T 5 ~ 7 3-1501 3 September 1985 L597 ‘A I. E E?» R is R Y ocr r "z r935 Tam Q An Econometnc Model of the U.SL. Wine Indufitry Ne P. Cla tor The Texas A&M versl S stem Coll ege S on, Te TEXAS AGRIC URAL EXPERIMENT STATION rke, Dirac TABLE OF CONTENTS Foreword . . . . . . . . . . . . . . . . . . . . . . . . . 2 Introduction . . . . . . . . . . . . . . . . . . . . . . . 2 Description of the Structural Model . . . . . . . . . . . 3 Econometric Results . . . . . . . . . . . . . . . . . . . 7 Model Validation . . . . . . . . . . . . . . . . . . . . . 9 Impact of Wine Imports . . . . . . . . . . . . . . . . . . 10 Conclusions . . . . . . . . . . . . . . . . . . . . . . . l3 Exhibit l . . . . . . . . . . . . . . . . . . . . . . . . 14 Identities . . . . . . . . . . . . . . . . . . . . . . . . 19 Endogenous Variables . . . . . . . . . . . . . . . . . . . 20 Exogenous Variables . . . . . . . . . . . . . . . . . . . 22 Tables . . . . . . . . . . . . . . . . . . . . . . . . . . 23 Figures . . . . . . . . . . . . . . . . . . . . . . . . . 29 Endnotes . . . . . . . . . . . . . . . . . . . . . . . . . 37 References . . . . . . . . . . . . . . . . . . . . . . . . 38 Keywords: econometric/wine industry/grapes/raisin/table grapes/wine grapes/supply/demand. AN ECONOMETRIC MODEL OF THE U.S. WINE INDUSTRY MICHAEL K. WOHLGENANT *Associate professor, Department of Agricultural Economics, Texas Agri- cultural Experiment Station and Texas A&M University. INIREVVCHUD This document summarizes an economic model of the U.S. wine industry. In the past 20 years, the wine industry has undergone sweeping changes. It is important for the public to recognize that a wine variety of factors influence the markets for wine and grapes, and that a knowledge of these factors can be important to ascertain- ing future trends in the industry. The purpose of this publication is not to advocate particular policies, but rather to summarize the factors influencing wine and grape prices and quantities in recent years, including the effects of increased wine imports on the domestic industry. This publication should be useful to those individuals primarily interested in the broader domestic and international dimensions of the U.S. wine and grape industries. INTRODUCTION Over the last 20 years the U.S. wine industry has undergone dra- matic changes. In the 1960's and early 1970's the industry was char- acterized by rapid growth in demand for wine combined with rising grape prices. Since the mid 1970's, growth in demand for U.S. wine has slowed as the growth in real consumer income has declined and as wine imports have made strong inroads into the domestic market. While growth in demand was declining, grape supplies continued to increase as new acreage planted in response to the initial wine boom reached bearing age. The effect has been a 60% drop for real grape prices (wine varieties) between 1972 and 1983. This report presents an econometric model of the U.S. wine industry. The model describes the behavior of 32 endogenous vari- ables in the markets for wine and grapes between 1947 and 1983. The model extends previous work by Wohlgenant (1978, 1982) to include equations for wine import demand and long—run supply response of grapes. The econometric model was developed primarily to quantify the effects of increased wine imports but also may be used to quan- tify the effects of changes in other variables including income, pop- ulation, and wine processing costs. DESCRIPTION OF THE STRUCTURAL MODEL The econometric model contains four blocks of equations: (a) consumption of domestic and imported wines, domestic wine prices, and production and inventories of domestic wine; (b) demand and supply of grapes crushed for wine production and grapes dried for raisins; (c) demand and supply of grapes sold for fresh use and grapes sold for canned use; and (d) acreage, yield, and production of grapes. There are three main types of grapes which have alternative uses. Raisin grapes (mainly Thompson Seedless) can be dried, crushed, sold for fresh consumption, or sold for canning. Table grapes (e.g., Tokay) can be sold for fresh consumption or for crush- ing. The most heterogenous category, wine grapes, includes such dis- tinctive varietals as Cabernet Sauvignon and Chenin Blanc. These varietals are used almost exclusively for wine production, although a small portion finds its way into the fresh market. Separate demand and supply equations were estimated for each grape type by market outlet. For taxation purposes, wines are classified as: still wine less than 14% alcohol by volume (mainly table wine), still wine over 14% alcohol by volume (mainly dessert wine and vermouth), and sparkling wine. Commercially, the most important wine is table wine which accounts for about 80% of all wine shipments. While there have been significant compositional changes over time, generally favoring table wine and sparkling wine, necessary price data were not available to estimate separate behavioral equations by product type. Thus only aggregate behavioral equations for all wine were estimated. In order to avoid aggregation bias, where possible, prices and quantities of individual components were used to develop index numbers for aggre- gate prices and quantities. Demand for wine was modeled by two equations: total demand for wine (domestic plus imports), and the ratio of domestic to imported wine consumption. The first equation relates per capita total wine consumption to an index of domestic and imported wine prices, per capita real income, and lagged per capita total wine consumption. Lagged consumption was included to account for dynamics in wine con- sumption resulting from habit persistence, exposure to new wine prod- ucts, advertising, etc. Plots of the data indicated an abrupt increase in demand beginning in 1970. This shift probably was asso- ciated with the introduction of "pop wines" and increased promotional effort by the wine industry. This effect was modeled with a dummy variable which takes the value l for 1970 and beyond, but zero other- wise. Plots of the data also suggested a curvilinear relationship between per capita wine consumption and price, so a double log (con- stant elasticity) functional form was used for this equation. The second wine demand equation relates the ratio of domestic to imported wine consumption to the ratio of domestic to imported wine prices and to the lagged quantity ratio. This specification assumes approximate validity of the two-stage consumer budgeting process whereby domestic and imported wines constitute a weakly separable group. In the first stage, the consumer allocates expenditures between wine (both domestic and imports) and all other goods.l In the second stage, expenditures on wine are allocated between domestic and imported wines. With similar income elasticities for domestic and imported wines, this is equivalent to specifying that the ratio of quantities consumed is determined by the price ratio of the two types of wine. This specification is typical of models employed in inter- national trade to predict commodity trade flows between different countries (Johnson, Grennes, and Thursby, 1979). The advantage of this type of model is that it overcomes the effects of extreme multi- collinearity from introducing prices of goods with similar quality as separate regressors in a demand equation. The lagged quantity ratio was included in the model to account for dynamics in shifting con- sumption from one wine type to the other. As in the case of total demand, graphical analysis of the data indicated a log linear speci- fication would be more appropriate for this specification. Graphical analysis also suggested an abrupt upward shift in this demand rela- tionship for the years 1981 and beyond, so a dummy variable was used to account for this shift. It is not clear what caused this shift; however, introduction of wine coolers by the domestic wine industry is a possibility. In both demand specifications, prices rather than quantities are assumed to be predetermined. Justification for this specification is towfold: domestic wine prices are determined mainly by lagged (endogenous) grape prices, and U.S. consumers face an essentially horizontal supply curve for imported wine. The reason wine prices lag changes in grape prices is because of the length of time required for wine processing and aging. Inventory-to-shipment ratios for wine historically have averaged between l and 1.5, suggesting current year supply is mainly from inventories. Wine prices also are influenced by other input prices (bottle prices, wage rates, etc.), but these prices are largely determined by forces exogenous to the industry. Prices for imported wines also are determined mainly by forces exoge- nous to the domestic industry. The main reason is that imports into the United States account for a small share of production and con- sumption in the major exporting countries. The behavioral equation explaining wine production was derived from a decision model of a representative producer who chooses the quantity of output to minimize the sum of expected production costs and expected inventory holding costs, given expected demand and beginning-of-the-period inventories. This results in a behavioral equation relating the quantity of wine produced to expected grape prices, expected demand for wine, and beginning inventories. Expected grape prices are assumed to be determined solely on the basis of current grape prices; expected wine demand is assumed to be based solely on lagged wine shipments. Time series analysis indi- cated that, aside from a linear time trend, expected grape prices could be explained by current-year prices and expected wine demand could be predicted by the level of wine shipments in the previous year. The final behavioral equation for wine production therefore relates the quantity of output to current-year grape prices, lagged wine shipments, beginning inventories, and a linear time trend. Given the quantity of output and wine shipments, ending inventories are predicted by the identity relating the change in wine stocks to the difference between production and consumption. Eight separate equations are used to explain derived demand for grapes. Prices of grapes crushed (raisin, table, and wine grapes) are specified as functions of the quantities crushed, lagged wine shipments, beginning wine inventories, and a linear time trend. Rel- ative quantities crushed reflect substitution among the three grape types in wine production. Lagged wine shipments and inventories, which are determinants of expected production, take into account the impact of demand and supply conditions in the wine market (Wohlge- nant, 1982). Demand for raisins was modeled by two equations: farm-level demand and inventory demand for raisins. The price of raisin grapes dried (the farm price for raisins) is specified as a function of the per capita quantity dried, per capita beginning inventories of rai- sins, per capita income, and lagged price of raisin grapes dried. Beginning inventories were converted to a fresh weight basis and were added to the current-year quantity of raisin grapes dried. Demand for the raw product consists of demand for current and future use; thus beginning inventories and lagged price, which are determinants of inventory demand, are included in the farm-level demand specifica- tion for raisins. Ending inventories of raisins, on a per capita basis, are related to per capita total supply of raisins (current-year produc- tion plus beginning inventories), current-year farm price of raisins, the rate of change in the farm price of raisins from the previous year, and a linear time trend. The rate of change in price is included in the specification to account for the speculative motive for inventory holding. Other things being equal, a larger supply of raisins, a lower current-year price, and a higher rate of price change each would be expected to lead to an increase in ending inven- tories. This inventory specification is intended to reflect the com- bined motives of raisin packers and the Raisin Administrative Commit- tee, who are permitted to allocate supplies between domestic and export markets and other noncompetitive outlets under the provisions of a Federal Marketing Order. Demand equations for grapes sold for fresh and canned use are based on the static theory of demand. Fresh prices for raisin grapes and table grapes are related to per capita quantities sold for fresh use and per capita income. Canned price for raisin grapes is speci- fied as a function of the per capita quantity canned and per capita income. Graphical analysis of the data indicated curvilinear rela- tionships between prices and quantities for fresh market and canned market demand, so double log functional forms were used for these demand specifications. Fresh prices for wine grapes have moved closely over time with prices on the crushed outlet, so fresh price on this outlet was estimated directly as a function of the crush price for wine grapes. There are two types of supply decisions to model: market allo- cation of predetermined grape supply, and long—run supply response of grapes. A key consideration in formulating market allocation equa- tions is the information available to firms at the time decisions are made. Cultural practices and contractual arrangements dictate that fresh market and canned market allocation decisions be made early in the year prior to determination of prices on other market outlets. This suggests that expected, rather than actual, prices on other out- lets influence these market allocations. These short-run price expectations are modeled simply as prices prevailing in the previous year. Thus the proportion of the grape supply allocated to a given outlet depends on the current—year price on that outlet and lagged prices on other outlets. Lagged quantity ratios also were included in the market allocation equations for raisin grapes to account for rigidities due to prior contractual commitments. In contrast to fresh market and canned market allocations, deci- sions on quantities to dry and to crush can be delayed until late in the season when the decision to dry must be made. Drying occurs in a very short period of time and yield is highly variable from year to year depending on weather conditions at the time drying occurs. Thus the decision to dry is based on expected, rather than the actual, dried price. This short-run price expectation is approximated by the price prevailing in the previous year. Therefore, quantities allo- cated for drying and crushing, as proportions of net supply of raisin grapes (total production less quantities allocated for fresh and canned use), depend on the current—year crushed price and lagged dried price. Supply response for each grape type is described by two equa- tions: one equation predicting bearing acreage and the other equa- tion predicting yield. Grapes are a perennial crop with production extending over a considerable number of years. An additional consid- eration is a fixed gestation period of 3 to 4 years between planting and the flow of production. The equations used to explain acreage adjustment are similar to those proposed by French and Matthews (1971). The main difference is that removals are assumed simply to be proportional to beginning bearing acreage. This simplification seemed reasonable because the productive capacity of vines changes very little as the vines become older. Given this simplification, bearing acreage for each grape type is related to lagged bearing acreage, price expectations formed 3 years ago, and a time trend to account for technological change. Price expectations were modeled by unrestricted distributed lags of prices from 4 to 6 years ago. Yields are influenced mainly by weather and technological change. With rapidly changing acreage, as in the case of wine varieties, average industry yields also can be influenced by the pro- portion of new bearing acreage. For this reason, the yield relation- ship suggested by French and Matthews (p. 486) was employed. This specification relates average industry yield to the change in bearing acreage from 3 years ago and to a linear time trend. ECONOMETRIC RESULTS In this section, econometric estimates of the structural model are presented. The model consists of 25 behavioral equations and 16 identities which describe the behavior over time of 32 endogenous variables. Individual equations and definitions of the variables used in the model are summarized in the Exhibit l. The method of estimation and years covered by estimation are indicated in parenthe- sis by each behavioral equation.4 Values in parentheses below the estimated parameters are standard errors of the coefficients. All price data were deflated by the consumer price index to remove the effects of inflation. Data for the wine variables were obtained from various publications and reports provided by the Wine Institute, U.S. Bureau of Labor Statistics, and U.S. Department of Comerce. Grape demand and supply relationships relate only to California, which accounts for about 90% of all grape varieties suitable for wine pro- duction. Grape price and quantity data were obtained from various published reports provided by the California Crop and Livestock Reporting Service and the Raisin Administrative Comittee. Together, Equations l and 2 predict changes in per capita con- sumption of domestic and imported wine in response to changes in wine prices and per capita income. All parameter estimates have the expected signs and are of reasonable magnitudes. Price and income elasticities of aggregate demand for wine are given directly by the coefficients of the price and income variables in Equation 1; the elasticity of substitution between domestic and imported wine is given directly as the negative of the coefficient of the price vari- able in Equation 2. These two equations can be combined to produce short-run estimates of own— and cross-price elasticities. For domes- tic wine, the own-price elasticity is -0.64 and the cross-price elas- ticity with respect to imported wine is 0.13. For imported wine, own— and cross-price elasticities of -1.05 and 0.54, respectively are indicated. In both demand relationships, significant lagged quantity variables are indicated. This suggests that changes in current-year prices and income affect future wine consumption. 4 Equation 3 predicts shipments of California wine into all mar- kets (domestic plus exports) as a share of shipments of domestically produced wine entering distribution channels in the United States. Graphical analysis indicated a U-shaped relationship with respect to time, so linear and squared trend variables were included in this specification. Equation 4 provides the linkage between domestic wine prices and crush grape prices. This price spread depends on wine processing and distribution costs, the most significant of which are wine bottle costs. These costs were proxied by the Bureau of Labor Statistics Producer Price index for glass containers (SIC 1380) since time-se- ries data were not available for wine bottle prices. A linear time trend also was included in the specification to account for increases in wine producing capacity and technological change. Wine prices were regressed on lagged rather than current-year grape prices because of time lags in wine processing. Different lagged specifica- tions were tested and a simple average of grape prices for the previ- ous 2 years was performed. While the estimate obtained might appear too small, it is statistically significant and entirely consistent with prior expectations. Theory suggests that, in the long run, the elasticity of wine price with respect to grape prices should be equal to the cost share of grapes in wine production. At the Sample means, this elasticity equals 0.07. This is entirely consistent with the data. On average, l ton of grapes yields 170 gallons of wine. With l2, 4/5 qt. bottles per case and an average (sample mean) price for grapes of $60 per ton, this would imply a farm equivalent of the wholesale value for grapes of about $0.84 per case. With an average price for wine of $12 per case, this would suggest an average cost share value of about 0.07. Wine output, crush demand for grapes, and demand and supply of raisins are simultaneously determined by Equations 5-8. Equations for wine production, demand for raisins, and crush supply of raisin type grapes were estimated by ZSLS. The crush demand equations were estimated by 3SLS with symmetry imposed on the grape quantity vari- ables. This restriction was suggested by theory. It was tested and not rejected statistically. All parameters have the expected signs. As hypothesized, wine production and crush grape prices are posi- tively related to lagged wine shipments and negatively related to beginning wine inventories. Wine output is negatively related to crush grape prices and negative price flexibilities are indicated for the three types of grapes used in wine production (estimated mean price flexibilities of -0.61, -0.26, and -0.99 for raisin, table, and wine type grapes, respectively). The signs and magnitudes of the cross-quantity parameters in the crush demand equations indicate that raisin and table grapes are close substitutes, but that wine grapes are complementary with raisin and table grapes. Equations 9, l0, and ll predict the farm-level price for rai- sins, ending inventories of raisins, and the proportion of the supply of raisin grapes crushed, respectively. All parameter estimates have the expected signs and are of reasonable magnitudes. A negative mean short-run price flexibility of -0.91 is indicated for the farm-level demand for raisins. The mean short-run supply elasticity for raisin grapes crushed is 0.40 and the mean cross-price elasticity with respect to (lagged) raisin price is -0.20. Given this supply equa- tion, the proportion of the supply of raisin grapes dried can be pre- dicted as one minus the proportion of the supply allocated for crush- ing. Equations l2 through l7 predict prices and quantities of raisin and table type grapes allocated for fresh and canned use. All the variables are hypothesized to be jointly determined. With the excep- tion of fresh market demand and fresh allocation of raisin type grapes, these equations were estimated by 2SLS. Fresh market demand relationships were estimated by 3SLS with symetry imposed at the sample mean relative budget shares of the two grape types. This restriction was suggested by theory, which indicates that the ratio of any two cross flexibilities is approximately equal to the recipro- cal ratio of budget shares for the two commodities (Houck, 1966). Inclusion of the current-year price variable in the fresh allocation equation for raisin type grapes led to a negative but insignificant coefficient estimate. Efforts to overcome this inconsistency proved unsuccessful, so the proportion of raisin type grapes allocated for fresh use was predicted simply by the lagged dried price and lagged quantity ratio (Equation l8). Overall, the parameter estimates of these equations have the correct signs and are of reasonable magni- tudes. Equations 19 through 25 predict bearing acreages and average industry yields of the three grape types. Note that the acreage equations for raisin and table type grapes are expressed in first- differences rather than levels. This was done in order to make these two data series stationary. Price changes in the current year do not influence acreage changes until 4 years from now due to a 3-year ges- tation period between new plantings and bearing age, and lagged price expectations. Lagged price variables through year t-6 were included in each equation. Additional price lags also were tested but were found to contribute little to the variation in bearing acreage. Short—run mean acreage elasticities (for year t-4) of 0.04, 0.02, and 0.20 for raisin, table, and wine type grapes, respectively, are indi- cated. Thus, the acreage response equations suggest extremely long lags in response to any sustained price changes. MODEL VALIDATION Having estimated an econometric model of the wine industry, the next step is to see how well the model predicts actual values of the endogenous variables (Exhibit lc). For each time period (1963-83), the estimated model is solved for the 32 endogenous variables (given actual values for the exogenous variables and lagged endogenous vari- ables). The predictive performance of the model is evaluated through visual comparison of actual with predicted values (Figures 1-32), and through evaluation of different goodness of fit statistics (Table l). In the figures, solid lines are actual values of the endogenous vari- ables while dashed lines are simulated values of the variables. The four evaluation statistics presented in Table l are: root mean square error (RMSE), root mean square percentage error (RMS%E), Theil's decomposition coefficient (Ud), and Theil's inequality coef- ficient (U1). For RMSE, RMS%E, and U1, the smaller the value, the 1O better the fit. Values for U1 and Ud range between 0 and 1. A value of 0 for U1 would mean a perfect forecast, while a value of 1 for Ud would mean completely unbiased forecasts (Kost, 1980). Overall, graphical comparisons of actual with predicted values and estimated evaluation statistics for the U.S. wine model indicate a highly satisfactory fit to the observed data. Graphical analysis reveals that predicted values generally follow the same pattern as actual values. The RMSE and RMS%E statistics are low except for IRNS and DPWGF. With the exception of QRGCA, Theil's decomposition coef- ficient (Ud) ranges between 0.67 and 1, suggesting relatively unbi- ased forecasts for the major quantity and price variables of the model. Theil's inequality coefficient (U1) is close to zero in all cases, indicating close approximation of predicted values with actual values. The smallest forecast errors occur in wine quantities and price, quantities and prices for grapes crushed and dried, and bear- ing acreages for the three grape types. Since these are the key variables linking wine imports to grape prices and quantities, this suggests that the model can provide accurate estimates of the impact of wine imports on the domestic industry. IMPACT OF WINE IMPORTS In this section, the U.S. wine model is used to estimate the effects of increased wine imports on prices and quantities of the domestic industry for the most recent five years, 1979-83. Since imports affect grape prices and quantities only after a l-year lag, quantitative estimates are provided for the years 1978-83. Over this period, imports increased approximately 70% from a volume of 78.367 million gallons in 1977 to 133.065 million gallons in 1983. The effects of increased imports are quantified by comparing the historically simulated time paths of the variables with the simulated values when the volume of imports is held constant at its 1977 level. The present analysis is concerned with the impact of import quanti- ties rather than prices. Thus, prior to conducting the simulations, the estimated import demand relation (Exhibit 1A, Equation 2) was inverted to obtain a price dependent specification. The policy ques- tion posed is: What would have been the expected levels of prices and quantities for U.S. wine and grapes if import supply had been restricted to the volume in 1977? In the simulations, grape acreages and yields were fixed at their historical levels until 1983, since this is the first year that a change in wine imports in 1978 would have an impact on bearing acreages and yields. Changes in acreages and production for 1983, resulting from a change in imports in 1978, were calculated by the estimated relations in Exhibit 1A.6 The estimated yield relationships for raisin and table type grapes (Exhibit 1A, Equations 23 and 24) indicate that changes in bearing acreage have small and insignificant ll effects on yields. Thus yields for these two grape types were assumed to be exogenously determined in the simulations. All simula- tions are dynamic in that they take into account feedback effects from changes in lagged endogenous variables. Historically simulated values for the major variables of the model are exhibited in Table 2. Table 3 shows the cumulative effects attributable solely to an increase in imports beginning in 1978. For example, a reduction of domestic wine shipments by 15.012 million gallons in 1983 (SWD for 1983) is the reduction in shipments for that year attributable to the increase in imports from 1978 to 1983. All effects take into account simultaneous relationships and lagged adjustments resulting from changed imports. Table 5 expresses the cumulative effects from Table 3 as percentage changes from the his- torically simulated values in Table 2. All effects have the expected signs. Increased imports reduced wine shipments, increased wine inventories (until 1983), reduced wine and grape prices, reduced uti- lization of grapes for crushing, increased utilization of grapes on other outlets, and finally, in 1983, reduced acreage and production of grapes. However, the timing and relative magnitudes of the effects differ depending on the particular market in question. The main factors leading to these differences are highlighted below. In 1978, increased wine imports of 15.098 million gallons reduced domestic wine shipments by 7.781 million gallons and increased wine inventories by 6.806 million gallons. Aside from the impact on California's share of the domestic market (a decrease of 6.806 million gallons, Exhibit 1A, Equation 3), these are the only effects initially. In 1979, reduced wine shipments and increased inventories (from an increase in imports in 1978) caused crush demand for grapes to decrease, which led to a decrease in crush prices for the three grape types and shifted utilization of raisin type grapes to the dried outlet. In turn, this increased supply of raisin type grapes for drying reduced the dried price, and increased raisin inventories. The impact of wine imports in 1978 on domestic wine prices does not show up until 1980. This is because domestic wine prices react to lagged rather than current year prices (Exhibit 1A, Equation 4). Likewise, because of lagged supply response in the fresh and canned markets (Table 1, Equations 15-17), price and quan- tity changes on these outlets first show up in 1980. Finally, because of a 4-year lag between changes in grape prices and bearing acreages (Exhibit 1A, Equations 20-22), increased imports in 1978, which first decreased grape prices in 1979, did not reduce total grape supplies until 1983. As indicated previously, effects in other years reflect the com- bined influence of imports in the current year and increased imports from previous years. For example, increased wine inventories of 5.974 million gallons in 1979 is the net effect of increased wine inventories of 6.806 million gallons in 1978, reduced wine shipments of 8.836 million gallons in 1979, and reduced wine production of 9.668 million gallons in 1979. This decrease in wine production resulted from a combination of decreased wine shipments, increased 12 inventories, and decreased grape prices from the increase in imports in 1978 (Exhibit lA, Equation 5). The increased responsiveness of wine production over time to changes in imports explains why the impact on inventories diminishes and eventually becomes negative in 1983. The fact that increased wine imports both reduce wine shipments and increase wine inventories is the key to understanding why crush grape prices are affected so drastically by increased wine imports (Table 4). Decreased wine shipments in the previous year (through increased wine imports) simultaneously reduce current crushing requirements through increased inventories, and reduce crush demand for future use through a decrease in expectations of future wine demand. This combination of reduced demand for current and future crushing requirements is reflected in the large and significant coef- ficients for lagged wine shipments and lagged wine inventories in the crush price (Exhibit lA, Equations 6-8). The market for raisins is adversely affected by wine imports as well since raisins are the main alternative outlet for raisin-type grapes. The results indicate that increased imports lead to some shift in utilization of grapes to the fresh and canned markets and, therefore, reduce price on these outlets. But these effects are small in comparison to the impact of imports on the crush and dried markets. Thus, in addition to strong effects of wine imports on crush demand for grapes, limited supply flexibility between crush/dry use and fresh/canned use is another factor contributing to large price reductions in the crush and dried markets. The final contrib- uting factor to large price reductions in response to increased wine imports is the highly inelastic nature of acreage response for grapes, which results from a combination of biological lags and lagged price expectations. The amounts by which increased wine imports have reduced average and total returns of grape producers are shown in Table 5 and 6. These effects were computed in the same manner as those shown in Table 3. Average returns for each grape type are total returns from sales on all outlets divided by total utilized production (Exhibit lA, Equations 34-36). These reduced returns were converted to 1984 real dollars by multiplying each value by 3.lll, which is the amount by which the general price level has risen since 1967 as shown by the consumer price index. Also, after the study was initiated, grape data for 1984 became available, allowing estimated reduced returns for 1984 to be included in these tables. The results in Tables 5 and 6 are broadly consistent with the findings thus far. Increased imports have the greatest impact on wine type grapes, as expected. Also, reduced returns for raisin-type grapes have shown a tendency to widen over time as utilization has shifted from crushing to drying and eventually to other market out- lets. Reduced average returns in 1979 and 1980, attributable to increased imports in l978 and 1979, had small effects on total grape supplies and, therefore, price reductions in 1983 and 1984. Reduced 13 grape supplies in 1983 from reduced grape prices in 1979 are shown in Table 3. Reduced grape supplies in 1984, attributable to reduced grape prices in 1979 and 1980, were estimated as 6,000, 1,000, and 42,000 tons for raisin, table, and wine type grapes, respectively. Total producer revenue reduction attributable to increased wine imports is substantial. For all three grape types the estimated loss in total revenue for the 6 years is $1.1 billion from an increase in total wine imports of 54.7 million gallons from 1978 through 1983. In 1984 alone, increased imports reduced total producer returns by an estimated $204.4 million. This was more than 27% of 1984 actual total revenue of $751.6 million. CONCLUSIONS Increased wine imports since 1978 have had significant impacts on segments of the U.S. wine industry. Increased wine imports have sharply reduced the growth in domestic wine sales, and reduced aver- age and total returns to grape producers. The effects have been greatest in the crush and dried markets for grapes. This is because. increased wine imports simultaneously reduce current and future crushing requirements through a decrease in expectations of future wine demand. In turn, reduced producer prices for crushing have resulted in a shift in supply of raisin-type grapes mainly to rai- sins, which has led to price decreases on this outlet as well. With long lags in supply response of grapes, producers can be expected to experience continued price declines into the future if wine imports continue to grow at the present rate. The model presented in this report can be used to quantify effects of changes in other variables of the model related to demand for wine, costs of producing wine, and supply response of grapes. For example, the model could easily be adapted to simulate the impact of increased grape supplies resulting from increased plantings in other regions of the country. The import case is especially inter- esting, however, because of current concern about the effects of imports on the economic welfare of segments of the U.S. wine indus- try. 14 EXHIBIT 1. ESTIMATED EXONOMETRIC MODEL OF U.S. WINE INDUSTRY A. Behavioral Equations Total Demand for Wine (OLS, 1963-1982) 1n(SWT/POP)t = 1.800 + 0.123 01 - 0.510 lnDPWt (0.750) (0.035) (0.157) +O.404 1nPDYt + 0.556 1n(SWT/P0P)t_1, (0.209) (0.095) R2 = 0.99, DW = 2.57, DH = -1.41. Ratio of Domestic to Imported Wine (GLS, 1964-1982) 1n(SWD/MW)t = 0.918 + 0.332 02 - 1.184 1n(DPWD/DPMW)t (0.370) (0.128) (0.334) +0.s80 1n(SWD/MW)t_1 (0.173) ' R2 = 0.89, r = 0.520. (0.201) California Share of Domestic Wine (OLS, 1947-1982) (sw/sw0)t = 0.793 + 0.168x10'2 T + 0.028x10'2 (T-22)2, (0.008) (0.o29x10'2) (0.002x10’2) R2 = 0.80, nw = 1.57. Domestic Wine Price (GLS, 1949-1982) nvwnt = 89.948 + 0.119 [.5(DPGCt_l + DPGCt_2)] (18.850) (0.058) +O.235 DIBCt - 1.048 T, (0.185) (0.247) R2 = 0.39, r = 0.798. (0.102) 10. 15 Wine Production (2SLS, 1950-1982) QWt = -6.476 - 1.165 npsct + 2.486 swt_1 (ll.239) (0.364) (0.402) (0.246) (1.011) Crush Demand for Raisin Type Grapes (3SLS, 1950-1983) DPRGCt = 58.680 - 0.045 QRGCt — 0.033 QTGCt + 0.008 QWGCt (9.262) (0.012) (0.009) (0.010) +00809 _ "' To (0.145) (0.107) (0.546) Crush Demand for Table Type Grapes (3SLS, 1950-1983) npwsct = 44.179 - 0.033 QRGC - 0.042 QTGCt + 0.001 QWGCt (7.370) (0.009) (0.010) (0.008) +0¢742 - - To (0.122) (0.084) (0.447) Crush Demand for Wine Type Grapes (3SLS, 1950-1983) npwcct = -1.906 + 0.008 QRGC + 0.001 QTGC - 0.089 QWGCt (10.151) (0.010) (0.008) (0.017) +1.43? swt_l - 0.501 1wt_1 - 1.129 T. (0.222) (0.138) (0.804) Farm Level Demand for Raisins (ZSLS, 1950-1983) DPRGD = 50.454 - 10.550 PQRNSt + 20.456 PDYt (l5.143) (1.853) (6.537) +0.s23 0Pn00t_1. (0.124) Inventory Demand for Raisins (2SLS, 1950-1983) PIRNSt = -0.604 + 0.220 PQRNSt (0.875) (0.121) -0.010 DPRGD + 0.017 (DPRGDt (0.006) (0.007)) (0.015) 16 ll. Crush Allocation of Raisin Type Grapes (ZSLS, 1950-1983) (QRGC/NQRG)t = 0.314 + 0.329x10‘2 npncct (0.070) (0.145x10'2) -0.111x10'2 DPRGDt_l + 0.0s5x10'2 T. (0.0a4x10'2> (0.20ex10'2> 12. Fresh Market Demand for Raisin Type Grapes (3SLS, 1948-1983) 1nDPRGFt = 4.113 - 0.947 ln(QRGF/POP)t (0.306) (0.153) -0.353 1n(QTGF/POP)t + 0.840 1nPDYt (0.196) (0.315) 13. Fresh Market Demand for Table Type Grapes (3SLS, 1948-1983) lnDPTGFt = 4.059 - 0.285 1n(QRGF/POP)t (0.378) (0.158) -0.847 1n(QTGF/POP)t + 0.983 lnPDYt. (0.248) (0.387) 14. Canned Market Demand for Raisin Type Grapes (ZSLS, 1948-1983) 1nDPRGCAt = 4.975 - 0.363 ln(QRGCA/POP)t + 0.523 1nPDYt (0.464) (0.167) (0.165) 15. Fresh Allocation of Raisin Type Grapes (OLS, 1948-1983) (QRGF/QRG)t = 0.009 - 0.033x10'2 DPRGDt_1 (0.020) <0.011x10'2> +0.3s1 (QRGF/QRG)t_l, (0.142) R2 = 0.44, nw = 1.95. 16. Fresh Allocation of Table Type Grapes (ZSLS, 1948-1983) (QTGF/QTG)t = 0.458 + 0.0s0x10‘2 DPTGFt (0.036) (0.031x10'2> -0.0s2x10‘2 DPTGCt_l - 0.195x10'2 T. (0.095x1o'2> (0.195x10'2> 17 17. Canned Allocation of Raisin Type Grapes (ZSLS, 1948-1983) (QRGCA/QRG)t = 0.662x10'2 + 0.021x10‘2 DPRGCAt (0.500x10'2) <0.009x1o'2) -0.036x10'2 DPRGCt_l + 0.603 (QRGCA/QRG)t_1 <0.009x10'2 (0.138) 18. Fresh Market Price for Wine Type Grapes (OLS, 1947-1983) npwsvt = -19.348 + 1.568 npwcct - 0.651 T, (7.310) (0.109) (0.295) R2 = 0.88, nw = 1.43 19. Crush Allocation of Wine Type Grapes (OLS, 1947-1983) QWGCt = —72.499 + 0.952 QWGt + 2.427 T, (3.651) (0.006) (0.283) R2 = 0.99, nw = 1.68. 20. Bearing Acreage of Raisin Type Grapes (GLS, 1954-1982) (ARGt - ARGt_l) = 0.866 + 0.135 (DPRGt_4 - DPRGt_5) (1.963) (0.058) (0.067) +0.004 (DPRGt_6 - DPRGt_7), (0.056) R2 = 0.20, r = 0.550. (0.155) 18 21. Bearing Acreage of Table Type Grapes (GLS, 1954-1982) (ATGt _ ATGt_l) = _Ou667 + 0.015 (DPTGt_4 - DPTGt_5) (0.639) (0.015) +0.035 (DPTGt_5 - DPTGt_6) (0.018) -0.005 (DPTGt_6 - DPTGt_7)| (0.015) R2 = 0.28, r = 0.453. (0.166) 22. Bearing Acreage of Wine Type Grapes (GLS, 1953-1982) AWGt = —l7036l + 00456 DPWGt_4 + 00180 DPWGt_5 (l8.l03) (0.167) (0.172) +0.185 DPWGt_6 + 0.521 AWGt_1 + 1.930 T, (0.146) (0.168) (1.268) R2 = 0.88, r = 0.544. (0.159) 23. Yield of Raisin Type Grapes (OLS, 1950-1982) YRGt = 6.904 + 0.019 (ARGt - ARGt_3) + 0.058 T, (0.496) (0.017) (0.023) R2 = 0.24, nw = 2.77. 24. Yield of Table Type Grapes (OLS, 1950-1982) YTGt = 6.783 + 0.054 (ATGt - ATGt_3) + 0.006 T, (0.477) (0.041) (0.021) R2 = 0.06, 0w = 2.41. 25. Yield of Wine Type Grapes (OLS, 1950-1982) YWGt = 30273 _ 00008 (AWGt _ AWGt_3) + 00098 T; (0.259) (0.003) (0.013) R2 = 0.67, nw = 2.11. 26. 27. 28. 29C 30. 31. 32. 33C 34. 35. 36. 37. 38. 39. 40. 41. 19 B. IDENTITIES 1nSWTt 0.8 1nSWDt + 0.2 1nMWt, 1nDPWt 0.8 lnDPWDt + 0.2 1nDPMWt, + DPWGCt ° QWGCt)/(QRGCt + QTGCt + QWGCt), = + - + DPRGCAt - QRGCAt)/(QRG¢t + QRGDt + QRGFt + QRGCAt), + DPRGFt - QRGFt DPTGt = (DPTGCt ° QTGCt + DPTGFt ' QTGFt)/(QTGCt + QTGFt), QRGt = QRGCt + QRGDt + QRGFt + QRGCAt, QTGt = QTGCt + QTGFt, QWGt = QWGCt + QWGFt PQRNSt (QRGDt + IRNSt_1)/POPt, PIRNSt (IRNS/POPt). 2O C. EN DOGEN OUS VARIABLES“ SWD = quantity of U.S. produced wine sold in the U.S. (mil. gal.), SW = quantity of California produced wine sold in all markets (mil. gal.), MW = quantity of imported wine sold in the U.S. (mil. gal.), DPWDb = deflated index of wholesale prices of domestic wines (1967 = 100)| QW = production of California wine (mil. gal.), IW = June 30 inventories of California produced wine (mil. gal.), DPRGC = deflated crush price of raisin type grapes ($/ton), DPTGC = deflated crush price of table type grapes ($/ton), DPWGC = deflated crush price of wine type grapes ($/ton), DPRGD = deflated dried price of raisin type grapes, fresh basis~ ($/ton), QRGC = quantity of raisin type grapes crushed (1000 tons), QTGC = quantity of table type grapes crushed (1000 tons), QWGC = quantity of wine type grapes crushed (1000 tons), QRGD = quantity of raisin type grapes dried, fresh basis (1000 tons), DPRGF = deflated fresh price of raisin type grapes ($/ton), DPTGF = deflated fresh price of table type grapes ($/ton), DPWGF = deflated fresh price of wine type grapes ($/ton), DPRGCA = deflated canned price of raisin type grapes ($/ton), QRGF = quantity of raisin type grapes sold for fresh use (1000 tons), QTGF = quantity of table type grapes sold for fresh use (1000 tons), QWGF = quantity of wine type grapes sold for fresh use (1000 tons), QRGCA = quantity of raisin type grapes canned (1000 tons), s?» cw bearing acreage of raisin type grapes (1000 acres), ATG bearing acreage of table type grapes (1000 acres), AWG YRG YTG YWG QRG QTG QWG IRNSC 21 bearing acreage of wine type grapes (1000 acres), yield of raisin type grapes (tons/acre), yield of table type grapes (tons/acre), yield of wine type grapes (tons/acre), production of raisin type grapes (1000 tons), production of table type grapes (1000 tons), production of wine type grapes (1000 tons), = August 31 inventories of raisins, fresh basis (1000 tons). 22 D. EXOGENOUS VARIABLES” PDY = per capita deflated total personal consumption expenditure ($1000/person), } DPMW = deflated wholesale price of imported table wine under ' $4/gal. (1967 = 100), POP = July 1 civilian population (mil.), D1 = zero—one dummy variable (D1 = 0 prior to 1970; D1 = 1 from 1970), D2 = zero-one dumy variable (D2 = O prior to 1981; D2 = 1 from 1981), T = linear time trend (1947 = 1, 1948 = 2, etc.), DIBC = deflated index of wholesale bottle prices (1967 = 100). aAl1 variables except DPWD, PDY, DPMW, and DIBC on crop year basis, July 1 through June 30; other variables on calendar year basis. bvariable DPWD constructed as geometric weighted average of deflated indexes for wholesale table wine and dessert wine prices; i.e., 1nDPWD = 0.6 lnDPTW + 0.4 lnDPDW, where DPTW is deflated index of wholesale prices of table wine and DPDW is deflated index of wholesale prices of dessert wine; weights are sample mean relative expenditure shares of table and dessert wine. °For 1962 and beyond, free plus reserve tonnage as designated by Federal Marketing Order for raisins; prior to 1962, packers’ stocks only. 23 Table 1. Summary of Goodness of Fit Statistics for Endogenous Variables of the Model -- Simulation, 1963-83 Endogenous variable Mean RMSE RMS%E Ud U1 SWD 280.351 8.856 0.029 0.98 0.0001 MW 59.123 4.981 0.064 1.00 0.0010 SW 240.426 8.289 0.031 0.93 0.0001 DPWD 95.72 2.44 0.026 1.00 0.0003 QW 258.588 20.684 0.079 0.96 0.0003 IW 260.615 19.678 0.079 0.95 0.0003 DPRGC 49 11 0.232 0.74 0.0042 DPTGC 48 10 0.226 0.84 0.0042 DPWGC 88 19 0.212 0.89 0.0022 DPRGD 81 16 0.208 0.96 0.0025 QRGC 762 185 0.317 0.99 0.0003 QTGC 258 66 0.272 0.89 0.0008 QWGC 1101 127 0.155 0.97 0.0001 QRGD 1095 221 0.311 0.95 0.0002 IRNS 280 115 1.885 0.67 0.0021 DPRGF 164 42 0.288 0.95 0.0016 DPTGF 165 44 0.348 0.67 0.0018 DPRQCA 76 12 0.160 0.70 0.0019 QRG8 200 38 0.210 0.78 0.0010 QTGF 221 41 0.197 0.73 0.0008 QRGCA 54 9 0.195 0.58 0.0035 Table 1 cont. 24 Endogenous Variable Mean RMSE RMS%E Ud U1 DPWGF 103 34 0.431 0.84 0.0033 QWGF 63 12 0.201 0.98 0.0028 ARG 245.775 3.775 0.015 1.00 0.0001 ATG 70.664 1.789 0.025 0.99 0.0004 AWG 195.504 9.134 0.055 0.79 0.0003 YRG 8.528 1.263 0.169 0.72 0.0175 YTG 6.826 1.066 0.168 0.91 0.0233 YWG 5.663 0.658 0.125 0.99 0.0206 QRG 2111 312 0.170 0.71 0.0001 QTG 483 79 0.179 0.89 0.0003 QWG 1165 129 0.152 0.97 0.0001 25 Table 2. Historically Simulated Values for Selected Endogenous Variables of the Model, 1978-1983“ Endogenous Year Variable 1978 1979 1980 1981 1982 1983 swn 353.341 368.678 371.589 387.214 398.550 421.016 1w 349.927 379.752 426.121 431.469 468.358 468.583 npwn 89.35 88.94 87.81 87.02 86.44 84.12 npncc 65 53 45 35 34 26 nvrcc 63 54 48 40 38 35 npwcc 86 82 74 73 65 88 npncn 117 102 81 90 72 69 npncr 251 191 156 213 136 152 09165 223 204 186 205 149 173 QRGC 647 789 890 564 800 713 QTGC 199 219 227 218 319 269 QRGD 834 1287 1510 998 1496 1358 IRNS 257 292 382 398 457 518 once 143 191 232 168 249 241 QTGF 194 198 201 202 263 235 QRGb 1670 2320 2692 1779 2624 2391 QTGb 393 417 428 420 582 504 QWGb 1706 1821 2004 1794 2152 1880 aDynamic simulation beginning in 1978 using actual values for exogenous variables including actual import volumes of 93.465, 95.865, 107.841, 118.656, 128.279, and 133.065 million gallons for 1978-1983, respectively. bActual utilized production. 26 Table 3. Cumulative Impact of Increased Wine Imports on Selected Endogenous Variables of the Model, 1978 - 83“ Endogenous Year ». Variable 1918 1919 1980 1981 1982 1983: swn -1.181 -10.011 —13.481 -15.829 -15.842 -15.012 1w 8.806 5.914 5.235 4.301 0.601 -4.349 npwn 0 0 -0.65 -1.38 -1.11 -2.05 nvnccb 0 -1 -8 -10 -10 -9 npwccb 0 -1 -1 -10 -10 -9 npwecb 0 -13 -16 -20 -24 -19 npncnb 0 -3 -4 -5 -6 -6 npncrb 0 0 -2 -4 -3 -s nprcrb 0 0 -2 -2 -2 -3 QRGC 0 -so -51 -s1 -1s -59 QTGC 0 0 -1 -1 -3 -3 QRGD 0 49 so 42 51 38 IRNS 0 29 14 20 25 21 once 0 0 2 2 s 8 QTGF 0 0 1 1 3 3 one 0 0 0 0 0 -3 QTG 0 0 0 0 0 0° owe 0 0 0 0 0 -24 aEach entry in the table shows the cumulative effect attributable to an increase in imports beginning in 1978. bPrice effects rounded to the nearest dollar. These price impacts are in 1967 dollars; multiply by 3 to find impacts in current dollars. chess than 500 tons. 27 Table 4. Cumulative Impact of Increased Wine Imports on Selected Endogenous Variab)les, 1978 - 83“ (Expressed as Percentage Changes from Historically Simulated Values Endogenous Year Variable 1978 1979 1980 1981 1982 1983 SWD -2.20 -2.72 -3.63 -4.09 -3.97 -3.57 IW 1.94 1.44 1.23 1.00 0.13 -0.93 DPWD 0 0 -0.73 -1.59 -1.99 -2.44 DPRGC 0 -13.71 -17.21 -30.28 -31.52 -36.30 DPTGC 0 -12.31 -14.99 -24.12 -26.23 -25.54 DPWGC 0 -16.60 -21.36 -27.86 -35.52 -22.18 DPRGD 0 -2.29 -4.83 -5.21 -8.35 -8.60 DPRGF 0 0 -1.10 -1.85 -2.57 -2.90 DPTGF 0 0 -0.84 -1.05 -1.54 -1.55 QRGC 0 -6.29 -6.48 -8.91 -9.33 -8.25 QTGC 0 0 -0.60 -0.61 -0.84 -0.93 QRGD 0 3.86 3.34 4.23 3.84 2.72 IRNS 0 2.40 3.95 4.97 5.49 5.21 QRGF 0 0 0.90 1.68 2.27 2.64 QTGF O 0 0.68 0.66 1.02 0.91 QRG 0 0 0 0 0 -0.14 QTG 0 0 0 0 0 -0.08 QWG 0 0 0 0 0 -1.26 Note: Each entry in the table shows the cumulative effect from Table 3 expressed a a percentage of the corresponding entry in Table 2. Percentages computed directly from Tables 2 and 3 may not be exactly equal to values in Table 4 due to rounding. 28 Table 5. Impact of Increase in Total Wine Imports, 1978-83, on Average Producer Returns for 1979-84 (Expressed in 1984 Real Dollars) Reduction in Average Returns (S/ton) Grape Type 1979 1980 1981 1982 1983 1984 Raisin 9 13 1s 21 21 21 Table 11 A 12 18 21 21 18 Wine 45 49 e9 80 79 s8 Table 6. Impact of Increase in Total Wine Imports, 1978-83, on Total Producer Returns for 1979-84 (Expressed in 1984 Real Dollars) Reduction in Total Returns ($1000) Grape Type 1979 1980 1981 1982 1983 1984 Raisin 20,880 34,996 28,464 55,104 50,880 48,291 Table 4,587 5,136 7,560 12,222 10,584 8,757 Wine 81,945 98,196 148,488 172,160 157,345 147,336 Total 107,412 138,328 184,512 239,486 218,809 204,384 29 Aooriwm: Aim .55 mmEB owmmEou *0 9.2a 2320c; *0 x005 UQEEQ .v 050E 39:9: =m c. 20m mc§v8o=uoa EcLQEmO *0 5.23:0 d 050E _mo> mm mu m» P» no mm .mm> mm on m~ wn Nw mm _ _ _ _ _ _ _ _ _ _ _ _ -Qm@ 1om_ |Qom 1oom |Qmm 10mm 5.2: loom r¢@Q_ Iomm a>>¢n_ >>w 2mm 4.2V w: 0E c. 20w ms; @269... *0 2.3.80 N wSQE 2mm .=Ev w: o5 c. Bow m2? vmoixza w: *0 2:580 ._. 950E .mw> mm m» mm F» Nw no .mm> mm mm wk P» Nw no _ _ _ _ _ _ 1o _ _ _ _ _ _1ooP \Iomr low Ioom vow Iomm loom -o~P Iomm woo? Ioov >><¢ 0>>m 3O Eoiwv wmQQm 8a wBS B $25 zwEo umflmcwo a wSQE _mw> mm 2 m» C. B 8 B _ _ _ _ _ ION 18 00KB 2% .25: m5; umoau9q mEBEmo B wm:9co>c_ om 93w .w mSQE Bm> mm 2 m» K 8 no A _ _ _ * _1OOP -oom |oom -¢Q< 5% >>_ .‘ E039 8.35 mar c_w_m. B wocq cmEo umfimzma K wSmE hmo> mm 2 m“ P» B 8 _ ~ _ _ _ _ IO oomnfi 2mm .__EV mczs mEBEmO B cozoso9m .m 9.6K B»; 3 2 m» K B mm _ _ _ _ _ _1ONP 58. -08 5% a? 2,0 31 3:9 ooo: umcwio woaflm mob wES B 2.28:6 or wSmE 8:9 ooo: @9620 mmafim 8a E29 “o bzcmzO .3 @591 .@@> no m~ m» PN ~@ no .@m> mm @~ m» P“ No no _ _ _ _ _ _ mm? _ _ _ _ _ _ loom loom loom [m- loom -omm Imwv »oo~P 06.5 00x0 Eoté 29E 59o amesm Q2 52m. 6 wsa v26 ufléwo .9 mama E95 wmama 8a m2; B 3.5 =86 uséwo .@ mama .@@> ow @¢ m» P» No no .m¢> mm on m» P» No no _ _ [ow on low |¢oP low .o¢F ,émF low? loom . JHJ>>|=J 1-1 0_< .41 00v 4:1. 32 E038 wmaflm 00b E29 *0 00:0 :00: 000230 6F 050E 3:9 coo: Emma cmw: .0529 00 woc9cm>c_ _.m 5:93 .m_. 050E 0m0> mm mm m» P» >0 mm _mw> mm mm mm Fm hm no _ _ _ _ _ _ [Cm _ _ _ _ _ _ IQ |¢¢P -¢¢~ loow ram? Iaow -¢¢@ |Qm~ ,oow mU¢&Q w2I_ E9 o8: wimp :00: 00:0 woafim 00b E29 00 a=¢m=0 .3 050E 3:2 coo: 00530 8.6a 00B 0E; 00 bzcmzO d? 059m 0mw> mm 0» 0» Fm >0 mm 0m®> mm 0» m» 0» >0 mm _ _ _ _ _ _ _ _ _ _ _ L oom ,¢¢@ loom IOOQ 100?? ;¢¢¢ 100v? 1o0~_ I00»? |OQmw looow 100w? Qomo 0030 33 HNQ> m» 3 F» B Aw¢o~¢¢o_V om: zwm: .2 20w 8.36 8a B oES B b.2230 dm wSQE ._Q0> _ _ -¢@ $9 6E Iovw $8 loom "65 Acoiwv woQSm 8a E29 B oocq uoccmo umfimzwa .2 059m m» m» I B _ \\l||||Il I¢@ rom $9 $2 .8 Aw¢o.o¢¢_V mm: 5m: .2 Bow mmafim 8a E29 B 8.28:0 .2 oSmE E 5 8 _ _ .~@> mm Eoiwv mmQSm .25 Q59 B 8.5 cmm: oofimcwo .5 QSmE E Q» I. E 10m? row w -¢<~ loom tome _ _ um» $9 nmmw 18m u_@._.n_Q 34 $98 8o: mmamawga s29 6 mmmwam 958m em Q59» $59 coo: mm: =8: § 28 wmaflm mar 0E3 B >=E~=o .8 mSQE hmm> mm E mm I h... 8 50> mm m» m» K 5 mm _ _ _ _ _ _ vmm _ _ _ _ _ _ .. 8 13w Iwmm 15m 1o: 10mm om< m0>>O Eoiwv wmamhm mqb oczs E 3.5 5mm: voymzoo mm mSmE A29 coo: nmccmo £69m wqb c_w_9 6 bzcmzfi .3 wSmE Ew> m_w m» E P» .8 mm bmo> mm 2 m» K 5 no _ _ _ _ _ _ _ _ _ _ om loo P 18F loom 10mm 105E dw 93ml Afiomhcos mmaSm .25 E29 3 22> sm 059... io> 8 2 m» K .5 mm 80> mm 2 m» K R. 8 _ _ _ _ _ _ . _ _ _ _ _ _ {QM um w -3 Iod -3 -3 um.» rod -2: A3 1o. Z 0.; om> 690m coo: wmQmB .25 m5; 3 ommmhom mccmom 6w mSQE 320m ooo 5 mwaflm 8a mES “o mmmmzom mccmmm mm 23mm 23> mm m» m» I 8 mm 89/ mm 2 m» E. 5 8 _ . IONP _ _ _ _ _ _ IOQ r03 $9 ... -8 -2: room Ion Iomm EX. LR $8 roww -8 $8 x .. -8 o>>< o? %. I C ,0 ‘ 36 3:9 ooo C mmamqm ma? 0E3 “o Eéguoi mm 930E 3:2 ooo 5 836 0gb oES 3 co=o:uo.n_ 9m mSQm .~@> mw @¢ mw PA Kw mm .m@> mm mh m» Fm “Q mm _ _ _ _ _ _ _ _ 1¢o@ Ioov Iowm »Q¢w :Q~¢ IOONF 1o? 1¢vm loom? |QQ@ IOOON lO©® O>>O 0P0 3:9 ooo: woQEm wab E29 B cozosu9m .8 mSmE $90596: £92m wwqb m5; B 29> ww mini Lm®> mm an mm PM Rm mm hm®> mm on mm P» Rm mm _ _ _ _ _ _ oomw _ _ _ _ _ _ ram loom? Iooww IOOPN Icovm IOQNN omnv U>>> 37 ENDNOTES 1The index numbers used are geometric weighted means, where the weights are approximately the sample mean averages of relative shares of individual components in the index. For example, the index for domestic wine prices is a geometric mean of table wine and dessert wine prices with weights of 0.6 and 0.4, respectively (Exhibit l, Footnote b). Other geometric indexes were constructed for total wine consumption and domestic and imported wine prices (Exhibit l, Equa- tions 26 and 27). This geometric index has been referred to by Star and Hall as an "Approximate Divisia Index." See their article for a discussion of the desirable theoretical properties of this index. 2For a discussion of the two—stage budgeting procedure in the context of consumer demand, see Barten. 3For example, 1983 total U.S. wine imports of 133.1 million gallons were only 3% of production and 5% of consumption for the two major exporting countries, France and Italy. 4The estimation methods employed were: ordinary least squares (OLS), generalized least squares (GLS), two—stage least squares (ZSLS), and three-stage least squares (3SLS). The GLS method was used when cor- recting for first-order serial correlation. In equations containing lagged dependent variables, the instrumental variable estimator described by Fuller (pp. 429-446) was used when correcting for serial correlation. The values for r reported in the Exhibit are estimated values for first-order serial correlation. The statistics DW and DH are the Durbin-Watson and Durbin-H statistics, respectively. 5The import wine price used is unit import value for table wines under $4 per gallon. This category is relatively homogenous and accounts for the bulk of the imports entering the U.S. Experimenta- tion with unit value measures for total imported table wine and all imported wine showed that the estimates would not be drastically altered. However, the value for R2 of the import share equation was substantially reduced when either one of the alternative unit value measures was used. This indicates a clear preference for unit import value for table wines under $4 per gallon. 6Acreage response was treated in this manner because bearing acreages for raisin and table type grapes are nonstationary, which causes dynamic simulated values to diverge from actual values (Chow, 1975). 38 REFERENCES Barten, A.P. "The Systems of Consumer Demand Functions Approach: A Review." Econometrica 45(l977):23-51. Chow, G.C. Analysis and Control of Dynamic Economic Systems. New York: John Wiley & Sons, Inc., 1975. French, B.C., and J.L. Matthews. "A Supply Response Model for Pere- nial Crops." Amer. J. Agr. Econ. 53(l97l):478-90. Fuller, W.A. Introduction to Statistical Time Series. The New York: John Wiley & Sons, Inc., 1976. Houck, J.P. "A Look at Flexibilities and Elasticities." J. of Farm Econ. 48(l966):225—32. Johnson, P.R., T. Grennes, and M. Thursby. "Trade Models with Dif- ferentiated Products." Amer. J. Agr. Econ. 6l(l979):120-27. Kost, W.E. "Model Validation and the Net Trade Model." Agr. Econ. Res. 32(l980):l—l6. Star, S., and R.E. Hall, "An Approximate Divisia Index of Total Fac- tor Productivity." Econometrica 44(l976):257—63. Wohlgenant, M.K. "An Econometric Analysis of the Dynamics of Price Determination: A Study of the California Grape-Wine Industry." Ph.D. thesis, University of Caifornia at Davis, June 1978. Wohlgenant, M.K. "Inventory Adjustment and Dynamic Winery Behavior." Amer. J. Agr. Econ. 64(l982):222—3l. [Blank Page in Original Bulletin] ' -'- ‘ u Iu-o [Blank Page in Original Bulletin] ‘ a- , v.- w.» [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 Agricultural Experiment Station 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, color, religion, sex, age, handicap, or national origin. 2M—9-85