5. MEAT DEMAND ANALYSIS

Introduction

Meat and poultry consumption is becoming more important in the Japanese diet. Some studies report that Japanese diets have become increasingly Westernized: as per capita income increases, Japanese people consume more meats and poultry than grain products. As there is a limit to total calorie intake, the ratio of meat products to other foods is becoming larger than that of grain commodities. As reported in Chapter 4, fish and meat are moderate substitutes for each other (see Table 4.7). Furthermore, in June 1988, Japan signed the Beef Market Access Agreement (BMAA) with the United States. This agreement may have accelerated the increasing trend of Japanese beef consumption owing to the influx of cheaper beef from abroad. According to Persaud and Chern (1999), Japan's meat production began first to stagnate and then to fall, while meat imports rapidly increased after the agreement.

The main objectives of the following analysis of meat demand are to analyse meat consumption patterns and to conduct an econometric analysis of the meat demand structure in Japan. Cross-sectional household data from the 1997 FIES are used and estimations are based on a total of 94 200 observations. The meat items included are beef, pork, poultry, ground meat, ham, sausage and bacon. Fish is not included in this model because there are many heterogeneous groups of fish and, when the substitution patterns of the aggregate groups of meat and of fish were analysed (see Chapter 4), it was found that fresh meat and fresh fish were mild substitutes. A study by Eales and Wessells (1999) also found that meat and fish have not been weakly separable in recent years. The results of separability tests tend to be highly dependent on a particular database. Eales and Wessells used the time series data generated from the FIES for 1981 to 1995, and did not include high-value processed meats in their study. The results for testing separability may be different for the present report, which uses household-level data and includes other processed meats in the model. In any case, a weak separability between the seven meats and other foods - including fish - is assumed.

Much research has been conducted on the structural change of meat and poultry demand in Japan using time series data (Hayes, Wahl and Williams, 1990; Capps et al., 1994; Eales and Wessells, 1999). After a decade of trade liberalization, it is extremely important to have an accurate estimate of Japanese meat and poultry demand elasticities. This report offers another estimation with cross-sectional data, which (as already noted) yield more reliable estimates of expenditure and income elasticities than time series data do. In this report, own-price, cross-price and expenditure elasticities are estimated by various single equation models and the almost ideal demand system (AIDS), developed by Deaton and Muellbauer (1980a; 1980b), as discussed in Chapter 3. These elasticity estimates are utilized to analyse Japanese consumption patterns of meat and its products.

Review of exising information and literature

Meat and poultry consumption in Japan over the last three decades has two main characteristics: an increasing trend in meat and poultry consumption; and a rapid increase of meat imports.

Historically, increases in per capita income in Japan have accelerated meat consumption. Figure 5.1 illustrates annual per capita beef, pork and poultry consumption and shows how per capita consumption of meats and poultry has been steadily increasing since 1960. Pork consumption increased until the early 1980s, but its growth seems to have levelled off thereafter. Over the last 30 years, the average consumption of beef has not increased as rapidly as that of pork and poultry. However, the increase in beef consumption has accelerated since the mid-1980s, and in recent years beef has been increasing more rapidly than pork and poultry consumption. In 1960, per capita consumption of beef, pork and poultry was 1.1, 1.1 and 0.8 kg, respectively. Ten years later, beef consumption had nearly doubled, and pork and poultry consumption had increased fourfold. By 1980, beef consumption had tripled, and pork and poultry consumption had increased tenfold. By 1995, beef consumption was seven and a half times its 1960 level, and poultry consumption had increased by nearly 14 times.

The Japanese Government protected beef producers with import quotas. Before the 1988 BMAA, all beef imports, except a small portion under the private quota, were monopolized by the Livestock Industries Promotion Corporation (LIPC) (see Hayami, 1979; Alston et al., 1990). Under the BMAA, quotas were raised by 60 000 tonnes each year from April 1988 to April 1991 (Persaud and Chern, 1999). Thereafter, the quota restriction was replaced by an import tariff. This agreement significantly curtailed LIPC's intervention in the beef market. Consequently, the Japanese beef market became freer; the Japanese cattle industry had to adjust its supply and pricing to the market situation, and consumers enjoyed the benefits of increasing competition from foreign countries.

The increases in meat imports since the 1988 agreement have been remarkable.^[8] Figure 5.2 illustrates the trends of beef, pork and poultry imports and shows how all meat imports increased rapidly after the mid-1980s, with beef imports surpassing pork imports during most of the 1990s. Poultry imports have been lower than those of beef and pork during the last three decades. As Figure 5.3 shows, domestic beef production has been stagnant since the mid-1980s, and beef imports have grown at an even faster rate than before. These figures clearly show that beef imports have exceeded domestic production since 1992. Given the historical restrictive beef quota and its trade liberalization in 1988, the relatively high growth rates of beef consumption were stimulated by increased imports in the late 1990s. Figure 5.3 also shows how the production of pork and poultry has decreased since the late 1980s. Figure 5.4 shows the yearly percentage change of nominal unit prices for beef, pork and poultry. Beef prices have been more volatile than those of pork and poultry. Since trade liberalization, beef prices have shown a downward trend, although the magnitude of this trend is not as large as might have been expected.

Food prices in Japan have generally been considered high, relative to other developed countries. Hayami (1979) reports the price differentials of food commodities based on 1977 data. Two decades ago, the retail price of beef in Tokyo was seven to nine times higher than in the capitals of other developed countries. In order to have an overview of the price differentials, MAFF regularly issues price comparisons for food items in major cities of the world. As shown on Table 5.1, which is based on 1996 data, beef is still more expensive in Tokyo than it is in other major cities, although the price differential ranges from twice to none. The pork price varies among major cities, and poultry is cheaper in Tokyo than in several other major cities. Combined with Hayami's survey, Table 5.1 shows that the price differentials among large cities in developed countries have decreased significantly over the last 20 years.

Estimation results on the own-price elasticities of meat demand in Japan, the United States and Canada vary considerably. Hayami (1979) investigated the consequences of Japanese beef trade liberalization and reported the results of own-price elasticities for beef from other studies: the results vary from -1.3 to -1.8. Applying the Rotterdam demand system, Sasaki (1995) estimated various meat demand elasticities in Japan. The own-price elasticities of beef, pork, poultry and meat products are -1.26, -1.53, -0.68 and -0.81, respectively. These results are consistent with Hayami's survey: beef and pork are highly price-elastic, while poultry is relatively inelastic.

Figure 5.1: Annual per capita beef, pork and poultry consumption, 1960-1999

Source: MAFF, 2001.

Figure 5.2: Japanese imports of beef, pork and poultry, 1960-1999

Source: MAFF, 2001.

Figure 5.3: Domestic production of beef, pork and poultry, 1960-1999

Source: MAFF, 2001.

Figure 5.4: Yearly percentage changes of beef, pork and poultry prices, 1970-1997

Source: FIES, various issues.

TABLE 5.1
International price comparison (November 1996)

Note: Price comparison with Tokyo (price in Tokyo = 100).
Source: MAFF, 1997.

There are many studies on United States and Canadian meat and poultry consumption. Eales and Unnevehr (1993) conducted a survey of meats and poultry demand elasticity studies in Canada. Table 5.2 shows excerpts from their results. In their survey, the estimates of own-price elasticity are considerably lower in absolute value than those from Hayami's study, particularly for beef and pork. Using the AIDS model estimated with United States data, Chen and Veeman (1991) reported own-price elasticities for beef, pork and poultry to be -0.77, -0.87 and -0.95, respectively. The Chen and Veeman study on United States meat demand yields estimates that are far closer to those obtained from Canadian data. The studies' comparisons between Western countries and Japan indicate that the own-price elasticities of meats and poultry are higher in Japan than in the United States and Canada.

TABLE 5.2
Canadian meat demand: own-price elasticity estimates (published in the 1990s)

Source: Eales and Unnevehr, 1993, Table 4.

Expenditure elasticity is one of the key determinants of future meat demand, and it is important to have an accurate estimate of expenditure elasticity in order to forecast the medium- to long-term demand for meat. Eales and Unnevehr (1993) surveyed the results of Canadian expenditure elasticity studies. Table 5.3 shows part of their survey results. Using United States data, Moschini and Meilke (1989) reported various expenditure elasticities on meat demand. The estimated expenditure elasticities of beef, pork and poultry were, respectively, 1.39, 0.85 and 0.21 in the post-structural change period of the fourth quarter of 1975 and in the 1967-1987 data set. There are only slight differences between the estimates made from Canadian and those made from United States data. In the Eales and Unnevehr study, expenditure elasticities for beef, pork and poultry are positive, but only the beef expenditure elasticity exceeds one in the Moschini and Meilke study.

TABLE 5.3
Canadian meat demand: expenditure elasticity estimates (published in the 1990s)

Source: Eales and Unnevehr, 1993, Table 4.

Sasaki (1995) reported the income elasticities of meats and poultry in Japan, and found elasticities for beef, pork, poultry and other meat products of 0.80, 1.29, 1.42 and 2.10, respectively. It is interesting to note that Sasaki's study shows beef to be a normal good, although pork, poultry and meat products are superior goods. These results are quite different from the findings of the Canadian and United States studies. Sasaki's estimates depend on the specification of the Rotterdam model and the use of time series data. In the present report, the AIDS model with cross-sectional data has been applied, which should yield more reliable estimates of expenditure elasticities.

Empirical results

The survey data used in this report consist of 94 200 observations for seven meat products. Price data are constructed for households with zero consumption (and thus zero budget share) according to their geographic location, the month of the survey and the household's income level, as described in Chapter 3. In order to have a consistent result, OLS, Heckman's two-step and Tobit estimators are compared. For system modelling, the LA/AIDS models with Stone and Laspeyres index is utilized, and the inverse Mills ratio is included to correct for the zero consumption problem. The full version of the AIDS model is also applied, and all results are compared. Both the LA/AIDS and the AIDS are estimated with homogeneity and symmetry conditions imposed.

Table 5.4 reports the comparison of estimated own-price elasticities: excluding sausage, beef has the most inelastic own-price elasticity among the seven meat commodities in the single equation models. These results indicate that Japanese consumers are insensitive to changes in the beef price. If beef were price-inelastic, the recent increases in beef consumption would not have been caused by price decreases, but either by per capita income increases or by substitution with other meat products. Table 5.5 reports the (Hicksian) compensated price elasticities obtained from the AIDS model with inverse Mills ratios for all seven variables. Most products are net substitutes. Only beef and ground meat show a negative Hicksian cross-price elasticity, which is caused by a net complementarity.

The estimates of expenditure elasticity are compared in Table 5.6. Among all meat products, only beef has an expenditure elasticity that exceeds one - those of other commodities are all less than one. This result is very similar to that found in the United States and Canadian studies reported by Eales and Unnevehr, and Moschini and Meilke. Sasaki's study, based on existing Japanese data, shows nearly the opposite results. From this study, it can be concluded that the Japanese meat consumption pattern has been Westernized in that expenditure elasticities have become similar to those estimated in North American countries. This preference change might have been caused by the increases in imported beef. The comparison of expenditure elasticities from various models reveals that results are robust. In addition, a high expenditure elasticity for beef indicates that beef consumption will not decrease dramatically as long as Japanese consumers maintain their high per capita income.

It is interesting to compare these estimates with those obtained by Eales and Wessells (1999) using the quarterly time series data for 1981 to 1995. Eales and Wessells tested the separability between meats and fish and estimated two models: one non-separable and one separable, including only meats (beef, pork and poultry). In fact, the estimated own-price and expenditure elasticities for beef, pork and poultry (Tables 5.4 to 5.6) estimated in this report are closer to the other estimates for the non-separable than for the separable model. For example, estimates of the own-price elasticities for beef range from -0.549 to -0.605, while Eales and Wessells obtained an estimate of -0.516 from their non-separable model and -0.166 from their separable model. There are other notable differences, including estimates of price elasticities in absolute value that are higher than those estimated by Eales and Wessells from their non-separable model, especially for pork and poultry. Expenditure elasticities are not strictly comparable because the models used by Eales and Wessells include other items in addition to beef, pork and poultry: the present report includes ground meat, ham, sausage and bacon, while Eales and Wessells include three categories of fish. Despite this difference, the estimates of expenditure elasticities are higher than those estimated by Eales and Wessells (1999).

Table 5.7 shows the estimation results of selected demographic variables obtained from the AIDS model. Considering the high expenditure elasticity for beef, it is reasonable to expect that the age of the household head and the number of wage earners in the household should have a positive impact on beef's budget share, while the expenditure on beef tends to decrease as the number of household members increases. The number of children shows negative impacts on the expenditure shares of beef, pork and poultry.

Concluding remarks

This report estimates the Japanese demand for beef, pork, poultry and processed meat products. The own-price, cross-price and expenditure elasticities of seven meat products are estimated, and some selected demographic variables are analysed. The results indicate that recent Japanese consumption behaviour is similar to that of United States and Canadian consumers, as measured by elasticity estimates.

This study is based on 1997 cross-sectional household survey data. A decade has passed since the beef trade was liberalized. Beef is inelastic to price, and the expenditure elasticity of beef exceeds one. As it is a cross-sectional analysis, this report does not make it clear whether a structural change has occurred. Estimation results from time series data will reveal further information about structural changes to consumer preferences. This report uses expenditure elasticity to show that beef is the only superior good among seven meat commodities. In this regard, as with other reported estimates, Japanese preference in the late 1990s had become closer to that of Western nations.

TABLE 5.4 - Comparison of own-price elasticities

Note: The numbers in parentheses underneeath the elasticity estimates are standard errors.

TABLE 5.5 - Hicksian compensated price elasticities of the AIDS model with inverse Mills ratio

TABLE 5.6 - Comparison of expenditure elasticities for meat products

Note: The numbers in parentheses underneath the elasticity estimates are standard errors.

TABLE 5.7 - Coefficients of demographic variables from the AIDS model

Note: An asterisk denotes statistically significance at the 5 percent level.