and Liberty Counties indicates a wide range in the cost of fertilizer and herbicides in Texas rice production (Gerlow 1982). Early results of a 1982/1983 water management survey (Griffin, Perry, and McCauley 1984) and research in progress at Beaumont, in TABLE 6. COMPARISON OF ALTERNATIVE STRATEGIES FOR FULL OWNER-OPERATORS IN THE WESTERN TEXAS RICE BELT f\ Current Policy Current Policy and 50% ' Marketing Margins Lower Marketing Margins Marginsl& 10% Lower 15% Lower Current 10% Lower 15% Lower Variable Variable Variable Variable Variable Variable Costs of Costs of Costs of Costs of Costs of Costs of Item Production Production Production Production Production Production After-Tax Net Present Value ($1,000? Mean —— 10.61 213.85 311.77 57.64 264.68 359.98 Std. Dev. 197.87 171.08 157.79 214.27 176.44 165.51 Minimum —537.69 —212.85 —77.46 -543.16 —165.37 -47.39 Maximum 493.13 579.43 663.19 546.89 648.03 732.73 r Present Value of f‘ Ending Net Worth in Year 10 ($1,000) “ Mean 961.62 1,184.65 1282.57 1029.49 1235.47 1330.77 Std. Dev. 194.19 171.08 157.79 211.37 176.44 165.51 Minimum 504.81 757.95 893.33 480.36 805.43 923.39 Maximum 1,463.93 1,550.22 1,633.99 1,517.68 1,618.83 1,703.50 Leverage Ratio in Year 10 (fraction) Mean 0.64 0.42 0.36 0.56 0.39 0.35 Std. Dev. 0.35 0.02 0.09 0.10 0.12 0.09 Minimum 0.27 0.23 0.22 0.28 0.22 0.21 Maximum 1.89 0.93 0.69 1.94 0.84 0.65 Probability 6f Success“ 0.56 0.84 0.96 0.66 0.88 0.98 Probability of Survival‘ 0.98 1.00 1.00 0.98 'i .00 1.00 aNet present value is the present value of net annual family withdrawals plus the present value of change in net worth over the 10-year planning horizon. After-tax net present value is largest for the tenant and smallest for the full owner due to the amount of initial equity each has invested, the amount of net gains each has from leasing idle land for pasture (none for the tenant), and the amount of retained earnings for each farm. Annual interest and principal payments on cropland for the full owner exceed the annual crop share rental cost of tenants who have greater annual retained earnings. bProbability of success is the probability that net present value will be greater than or equal to zero, assuming a discount rate of 4%. ‘Probability of survival is the probability that the farm will maintain its leverage ratios at less than maximum levels established for local financial institutions. Eagle Lake, and other locations (McCauley, personal communication) indicate similar ranges of costs are as- sociated with rice producers’ water management prac- tices. Current financial and economic conditions warrant recognition that maximizing physical rice yield does not maximize profit. Additional variable inputs such as in- secticides, fungicides, and machinery operations as well as those discussed above should be applied only when the additional revenue generated exceeds the additional costs incurred. Changing Marketing Alternatives Several alternatives are available to the rice indus- try to mitigate the impact of price risk on the marketing margins. First, the rice milling industry, through hedg- ing in a viable futures market, could pass most of the costs of price risk to speculators. Since the relative change in frnargin was associated with risk (refer to Equation l), hedging by mills should lower the margin to nearly the same relative level as in the old policy period. Transaction costs in the futures market, howev- er, would prevent the margin returning to the same level. Brorsen et al. (1984) found the relative margin changes between farm and export wheat prices were small after the wheat policy shift occurred in the early 1970s. Price variability and the accompanying risk were similar to rice. The wheat industry, however, had three viable wheat futures markets when this policy change occurred. The rice producers did not have this benefit. 12 A second alternative is for rice producers to collec- tively mill and market their rice, thus bypassing one pricing point, i.e. , pooled cooperative marketing of mill- ed rice. Since the producers never relinquish ownership of the rice until it is sold as milled rice, the margin shift due to price risk would be returned to the cooperatives producers as part of the rough rice price received. The producefs improved position in the market would be partially offset, however, by the investment cost of membership in a cooperative. A third alternative is for a policy change to occur such that the cost of price risk is shifted to the taxpayer. This could be accomplished by raising domestic prices to a stabilized level slightly above the general world rice price and utilizing an export subsidy to maintain export levels. 12The New Orleans market for rice futures trading opened in April 1981 and closed in Iune 1983. The rough rice contracts have been transferred to the Mid-American Commodity Exchange in Chicago as of September 1983. 11 TABLE 7. COMPARISON OF ALTERNATIVE STRATEGIES FOR PART OWNER-OPERATORS IN THE WESTERN TEXAS RICE BELT Current Policy Current Policy and 50% Current Marketing Margins Lower Marketing Margins Policy, \ Margins 8t 10% Lower 15% Lower Current 10% Lower 15% Lower Variable Variable Variable Variable Variable Variable Costs of Costs of Costs of Costs of Costs of Costs of Item Production Production Production Production Production Production After-Tax Net Present Value ($1 ,000)"“ Mean 367.19 585.78 674.65 419.31 630.78 714.47 Std. Dev. 207.79 145.26 135.23 230.65 147.57 143.23 Minimum -87.06 — 32.90 324.28 — 126.77 286.25 358.61 Maximum 692.17 831.08 986.92 858.34 969.46 1,048.09 Present Value of Ending Net Worth in Year 10 ($1,000) Mean 578.18 781.18 867.81 627.62 823.94 907.63 Std. Dev. 173.55 136.10 135.23 199.75 147.57 143.23 Minimum 249.31 272.28 517.44 222.01 479.42 551.78 Maximum 885.33 1,024.24 1,180.08 1,051.50 1,162.63 1,291.25 Leverage Ratio in Year 1O (fraction) Mean 0.57 0.28 0.22 0.55 0.23 0.21 Std. Dev. 0.68 0.09 0.09 0.71 0.09 0.08 Minimum 0.14 0.13 0.09 0.13 0.09 0.09 Maximum 2.74 2.19 0.46 3.10 0.48 0.44 Probability of Successb 0.88 0.98 1.00 0.90 1.00 1.00 Probability 0f Survival‘ 0.88 0.98 1.00 0.90 1.00 1.00 “Net resent value is the resent value of net annual famil withdrawals plus the present value of chan e in net worth over the 10-year lannin horizon. After-tax net P P Y . __ 8 P 8 _ present value is lar est for the tenant and smallest for the full owner due to the amount of initial equity each has invested, the amount of net ains each has from leasin - g a 1 g g idle land for pasture (none for the tenant), and the amount of retained earnings for each farm. Annual interest and principal payments on cropland for the full owner exceed the annual crop share rental cost of tenants who have greater annual retained earnings. bProbability of success is the probability that net present value will be greater than or equal to zero, assuming a discount rate of 4%. ‘Probability of survival is the probability that the farm will maintain its leverage ratios at less than maximum levels established for local financial institutions. Combination of Marketing . Alternatives/Cost of Production Changes The simulation model indicates that the combined effects of reducing the marketing margin and reducing variable costs result in significantly greater producer viability (success and survival) than the current policy conditions alone (Tables 6, 7, and 8). After-tax net present value and present value of ending net worth rise sharply for each tenure group simulated. Probability of remaining solvent for full and part owners is at 1.0; but it is only 0.88 for the full tenant. Reverting t0 1960s Policy Reversal of the 1970s rice program to the 1960s policy would lower both marketing margins and price risk. Such a shift would improve producers’ chances of survival and success for all three tenure arrangements provided the producer had not shifted to a different marketing alternative or lowered his variable cost of production. Full and part owners would be better oif with the 1970s program if relative marketing margins are at the same level as under the 1960s policy or variable production costs are lower. Likewise, the tenant operator would be financially better off, if he can sur- 12 vive. His chances of surviving, however, are lower with the 1970s policy than with the 1960s policy even with shifts in marketing alternative and variable costs. Low beginning equity coupled with price variability result in tenant producers being more vulnerable under the cur- rent policy during adverse periods. LIMITATIONS AND CONCLUSIONS This study has several important limitations. First is the 5 years of field observations used for estimating rice yield variability. Additional years would strengthen this distribution; unfortunately, additional data are not avail- able. Second, simulation results on producers’ viability in this study assume the operator. does not change his marketing practices from the 1960s policy to the 1970s policy despite increases in price risk and marketing margins. A shift in marketing practices was not analyzed. Finally, the simulations were analyzed using typical i farms in the Texas rice area. The results for a given producer could differ. The shift from a rice program with emphasis on supply control through marketing quotas and allotments‘! to a market-oriented program has impacted the U. S. rice industry in the following ways: '\ \ TABLE 8. COMPARISON OF ALTERNATIVE STRATEGIES FOR TENANT OPERATORS lN THE WESTERN TEXAS RICE BELT Current Policy Current Policy and 50% gglrirfyt Marketing Margins Lower Marketing Margins Marginsl& 10% Lower 15% Lower Current 10% Lower 15% Lower Variable Variable Variable Variable Variable Variable Costs of Costs of Costs of Costs of Costs of Costs of Item Production Production Production Production Production Production After-Tax Net Present Value ($1 ,0O0)“ Mean 551.84 692.41 730.17 697.41 853.55 924.49 Std. Dev. 363.60 416.63 457.89 337.15 361.24 381.58 Minimum —36.46 —7.16 7.48 —35.87 —6.56 8.08 Maximum 1,030.45 1,189.82 1,241.27 1,211.95 1,361.16 1,435.12 Present Value of i Ending Net Worth in Year 1O ($1,000) Mean 551.26 689.59 731.38 672.22 826.16 897.10 Std. Dev. 287.20 339.00 377.59 281.63 304.03 323.52 Minimum 89.86 119.16 133.81 67.89 119.76 134.41 Maximum 978.04 1,137.41 1,188.86 1,159.54 1,308.75 1,382.71 Leverage Ratio in Year 10 (fraction) Mean 1.07 0.82 0.81 0.69 0.51 0.47 Std. Dev. 1.45 1.27 1.12 1.23 0.92 0.82 Minimum 0.04 0.05 0.04 0.04 0.05 0.06 Maximum 4.00 3.28 2.82 4.00 3.26 2.79 Probability of Successb 0.70 0.72 0.70 0.84 0.88 0.88 Probability of Survival‘ 0.70 0.72 0.72 0.84 0.88 0.88 "Net present value is the present value of net annual family withdrawals plus the present value of change in net worth over the 10-year planning horizon. After-tax net present value is largest for the tenant and smallest for the full owner due to the amount of initial equity each has invested, the amount of net gains each has from leasing idle land for pasture (none for the tenant), and the amount of retained earnings for each farm. Annual interest and principal payments on cropland for the full owner exceed the annual crop share rental cost of tenants who have greater annual retained earnings. “Probability of success is the probability that net present value will be greater than or equal to zero, assuming a discount rate of 4%. “Probability of survival is the probability that the farm will maintain its leverage ratios at less than maximum levels established for local financial institutions. 1. The time for mill prices to adjust to new informa- tion shortened from about 20 weeks to about 13 weeks. 2. Markets at the mill level were still inefficient in the weak form sense (information was received from historical prices only). 3. The leading market at the mill level shifted from Louisiana to Arkansas. 4. The industry had to contend with increased price variability. Coefficients of variation for farm and mill prices increased fourfold or more. 5. Margins between the mill and farm prices in- creased, lowering the farmers’ share of the adjusted mill price about 7 percentage points. 6. The margin increase was related to increased price variability. Changes in quantity milled and mill costs had an insignificant impact on the margin increase. *The amount of the margin change passed back to the producer through a discounted price between the two periods was determined to be $0.52/cwt. 7. The increased price variability combined with a discounted farm price decreased producers’ probability of survival from 0.98 to 0.88 for the part owners, and 0.86 to 0.70 for full tenant operators. 8. The 1970’s rice policy increased the absolute and relative variance in after-tax present value for rice pro- ducers. 9. Program benefits under the l970’s rice policy were not sufficient to fully compensate part owners and tenants for the increased price variability and the mar- keting margin change. 10. The shift in farm policy from the 1960’s to the l970’s was biased against the tenant operator and will likely contribute to major structural changes among all rice producers. ll. The relative marketing margin could be lowered to near that occurring during the l960’s if the milling industry shifted price risk to the futures market or rice producers vertically integrated through a cooperative mill, among other alternatives. Such marketing alterna- tive changes would substantially improve the rice pro- ducer’s viability. Price variation, however, still impacts negatively on the producers. l2. A combination of lowering variable costs and shifting marketing alternatives puts the producer in a more viable position. 13 APPENDIX MARKET EFFICIENCY METHODS The statistical theory of time series analysis assumes the series t0 be investigated are mean and covariance stationary, i.e., the mean and covariance are not a function of time (Chow 1975, p. 62). Time series analysis methods assume the time series (y) can be decomposed as Yt:Dt+5t+6t (l) where Dt is the deterministic component which contains nothing about the impacts of new information and (st + et) is a stationary stochastic process which is mod- eled here using time series analysis. The first step in the time series procedure was to remove the deterministic part) of the series, leaving only stationary stochastic components (Bessler and Schrader 1980; Grant et al. 1983). The weekly milled rice prices of this study have trend components because of inflation, and seasonality because of storage costs. Trend components were re- moved by first differencing. Seasonal patterns were re- moved using a quarterly spline function (Judge et al. 1980; Brorsen 1983).l If all the deterministic components of the series are removed, the remaining short memory portion (st + et) should reflect how rapidly new information is procured by each market. Autoregressive (AR) modeling of the stationary data can provide a test of market efficiency in the weak form sense (Fama 1970). If the stationary data follow an AR model of order zero, the corresponding market would be efficient in the sense that it would adjust instantaneously to the information reflected by its own prices. These tests of market efficiency need to be interpreted with caution since they assume that (a) trans- action costs are zero, (b) all traders are risk neutral, (c) information is transmitted to all traders instantaneously, (d) all traders agree about the influence of new informa- tion on current prices, and (e) the cost of information is zero for all traders (Danthine 1977; Lucas 1978). Haugh (1972) developed a method of pairwise comparison using the cross-correlations of the univariate residuals. In the current case, it is useful to extend beyond these ap- proaches by considering several price series simultane- ously. Selected AR models were fitted to the stationary weekly price data to investigate the dynamic interactions among rice prices at different locations and between the two policy periods. AR orders (p) were determined using Akaike’s information criterion (AIC) (Akaike 1969). Auto- regressive models using these AR orders were then estimated using ordinary least squares. The autoregres- sive model was of the formzz lThe spline function used a cubic polynomial in time with continuity and differentiability restrictions imposed at each of the four switching points. 2Note this is a reduced form model. Therefore, current endogenous variables do not appear on the right-hand side of the equation. 14 V10) a110)--,fl11<0)p Yift-j) 61ft) <2}, || ll P-i + via-j) elm a» ...ékk where yl through yk are prefiltered versions of the k original price series, p is the order of the multivariate autoregressive scheme, the aimfi), i,m = 1, . . .(k)j = 1, .,p are coefficients and the errors are multivariate white noise. Market efficiency in each market can be evaluated by testing the restriction that all coefficients for the lagged values of a particular variable are zero. The test is equivalent to causality tests within the con- text of a multivariate AR model (Tjostheim 1981). If the efficiency of the first market is of interest, then for channel one: p Ylft) = _2lfa110) V101’) + J: + a11<0)Y1<>¢ GPJ-(t) LRMU I (Brorsen 1983) (4). where LRMij is the long-run multiplier measuring the?! impact on the expected value of price i from a change in price j and PJ-(t) is price j in time t. i: IMPACT OF PRICE VARIABILITY ON MARKETING MARGIN METHODS A theoretical model of price determination between levels of a marketing channel has been developed by Gardner (1975). He contends that prices are determined by retail demand, farm supply, and the supply of mar- keting services. If mills are risk averse, then a change in price variability would be expected to shift the supply of marketing services. Gardnens model assumes a competi- tive market. Due to the concentration of rice mills in the Texas rice area, rice milling is probably not perfectly competitive. The assumptions of perfect competition are stronger than are needed for a firm to behave as a price taker. Baumol, Panzer, and Willig (1982) have proposed perfect contestability as a generalization of perfect com- petition. They showed that a market is contestable if entrants can reverse their investments without loss and suffer no disadvantages relative to incumbents. Although the assumptions of a contestable market are rather de- manding, especially in the short run, they may provide plausible approximation for many concentrated indus- tries (see Baumol et al. 1982). Under the maintained hypothesis that the mere threat of entry makes firms behave as if they were price takers, it is appropriate to explore the implications of price uncertainty in the rice marketing channel. The supply of marketing services is written in price dependent form as 5 = f1 (Q»V.Z) (5) where S is the margin, Q is quantity milled, V is a measure of price variability, and Z is a set of exogenous shifters (in this case, milling costs). The quantity sup- plied at the farm (Q?) is Q? = f2 (Pr, X) (6) where P; is the farm price and X is a set of exogenous shifters (e.g., yield). The quantity demanded at the mill level is Q91 = fa (PIIDY) (7) where Pm is mill price and Y is a set of exogenous shifters (e.g., population, income, world rice production). The system is completed by the following identities: Pf = Pm — S (8) Q = Q? = Qfi (9) The inverse supply of marketing services (5) can be estimated directly assuming quantity is determined ex- ogenously since the majority of rice production is milled in the same crop year it is produced (USDA, Rice Situation and Outlook). In addition, rice production, as with most agricultural crops, is related to lagged price rather than current price and thus quantity is exogenous- ly determined (Wold 1964; Grant and Leath 1979). The incidence of a change in margin can be determined by a method similar to that of Fisher (1981). After estimates for (5) are obtained, the impact of increased price varia- bility on the margin can be obtained by totally diifer- entiating (5): ds =ifkao a-iflav +if1az <10) aQ av az If dQ and dZ are assumed to be zero, then (10) can be solved for the change in margin associated with a change in price variability. By equating the quantities in (6) and (7) and totally differentiating, the following is obtained: 35am + iéax - Lfaapm Jifgar = o (n) 6P; 8X 813m dY By assuming dX and dY to be zero and writing (11) in elasticity form, the following is obtained: dP dPm Si“ 6a-—— = 0 (12) Pf Pm where es is the elasticity of farm supply and ed is the elasticity of mill demand. By totally differentiating (8), the following is obtained: dPf = dPm — dS (13) Equations and (13) can be solved for dP; and dPm, given es, ed, Pm, Pf, and dS. APPENDIX TABLE 1. SUMMARY OF CASH PRODUCTION COSTS FOR RICE USED IN THE BASE SIMULATION Item First Crop Second Crop ($/acre) Seed 29.1 1 0.00 Fertilizer 41.10 12.15 Chemicals 74.00 0.00 Fuel and Lubea 76.60 12.00 Machinery Repairs 13.87 0.00 Variable Harvesting Expensesb 73.35 18.52 Labor‘ 25.63 10.57 lnterestd 12.51 2.00 Totale 346.61 55.24 aFuel and lube costs include costs for pumping water. “Harvesting costs of $1.63/cwt include combining, hauling, drying, and other handling costs. FLabor costs were derived by allocating one full-time employee and part-time labor to the first and second crop of rice. Part-time labor was hired at $4.50/hour and the full-time employee received $12,000/year. dlnterest was charged using a real rate of interest (5%) paid for 9 months. eAll nonlabor costs were decreased 5% and 10% for the alternative production cost scenarios. 15 REFERENCES Akaike, H. “Fitting Autoregressive Models for Prediction.” Annals 0f the Institute of Statistical Mathematics, 21 (1969): 826-39. Baumol, William ]., Iohn C. Panzar, and Robert D. Willig. Contest- able Markets and the Theory 0f Industry Structure. San Diego: Harcourt Brace Iovanovich, 1982. Bessler, D. A. and L. F. Schrader. “Measuring Leads and Lags Among Prices: Turkey Products." Agricultural Economics Research, Vol. 32, No. 3, Iuly 1930. Brorsen, B. W. “A Study of the Efficiency and Dynamics of the Rice Market.” Unpublished Ph.D. dissertation, Texas A&M Universi- ty, Dept. of Agri. Econ., 1983. Brorsen, B. W., W. R. Grant, I. W. Richardson, and L. D. Schnake. “Impact of Increased Price Variability: A Case Study with Wheat." Western journal of Agricultural Economics, Vol. 9, No. 2, December 1984 (forthcoming). Chow, Gregory. Analysis and Control of Dynamic Economic Systems. Iohn Wiley and Sons, New York, 1975. Danthine, I. P. “Martingale Market Efficiency and Commodity Prices." European Economic Review, 10 (1977): 1-17. Dhrymes, Phoebus I. “Restricted and Unrestricted Reduced Forms: Asymptotic Distribution and Relative Efficiency.” Econometrica, 41 (1973). 119-134. Fama, E. F. “Efficient Capital Markets, A Review of the Theory and Empirical Works." The journal of Finance, 25 (1970): 383-417. Fisher, B. S. “The Impact of Changing Marketing Margins on Farm Prices.” American journal of Agricultural Economics, 63 (1981): 261-263. Gardner, Bruce L. “The Farm-Retail Price Spread in a Competitive Food Industry." American journal of Agricultural Economics, 57 (1975). 399-409. Gerlow, Arthur, Tex. Agr. Ext. Ser., Bryan, Texas, personal communi- cations, 1982. Grant, Warren R. “Regional Adjustments and Regional Competitive Advantages in U.S. Rice Productions.” Rice Outlook and Situa- tion, ERS, USDA, RS-39, March 1982. Grant, Warren R., Shelby H. Holder, Ir., and Milton H. Ericksen. “Rice: Analytical Base and Policy Issues.” NED, ESCS, USDA, Staff Report, December 1980. Grant, W. R. and M. W. Leath. “Factors Affecting Supply, Demand, and Prices of U.S. Rice.” ESCS, USDA, ESCS-47, March I979. Grant, W. R., A. W. Ngenge, B. W. Brorsen, and I. P. Chavas. “Grain Price Interrelationships.” Agricultural Economics Research, Vol. 35, No. 1, Ianuary 1983. Griffin, Ronald, Gregory Perry and Garry McCauley. “Water Use and Management in the Texas Rice Belt Region.” Tex. Agr. Exp. Sta. Bul. 1559, Iune 1984. Haugh, L. D. “The Identification of Time Series Interrelationships p, With Special Reference to Dynamic Regression Models." Unpub-w lished Ph.D. dissertation, Statistics Dept., Univ. of Wisconsin, Madison, 1972. Holder, Shelby H., Ir., and Warren Grant. “U.S. Rice Industry." ESCS, USDA, AER N0. 433, Aug. 1979. Iudge, George G., William E. Griffiths, R. Carter Hill, and Tsoung- Chao Lee. The Theory and Practice of Econometrics, Iohn Wiley and Sons, New York, 1980. l‘ Lucas, R. E. “Asset Prices in an Exchange Economy." Econometrica, 46 (1973). 1429-45. McCauley, Garry. Tex. Agr. Exp. Sta., Beaumont, Tex., personal communications, 1981-1983. Meyer, Donald I. “The Effect of Competition in Sealed Bid Auction: Certainty and Uncertainty.” Unpublished Ph.D. dissertation, Texas A&M University, 1983. Mullins, T., W. R. Grant, and R. D. Krenz. “Rice Production Prac- tices and Costs in Major U.S. Rice Areas, 1979.” Univ. of Ark., Fayetteville, Agri. Exp. Sta., Bul. 851, Mar. 1981. Richardson I. W. and D. V. Bailey. "Debt Servicing Capacity of Producers in the Lower Gulf Coast.” Tex. A&M Univ., Dept. of Agri. Econ. (Mimeo), Nov. 1982. Richardson, I. W. and G. D. Condra. “Farm Size Evaluation in the El Paso Valley: A Survival/Success Approach." American journal of Agricultural Economics, 63 (1981): 430-437. Richardson, I. W. and C. I. Nixon. “Producers Preference for a Cot- ton Farmer Owned Reserve: An Application of Simulation and Stochastic Dominance. " Western journal of Agricultural Econom- ics, 7 (1982). 123-132. Schmidt, Peter, “The Asymptotic Distribution of Dynamic Multip- liers.” Econometrica, 41 (1973): 161-164. Sporleder, T. L. and I. P. Chavas. “Aspects of Pricing Efficiency and the Value of Information.” Unpublished working paper, Texas A&M University, May 1979. Tjostheim, Dag. “Granger Causality in Multiple Time Series.” journal of Econometrics, 17 (1981): 151-76. United States Department of Agriculture. Agricultural Statistics. U.S. Government Printing Office, selected issues. . Rice Situation and Outlook. ERS, Washington, D.C., selected issues. . Rice Market News. AMS, San Francisco, selected issues. . Agricultural Prices, SRS, Washington, D.C., selected issues. Wold, H. Econometric Model Building: Essays on Causal Chain Approach. Amsterdam, Holland, North Holland Pub. Co., 1964. ACKNOWLEDGMENTS Helpful suggestions on earlier drafts from David Bessler, Arthur Gerlow, Ronald Griffin, Mack Leath, Carl Shafer, and Edward Smith are appreciated. Typing and editorial assistance from Sue Durden, Anna Halla- ran, Nina Nobles, and Karen Pilant are gratefully ac- knowledged. The assistance of Curtis Leonhardt of the Rice Council and Ron Griffin and Iames Stansel of the Texas Agricultural Experiment Station in providing photographs for the cover is gratefully acknowledged. 16 " ‘$11., [Blank Page in Original Bulletin] Mention of a trademark or a proprietary product does not constitute a guarantee or a warranty of the product by the Texas Agricultura§ Experiment Station or the U.S. Department of Agriculture and does not imply its approval to the exclusion of other products that also may be suitable. All programs and information of the Texas Agricultural Experiment Station are available to everyone without regard to race, colo,“ religion, sex, age, national origin, or handicap. 2.5M—8/84