November 1998 SCSB# 390
Chapter 3 Broiler Exports: A Structural Time Series Approach Crispin Mutshipayi Kapombe AbstractThe goal of this research paper is to provide information that can increase U.S. exports of poultry products. The analysis focuses on individual market responsiveness to U.S. export price, the impact of currency realignments on imports from the United States (e.g., exchange rate impacts), per capita income of importing countries, the price of substitutes, and different government interventions in the poultry market (e.g., import quotas, etc.). The analysis indicates that in addition to these explanatory variables that a trend component has been vital in the expansion of the broiler industry during the study period. The results suggest that the United States can increase its broiler exports more effectively by (1) extending efforts on international macroeconomic policy coordination rather than depending on domestic sectoral policies, and (2) working toward elimination of trade distortion practices through the North American Free Trade Agreement (NAFTA), the General Agreement on Tariffs and Trade (GATT), and bilateral trade negotiations. Given these results, the advantages of a structural time series model (STSM), formulated directly in terms of unobserved components such as trends and seasonals or cycles, make it an attractive methodology in the context of modeling broiler exports. IntroductionNAFTA, GATT and similar international trade pacts, trading blocks, and economic integration schemes are altering the international trade environment. Thus, information on trade and the impacts of the changes on U.S. agricultural trade is essential to developing policies that will enable the country to maintain its competitive edge in exports of many agricultural products. While U.S. poultry product exports are relatively small, they have been growing and in 1995 were about 4 percent of the total value of agricultural exports; only 15 percent of poultry were exported in 1995 (U.S. Agricultural Trade Update, ERS, USDA, March 4, 1996). International trade in poultry products has been increasing slowly and has the potential for adding substantially to U.S. agricultural exports. The attainment of the full potential, however, requires a more thorough understanding of international markets for poultry, a determination of the factors affecting trade, and an assessment of the competitive position of the U.S. poultry industry vis-à-vis other actual and potential exporters as well as the policies and programs of actual importers of poultry products. Exports of broiler meat have become important to the broiler industry. From 1979 to 1989, they were equal to 3.7 percent and 4.7 percent of total production, respectively. During the 1990s, exports have been increasing as a percent of production, from 7.1 percent in 1992 to 15.2 percent in 1995 (USDA, Poultry Outlook, Feb. 16, 1996). Export growth has been recorded in neighboring countries as well as in many smaller countries. Implementation of the Export Enhancement Program (EEP) in 1986 and a lower valued dollar have contributed to increased broiler exports since 1985. Also, growth in the export market is fueled by large domestic supplies, a general easing of trade restrictions, attractive prices of U.S. broilers–especially for leg quarters and dark meat parts, and by development of foreign markets by individual companies and export promotion groups. In 1994, the major importing countries for U.S. poultry products were as follows: Russia (820 million pounds), Hong Kong (710 million pounds), Japan (250 million pounds), Mexico (220 million pounds), Poland (120 million pounds) Canada (79 million pounds), and China (72 million pounds) (USDA, Poultry Outlook, Feb. 28, 1995). Data for 1995 show broiler exports at 3.8 billion pounds, up almost 1.0 billion from 1994. Substantial benefits are expected to accrue to the U.S. poultry industry from expanding world poultry meat consumption. Low U.S. prices for poultry products will ensure a large market share, but exports of further-processed poultry meat are expected to remain relatively small. Future growth may be tempered by competition from China, Brazil, Thailand, and the European Union (EU), but U.S. sales to Japan, Hong Kong, Mexico, China, South Korea, and other Pacific countries are still expected to grow. Sales to Latin America, the Caribbean, and the Middle East also may expand, but export growth to the former Soviet Union (FSU), mainly Russia and Eastern Europe, may only continue for a few more years. Longer term, the countries of the FSU and Eastern Europe have the potential to produce all of their domestic requirements and possibly export a surplus from time to time (USDA, CED, World Baseline Projections, September 1994, pp. 25-31). The goal of the research on which this paper is based is to provide information that can increase U.S. exports of poultry products. To help achieve this goal the study has the following specific objectives: (1) determination of factors that affect poultry exports and (2) evaluation of import policies and restrictive practices of importers and potential importers of U.S. poultry products. The study identifies economic and policy variables entering into the import demand for U.S. poultry products by individual countries and estimation of the impacts of each variable using the Kalman filter method, so as to obtain more reliable estimates of the market relationships. Specifically, the analysis focuses on the individual markets' responsiveness to U.S. export price, the impact of currency realignments on their respective imports from the United States (e.g., exchange rate impacts), per capita income of importing countries, the price of substitutes, and different government interventions in the poultry market (e.g., import quotas, etc.). To achieve these objectives, an econometric model was conceptualized and assumptions made concerning the characteristics of the individual equations, the relationships between the equations within a sector, and the associations between model sectors. This model involves (1) specification of supply side equations of poultry exports, including production, inventories, domestic demand, export supply, and price linkage equations and (2) specification of total import demand equations in four major importing countries of U.S. poultry products (Japan, Hong Kong, Canada, and Mexico). It consists of the specification of U.S. poultry export models in the form of structural time series models (STSMs) that can be recast in state space systems and estimated by means of the Kalman filter method. The results permit the analysis of the impacts of selected exogenous factors on U.S. poultry exports. This publication covers the following: BackgroundStudies dealing with the broiler industry date from an analysis of the apparent seasonal production and consumption patterns of the broiler meat industry by Boutwell and Seale (1969) and Reece (1974). Roys and Johnson (1973), Malone and Reece (1976), and Chavas and Johnson (1981, 1982) estimated the impact of economic variables on production, consumption, and prices in the U.S. poultry and egg sector. Most of these studies have used annual data (Heien, 1976) or quarterly data (Chavas and Johnson, 1981, 1982; Roys and Johnson, 1973) and one (Malone and Reece, 1976) used monthly data. Several authors have confirmed the importance of the rational expectations hypothesis (REH) in poultry (broiler) supply response (Huntzinger, 1979; Goodwin and Sheffrin, 1982; Aradhyula and Holt, 1989). Goodwin and Sheffrin (1982, p. 660) have indicated that "the decision to supply broilers is, of course, made under uncertainty, and, in principle, other moments of the probability distribution of prices besides the mean could affect behavior." Some studies have analyzed poultry export models. Haley (1990) evaluated the export enhancement program (EEP) for poultry. This study attempted to construct a trade flow model. It emphasized the Armington (1969) variant of the constant elasticity net trade model. This model differentiates poultry products by countries of export and of import. Demand analysis in the importing countries is represented by a two-stage process: the quantity choice of consumers and then the choice among competing suppliers. This kind of Armington approach also had been applied earlier in the context of the Static World Policy Simulation (SWOPSIM) modeling framework by Dixit and Roningen (1986). The cross-country approach to trade flow analysis has also been applied by Alston and Scobie (1987) and Harling and Thompson (1985). Also, of related importance are the models of Martinez, et al. (1986), Leong and Elterich (1985), and Yasin (1992) which concentrate primarily on modeling poultry export and import demands. Background for export model specification can be found in the study by Bishop et al. (1990). That study evaluated markets for poultry in a variety of countries and showed how trade liberalization could lead to an expansion of these markets and probably decrease prices. Thus, efficient producing countries such as the United States, Thailand, and Brazil could increase exports to currently protected importing countries such as Japan, Canada, and members of the European Community. Other authors (Rausser and Cargill, 1970; Chavas, 1983) have built poultry models that allow coefficients to vary with time, so as to solve parametric variation problems. Rausser and Cargill analyzed broiler prices by applying a spectral analysis approach. They concluded that the continuous dynamic adjustment of the broiler industry to changing technology and growing markets cannot readily be represented by periodic cycles. They stated that the inherent cyclical tendencies of the broiler industry appear to be altered by such factors as improved technology, market growth, and vertical integration. Chavas (1983) analyzed structural changes in the demand for meat using a random coefficient model and based on the Kalman filter estimation technique. To estimate the variance of the random coefficients, the one-step-ahead prediction error approach was used. The study identified the structural changes that occurred for beef and poultry, but not pork, in the last part of the 1970s. The empirical results suggest that the price and income elasticities of beef have been decreasing in the last few years, while the income elasticity for poultry has been increasing (+ 0.012 in 1975 and + 0.275 in 1979). Poultry consumption appeared to be increasingly less affected by the pork market. These results, especially the rise in income elasticity, indicate strong long-term growth prospects for the poultry industry. The Structural Time Series ApproachStructural time series models (STS) have been proposed by Harvey and others in a number of papers and monographs (Harvey and Todd 1984; Harvey 1985; Harvey and Durbin 1986; Harvey and Souza 1987; Harvey and Koopman 1992; and Gonzalez and Moral 1995). A thorough discussion of the methodological and technical ideas underlying structural time series models is contained in the monographs by Harvey (1989, 1994). Structural time series models are models which are formulated directly in terms of components of interest (e.g., the trend, seasonal, cycle, and the residual-irregular-components). In a structural time series model the explanatory variables enter into the model side by side with the unobserved components. In the absence of these components, the model reverts to a regression. The stochastic formulation proposed for the trend component is more flexible (see Harvey, 1985, 1991, and 1994) since it allows the level, µt, and the slope, bt, to evolve over time:
where ht and zt are normally independent white noise processes with zero mean and variance s2h and s2z, respectively. This model is a local approximation to a linear trend; it collapses to a deterministic global trend when s2 h = s2z = 0. A stochastic trend component (µt) is included in each equation to capture the changes in variables such as productivity, technical progress, and structural change. Annual data (1970 - 1991) have been used for broiler exports, inventories, and imports demand functions. These functions have only the trend component included in their models specification. A stochastic seasonal component (gt) is added to the model to account for seasonal fluctuations in the demand and supply schedules due to changes in prices, tastes, preferences, weather, etc. The process generating the seasonal component is:
where wt is a normally distributed independent white noise process with zero mean and variance s2w, and s is the number of "seasons" in the year. Seasonality changes slowly by means of a mechanism that guarantees that the sum of the seasonal factors over any consecutive s time periods has an expected value of zero and a variance that remains constant over time. The smaller the variance, the more stable the component. The trend (µt) and seasonal (gt) components are included in quarterly models for broiler supply, demand, and price functions. In addition, a stochastic cyclical component (yt) is added to the model for the retail choice beef price to capture the presence of a cyclical pattern in the price response. Walters (1965), cited by Okeyere and Johnson (1987, p. 1457), argues that "the long production period gives rise to cycles that must be reflected in the models." The specification of a stochastic cycle is:
(3.3) where kt and kt* are normally distributed, mutually independent, white noise disturbances with the same variance, s2k, and 0 < or = r < or = 1. Here l can be thought of as the frequency of the cycle and r as the damping factor of the amplitude. If r is strictly less than one, the process generating yt is stationary. If the ris equal to unity and s2k is equal to zero then, of course, (3.3) collapses to a deterministic cycle: yt = acoslct + bsinlct. Also, a disturbance term (et) is added to each model specification. Economic Model SpecificationFollowing Chavas (1983), Harvey et al. (1986), Dorfman and Havenner (1991), and Knapp and Konyar (1991), this study consists of the evaluation of U.S. poultry exports in the form of structural time series models (STSMs) to be recast in state space systems and estimated by means of the Kalman filter method (see Harvey 1991, 1994). The state space approach emphasizes dynamic rather than static relationships between variables. The U.S. broiler model consists of three major components: (1) supply side equations (supply, demand, inventories, and export demand functions); (2) price linkages (prices of broilers, beef, pork, turkey, and real food expenditure functions); and (3) import side equations (import demand functions for Japan, Hong Kong, Canada, and Mexico). Data used for model estimation cover a more recent period, 1970 through 1991, than previous studies. Available data from January 1992 to December 1993 are used to check the forecasting performance of the model. Data are quarterly values for the following: U.S. broiler production, demand and prices, and annual values for U.S. broiler export supplies; inventories demand; and broiler imports demand for Japan, Hong Kong, Canada, and Mexico. All monetary values are deflated by the gross national product deflator of the United States or importing countries. All prices are deflated by the consumer price index of the United States or importing countries. All quantity variables in the demand equations (U.S. broiler demand and U.S. broiler import demands) are expressed in per capita. Table 3.1 identifies variables and Table 3.2 presents the specified broiler model equations and expected signs for each exogenous variable included in the model. See appendix A for the sources of data. Supply SectorThis sector consists of four equations: broiler supply, demand, export, and inventory equations. Following Labys and Pollack (1984, p. 53), Martinez et al. (1986) and Aradhyula and Holt (1989), the broiler supply function (QBP) (equation 4) is specified as a function of the changes in the lagged one quarter wholesale broiler price (WPBr), real price of broiler feed (PBF), the hatch of broilers in commercial hatcheries (HATCHt-1) and its four-quarter lag (QBPt-4). Because the biological production lag for broilers is approximately two months, it follows that current quarter production depends on the expectations formed by producers in the previous quarter. The only input price included is for feed, PBFt-1; it is determined as a weighted average of the prices of corn and soybean meal. As Rogers (1979) indicates, feed costs have historically accounted for 64 to 73 percent of total broiler production costs. Consequently, the feed cost variable, lagged one quarter, should reflect the more important changes in short-run production costs. Feed cost (PBF) and wholesale broiler price (WPB) lags assume a one-quarter delay in response to changing profitability. In the short-run, the HATCH variable reflects the dependency of broiler production on the number of broiler-type chicks available. Broiler producers also may not fully adjust production to a desired level during any given quarter because of capital constraints, adjustment costs, and fixed contract periods. Thus, a dependent variable with a one-quarter lag is included in the supply equation (see Chavas and Johnson, 1982; Aradhyula and Holt, 1989). Per capita broiler demand (PCQBD) function is given by equation (5) and specified as a function of the retail prices of beef (CRBP), pork (RPP), turkey (RPT), wholesale price of broilers (WPBr), real food expenditures (FEXP) and its one quarter lag (PCQBDt-1). The broiler demand formulation does not represent a consumer demand curve per se because the wholesale price of broilers has been used in place of the retail price. Price determination in the broiler market occurs at the wholesale level (see Chavas, 1982; and Aradhyula and Holt, 1989). Previous studies have established that pork, beef, and turkey are substitutes in consumption for broilers, while demand for all meats is sensitive to income (Aradhyula and Holt, 1989; Udah, 1990). Additionally, there is a seasonal pattern to consumer demand, peaking in the summer months and bottoming out in the winter (Goodwin and Sheffrin, 1982). The broiler inventories demand (QBID) function is given by equation (6) and specified as a function of broiler price obtained by exporters ('international price') (BIP), wholesale price of broilers (WPBr), the quantity of broilers produced (QBP) and demanded (PCQBD) and its one quarter lag (QBIDt-1). The estimated inventories demand equation is based on the hypothesis that inventories are inversely related to own-price and are consistent with the partial adjustment inventory model suggested by Labys (1973, pp. 70-71). Logically, as the own-price rises, the opportunity cost of carrying broilers out for a given period rises and the incentives for profit taking increase. The inverse is true as price falls. Following Chambers and Just (1981), Martinez et al. (1986), and Conway (1987), the broiler export supply (QBX) function is given by equation (7) and specified as a function of wholesale price (WPBr), real exchange rate (SDR), quantity of broiler exports from Brazil (BBEXP), gross domestic product for the world (WI), and its one quarter lag variable (QBXt-1). Broiler exports demand is expected to be negatively related to wholesale price, exchange rate, and to Brazilian broiler exports. Brazil is a major competitor of the United States in the world market. It is hypothesized that higher world income results in larger broiler purchases, and thus, a positive sign on WI variable is expected. The exchange rate between the dollar and the Japanese yen is used as a proxy for the exchange rate with other countries. It is hypothesized that an increase in the value of the dollar would make U.S. exports more expensive, reducing broiler demand abroad. Price Linkages SectorThe supply side equations form a system representing a U.S. meat demand model by introducing price linkage equations which connect broiler prices with beef, pork, turkey, and real food expenditures. The broiler feed price did not show any linkage with other prices and quantities in the model and was dropped from the price linkages sector. Equations (8), (9), (10), (11), and (12) specify the price linkage functions for broilers (8), retail choice beef (9), pork (10), turkey (11), and real food expenditures (12), respectively. Broiler Imports Demand SectorThis study estimates import demands for U.S. broilers in four major markets: Japan, Hong Kong, Canada, and Mexico. Four import demand functions are specified and estimated by the Kalman Filter technique based on the sample period 1970 to 1991. Following Masayoshi and Heady (1984), Leong and Elterich (1985), and Ortalo-Magne and Goodwin (1992), the import demand for U.S. broiler chicken in the ith country (BIDit) (I = 1, 2, 3, 4) is given by equation (13) and specified as a function of one quarter lagged dependent variable (BIDit-1), the real FOB prices of broiler (RFOPBrit), beef (RFOBIPit), and pork (RFOPOIP) imported by country i, the real per capita gross domestic product (RYEit) of the importing country, per capita production of broilers (PCQBPit) in the ith importing country, and the real exchange rate (RSDit) between the currency of the ith importing country and the U.S. dollar. Also, a binary variable (wt83) was included in the Mexican import demand equation to account for the persistent effect of keeping high trade tariff rate quotas on poultry imports following the liberalization that began after 1982. A positive relationship is expected between broiler import demand and FOB prices of beef and pork, the quantity of broilers demanded in the previous period (BIDt-1), and the importing country's per capita income (RYEt). As the price of a normal good increases, consumers are likely to buy less of it and more of its substitutes. Also, expenditures for a normal good are expected to increase with increases in income. A negative relationship is expected between broiler import demand and its own price (RFOPBr), the per capita production of broilers, and the exchange rate between currencies. The sign of the coefficient on the binary variable (wt83) is not predetermined. The STSM with explanatory variables (Table 3.2) is a generalization of the classical linear regression model. It can be observed that if s2h = s2z = s2w = s2k = 0, all the equations in Table 3.2 collapse to standard regression equations with a linear deterministic time trend, seasonal, and cyclical components in addition to the explanatory variables. The specification of broiler models in Table 3.2 assumes that the dependent variables are not stationary and that first differences and seasonal and cyclical differences should be taken to achieve stationary. Therefore, these models also include, as a particular case, the models with different dependent variables, see Gonzalez and Moral (1995), Harvey and Todd (1984). Estimation and Forecasting ResultsThis section presents (1) estimation and interpretation of results, and (2) forecasting results. Estimation of ResultsThe sample period chosen to estimate the broiler models (Table 3.2) was from 1970 to 1991 (1970.I - 1991.IV for quarterly data and 1970 - 1991 for annual data), while available data from 1992 to 1993 have been kept to check the forecasting performance of the model. The results of estimating broiler models (Table 3.2) with the STAMP 5.0 statistical package are presented in Table 3.3 and Table 3.4 (see appendix B for information on the Stamp software). Broiler Production Equation. Commercial broiler slaughter (RTC weight) is closely related to the hatch in the same quarter. A 1 percent increase in broiler hatch (HATCHt) would increase broiler supply by +.25 percent or .25 million pounds. The elasticity on HATCHt-1 was close to zero and not significant, as in previous studies (see Chavas and Johnson 1982; Aradhyula and Holt 1989). There is usually enough excess capacity maintained at the hatchery supply flock level to assure a large degree of production flexibility. It is possible that baby chicks hatched in the quarter, could be used in the production process to produce a 3.5 to 4 pound broiler in the same quarter. The broiler grower feed price variable, lagged one quarter (RPBFt-1), is negatively related to broiler supply. The supply elasticity with respect to feed price is -0.052 and well within the range of previously reported estimates (e.g., Aradhyula and Holt 1989; Goodwin and Sheffrin 1982; Chavas and Johnson 1982). A 1 percent increase in the broiler grower feed price would decrease broiler supply by 52,000 pounds of broiler meat. Also, changes in productivity and technology, represented by a stochastic trend component, are positively related to broiler supply. The trend variable estimates, level and slope combined, indicate that broiler production could be cut by approximately 6.68 million pounds and still maintain a constant level of supply over time. The supply elasticity with respect to the trend variable is +6.67 for the level and +0.01 for the slope, well above the range of previously reported estimates when the trend is deterministic (e.g., Goodwin and Steven 1982; Chavas and Johnson 1982). Technical advancements are evident from lower input requirements for poultry production. A ton of feed now produces 37 percent more broilers and 54 percent more turkey than in 1955. Labor input requirements per pound of broiler and turkey production are 2.6 percent and 3.8 percent, respectively, of 1945-49 requirements (Lasley 1983). Finally, the broiler supply appears to be responsive to its own-price movements in the third-quarter, although in an inelastic sense. The elasticity of supply with respect to wholesale price is 0.02. In studies related to agricultural supply response, it is typical to find that agricultural supply exhibits an inelastic response to market prices (e.g., Childs and Hammig, 1989). The lagged supply variable is significant in the supply equation and implies an elasticity of 0.49. This indicates a significant degree of partial adjustment in quarterly broiler production. Broiler Demand Equation. The demand for broilers is positively influenced by real food expenditures and negatively related to own-price. Other variables were not statistically significant. The elasticity of demand with respect to real food expenditures (income) was +.26, implying that a 1 percent increase in real food expenditures increases broiler consumption by .26 percent. This result is within the range of previously reported estimates. Chavas (1983) found that "structural change in poultry demand is reflected in an income elasticity increase from +.012 in 1975 to +.275 in 1979." He concluded that "the rise in income elasticity indicates strong long-term growth prospects for the poultry industry." The own-price effect for the broiler demand implies an elasticity of -.12, which indicates an inelastic per capita demand. Martinez et al. (1986) found a price demand elasticity of -.19, smaller than those reported by George and King (1971) and by Malone and Reece (1976), who estimated demand elasticities to be around -.6. Consequently, increases in per capita demand for broilers in recent years may be due more to changes in tastes and preferences (perhaps against red meats) than to lower prices. The model results show that the demand for broilers is positively influenced by changes in tastes and preferences, as represented by a stochastic trend. The elasticity of demand with respect to the trend component, level and slope combined, is +2.74. A positive change in tastes and preferences for broiler meat, say by a 1 percent margin, would boost broiler demand by 2.74 million pounds of ready to cook (RTC) meat. Broiler Inventories Demand Equation. Broiler inventories demand is negatively influenced by the price obtained by exporters (RBIP). When the 'international price' increases, it will be more profitable to sell on the world markets; therefore, exports will increase and the level of inventories (stocks) will decrease. A 1 percent increase in the 'international price' would decrease broiler inventories by +.12 percent. On the other hand, broiler inventories demand is positively influenced by the broiler own-price (WPBr) and lagged inventories variables. When the wholesale price of broilers increases, broiler demand decreases and the level of inventories increases. A 1 percent increase in the wholesale price of broilers would increase broiler inventories demand by +.81 percent. Also, it is clear that when inventories of the preceding period are larger, the level of inventories in the current period is bound to be higher. Previous studies have formulated broiler inventories demand as an identity equation; no estimates were available for comparison. Broiler Export Demand Equation. Results indicate that exchange rate (SDRJP) and wholesale price (RWPBr) have a significantly negative influence on U.S. broiler exports. The elasticity of exports with respect to the exchange rate and wholesale price were -.76 and -.86, respectively, implying that a 1 percent increase in real exchange rate and wholesale price would decrease U.S. broiler exports by .76 and .86 percent, respectively. The only surprising result is the sign on Brazilian broiler exports. USBX were expected to be negatively related to Brazilian exports because Brazil is a major competitor of the United States in the world broiler market. World demand for broilers and the quantity exported by the United States and Brazil have been increasing. The relationship between USBX and BBEXP is positive because, apparently, of a dynamic demand and an increase in population and income in old and new importing countries of poultry products. Substantial benefits are expected to accrue to the U.S. poultry industry from opening up new markets for poultry products and expanding world poultry meat consumption. The elasticity of U.S. broiler exports with respect to Brazilian exports was +.12, implying that a 1 percent increase in Brazilian broiler exports would be associated with an increase of U.S. exports by .12 percent. Lower prices of U.S. broilers, especially for leg quarters and dark meat parts, are key elements of the U.S. competitiveness in world broiler markets. Martinez et al. (1986) found a negative relationship between USBX and BBEXP. Broiler Wholesale Price Equation. Broiler wholesale price is positively influenced by the prices of pork and turkey. More specifically, the price response of broilers to pork and turkey prices indicates that for each 1 percent change in these parameters, the broiler price would change in the same direction by +.21 and +.54 percent, respectively. Both, pork and turkey price increases would positively influence the demand for broiler meat. While pork is generally considered to be the major competitor with broilers, pork is also a strong substitute for beef. Thus, both red meats will influence the demand for broiler meat. This is especially evident in an inflationary time period. The price of broilers is negatively influenced by the quantity available for market. Demand price flexibility, or the responsiveness of price to quantity changes, indicates that price would change by -.56 percent in the opposite direction for each 1 percent change in quantity. Choice Retail Beef Price Equation. The choice retail beef price is positively influenced by the price of pork. While pork is generally considered to be the major competitor with chicken, pork is also a strong substitute for beef. A 1 percent increase in the price of pork would increase the price of beef by +.14 percent. Pork Retail Price Equation. The price of pork is positively influenced by increased beef and turkey prices. A 1 percent increase in beef and turkey prices would increase the price of pork by +.32 and turkey by +.24 percent. Turkey Retail Price Equation. The price of turkey is positively influenced by the price of pork. A 1 percent increase in pork price would increase turkey price by +.18 percent. Real Food Expenditures Equation. Real food expenditures are positively influenced by increased broiler prices. A 1 percent increase in broiler price would increase real food expenditures by +.012 percent. Japanese–U.S. Broiler Imports Demand Equation. Japanese demand for U.S. broiler imports is negatively influenced by the exchange rate (SDRJP) and per capita Japanese broiler production. The exchange rate plays a significant role in the export of U.S. broilers to Japan. Import demand decreases by 1.06 percent with a 1 percent increase in the yen-dollar exchange rate. Japanese production of broilers is a significant determinant of import demand. A 1 percent increase in Japanese production of broilers would decrease broiler imports by 1.1 percent. This result is consistent with the presence of a strong association between production and import demand for U.S. broilers over the time used in this study. Other variables were not significant. Hong Kong–U.S. Broiler Imports Demand Equation. Hong Kong demand for U.S. broiler imports is negatively influenced by the exchange rate (HORSD) and broiler import price (HOFOBP). A 1 percent increase in the Hong Kong–U.S. dollar exchange rate and broiler import price decreases imports by .55 and .47 percent, respectively. Hong Kong–U.S. broiler import demand is own-price inelastic. Canada–U.S. Broiler Imports Demand Equation. Canadian demand for U.S. broiler imports is positively influenced by pork (CAFOPOIP) and beef (CAFOBIP) import prices. A 1 percent increase in pork and beef import prices would increase Canadian imports of U.S. broilers by +.76 and +.41 percent, respectively. Canada has retained import quotas under the U.S.–Canada Free Trade Agreement and the quantity of broiler imports from the United States has been decreasing since 1984. An intervention variable (wt84), included in the model to capture the effect of import quotas, was not statistically significant. Mexican–U.S. Broiler Imports Demand Equation. Mexican demand for U.S. broiler imports is negatively influenced by the exchange rate (MERSD), per capita Mexican broiler production (PCMEQBP), broiler import price (MEFOBP), and the trade policy measures taken after 1982 (wt83). Mexican tariff rate quotas and broiler import prices play a significant role in the import of U.S. broilers by Mexico. After liberalizing its trade policy in 1982, Mexico kept high trade tariff rate quotas on poultry imports from the United States The impact of this policy has become significant and persistent over time. A 1 percent increase in tariff rate quotas, since 1983, thereafter, decreased Mexican imports of U.S. broilers by 1.3 percent. A 1 percent increase in broiler import price would decrease imports by 1.04 percent. Also, a 1 percent increase in the peso-dollar exchange rate and per capita Mexican broiler production would decrease imports by .57 and .47 percent, respectively. Forecasting ResultsThe extrapolation of the models beyond the end of the sample in 1991 represents the forecasting of broiler supply, demand, prices, etc., which can be made from STS model results shown in Table 3.3. The calculations were carried out using the Kalman filter technique. The forecasts, together with their estimated root-mean square errors (RMSE), are shown in Table 3.5. As can be seen, fitting quarterly STS models leads to better forecasts with a 20 to 60 percent savings on RMSE for the first one-quarter ahead forecast. This saving tends to zero as the forecast horizon increases, and the forecasts tend towards the mean values of the time series variables. Also, Table 3.5 shows the actual figures for each variable. The forecasts values for 1992.I to 1993.IV are close to the average for each series, and they show that the models were predicting the real situation for each quarter. For instance, a directional change was correctly forecast in 1992.IV and 1993.IV for broiler supply and demand, etc.. The forecasts indicate a net increase in broiler supply and demand, a stagnation in meat prices, and a net decrease in real food expenditures by year 2000. Yearly models performed quite well with high RMSE. The forecast values were below or above the average for each series but within an acceptable range. Broiler exports, inventories and imports demand are expected to increase by the year 2000. Finally, fitting the STS models to broiler variables leads to a small gain in forecasting precision when forecasting one quarter or year ahead, but this gain rapidly falls to zero as the forecast horizon increases. Policy ImplicationsFrom the broiler export model results, specific policy inferences can be drawn. (1) Real exchange rate movements have been a significant factor in determining trade in the broiler market. First, because the real exchange rate movements were found significant in the U.S. export supply model and in all but one of the four import markets. Second, a rising value of the dollar could cause a decline in broiler exports and force U.S. producers out of foreign markets. Third, a lower U.S. dollar would increase broiler exports, ensure a larger market share and substantial benefits to U.S. broiler industry. (2) Since 1983, there has been a permanent and persistent effect of tariff rate quotas imposed on U.S. broiler exports by Mexico. These tariffs have decreased Mexican imports of U.S. broilers each year. A progressive reduction of tariff rate quotas, under NAFTA negotiations, would benefit the U.S. broiler industry. (3) The response of broiler export supplies to own-price changes was found to be inelastic. (4) Foreign production of broilers was found to be significant in two of the four major import markets for U.S. broilers. Increased foreign production together with trade policies that restrict trade and subsidize local production would be likely to cause a contraction in the U.S. market share. (5) Pork and beef are substitutes for broilers in Canada, suggesting that the country would tend to import more broilers in the event of a relative rise in pork and beef prices. (6) Feed costs, productivity, and technological progress are significant determinants of U.S. broiler supplies. (7) A substantial portion of the increases in per capita demand for broilers in recent years seems to have been caused by changes in tastes and preferences rather than by lower prices. (8) Demand for broilers was found to be inelastic to income and price changes. (9) The broiler wholesale price is positively influenced by pork and turkey prices. Both pork and turkey prices would positively influence the demand for broiler meat. While pork is generally considered to be the major competitor with broilers, pork is also a strong substitute for beef. Thus, both red meats will influence the demand for broiler meat. Concluding RemarksKnowledge of the variables that influence U.S. broiler exports could be of assistance to policy-makers in planning growth and strategies for the broiler industry, and particularly important to broiler producers, processors, and others connected with the industry and trade. An econometric analysis of U.S. broiler exports indicates that the broiler export market is very sensitive to changes in real exchange rate and trade distortion policies. These results suggest that the United States can increase its broiler exports more effectively by the following: (1) extending efforts on international macroeconomic policy coordination rather than depending on domestic sectoral policies, and (2) working toward elimination of trade distortion practices through NAFTA, GATT, and bilateral trade negotiations. Given these results, the advantages of STSM, formulated directly in terms of unobserved components such as trends and seasonals or cycles, make it an attractive methodology in the context of modeling broiler exports. As a matter of fact, the introduction of a stochastic trend component in the econometric models has turned out to be a powerful tool to proxy the influence of unobserved variables such as changes in tastes, preferences, productivity, technology, etc. The analysis of the contribution of the explanatory variables to the growth of broiler exports has shown that the trend component has been vital in the expansion of the broiler industry during the study period. Also, quarterly models have provided accurate short-term forecasts of future values of the series such as supply, demand, and meat prices. More work is required to extend this study's framework to (1) analyze the competitive position of U. S. broiler industry, and (2) formulate an international trade model of poultry markets by including other poultry producers and importers. ReferencesAppendixesDocument prepared by: Leigh H. Stribling, lstribli@acesag.auburn.edu Alabama Agricultural Experiment Station Auburn University |
||||||||||||||||||||||||
|
|
|