- Email: [email protected]
Price elasticities in international food trade: synthetic estimates from a global model
lasticities i Synthetic
e:
Rod Tyers and Kym Anderson, University of Adelaide
Govenunent intervention, particularly in agncuhural commodity markets, is frequently justified on the grounds that it favorably affects the terms of trade. In this paper an established dynamic simulation model of seven food commodity markets is used to provide synthetic estimates of the elasticities of net export demand and import supply on which such justifications rest. Estimates are presented for both the short and long runs, and the effects of market-insulating agricultural policies on these elasticities are investigated. The results cast doubt on ti F proposition that any individual economy has strong monopoly or monopsony power in international food markets in anything other than the very short run. But effective cooperation by groups of exporting countries, such as the Cairns Group of “nonsubsidizing” agricultural exporters, along with the United States or even the EC, could yield substantial market power to those groups in both the short and long runs. Nevertheless, the major part of the power such groups might wield stems from self-imposed domestic-market-insulating agricultural policies in the rest of the world.
Estimates of elasticities in international trade are continually in demand. They are of particular importance to policy analysts seeking to measure the likely effects of government intervention in countries that have some monopoly or monopsony power in the international market. Food price elasticities are especially sought after because (1) most governments intervene in their food markets and (2) a number of those countries have-or believe they have-substantial market power in international food trade, in which case their perceived optimal food trade tax may be positive rather than zero. There are two alternative approaches to obtaining price elasticities in international trade: direct estimation, which involves regressing trade
Address correspondence to Rod Tyers, Department of Economics, University of Adelaide, Adelaide. S.A. 5001, Australia. This paper stems from research on trade elasticities completed in association with the International Agricultural Trade Research Consortium Symposium on Elasticities in International Trade, Dearborn, Michigan, July 1987. It is one of two papers based on the same food trade model that emerged from that workshop. For useful comments and suggestions thanks are due to F. Gerard Adams and to the participants in the Michigan symposium. ; final draft accepted. Received
Journal of Policy Modeling Il(3):315-344 (1989) 0 Society for Policy Modeling, 1989
315 0161-8938/89/$03.50
Rod Tysrs and Kym Anderson
316 volumes
against
border prices, and synthetic estimation. Serious econometric difficulties arise with direct estimation methods, as was pointed out long ago by Orcutt (1950). It is appropriate, then, to consider various synthetic estimation methods. Ideally, synthetic estimates of trade elasticities for particular commodities should be obtained from a multicommodity, multicountry dynamic general equilibrium model of the world economy. In the absence of such models, however, analysts have relied on lesscomu plete models’ or world markets. Single-commodity models are often used, but *hey miss the important interactions that exist between closely related ~smmodities. In this paper we use a multicommodity, multicountry &inamic simulation model of markets for grains, livestock products, and sugar. This model has been widely applied in the analysis of distortions affecting brade in food commodities (see, for example, Anderson and Tyers 1; 35, 1886, 1988; Tyers and Anderson 1988a; World Bank 1986, Ch. 6), although its implicit trade elasticities have not been examined previously. Section 1 describes the features of the model. From the viewpoi;!nt of obtaining trade elasticities, its key features are its coverage of multiple commodity markets, which ensures that the interactions between various food markets are explicitly represented; its dynamics, which permit both short- and long-run trade elasticities to be obtained; and its inclusion of estimated short- and long-run international-to-domestic price transmission elasticitie s (distinguished as between producer and. consumer prices) to represent the stabilizing and insulating behavior of government policies. In Section 2 the approach used to obtain the trade elasticities directly from the parameters of the model is outlined and the elasticities are reported. Included are values for the short and long run (with and without market-insulating policies) and for both the early and late 1980s. Section 3 presents the magnitudes of trade elasticities facing a number of country groups and assesses their potential market power. Section 4 then lists some caveats that should be borne in mind when interpreting the elasticities; concluding comments are provided in Section 5.
The model we use provides dynamic, stochastic simulations of the world markets for grains, livestock products, and sugar (CLS). It is an extended and updated version of the model presented by Tyers (1984, 1985). To keep it manageable, the level of commodity disaggregation is restricted to seven groups: wheat, coarse grain, rice, meat
PRICE ELASTICITIES
IN INTERNATIONAL
FOOD TRADE
317
of 6ruminants (cattle and sheep), meat of nonruminants (pigs and poul4ry), dairy products, and sugar. These account for about one half of the world food trade (e%ble oils and beverages accozni 5~ -nest of the rest) and one tenth of global trade in all commodities. Salient features of the model include the following: 1. It is global in coverage, identifying 30 countries and counq groups. 2. It incorporates the cross effects, in both production sad consumption, between the seven interdependent commodity groups and separates the direct demand for grains as food from the indirect demand for animal feeds. 3. Stockholding behavior is included endogenously, based on empirical analysis of stock level responses to price and quantity changes in each country. 4. Lags in supply adjustment to price changes are included. 5. Policies in each country are assumed to allow domestic pries to change only gradually (and in some cases not at all) m response to changes in intemational food prices. Production behavior is represented by Nerlovian reduced-form partial-adjustment equations th~;tare log-linear, resulting in constant shortand long-run supply elasticities. Special shifters are added to allow for the effect on production of land set-aside policies such as those used in the United States. Direct human consumption is characterized by income and price elasticities of demand, which are constant in any year but decline over time, whereas feed consumption by animals is determined by input-output coefficients that relate feed use to a steadystate level of production for each livestock product. Policies that affect domestic prices are incorporated via econometrically estimated international-to-domestic price transmission equations for each country and commodity. These equations capture both the protection and the stabilization components of food price and trade policies. They are based on estimates of reduced-form Nerlovian partial-adjustment equations that distinguish short-run from long-run elasticities of price transmission. Separate elasticities are used for producer and consumer prices. In general, even the long-run price transmission elasticities are less than unity, reflecting the prevalence of nontariff protection instruments in food markets. In the face of volatile and declining real prices in international commodity rr,arkets, governments limit the extent to which both the long-run trend and the short-run changes in domestic prices follow those of international prices. The smaller the short-run elasticity of price transmission in relation to its
318
Rod Tyers and Kym Anderson
long-run counterpart, the greater the degree of market insulation and the more sluggish the eventual transmission of any sustained change in the international price. In a few extreme cases domestic prices are long-run elascompletely insulated, which means both the shortticities of price transmission are zero. Storage behavior is represented by a combin competitive speculation and the actions of public only to quantity triggers. Target levels of closing each commodity at a fixed proportion of trend c porting countries and trend production in exporting countries. Interon expected temporal deviations from these t;irget levels de speculative profits and the deviation in domestic y (production plus carry-in) from the long-run trend. Structurally, the model is a set of expressions for quantities consumed, produced, and stored, each of which is a function of known past prices and endogenous current prices. The model is solved iteratively by starting from the 198042 base period and beginning each subsequent year with the assumption that all prices are the same as those in the preceding year, generating random disturbances in production, and calculating new production, consumption, and closing stock levels in each country. The resulting excess demands are then totaled and international prices are adjusted to move world markets toward clearance. The procedure is then repeated until a satisfactory degree of market clearance has been achieved for each commodity. Thus, the model selects that series of international and domestic prices, production, consumption, and closing stock levels that simultaneously clears all markets in each successive year, from 1983 to 1995. Once 100 simulations of this type have been completed, each using a different set of generated random disturbances from thy distributions of each error term; forecast means and standard deviations are calculated for all key variables in the model for each year of the simulation period. They can also be calculated for the base period (1980-82) simply by replacing the values of all time-dependent parameters with their 1980-82 values. The solution procedure is conventional, but it is not based on a standard software package. A key assumption with important implications for ‘the subject of this paper is that of product homogeneity. The work of Brown (1987) and others has demonstrated that the degree of market power afforded by imperfect substitutability between like products from different countries is sufficient to substantially affect outcomes from government intervention. This * found to be particularly importan: in economywide models with ghly aggregated product categ ties. We see this
PRICE ELASTICITIES IN INTERNATIONAL
FOOD TRADE
319
as comparatively unimportant in the case of the ptial-equilibrium GLS model. Although disaggregation into seven commodities (the most important of which are the individual grains) does not in itself ensure perfect homogeneity, it renders the assumption more valid. The parameters of the model are based on time-series analysis with estimation intervals in the period 1962-83. Alternative estimates of the price transmission (policy) and storage parameters tend not to be readily avaiiable eisewhere in the literature and so their original values have remained largely unrevised. Many alternative estimates are available for the parameters governing domestic supply and demand behavior, however. Accordingly, a number of our original estimates have been revised where alternative values are based on more sophisticated country-level analysis. All of the parameters are documented in Tyers and Anderson (forthcoming). Because our purpose here is to examine the trade elasticities implicit in our model, and because these are only as good as the structural parameters on which they are based, we provide a sampling of key demand, supply, and price transmission elasticities in Appendix I.
2. IMPLIED TRADE ELXTICITIES
IN THE GLS MODEL
It is clear from the above description that the parameter set of the GLS model is sufficiently rich to imply potentially useful estimates of trade elasticities. They depend on domestic own- and cross-price elasticities of supply and demand, on the price responsiveness of storage, and on price transmission elasticities in the various countries, as well as on the shares of each country in different markets: they vary according to the time allowed for adjustment and the year chosen. Indirect estimates can therefore be derived oanalytically from the parameter set. This is done by calculating the extent to which the excess demand in the rest of the world for commodity i produced by country K adjusts in response to a change in the price at which country K trades commodity i. (The excess demand of the rest of the world may be a positive or negative quantity, the latter indicating that country K is a net importer of commod.ity i.) The detailed derivation of these elasticities based on the equations of the model provided in Tyers and Anderson (1988b). A summary of the key analytical results is given in Appendix II. With 30 countries/country groups and seven commodity groups, the total number of own-price trade elasticities is 210 for each year and for each period of adjustment. It is therefore necessary to be selective in presenting these parameters. Only the very-short-run (same year), short-run (after one year), and long-run (after full adjustment) trade
320
RodTyersandKymAnderum
elasticities are reported here.’ In Table 1 the own-price elasticities a= shown for the base period 1980-82 for large participants in world food markets wherever the magnitudes of those elasticities are less than 40. The key points to note from the implied trade elasticities in Table 1 are as follows: 1. The long-run trade elasticities have magnitudes larger than 7.5
2.
3.
4.
5.
6.
7.
8.
for all countries and commodities other than wheat and coarse grain in the United States and wheat in the EC and the USSR. The short-run trade elasticities (after one year’s adjustment) are never smaller than unity, suggesting that by reducing the volume of its food imports, no individual economy can increase its export revenue and sustain the increase through a full year or more. Following the United States, the smallest short-run export demand elasticities are faced by Canada, Australia, and the EC in the wheat market and by the EC in the market for dairy products. The smallest import supply elasticities are faced by the USSR and China in the wheat market and by 3apan and the USSR in the market for coarse grain. In the rice market, Thailand faces the lowest export demand elasticity, whereas Japan’s small share of the international rice trade results in a comparatively high import demand elasticity at the margin. Australia and New Zealand have the lowest export demand elasticities for ruminant meat, followed by Argentina, and the EC and New Zealand have the lowest for dairy products. In the sugar market, 10 of the countries listed in Table 1 face relatively low trade elasticities, with Brazil and the EC facing the lowest on the export side and the USSR facing the lowest on the import side. With the exception of coarse grain in the short run, Japan’s import supply elasticities are very large, suggesting that Japan has little monopsony power in international food markets.
These trade elasticities would be considerably larger, especially in the short run, if countries did not insulate their domestic markets ‘In our analysis the trade elasticities are calculated in matrices that include non-zero cross elasticities. Because our primary interest here is in these elasticities as measures of comparative market power at the margin, only the own-price elasticities are presented. More complete tabylations are available on request from the authors.
h
VSR SR LR VSR SR LR VSR SR LR VSR SR LR (14.9) -5.2 -11.5 - 30.0 (9.0) (19.8)
(6.1)
(2.8
- 19.3
-5.1 - 15.5 -38.1
Coarse Grain
-4.5 -8.3 - 16.5
-6.8 -11.9 -21.2
Rice
- 22.7
- 13.4 - 24.7
Ruminant Meat
Nonruminant Meat
Dairy Products
-4.5 -5.9 -8.5 -11.5 - 15.3 - 20.6
(9.9) (13.0) (18.0) - 30.0
Sugar
adjustment.
“Elasticities not shown have magnitudes greater than 40. Values in parentheses are the excess supply elasticities faced by the coun+q; the other values are excess demand elasticities. VSR refers to the very-short-run adjustment within the first year, SR to adjustment after one year, and LR to the long run after full
Source: Derived from the model described in the text.
Thailand
Brazil
Argentina
China
Wheat
PRICE ELASTICITIES IN INTERNATIONAL
FOOD TRADE
323
from changes in international prices. To see this, the elasticities were reestimated based on unit price transmission elasticities. The implied trade elasticities that emerge with this respecification, summarized in Table 2, are in most cases more than twice those in Table 1. This result suggests that the insulating component of agricultural policies has a very substantial depressing impact on the size of food trade elasticities and thus cannot be ignored in the process of estimating those elasticities, a point stressed by Abbott (1979) and Bredahl et al. (1979).2 The trade elasticities in Tables 1 and 2 refer to the base period 198082. Since then, however, the degree of market insulation has changed, particularly in the United States. There, increased participation by farmers in grain price support programs under the Food Security Act of 1985 has substantially reduced the responsiveness of output to international price changes. In addition, the shares of each country in world production and trade have changed. Both these effects are represented in a “reference’ 9 projection to 1988. In this projection, policy regimes other than that of the United States, as represented in the model, are assumed to remain unchq+l (see Tyers and Anderson (1988a) for a detailed account of this projection). Assuming that this projection turns out to be reasonable in retrospect, it allows us to gauge how implicit trade elasticities may be changing during the 1980s. Table 3 illustrates these trends for the very-short-run trade elasticities. In most cases the projected trade elasticities are somewhat smaller in 1988 than in the base period, the only exception being for sugar. The differences appear to be larger for the case where the stabilization component of policy is removed in all economies (price transmission elasticities set at unity). The latter observation suggests that the changes in the global geographic distribution of food production and consumption have led to a slight decline in the aggregate responsiveness of private agents such as consumers, farmers, and stockholders. More generally, the point that is clear from comparing Tables 1 and 2 with Table 3 is that, even when most underlying parameters are stable, trade elasticities are by no means constant over time. How do these implicit trade elasticities compare with those adopted ‘Removal of the protection component of agricultural policies may or may not raise the food trade elasticities in Table 2: it depends on whether production and consumption in the rest of the world is relocated to more- or less-elastic markets and whether the country in question becomes a larger or smaller participant in the world food economy as a consequence of the liberalization.
(6.1) (13.0) (12.2) (23.2) - 26.6
-9.2 - 18.2 - 15.2 - 30.0
-3.8 -6.9 - 15.1 - 27.8 (34.6)
- 13.3
(33.3)
03.5)
(7.8) (31.2) - 22.7
- 1.9 -6.4
Grain
Coarse
- 27.5 - 38.8
- 35.4
-37.1
Rice
Nonruminant Meat
- 20.0 -31.2
Dairy Prducts
- 24.4 - 37.7
(12.4) (18.9) (38.8)
-31.1
- 19.2 - 28.6
(27.8)
Sugar
price transmission elasticities set to unity (no insulation).
- 38.6
-36.3
Ruminant Meat
1980-
“Elasticities not shown have magnftudes greater than 40. Values in parentheses are the excess supply elasticities faced by the country; the other values are excess demand elasticities. VSR refers to the very-short-run adjustment within the first year and LR refers to the long run after full adjustment.
Source: Derived from the model described in the text, with all international-to-domestic
Thiland
Brazil
Argentina
China
USSR
New Zealand
Australia
Canada
Japan
European Community ( 10)
VSR LR VSR LR VSR LR VSR LR VSR LR VSR LR VSR LR VSR LR VSR LR VSR LR VSR LR
wheat
: Excess Demand (Supply) Elasticities Facing Large Participants in World Food Markets: The Case of No Muket Insulation,
United States
Ta 82”
(10)
ref. PTE=l ref. lTE=l ref. PTE=l ref. FTE=l ref. PTE=l ref. PTE=l ref. PIE=1 ref. PTE=l ref. F’TE=l ref. PTE=l ref. PTE=l (2.9) (11.4) (2.5) (10.3) -5.0 -24.3 (9.0)
-0.51 -3.1 -2.1 -8.9 (8.2) (37.6) - 1.8 -8.5 -2.9 - 13.9
Wheat
-3.6 -22.0
- 24.0
- 12.7
(28.3)
-11.2
- 3.9 - 20.4
-9.2 - 24.3
-9.3
-5.1 - 32.6 -6.4
(25.3)
-11.3
- 3.6 - 20.0 - 13.8
(3.3) (14.4) (7.2) (27.6) -26.1
-7.5 - 33.4
(24.1)
-3.4 - 14.1 (9.9)
(13.2)
-4.0 - 32.3 - 1.7 - 14.3 (31.0)
(33.6)
(20.7)
-4.7 -29.1
Sugar
Products
Meat
Meat
“Elasticities not shown have magnitudes greater than 40. Values in parentheses are the very-short-run excess supply elasticities faced by the country; the other values are very-short-run excess demand elasticities. Ref. refers to the reference scenario (corresponding with Table 1 for the base period), PTE= 1 refers to the scenario in which all price transmission elasticities are set at unity (corresponding with Table 2 for the base period).
- 18.7
(5.6) (13.3) (8.6) (19.0) -4.7 -11.8
(2.8) (7-O) -7.0 - 17.5 - 13.7 - 33.9
-0.27 -1.1
Rice
Dairy
Nonruminant
Ruminant
Facing Large Participants in World Food Markets, Projected to 1988”
Coarse Grain
Excess Demand (Supply) Elasticitk
‘ource: Derived from the model described in the text.
Thailand
Brazil
Argentina
China
USSR
New Zealand
Australia
Canada
Japan
European CornmuG;;
United States
Table 3: Very-Short-Run
326
Rod Tyers and Kym Anderson
in other studies? The most commonly estimated food trade elasticities are the excess demand elasticities facing the United States grain export sector. A sample of such empirical studies over the past decade is summarized in Table 4. The GLS model suggests that the elasticities are much higher than those used by Johnson et al. ( 1985), whereas our coarse grain elasticities are similar to theirs. Our long-run elasticities are also somewhat larger than those suggested by Bredahl et al. (1979), but they are substantially lower than those suggested by Johnson (1977). The work by Johnson et al. (1985) is evidently based on national models for the United States developed at the Food and Agricultural Policy Research Institute. These models use estimated trade elasticities to characterize the rest of the world in aggregate. A difficulty with the reliance on such estimates is that they are invariably based on timeseries analysis, which attempts to explain changes in the level of exports in terms of border price changes. Because exports have other determinants frequently omitted from such studies, specification error leads to the underestimation of the true (ceteris paribus) responsiveness of exports to border price. In the recent work of Meyers et al. (198?), the e!asticities quo:& are from a single-commodity analysis using an eight-region global model. In that study, trade elasticities are derived by comparing model solutions in which yield functions in the United States have been shifted. The estimates of trade elasticities that result are substantially larger than those from the earlier Johnson et al. (1985) and Meyers and Helmar (1986) studies. In the case of coarse grain, their one-year elasticity is even larger than our estimate from the GLS model. The Meyers et al. estimates would, of course, have been larger still had the adjustment to a supply shock in one commodity market been permitted to spread across several interacting markets. Although an experiment similar to this is discussed in their article, it was based on simultaneous supply shocks in all markets and hence sheds no light on the (ceteris parihus) trade elasticities that are the subject of this paper. 3. TRADE ELASTICITIES FACING COUNTRY GROUPS The magnitudes of trade elasticities are smaller, and hence market power is greater, for aggregates of either exporting or importing countries. To quantify this, we have calculated these elasticities for a number of count*y groups featuring prominently in t international trade negotiations. The results are present
-0.60 -1.2 -3.3 - 0.46 - 1.2 -3.7 -5.9 -11.1 - 22.7
-0.51 -1.0 -2.9 - 0.27 - 0.74 - 2.3 -4.7 -8.9 - 18.1
19*’
-0.00
-0.09
-6.7
- 10.2
to - l.3b
to - 1.7
Bredahl et al. Johnson” (1977, Table 1) (1979, Table 4)
- 0.25 “near 1 .O”
-0.16 “near 1.O”
-0.15 “near 1.O” -0.29
-0.11
-0.46
-0.30 -0.90
-0.14 -0.23
to -0.68 - ‘7.0
to - 1.5 to -3.3
to -3.1 to -5.0
“Assuming price transmission elasticities are unity. “For corn only; -0.29 to -2.6 for sorghum.
- 1.38 - 1.59
-0.9 - 1.23
Other Studies Meyers and Surveyed by Gardiner Meyers et al. and Dixit Johnson et al. Helmar (1985, Table 1) (1986, Table 5) (1987, Tables 2, 3,s) (1987, Tables 1, 2)
Source: The first two columns are derived from the model described in the text; other values ue from the studies specified.
VSR SR LR Coarse VSR grain SR LR Rice VSR SR LR
Wheat
1983
Thii study (Tables 1, 3)
7M_~le4: Comparison of Excess Demand Elasticities Facing U.S. Grain Exporters
328
Rod Tyers and Kym Anderson
which lists, for comparison, the trade elasticities facing the United States and the EC-10 (provided originally in Table 1). Other couutry aggregates for which trade elasticities are presented include the larger West European aggregates, the EC- 12, and the combination of the EC with the European Free Trade Association (EFTA) countries (Austria, Finland, Norway, Sweden, and Switzerland). Also listed are the trade elasticities facing the Cairns Group of “nonsubsidizing” agricultural exporting countries (Argentina, Australia, Brazil, Canada, Chile, Colombia, Hungary, Indonesia, Malaysia, the Philippines, New Zealand, Thailand, and Uruguay), along with those facing various combinations of country groups with the United States. The larger European groups are generally less powerful in the GLS markets than the EC-lo. This is because Spain, Portugal, and the EFTA countries tend to be net importers of GLS products. In combination, therefore, Western Europe has smaller net exports than the EC-10 and faces higher trade elasticities. The key exception to this pattern is in the market for dairy products, where the export demand elasticity faced by Western Europe is slightly lower than that faced by the EC-10 alone. Nevertheless, more widespread cooperation in West European agricultural policy is unlikely to enhance the region’s international market power3 The principal rivals of Western Europe in the current agricultural trade negotiations are the United States and the Cairns Group. The results in Table 1 suggest that, despite its numbers, the Cairns Group lacks the power of the United States in the key wheat and coarse grain markets. It is more powerful, however, in the markets for rice, ruminant meats, dairy products, and sugar. Were the United States ever to join forces with the Cairns Group in acting to restrict food supply, the resulting bloc would be particularly powerful. Even in the long run it would face elasticities less than two in the key grains, less than five in the sugar market, and less than ten in the markets for rice and ruminant meats. The ultimate supply-side bloc would combine the United States and the Cairns Group with the EC. Improbable though it may be, collusion to restrict the excess supply of such a bloc would have very substantial effects on world food prices in both the short and long runs. As we saw in Section 2, however, this immense power in world ‘Western Europe’s market power might be enhanced if increased cooperation on agricultural policy led to higher protection and hence increased subsidized exports. At present, the EFTA countries have higher average rates of protection on GLS products (see Tyers and Anderson 1988a).
+ USA
VSR LR VSR LR VSR LR VSR LR VSR LR VSR LR VSR LR VSR LR -3.3 -3.3 - 18.2 -3-7 - 19.8 -3.7 - 19.1 - 1:o -5.4 - 0.27 -1.4 - 0.46 - 2.4 - 0.22 -1.1
-0.60
(5.3) (36.3) (5.3) (34.5) -2.1 - 15.2 -0.24 - 1.9 -0.52 -3.6 -0.25 -1.6
- 0.46 -3.7 (38.6)
-3.5 - 14.1 -1.9 -7.5 -6.5 -25.0 -1.9 -7.8
-5.9 - 22.7
Rice
-1.5 - 10.5 -2.1 -9.1 (9.3) (35.5) - 1.9 -7.8
(10.0)
Ruminamt Meat
-9.4 - 32.5 -4.6 - 17.7
-25.6
- 19.8
- 20.2
-21.4
Nonruminant Meat
-2.5 -7.6 -2.5 -7.5 -2.1 -6.3 -6.6 -25.7 -4.0 - 13.1 -2.0 -4.8 -1.1 -2.8
- 14.2
Dairy Prducts
- 1.4 -2.2
(6.5) (12.2) -4.6 -8.8 -5.1 -9.5 -5.7 - 10.6 - 1.7 -2.8 -2.6 -4.2 -22.5
Sugar
adjustment.
“Elasticities not shown have magnitudes greater than 40. Values in parentheses are the excess supply elasticities faced by the country; the other values are excess demand elasticities. VSR refers to the very-shost-run adjustment within the first year, SR to adjustment after one year, and LR to the long nm after full
Source: Derived from the model described in the text.
USA + EC - 12 + CAIRNS Group
EC-12
USA + CAIRNS Group
CAIRNS Group
EC-12 + EFTA
EC- I2
EC-IQ
United States
T&;:le 5: Trade Elasticities Facing Groups of Countries, 1980-82” _Ip__ COarSe Grain wheat
330
Rod Tyers and Kym Anderson
food markets stems primarily from the insulating policies of countries in *be rest of the world. Were they to collectively expose all their private agents to movements in international food prices, the power of such a supply-side bloc to raise the price would be cut by more than half. This is demonstrated in Table 6, where the trade elasticities are listed for the same country groups but under the assumption that no insulation occurs in the rest of the world. For the most powerful group, this change raises trade elasticities between two- and sevenfold. 4. SOME CAVEATS The model we use has a number of useful features from the viewpoint of obtaining trade elasticities. The main ones are its inclusion of interdependent commodity markets, its use of separate price transmission equations for producer and consumer prices, and its dynamics, which allow distinctions between short- and long-run adjustment in both price transmission and production in each commodity market in each country. It does lack a number of desirable features, however. From the list of such features identified by Abbott (1988), those that are rIGssing from our approach include the capacity to differentiate responses to upward from downward movements in prices. This applies both to price transmission, where asymmetric political pressures result in greater insulation against either upward or downward movements in international prices, and to production, where gains achieved in part through technical change or fixed capital formation are not reversible if prices decline. The latter point is less problematic, however, because production increases due to technical change are represented in the GLS model by irreversible shifts in supply curves (see Tyers and Anderson 1987 for a discussion on this point). Nevertheless, because the elasticities in the model are based on data that include both upward and downward price movements, the implicit trade elasticities are likely to be biased. In the case of industrial counties where farmers are protected, upward price movements are likely to be more fully transmitted than downward movements, and the same bias might be expected to apply to production. Thus, in commodity markets where these countries dominate, the decline in global excess demand in response to an increase in a country’s offer price would be larger than our estimates suggest, while the response to a decrease would be smaller. On the vther hand, in markets dominated by developing countries with consumer-oriented policies, such as that for rice, we would expect our estimates to have the opposite bias.
+ EFTA
+ USA
VSR LR VSR LR VSR LR VSR. LR VSR LR VSR LR VSR LR VSR LR
-3.8 -6.9 - 15.1 -27.8 - 16.7 - 30.6 - 16.3 - 30.0 -5.4 - 10.3 -2.1 -3.4 -2.8 -4.4 -1.6 -2.3 -5.7 -21.9 -1.2 -3.8 -2.1 -6.0 - 1.3 -3.2
(13.0)
(13.3)
- 1.9 -6.4
Grain
Coarse
12.2 28.2 14.1 29.2 (38.4) - 10.8 -20.7
-
Rumlmant Meat
- 24.9
Nonruminant Meat
15.0 --21.4 - 10.6 - 14.8 -
-39.1
- 20.0 -31.2 - 20.0 -31.0 - 16.9 - 26.0
Dairy Products
-7.3 - 10.2
- 19.2 - 28.6 -21.5 -31.7 - 23.8 - 35.0 -8.9 - 13.2 - 13.9 - 20.7
(27.8)
Sugar
“Perfect price transmission is assumed in all countries (price transmission elasticities set to unity). Elasticities not shown have magnitudes greater than 40. Values in parentheses are the excess supply elasticities faced by the country; the other values are excess demand elasticities. VSR refers to the very-short-run adjustment within the first year and LR refers to the long nut after full adjustment.
price transmission elasticities set to unity (no insulation).
- 13.3 - 18.5
-23.4 - 32.9 - 13.0 - 17.9
-37.1
Rice
&urce: Derived from the model described in the text, with all international-to-domestic
USA + EC - 12 + Cairns Group
EC-12
USA + Cairns Group
Cairns Group
EC-12
EC-12
EC-10
United States
Wheat
Tabk 6: Trade Elasticities Facing Groups of Countries: The Case of No Market Insulation”
332
RodTyersand KymAnderson
A second concern is with the assumption of intercountry homogeneity within each of the seven identified food commodity markets. Even though the model suffers less from the aggregation biases common in the economy-wide models reviewed by Brown (1987), each commodity is traded in numerous grades and degrees of processing. Because the exports of each country are differently specialized, these markets do behave in ways consistent with some degree of differentiation by country of origin. This suggests that the true trade elasticities may be smaller than our estimates; however, this bias is generally unimportant where substitutability within commodity aggregates is high compared with that between commodities. Finally, the capacity of the model to project changes in trade elasticities over time is dependent on the stability of market-insulating policies. It is likely, for example, that domestic markets tend to be more insulated from large changes in international prices than they are from small changes. The degree of insulation is also very sensitive to changes in policy type. An important concern of Abbott is with the increasing prevalence of state trading in food commodity markets. This practice is usually highly insulating; as a result, price transmission elasiicities may be smaller than those implied by any analysis of historical time series. If this is true, it is also a source of upward bias in our estimates. 5. CONCLUSICINS Bearing in mind the features of the model and its important omissions, the key points to note from the implied trade elasticities reported in this paper are: 1. Our estimates of short-run (one year response) and long-run trade elasticities facing all individual economies exceed unity, thus ruling out sustained increases in export revenue from restricting excess supply in any individual commodity market. 2. Except for wheat and coarse grain in the United States and wheat in the USSR, the estimated long-run trade elasticities facing all individual economies have magnitudes larger than 7.5. 3. Some small economies have low export demand elasticities in certain commodities (rice in Thailand, other grains and beef in Argentina, ruminant meats and dairy products in New Zealand), whereas large econ.omies do $101necessarily have low trade elasticities (Japan, for example). 4. Trade elasticities would be much larger (in most cases more
’
PRICE ELASTICITIES
IN INTERNATIONAL
FOOD TRADE
333
than double) were it not for the domestic-market-stabilizing behavior of government policies. 5. These elasticities vary over time because food production and demands are changing at different rates in countries with different degrees of insulation and because insulation in some countries is increasing, so that they are smaller in the late 1980s than in the early 1980s. 6. For the United States, the grain export demand elasticities obtained from our model appear to be somewhat larger than those from other studies, in part because it allows for more substitution in production and consumption between related products. Because these synthetic estimates are necessarily less than perfect, it is difficult to draw strong conclusions about the precise magnitude of food trade elasticities. They certainly cast doubt, however, on the proposition that any country has strong monopoly or monopsony power in international food markets in anything other than the very short run. While some aggregates of the larger exporting countries have substantial potential market power, even in the long run, this stems primarily from insulating policies in the predominantly food-importing countries (and hence from a lack of preparedness on the part of gocremments to permit their domestic consumers, producers, and/or stockholders to share in the costs of adjustment to international price changes). This result suggests that those countries would be less vulnerable to the exploitation of market power were they to adopt less distortionary (and, particularly, less insulating) agricultural and food policies. REFERENCES Abbott, P.C. (1979) Modelling International Grain Trade With Government-Controlled Markets, American Journal of AgriculturaZ Economics 6 I( 1):22-3 1. -. (1988) Estimating U.S. Agricultural Export Demand Elasticities: Econometric and Economic Issues. In Elasticities in International Agricultural Trade (C. Carter, Ed.). Boulder: Westview Press. Anderson, K., and Tyers, R. (1985) European Community Grain and Meat Policies: Effects on International Prices, Trade and Welfare, European Review of Agricultural Economics 11(4):267--294. -. (1986) Agricultural Policies of Industrial Countries and Their Effects on Traditional Food Exportersi The Economic Record 62( 179):385-399. -. (1988) Global Effects of Liberalising Trade in Agriculture, Thames Essay No. 55. London: Trade Policy Research Centre. Bredahl, M.E., Meyers, W.H., and Collins, K.J. (1979) The Elasticity of Foreign Demand for
334
Rod Tyers and Kym Anderson
US Agricultural Prod xzts‘The Importance of the Price Transmission Elasticity, Americwt Journal of Agricultu -al Economics 61( 1):58-63. Brown, D.K. (1987) Tariff>, thz Terms of Trade and National Product Diffemntiation, Journal of Policy Modeling 9(3):503-526. Gardiner, W.H., and Dixrt, P.M. (1987) Price Elasticity of Export Demand: Concept and Estimates. FAER ‘Jo. 228, U.S. Department of Agriculture, Washington, D.C. Johnson, P.R. (1977) The Elasticity of Foreign Demand for U.S. Agricultural Products, American 1 Economics 59(4):735-736. Journal of Agricu Johnson, S.R., Womack . A. WT.,Meyers, W.H., Young, R.E. II, and Brandt, J. (1985) Options for the 1985 Faru Bill: An Analysis and Evaluation. FAPRI Report No. l-85, Food and Agricultural Polic ;, Research Institute, University of Missouri-Columbia and Iowa State University. Meyers, W.H., and Helrudt, M.D. ( 1986) Trade Implications of the Food Security Act of 1985. Staff Report No. 86SX4, Center for Agricultural and Rural Development, Iowa Siate University. Meyers, W&I., Devados,, S., and Helmar, M.D. (1987) AgriculturalTrade Liberalisation: CrossCommodity and ,Jross-Country Impacts, Jolcnal of Poticy Modeling 9(3):455-484. Grcutt, G.H. (1950) Measlurementof Price Elasticities in International Trade, Review of Economics and Stat ‘,-rics32(2): 113- 132. Tyers, R. (1984) Agrict&ual Protection and MarketInsulation: Analysis of International Impacts by Stochastic Simulation. Research Paper No. 111, Australia-Japan Research Centre, Canberra. -. (1985) Interr&onal impacts of Protection: Model Structure and Results for EC Agricultural Policy Journal of Policy Modeling 7(2):219-251. Tyers, R., and Anders m, K. (1987) On Modelling the Effects of Agricultural Policies, T&kwhrifi voor Sociaai H.~st~nschappslijkonderzoek van de Ltmdbouw Jaargang 2, No. 2, pp. 150158. -. (1988a) “Li’&lising OECD Agricultural Policies in the Uruguay Round: Effects on Trade and V&!fae, Journal of Agricultural Economics 39(2):197-216. -. (1988b) ln-+&ect Price Transmission and Implied Trade Elasticities in a Multicommodity Wc.rldFcod Model. In Elasticities in International Agricultural Trade (C. Carter, Ed.) Boulder: Westview Press. -. (for&con+) Distortions in Werld Food Markets. Cambridge: Cambridge University Press. World Bank (1986) WSrld Development Report 1986. Oxford: Oxford University Press.
Rice Wheat CGrain sugar Dairy RMeat NRMeat
699 57772 6299 14164 118757 7520 14813
Rice 0.40
0.38 5.00 0.38 6.00 0.38 0.40
- 0.66
0.92 -0.10 -0.01 -0.01 -0.37
0.90
CGrain
-0.51 -0.10
Wheat
- 0.06 -0.05 0.50
Sugar
0.51 0.12
Dairy
- 0.03 1.02 -0.30
RMeat
- 0.48 1.14
NRMeat
0.02 0.25 - 0.90
NRMeat
0.02 -0.60 0.26
RMeat
-0.40 0.02 0.02
Dairy
Long-run elasticity of supply with respect to the price of:
0.05 -0.12
0.10 0.02 -0.20 0.01
0.25 945 -0.80 Rice -0.30 47850 0.01 Wheat 0.17 70195 CGrain 10533 sugar 107187 Dairy 7632 RMeat 14029 NRMeat Indirect demand parameters for coarse grain: Shares of livestock sectors grain-fed Grain use per unit of output
Reference production (kt)
Sugar
CGrain
Rice
Elasticity of Demand with Respect to the Price of: _.Wheat
Trend Consumption @t)
(a) THE EUROPEAN COMMUNITY
Table A.$: A Sampling of Key Demand, Supply, and Price Transmission Elasticities for Major Industrial Countries
(b) JAPAN
C6rain sugar Dairy RMeat NRMeat
wkak
Rice
le A.1:
-0.03
/,,
,,,, ,,
,,
0.01 0.14 - 0.40
Rice
0.04
0.10
t-2 t
0.12 -0.06
0.02
t-1
0.22 -0.14
-0.03
t-2
Ruminant meat
,,,,,,,, ,,,,,,,,,
- 0.05
Sugar
0.46 0.40
-0.80
Dairy
0.46 6.00
- 1.40 0.25
RMeat
Elasticity of Demand with Respect to the Price of:
0.07
CGrain
-0.01
0.10
- 0.02 - 0.02
t-1
Wheat
Reference Consumption (kt)
-0.22
-0.22 0.40 - 0.02
t
Dairy
Rice 10472 - 0.23 0.03 Wheat 633 1 0.24 -0.60 CGrain 19436 0.16 0.25 Sugar 285 1 O.@l Dairy 8113 RMeat 706 NRMeat 2904 Indirect demand parameters for coarse grain: Shares of livestock sectors grain-fed Grain use per unit of output
0.30 -0.22 -0.02
t-1
t-2
t-1
t
t-1
t-1
0.20
Sugar
coarse grak
Wheat
Short-run elasticity of supply with respect to the price of:
Rice
(Japan continued)
1.00 5.00
0.40 -1.00
NRMeat
t
0.76
t-1
t-2
-0.16
Nonruminant meat
t-1
0.08
0.30 -0.20
t-1
Rice Wheat CGrain
(c) THE UNfTED STATES -
Rice Wheat CGrain sugar Dairy RMeat NRMeat
Wheat
Rice Wheat CGrain Sllgti Dairy RMeat NRMeat
-0.15 0.30
t-1
Reference Consumption (kt) 2015 26958 155456
-0.01 0.01 -0.01 -0.03 -0.05
t t-2
-
Rice 0.20 0.60 -0.40
Wheat -0.30 0.60
CGrain Sugar
Dairy RMeat
Long-run elasticity of supply with respect to the price of:
Rice -0.20 0.01 0.01
0.10
t-1
Sugar
0.35
t-1
0.30 0.02
t-2
-0.10
t
-0.02 0.10 - 0.02
t-1
-0.02 0.40
t-2
Ruminant meat
Wheat 0.08 -0.12 0.08
CGrain 0.04 0.06 -0.20
0.07
Sugar
Dairy
RMeat
Elasticity of Demand with Respect to the Price of:
t
Dairy
0.50 - 0.06 0.80 -0.09 - 0.06 0.04 0.80 -0.23 - 0.06 Short-run elasticity of supply with respect to the price of:
- 0.02 -0.04
Coarse grain
Reference production (kt) 9375 675 399 853 6798 478 2619
0.33
t-1
t-2
- 0.05
Nonruminant meat
NRMeat
t
-0.10 0.99
NRMeat
t-1
-0.01
-0.15
-0.09 0.45
t-1
0.35 -0.02
-0.20
-0.01
t
-0.01 -0.02 -0.01
-0.30 0.40
t-1
0.02 - 0.50 0.20 0.67 6.00
-0.30 0.02 0.01 0.67 0.40
Wheat -0.20 0.80 -0.28 -0.53 0.75
CGrab
Sugar -0.04 Dairy
t-2
-
0.07
t-1 - 0.02
Sugar
0.07
t
0.02 0.01
t-1
Dairy
0.08
t-2
0.03 -0.20
t
- 0.20 0.72 -0.13
RMeat
t-2
t
-0.16 1.12
0.61
t-1
t-2
- 0.08
Nonruminant meaat
NRMeat
1.00 5.00
0.01 0.20 -0.80
A more complete list is also provided in
0.03 -0.10 0.24 0.32 - 0.05 - 0.02
t-1
Ruminant meat
0.28 -0.08 0.85 -0.24 0.03 -0.38 Short-run elasticity of supply with respect tenthe price of:
-0.04
Rice 0.75 -0.04
-0.10
Coarse grain
(W 4713 72301 211494 5321 61807 10578 13991
-0.20
!L@q-run elasticity of supply with respect to the price of:
0.05
Swrce: The parameters of the model are presented as detailed estimates in Tyers and Anderson (forthcoming). Tyers and Anderson ( 1988b).
CGrain sugar Dairy RMeat NRMeat
Rice Wheat
Wheat
Rice
Rice Wheat CGrain Sugar Dairy RMeat NRMeat
Reference production
8693 Dairy 60503 RMeat 11890 NRMeat 13825 Indirect demand parameters for coarse grain: Shares of livestock sectors grain-fed Grain use per unit of output
sugar
USSR
United States
Japan
EFTA
EC-10
Canada
Australia
SR LR SR LR SR LR SR LR SR LR SR LR SR LR
0.78 1.00 0.68 1.00 0.09 0.20 0.11 0.79 0.20 1.00 1.00 1.00 0.05 0.45
0.11 0.63 0.68 1.00 0.08 0.11 0.11 0.79 0.06 0.25 1.00 1.00 .os 0.45 0.69 0.96 1.00 1.00 0.24 0.58 0.15 1.00 0.20 1.00 1.00 1.00 0.02 0.17
P 0.69 0.96 1.00 1.00 0.13 0.26 0.15 1.00 0.02 0.12 1.00 1.00 ox!2 0.17
c
P
Wheat c
Coarse grain
0.62 .84 0.90 0.90 0.11 0.46 1.00 1.00 0.06 0.55 0.82 1.00 0.06 0.30
P
C 0.23 1.00 0.90 0.90 0.11 0.22 0.30 0.30 0.03 0.12 0.71 1.00 0.06 0.30
Rice
0.73 1.00 0.27 0.46 0.24 0.45 0.01 0.04 0.10 0.24 0.60 0.61 0.05 0.20
P 1.00 1.00 0.08 0.40 0.14 0.45 0.01 0.04 0.10 0.24 0.21 0.53 0.05 0.20
c
Ruminant meat
0.46 0.52 0.08 0.40 0.12 0.76 0.13 0.68 0.49 0.63 1.00 1.00 0.05 0.20
P
0.25 0.34 0.83 0.85 0.62 0.76 0.16 0.16 0.47 0.86 1.00 1.00 0.05 0.20
C
Nonruminant meat
Table Q-2.: A Sampling of Key Elasticities of Transmission of International Price Changes to Domestic Prices”
0.40 0.45 0.96 0.40 0.08 0.30 0.06 0.19 0.03 0.08 0.07 0.36 0.05 0.13
P
0.13 0.39 0.96 0.40 0.08 0.30 0.06 0.19 0.03 0.08 0.06 0.18 0.05 0.13
c
Dairy products
0.49 0.54 0.07 0.25 0.00 0.00 G.%I 0.00 0.00 0.00 0.10 0.48 0.02 0.04
P
0.00 0.00 0.12 0.60 0.00 0.00 o.GG 0.00 0.00 0.00 0.10 0.48 0.02 0.04
c
Sugar
SR LR SR LR SR LR SR LR SR LR SR LR 0.05
0.60 0.15 0.90 0.09 1.00 0.40 0.60 0.80 1.00 0.42 0.79
0.44
0.60 0.15 0.90 0.09 1.00 0.40 0.60 0.80 1.00 0.42 0.79
0.54 0.87 0.14 0.80 0.47 0.94 0.85 1.00 0.70 0.80 0.57 1.00
0.05 0.70 0.14 0.80 0.46 1.00 0.85 1.00 0.70 0.80 0.35 0.42
c
P
P
c
COtUW? grain
Wheat
0.35 0.58 0.17 0.26 0.20 0.60 0.49 0.74 0.56 0.56 0.16 0.46
P
.,,,,,
,,
,,
/,.,
,,,,,,
,,,
,//
consumer prices, respectively.
88,
,,
3,
,,,,
,,/
3,
,,,,
8,
,,,,
c 0.48 0.66 0.15 0.40 0.05 0.40 0.17 0.30 0.58 0.63 0.44 0.60
P 0.05 0.50 0.15 0.40 0.05 0.40 0.17 0.30 0.77 0.90 0.44 0.60
c
Ruminant meat
0.17 0.25 0.15 0.60 0.05 0.40 0.18 0.50 0.43 0.46 0.72 0.77
P
0.05 0.22 0.15 0.60 0.20 0.40 0.18 0.50 0.66 0.80 0.72 0.77
C
Nonruminant meat
0.10 0.16 0.15 0.25 0.05 0.20 0.01 0.20 0.34 0.35 0.54 0.51,
P
0.05 0.12 0.15 0.25 0.02 0.20 0.01 0.20 0.34 0.35 0.54 0.54
c
Dairy products
0.19 0.23 0.09 0.20 0.02 0.20 0.24 1.00 0.00 0.00 0.24 0.90
P
0.05 0.20 0.09 0.20 0.02 0.20 0.24 1.00 0.00 0.00 0.24 0.90
c
Sugar
,,,
./,,,,,m,
,,,,
geometric lag structure connecting them); P and C refer to domestic producer and
0.05 0.40 0.17 0.26 0.05 O.?O 0.31 0.58 0.56 0.56 0.26 0.32
Rice
“SR and LR refer to short-run and long-run elasticities (with a Nerlovian
Brazil
Argentina
Thailand
Indonesia
India
China
Table A.2. (continued)
PRICE ELASTICITIES
IN INTERNATIONAL
FOOD TRADE
341
APPENDIX HI: DERIVATION OF IMPLIED TRADE ELASTICITIES FROM THE GLS MODEL The model treats each of its 30 countries or country groups as “large,” in that changes in the trade volumes of each induce nonzero changes in international prices. Of special interest are the excess supply or demand elasticities (of “the rest of the world”) that face each country. Where these are comparatively small in magnitude for particular countries, those countries have the capacity to increase domestic welfare at the expense of the rest of the world by imposing trade taxes. Consider a world comprising N countries (K = 1, . . . , N) and several commodities. Let mf be the excess demand in the rest of the world for commodity i produced by country K. This excess demand might be either positive or negative, the latter indicating that country K is a net importer. The excess demand elasticity es is then the proportional adjustment that occurs in rnfin response to a given proportional change in the price at which country K trades commodity j (Pi, where j and i may or may not refer to the same commodity): dm% dP. et =2/I mf Pjm
(A.11
We summarize here an analytical approximation to the elements of lc. structural equations and parameters of the mauix [et], based on tPa the model (a more detailed account is provided in TyerS atid Anderson 1988b and forthcoming). First, we seek an expression for the adjusment in excess demand in the rest of t:rc world, L%v$~ in terms of changes in prices. This adjusment is simply the sum of the adjustments made in the countries, k, of the rest of the world group: dmf = Z (t&z + dcz - &a + dSia. k#k
(A.2)
first term is the adjustment in direct human demand, &, the second is that in livestock feed demand, cc, the third is that in production, qjk, and the final term is the adjustment in stock levels~ s,. It is convenient to express each of these terms as functions of domestic prices in country k (#A). The direct human consumption adjustment is readily derived from the differentiation of the direct consumption equation to give
The
(A.3)
where aijkis the elasticiity of direct human demand for commodity i with respect to the consumer price of j and & is the consumer price of j, both in country k. The corresponding adjustments in feed demand, production, and
RodTyersandKymAnderson
342
stock levels depend on the length of run. Here we consider three extreme cases discussed in the paper: adjustment in the same year, after one year, and in the long run. tintheSameYear The relationship between feed demand and livestock production is based in the model on fixed input-output coefficients that link that demand to an index of steady state output for each livestock product. Taking derivatives of the terms in the feed demand equations, which depend on prices in the curre year, an summing over the feedconsuming sectors, n, we obtain following expression: (A.41
where 0~~~is the fixed input-output coefficient linking feed demand to onB is a coefficient indicating the deviation of the steady-state output and +T same-year production response from that in the long run (steady state), as explained in detail in Tyers and Anderson (1988b). ISnR is the partial adjustment elasticity for the production of livestock product n in country k (if Sti is unity, the entire response is immediate). The product of ankand bhjk is then the same-year output response elasticity for livestock product n with respect to the price of c ity j in country k. Because livestock producers respond as consumers of feed grains, the.prices in this equation are consumer prices where commodity j is a livestock feed and producer prices where it is a livestock product. The corresponding production adjustment is derived in the same way: (A.3
where the same xule applies regard ng the prices in the final term of equation (AS) as applies for (A.4). The stock level adjustment dsikis a complex function of the changes in both production and consumer prices, dqikand dpz (derived in Tyers and Anderson 1988b). When it is combined withequations (A.3), (A.4), and (AS), one obtains the following expression for the adjustment in the overall excess demand of country k in terms of price changes: (A.6)
are constants in any given year. de elasticities via e
ion (A.
1) 9
expressions relating
PRICE ELASTICITIES IN INTERNATIONAL FOOD TRADE
343
domestic prices, &’ andp$, to international prices, Pi, are also lzeded. These are given by (A.71
where +iM and +F are the same-year international-to-domestic price transmission elasticities for consumers and producers, respectively. The adjustment in the excess demand of “the rest of the world” for the exports of focus country, K, is then (A-8)
where =he sum over countries k excludes the focus country, K. From equation (A. l), the excess demand elasticity facing country K is then
Adjustment after One Year In this case the terms in equation (A.2) depend on the sum of the same-year and the one-year lagged adjustments in consumption, production, and stocks. It is convenient to begin with the production adjustment, where the one year production response is (A. 10)
where b,, is the one-year component of the long-run elasticity of production response of commodity i with respect to the price of commodity j in country k. Once again, the prices in this equation may be either consumer or producer prices, as in equations (A.4) and (AS). The response of feed demand after one year then depends on the adjustment of steady-state livestock production to changes in price. In this case the feed consumption adjustment is
where ai& is the same adjusted input-output coefficient as in equation are coefficients indicating the deviation fo the same(A-4), TOnk and +rlnk year and one-year response in the production of n from that in the long run (steady state), d& is the same-year production adjustment given in equation (AS), and && is the one-year adjustment given in equation (A. 10).
Rod Tyers and Kym Anderson
344
The adjustment in country k’s overallexcess demand can then be formulated in terms of changes in domestic producer and consumer prices, as in equation (A.@. It remains to link these domestic price changes to changes in international prices. This is readily achieved by using the formulation of the one-year price transmission elasticity in terms of the corresponding same-year and long-run elasticities, which are both parameters in the model. Because the expression is identical for domestic consumer sand producer prices, we drop the P and C superscripts: (A. 12)
where $y is the long-run price transmission elasticity. An expression similar to equation (A.@ readily follows. The one+ies can then be calculated from a further equation year trade @ g Run
Adjustment in t
In this case, the global excess demand elasticities embody all longrun quantity adjustments to price changes. Again, equations (A. 1) and (A.2) apply, except that changes in stock levels in the long run do not effect annual excess demand. Accordingly, the third term in equation (A.l) is dropped. The changes in direct and feed consumption are now (A. 13)
dcf, =
xn %t
qnk
2 @tijk+ btnjk+ bad i
dP*
F-
(A. 14)
Jr:
Note that the three components of long-run supply elasticities are summed in the second term of equation (A. 14). This same total longrun supply elasticity is the basis for the change in production: dq,&= qi&Z (bwk + b,, + b& 91s J
(A. 15)
Pjk
Substituting equations (A.13), (A.14), and (A.15) into (Al), and setting dsikto zero, we derive and expression like equation (A.@. The expression for the long-run trade elasticity then differs from equation (A.9) only in that the constants Ajk and Bjkdiffer and the price transmission elasticities that appear are the long-run values +i?tnd +s?
e:
Rod Tyers and Kym Anderson, University of Adelaide
Govenunent intervention, particularly in agncuhural commodity markets, is frequently justified on the grounds that it favorably affects the terms of trade. In this paper an established dynamic simulation model of seven food commodity markets is used to provide synthetic estimates of the elasticities of net export demand and import supply on which such justifications rest. Estimates are presented for both the short and long runs, and the effects of market-insulating agricultural policies on these elasticities are investigated. The results cast doubt on ti F proposition that any individual economy has strong monopoly or monopsony power in international food markets in anything other than the very short run. But effective cooperation by groups of exporting countries, such as the Cairns Group of “nonsubsidizing” agricultural exporters, along with the United States or even the EC, could yield substantial market power to those groups in both the short and long runs. Nevertheless, the major part of the power such groups might wield stems from self-imposed domestic-market-insulating agricultural policies in the rest of the world.
Estimates of elasticities in international trade are continually in demand. They are of particular importance to policy analysts seeking to measure the likely effects of government intervention in countries that have some monopoly or monopsony power in the international market. Food price elasticities are especially sought after because (1) most governments intervene in their food markets and (2) a number of those countries have-or believe they have-substantial market power in international food trade, in which case their perceived optimal food trade tax may be positive rather than zero. There are two alternative approaches to obtaining price elasticities in international trade: direct estimation, which involves regressing trade
Address correspondence to Rod Tyers, Department of Economics, University of Adelaide, Adelaide. S.A. 5001, Australia. This paper stems from research on trade elasticities completed in association with the International Agricultural Trade Research Consortium Symposium on Elasticities in International Trade, Dearborn, Michigan, July 1987. It is one of two papers based on the same food trade model that emerged from that workshop. For useful comments and suggestions thanks are due to F. Gerard Adams and to the participants in the Michigan symposium. ; final draft accepted. Received
Journal of Policy Modeling Il(3):315-344 (1989) 0 Society for Policy Modeling, 1989
315 0161-8938/89/$03.50
Rod Tysrs and Kym Anderson
316 volumes
against
border prices, and synthetic estimation. Serious econometric difficulties arise with direct estimation methods, as was pointed out long ago by Orcutt (1950). It is appropriate, then, to consider various synthetic estimation methods. Ideally, synthetic estimates of trade elasticities for particular commodities should be obtained from a multicommodity, multicountry dynamic general equilibrium model of the world economy. In the absence of such models, however, analysts have relied on lesscomu plete models’ or world markets. Single-commodity models are often used, but *hey miss the important interactions that exist between closely related ~smmodities. In this paper we use a multicommodity, multicountry &inamic simulation model of markets for grains, livestock products, and sugar. This model has been widely applied in the analysis of distortions affecting brade in food commodities (see, for example, Anderson and Tyers 1; 35, 1886, 1988; Tyers and Anderson 1988a; World Bank 1986, Ch. 6), although its implicit trade elasticities have not been examined previously. Section 1 describes the features of the model. From the viewpoi;!nt of obtaining trade elasticities, its key features are its coverage of multiple commodity markets, which ensures that the interactions between various food markets are explicitly represented; its dynamics, which permit both short- and long-run trade elasticities to be obtained; and its inclusion of estimated short- and long-run international-to-domestic price transmission elasticitie s (distinguished as between producer and. consumer prices) to represent the stabilizing and insulating behavior of government policies. In Section 2 the approach used to obtain the trade elasticities directly from the parameters of the model is outlined and the elasticities are reported. Included are values for the short and long run (with and without market-insulating policies) and for both the early and late 1980s. Section 3 presents the magnitudes of trade elasticities facing a number of country groups and assesses their potential market power. Section 4 then lists some caveats that should be borne in mind when interpreting the elasticities; concluding comments are provided in Section 5.
The model we use provides dynamic, stochastic simulations of the world markets for grains, livestock products, and sugar (CLS). It is an extended and updated version of the model presented by Tyers (1984, 1985). To keep it manageable, the level of commodity disaggregation is restricted to seven groups: wheat, coarse grain, rice, meat
PRICE ELASTICITIES
IN INTERNATIONAL
FOOD TRADE
317
of 6ruminants (cattle and sheep), meat of nonruminants (pigs and poul4ry), dairy products, and sugar. These account for about one half of the world food trade (e%ble oils and beverages accozni 5~ -nest of the rest) and one tenth of global trade in all commodities. Salient features of the model include the following: 1. It is global in coverage, identifying 30 countries and counq groups. 2. It incorporates the cross effects, in both production sad consumption, between the seven interdependent commodity groups and separates the direct demand for grains as food from the indirect demand for animal feeds. 3. Stockholding behavior is included endogenously, based on empirical analysis of stock level responses to price and quantity changes in each country. 4. Lags in supply adjustment to price changes are included. 5. Policies in each country are assumed to allow domestic pries to change only gradually (and in some cases not at all) m response to changes in intemational food prices. Production behavior is represented by Nerlovian reduced-form partial-adjustment equations th~;tare log-linear, resulting in constant shortand long-run supply elasticities. Special shifters are added to allow for the effect on production of land set-aside policies such as those used in the United States. Direct human consumption is characterized by income and price elasticities of demand, which are constant in any year but decline over time, whereas feed consumption by animals is determined by input-output coefficients that relate feed use to a steadystate level of production for each livestock product. Policies that affect domestic prices are incorporated via econometrically estimated international-to-domestic price transmission equations for each country and commodity. These equations capture both the protection and the stabilization components of food price and trade policies. They are based on estimates of reduced-form Nerlovian partial-adjustment equations that distinguish short-run from long-run elasticities of price transmission. Separate elasticities are used for producer and consumer prices. In general, even the long-run price transmission elasticities are less than unity, reflecting the prevalence of nontariff protection instruments in food markets. In the face of volatile and declining real prices in international commodity rr,arkets, governments limit the extent to which both the long-run trend and the short-run changes in domestic prices follow those of international prices. The smaller the short-run elasticity of price transmission in relation to its
318
Rod Tyers and Kym Anderson
long-run counterpart, the greater the degree of market insulation and the more sluggish the eventual transmission of any sustained change in the international price. In a few extreme cases domestic prices are long-run elascompletely insulated, which means both the shortticities of price transmission are zero. Storage behavior is represented by a combin competitive speculation and the actions of public only to quantity triggers. Target levels of closing each commodity at a fixed proportion of trend c porting countries and trend production in exporting countries. Interon expected temporal deviations from these t;irget levels de speculative profits and the deviation in domestic y (production plus carry-in) from the long-run trend. Structurally, the model is a set of expressions for quantities consumed, produced, and stored, each of which is a function of known past prices and endogenous current prices. The model is solved iteratively by starting from the 198042 base period and beginning each subsequent year with the assumption that all prices are the same as those in the preceding year, generating random disturbances in production, and calculating new production, consumption, and closing stock levels in each country. The resulting excess demands are then totaled and international prices are adjusted to move world markets toward clearance. The procedure is then repeated until a satisfactory degree of market clearance has been achieved for each commodity. Thus, the model selects that series of international and domestic prices, production, consumption, and closing stock levels that simultaneously clears all markets in each successive year, from 1983 to 1995. Once 100 simulations of this type have been completed, each using a different set of generated random disturbances from thy distributions of each error term; forecast means and standard deviations are calculated for all key variables in the model for each year of the simulation period. They can also be calculated for the base period (1980-82) simply by replacing the values of all time-dependent parameters with their 1980-82 values. The solution procedure is conventional, but it is not based on a standard software package. A key assumption with important implications for ‘the subject of this paper is that of product homogeneity. The work of Brown (1987) and others has demonstrated that the degree of market power afforded by imperfect substitutability between like products from different countries is sufficient to substantially affect outcomes from government intervention. This * found to be particularly importan: in economywide models with ghly aggregated product categ ties. We see this
PRICE ELASTICITIES IN INTERNATIONAL
FOOD TRADE
319
as comparatively unimportant in the case of the ptial-equilibrium GLS model. Although disaggregation into seven commodities (the most important of which are the individual grains) does not in itself ensure perfect homogeneity, it renders the assumption more valid. The parameters of the model are based on time-series analysis with estimation intervals in the period 1962-83. Alternative estimates of the price transmission (policy) and storage parameters tend not to be readily avaiiable eisewhere in the literature and so their original values have remained largely unrevised. Many alternative estimates are available for the parameters governing domestic supply and demand behavior, however. Accordingly, a number of our original estimates have been revised where alternative values are based on more sophisticated country-level analysis. All of the parameters are documented in Tyers and Anderson (forthcoming). Because our purpose here is to examine the trade elasticities implicit in our model, and because these are only as good as the structural parameters on which they are based, we provide a sampling of key demand, supply, and price transmission elasticities in Appendix I.
2. IMPLIED TRADE ELXTICITIES
IN THE GLS MODEL
It is clear from the above description that the parameter set of the GLS model is sufficiently rich to imply potentially useful estimates of trade elasticities. They depend on domestic own- and cross-price elasticities of supply and demand, on the price responsiveness of storage, and on price transmission elasticities in the various countries, as well as on the shares of each country in different markets: they vary according to the time allowed for adjustment and the year chosen. Indirect estimates can therefore be derived oanalytically from the parameter set. This is done by calculating the extent to which the excess demand in the rest of the world for commodity i produced by country K adjusts in response to a change in the price at which country K trades commodity i. (The excess demand of the rest of the world may be a positive or negative quantity, the latter indicating that country K is a net importer of commod.ity i.) The detailed derivation of these elasticities based on the equations of the model provided in Tyers and Anderson (1988b). A summary of the key analytical results is given in Appendix II. With 30 countries/country groups and seven commodity groups, the total number of own-price trade elasticities is 210 for each year and for each period of adjustment. It is therefore necessary to be selective in presenting these parameters. Only the very-short-run (same year), short-run (after one year), and long-run (after full adjustment) trade
320
RodTyersandKymAnderum
elasticities are reported here.’ In Table 1 the own-price elasticities a= shown for the base period 1980-82 for large participants in world food markets wherever the magnitudes of those elasticities are less than 40. The key points to note from the implied trade elasticities in Table 1 are as follows: 1. The long-run trade elasticities have magnitudes larger than 7.5
2.
3.
4.
5.
6.
7.
8.
for all countries and commodities other than wheat and coarse grain in the United States and wheat in the EC and the USSR. The short-run trade elasticities (after one year’s adjustment) are never smaller than unity, suggesting that by reducing the volume of its food imports, no individual economy can increase its export revenue and sustain the increase through a full year or more. Following the United States, the smallest short-run export demand elasticities are faced by Canada, Australia, and the EC in the wheat market and by the EC in the market for dairy products. The smallest import supply elasticities are faced by the USSR and China in the wheat market and by 3apan and the USSR in the market for coarse grain. In the rice market, Thailand faces the lowest export demand elasticity, whereas Japan’s small share of the international rice trade results in a comparatively high import demand elasticity at the margin. Australia and New Zealand have the lowest export demand elasticities for ruminant meat, followed by Argentina, and the EC and New Zealand have the lowest for dairy products. In the sugar market, 10 of the countries listed in Table 1 face relatively low trade elasticities, with Brazil and the EC facing the lowest on the export side and the USSR facing the lowest on the import side. With the exception of coarse grain in the short run, Japan’s import supply elasticities are very large, suggesting that Japan has little monopsony power in international food markets.
These trade elasticities would be considerably larger, especially in the short run, if countries did not insulate their domestic markets ‘In our analysis the trade elasticities are calculated in matrices that include non-zero cross elasticities. Because our primary interest here is in these elasticities as measures of comparative market power at the margin, only the own-price elasticities are presented. More complete tabylations are available on request from the authors.
h
VSR SR LR VSR SR LR VSR SR LR VSR SR LR (14.9) -5.2 -11.5 - 30.0 (9.0) (19.8)
(6.1)
(2.8
- 19.3
-5.1 - 15.5 -38.1
Coarse Grain
-4.5 -8.3 - 16.5
-6.8 -11.9 -21.2
Rice
- 22.7
- 13.4 - 24.7
Ruminant Meat
Nonruminant Meat
Dairy Products
-4.5 -5.9 -8.5 -11.5 - 15.3 - 20.6
(9.9) (13.0) (18.0) - 30.0
Sugar
adjustment.
“Elasticities not shown have magnitudes greater than 40. Values in parentheses are the excess supply elasticities faced by the coun+q; the other values are excess demand elasticities. VSR refers to the very-short-run adjustment within the first year, SR to adjustment after one year, and LR to the long run after full
Source: Derived from the model described in the text.
Thailand
Brazil
Argentina
China
Wheat
PRICE ELASTICITIES IN INTERNATIONAL
FOOD TRADE
323
from changes in international prices. To see this, the elasticities were reestimated based on unit price transmission elasticities. The implied trade elasticities that emerge with this respecification, summarized in Table 2, are in most cases more than twice those in Table 1. This result suggests that the insulating component of agricultural policies has a very substantial depressing impact on the size of food trade elasticities and thus cannot be ignored in the process of estimating those elasticities, a point stressed by Abbott (1979) and Bredahl et al. (1979).2 The trade elasticities in Tables 1 and 2 refer to the base period 198082. Since then, however, the degree of market insulation has changed, particularly in the United States. There, increased participation by farmers in grain price support programs under the Food Security Act of 1985 has substantially reduced the responsiveness of output to international price changes. In addition, the shares of each country in world production and trade have changed. Both these effects are represented in a “reference’ 9 projection to 1988. In this projection, policy regimes other than that of the United States, as represented in the model, are assumed to remain unchq+l (see Tyers and Anderson (1988a) for a detailed account of this projection). Assuming that this projection turns out to be reasonable in retrospect, it allows us to gauge how implicit trade elasticities may be changing during the 1980s. Table 3 illustrates these trends for the very-short-run trade elasticities. In most cases the projected trade elasticities are somewhat smaller in 1988 than in the base period, the only exception being for sugar. The differences appear to be larger for the case where the stabilization component of policy is removed in all economies (price transmission elasticities set at unity). The latter observation suggests that the changes in the global geographic distribution of food production and consumption have led to a slight decline in the aggregate responsiveness of private agents such as consumers, farmers, and stockholders. More generally, the point that is clear from comparing Tables 1 and 2 with Table 3 is that, even when most underlying parameters are stable, trade elasticities are by no means constant over time. How do these implicit trade elasticities compare with those adopted ‘Removal of the protection component of agricultural policies may or may not raise the food trade elasticities in Table 2: it depends on whether production and consumption in the rest of the world is relocated to more- or less-elastic markets and whether the country in question becomes a larger or smaller participant in the world food economy as a consequence of the liberalization.
(6.1) (13.0) (12.2) (23.2) - 26.6
-9.2 - 18.2 - 15.2 - 30.0
-3.8 -6.9 - 15.1 - 27.8 (34.6)
- 13.3
(33.3)
03.5)
(7.8) (31.2) - 22.7
- 1.9 -6.4
Grain
Coarse
- 27.5 - 38.8
- 35.4
-37.1
Rice
Nonruminant Meat
- 20.0 -31.2
Dairy Prducts
- 24.4 - 37.7
(12.4) (18.9) (38.8)
-31.1
- 19.2 - 28.6
(27.8)
Sugar
price transmission elasticities set to unity (no insulation).
- 38.6
-36.3
Ruminant Meat
1980-
“Elasticities not shown have magnftudes greater than 40. Values in parentheses are the excess supply elasticities faced by the country; the other values are excess demand elasticities. VSR refers to the very-short-run adjustment within the first year and LR refers to the long run after full adjustment.
Source: Derived from the model described in the text, with all international-to-domestic
Thiland
Brazil
Argentina
China
USSR
New Zealand
Australia
Canada
Japan
European Community ( 10)
VSR LR VSR LR VSR LR VSR LR VSR LR VSR LR VSR LR VSR LR VSR LR VSR LR VSR LR
wheat
: Excess Demand (Supply) Elasticities Facing Large Participants in World Food Markets: The Case of No Muket Insulation,
United States
Ta 82”
(10)
ref. PTE=l ref. lTE=l ref. PTE=l ref. FTE=l ref. PTE=l ref. PTE=l ref. PIE=1 ref. PTE=l ref. F’TE=l ref. PTE=l ref. PTE=l (2.9) (11.4) (2.5) (10.3) -5.0 -24.3 (9.0)
-0.51 -3.1 -2.1 -8.9 (8.2) (37.6) - 1.8 -8.5 -2.9 - 13.9
Wheat
-3.6 -22.0
- 24.0
- 12.7
(28.3)
-11.2
- 3.9 - 20.4
-9.2 - 24.3
-9.3
-5.1 - 32.6 -6.4
(25.3)
-11.3
- 3.6 - 20.0 - 13.8
(3.3) (14.4) (7.2) (27.6) -26.1
-7.5 - 33.4
(24.1)
-3.4 - 14.1 (9.9)
(13.2)
-4.0 - 32.3 - 1.7 - 14.3 (31.0)
(33.6)
(20.7)
-4.7 -29.1
Sugar
Products
Meat
Meat
“Elasticities not shown have magnitudes greater than 40. Values in parentheses are the very-short-run excess supply elasticities faced by the country; the other values are very-short-run excess demand elasticities. Ref. refers to the reference scenario (corresponding with Table 1 for the base period), PTE= 1 refers to the scenario in which all price transmission elasticities are set at unity (corresponding with Table 2 for the base period).
- 18.7
(5.6) (13.3) (8.6) (19.0) -4.7 -11.8
(2.8) (7-O) -7.0 - 17.5 - 13.7 - 33.9
-0.27 -1.1
Rice
Dairy
Nonruminant
Ruminant
Facing Large Participants in World Food Markets, Projected to 1988”
Coarse Grain
Excess Demand (Supply) Elasticitk
‘ource: Derived from the model described in the text.
Thailand
Brazil
Argentina
China
USSR
New Zealand
Australia
Canada
Japan
European CornmuG;;
United States
Table 3: Very-Short-Run
326
Rod Tyers and Kym Anderson
in other studies? The most commonly estimated food trade elasticities are the excess demand elasticities facing the United States grain export sector. A sample of such empirical studies over the past decade is summarized in Table 4. The GLS model suggests that the elasticities are much higher than those used by Johnson et al. ( 1985), whereas our coarse grain elasticities are similar to theirs. Our long-run elasticities are also somewhat larger than those suggested by Bredahl et al. (1979), but they are substantially lower than those suggested by Johnson (1977). The work by Johnson et al. (1985) is evidently based on national models for the United States developed at the Food and Agricultural Policy Research Institute. These models use estimated trade elasticities to characterize the rest of the world in aggregate. A difficulty with the reliance on such estimates is that they are invariably based on timeseries analysis, which attempts to explain changes in the level of exports in terms of border price changes. Because exports have other determinants frequently omitted from such studies, specification error leads to the underestimation of the true (ceteris paribus) responsiveness of exports to border price. In the recent work of Meyers et al. (198?), the e!asticities quo:& are from a single-commodity analysis using an eight-region global model. In that study, trade elasticities are derived by comparing model solutions in which yield functions in the United States have been shifted. The estimates of trade elasticities that result are substantially larger than those from the earlier Johnson et al. (1985) and Meyers and Helmar (1986) studies. In the case of coarse grain, their one-year elasticity is even larger than our estimate from the GLS model. The Meyers et al. estimates would, of course, have been larger still had the adjustment to a supply shock in one commodity market been permitted to spread across several interacting markets. Although an experiment similar to this is discussed in their article, it was based on simultaneous supply shocks in all markets and hence sheds no light on the (ceteris parihus) trade elasticities that are the subject of this paper. 3. TRADE ELASTICITIES FACING COUNTRY GROUPS The magnitudes of trade elasticities are smaller, and hence market power is greater, for aggregates of either exporting or importing countries. To quantify this, we have calculated these elasticities for a number of count*y groups featuring prominently in t international trade negotiations. The results are present
-0.60 -1.2 -3.3 - 0.46 - 1.2 -3.7 -5.9 -11.1 - 22.7
-0.51 -1.0 -2.9 - 0.27 - 0.74 - 2.3 -4.7 -8.9 - 18.1
19*’
-0.00
-0.09
-6.7
- 10.2
to - l.3b
to - 1.7
Bredahl et al. Johnson” (1977, Table 1) (1979, Table 4)
- 0.25 “near 1 .O”
-0.16 “near 1.O”
-0.15 “near 1.O” -0.29
-0.11
-0.46
-0.30 -0.90
-0.14 -0.23
to -0.68 - ‘7.0
to - 1.5 to -3.3
to -3.1 to -5.0
“Assuming price transmission elasticities are unity. “For corn only; -0.29 to -2.6 for sorghum.
- 1.38 - 1.59
-0.9 - 1.23
Other Studies Meyers and Surveyed by Gardiner Meyers et al. and Dixit Johnson et al. Helmar (1985, Table 1) (1986, Table 5) (1987, Tables 2, 3,s) (1987, Tables 1, 2)
Source: The first two columns are derived from the model described in the text; other values ue from the studies specified.
VSR SR LR Coarse VSR grain SR LR Rice VSR SR LR
Wheat
1983
Thii study (Tables 1, 3)
7M_~le4: Comparison of Excess Demand Elasticities Facing U.S. Grain Exporters
328
Rod Tyers and Kym Anderson
which lists, for comparison, the trade elasticities facing the United States and the EC-10 (provided originally in Table 1). Other couutry aggregates for which trade elasticities are presented include the larger West European aggregates, the EC- 12, and the combination of the EC with the European Free Trade Association (EFTA) countries (Austria, Finland, Norway, Sweden, and Switzerland). Also listed are the trade elasticities facing the Cairns Group of “nonsubsidizing” agricultural exporting countries (Argentina, Australia, Brazil, Canada, Chile, Colombia, Hungary, Indonesia, Malaysia, the Philippines, New Zealand, Thailand, and Uruguay), along with those facing various combinations of country groups with the United States. The larger European groups are generally less powerful in the GLS markets than the EC-lo. This is because Spain, Portugal, and the EFTA countries tend to be net importers of GLS products. In combination, therefore, Western Europe has smaller net exports than the EC-10 and faces higher trade elasticities. The key exception to this pattern is in the market for dairy products, where the export demand elasticity faced by Western Europe is slightly lower than that faced by the EC-10 alone. Nevertheless, more widespread cooperation in West European agricultural policy is unlikely to enhance the region’s international market power3 The principal rivals of Western Europe in the current agricultural trade negotiations are the United States and the Cairns Group. The results in Table 1 suggest that, despite its numbers, the Cairns Group lacks the power of the United States in the key wheat and coarse grain markets. It is more powerful, however, in the markets for rice, ruminant meats, dairy products, and sugar. Were the United States ever to join forces with the Cairns Group in acting to restrict food supply, the resulting bloc would be particularly powerful. Even in the long run it would face elasticities less than two in the key grains, less than five in the sugar market, and less than ten in the markets for rice and ruminant meats. The ultimate supply-side bloc would combine the United States and the Cairns Group with the EC. Improbable though it may be, collusion to restrict the excess supply of such a bloc would have very substantial effects on world food prices in both the short and long runs. As we saw in Section 2, however, this immense power in world ‘Western Europe’s market power might be enhanced if increased cooperation on agricultural policy led to higher protection and hence increased subsidized exports. At present, the EFTA countries have higher average rates of protection on GLS products (see Tyers and Anderson 1988a).
+ USA
VSR LR VSR LR VSR LR VSR LR VSR LR VSR LR VSR LR VSR LR -3.3 -3.3 - 18.2 -3-7 - 19.8 -3.7 - 19.1 - 1:o -5.4 - 0.27 -1.4 - 0.46 - 2.4 - 0.22 -1.1
-0.60
(5.3) (36.3) (5.3) (34.5) -2.1 - 15.2 -0.24 - 1.9 -0.52 -3.6 -0.25 -1.6
- 0.46 -3.7 (38.6)
-3.5 - 14.1 -1.9 -7.5 -6.5 -25.0 -1.9 -7.8
-5.9 - 22.7
Rice
-1.5 - 10.5 -2.1 -9.1 (9.3) (35.5) - 1.9 -7.8
(10.0)
Ruminamt Meat
-9.4 - 32.5 -4.6 - 17.7
-25.6
- 19.8
- 20.2
-21.4
Nonruminant Meat
-2.5 -7.6 -2.5 -7.5 -2.1 -6.3 -6.6 -25.7 -4.0 - 13.1 -2.0 -4.8 -1.1 -2.8
- 14.2
Dairy Prducts
- 1.4 -2.2
(6.5) (12.2) -4.6 -8.8 -5.1 -9.5 -5.7 - 10.6 - 1.7 -2.8 -2.6 -4.2 -22.5
Sugar
adjustment.
“Elasticities not shown have magnitudes greater than 40. Values in parentheses are the excess supply elasticities faced by the country; the other values are excess demand elasticities. VSR refers to the very-shost-run adjustment within the first year, SR to adjustment after one year, and LR to the long nm after full
Source: Derived from the model described in the text.
USA + EC - 12 + CAIRNS Group
EC-12
USA + CAIRNS Group
CAIRNS Group
EC-12 + EFTA
EC- I2
EC-IQ
United States
T&;:le 5: Trade Elasticities Facing Groups of Countries, 1980-82” _Ip__ COarSe Grain wheat
330
Rod Tyers and Kym Anderson
food markets stems primarily from the insulating policies of countries in *be rest of the world. Were they to collectively expose all their private agents to movements in international food prices, the power of such a supply-side bloc to raise the price would be cut by more than half. This is demonstrated in Table 6, where the trade elasticities are listed for the same country groups but under the assumption that no insulation occurs in the rest of the world. For the most powerful group, this change raises trade elasticities between two- and sevenfold. 4. SOME CAVEATS The model we use has a number of useful features from the viewpoint of obtaining trade elasticities. The main ones are its inclusion of interdependent commodity markets, its use of separate price transmission equations for producer and consumer prices, and its dynamics, which allow distinctions between short- and long-run adjustment in both price transmission and production in each commodity market in each country. It does lack a number of desirable features, however. From the list of such features identified by Abbott (1988), those that are rIGssing from our approach include the capacity to differentiate responses to upward from downward movements in prices. This applies both to price transmission, where asymmetric political pressures result in greater insulation against either upward or downward movements in international prices, and to production, where gains achieved in part through technical change or fixed capital formation are not reversible if prices decline. The latter point is less problematic, however, because production increases due to technical change are represented in the GLS model by irreversible shifts in supply curves (see Tyers and Anderson 1987 for a discussion on this point). Nevertheless, because the elasticities in the model are based on data that include both upward and downward price movements, the implicit trade elasticities are likely to be biased. In the case of industrial counties where farmers are protected, upward price movements are likely to be more fully transmitted than downward movements, and the same bias might be expected to apply to production. Thus, in commodity markets where these countries dominate, the decline in global excess demand in response to an increase in a country’s offer price would be larger than our estimates suggest, while the response to a decrease would be smaller. On the vther hand, in markets dominated by developing countries with consumer-oriented policies, such as that for rice, we would expect our estimates to have the opposite bias.
+ EFTA
+ USA
VSR LR VSR LR VSR LR VSR. LR VSR LR VSR LR VSR LR VSR LR
-3.8 -6.9 - 15.1 -27.8 - 16.7 - 30.6 - 16.3 - 30.0 -5.4 - 10.3 -2.1 -3.4 -2.8 -4.4 -1.6 -2.3 -5.7 -21.9 -1.2 -3.8 -2.1 -6.0 - 1.3 -3.2
(13.0)
(13.3)
- 1.9 -6.4
Grain
Coarse
12.2 28.2 14.1 29.2 (38.4) - 10.8 -20.7
-
Rumlmant Meat
- 24.9
Nonruminant Meat
15.0 --21.4 - 10.6 - 14.8 -
-39.1
- 20.0 -31.2 - 20.0 -31.0 - 16.9 - 26.0
Dairy Products
-7.3 - 10.2
- 19.2 - 28.6 -21.5 -31.7 - 23.8 - 35.0 -8.9 - 13.2 - 13.9 - 20.7
(27.8)
Sugar
“Perfect price transmission is assumed in all countries (price transmission elasticities set to unity). Elasticities not shown have magnitudes greater than 40. Values in parentheses are the excess supply elasticities faced by the country; the other values are excess demand elasticities. VSR refers to the very-short-run adjustment within the first year and LR refers to the long nut after full adjustment.
price transmission elasticities set to unity (no insulation).
- 13.3 - 18.5
-23.4 - 32.9 - 13.0 - 17.9
-37.1
Rice
&urce: Derived from the model described in the text, with all international-to-domestic
USA + EC - 12 + Cairns Group
EC-12
USA + Cairns Group
Cairns Group
EC-12
EC-12
EC-10
United States
Wheat
Tabk 6: Trade Elasticities Facing Groups of Countries: The Case of No Market Insulation”
332
RodTyersand KymAnderson
A second concern is with the assumption of intercountry homogeneity within each of the seven identified food commodity markets. Even though the model suffers less from the aggregation biases common in the economy-wide models reviewed by Brown (1987), each commodity is traded in numerous grades and degrees of processing. Because the exports of each country are differently specialized, these markets do behave in ways consistent with some degree of differentiation by country of origin. This suggests that the true trade elasticities may be smaller than our estimates; however, this bias is generally unimportant where substitutability within commodity aggregates is high compared with that between commodities. Finally, the capacity of the model to project changes in trade elasticities over time is dependent on the stability of market-insulating policies. It is likely, for example, that domestic markets tend to be more insulated from large changes in international prices than they are from small changes. The degree of insulation is also very sensitive to changes in policy type. An important concern of Abbott is with the increasing prevalence of state trading in food commodity markets. This practice is usually highly insulating; as a result, price transmission elasiicities may be smaller than those implied by any analysis of historical time series. If this is true, it is also a source of upward bias in our estimates. 5. CONCLUSICINS Bearing in mind the features of the model and its important omissions, the key points to note from the implied trade elasticities reported in this paper are: 1. Our estimates of short-run (one year response) and long-run trade elasticities facing all individual economies exceed unity, thus ruling out sustained increases in export revenue from restricting excess supply in any individual commodity market. 2. Except for wheat and coarse grain in the United States and wheat in the USSR, the estimated long-run trade elasticities facing all individual economies have magnitudes larger than 7.5. 3. Some small economies have low export demand elasticities in certain commodities (rice in Thailand, other grains and beef in Argentina, ruminant meats and dairy products in New Zealand), whereas large econ.omies do $101necessarily have low trade elasticities (Japan, for example). 4. Trade elasticities would be much larger (in most cases more
’
PRICE ELASTICITIES
IN INTERNATIONAL
FOOD TRADE
333
than double) were it not for the domestic-market-stabilizing behavior of government policies. 5. These elasticities vary over time because food production and demands are changing at different rates in countries with different degrees of insulation and because insulation in some countries is increasing, so that they are smaller in the late 1980s than in the early 1980s. 6. For the United States, the grain export demand elasticities obtained from our model appear to be somewhat larger than those from other studies, in part because it allows for more substitution in production and consumption between related products. Because these synthetic estimates are necessarily less than perfect, it is difficult to draw strong conclusions about the precise magnitude of food trade elasticities. They certainly cast doubt, however, on the proposition that any country has strong monopoly or monopsony power in international food markets in anything other than the very short run. While some aggregates of the larger exporting countries have substantial potential market power, even in the long run, this stems primarily from insulating policies in the predominantly food-importing countries (and hence from a lack of preparedness on the part of gocremments to permit their domestic consumers, producers, and/or stockholders to share in the costs of adjustment to international price changes). This result suggests that those countries would be less vulnerable to the exploitation of market power were they to adopt less distortionary (and, particularly, less insulating) agricultural and food policies. REFERENCES Abbott, P.C. (1979) Modelling International Grain Trade With Government-Controlled Markets, American Journal of AgriculturaZ Economics 6 I( 1):22-3 1. -. (1988) Estimating U.S. Agricultural Export Demand Elasticities: Econometric and Economic Issues. In Elasticities in International Agricultural Trade (C. Carter, Ed.). Boulder: Westview Press. Anderson, K., and Tyers, R. (1985) European Community Grain and Meat Policies: Effects on International Prices, Trade and Welfare, European Review of Agricultural Economics 11(4):267--294. -. (1986) Agricultural Policies of Industrial Countries and Their Effects on Traditional Food Exportersi The Economic Record 62( 179):385-399. -. (1988) Global Effects of Liberalising Trade in Agriculture, Thames Essay No. 55. London: Trade Policy Research Centre. Bredahl, M.E., Meyers, W.H., and Collins, K.J. (1979) The Elasticity of Foreign Demand for
334
Rod Tyers and Kym Anderson
US Agricultural Prod xzts‘The Importance of the Price Transmission Elasticity, Americwt Journal of Agricultu -al Economics 61( 1):58-63. Brown, D.K. (1987) Tariff>, thz Terms of Trade and National Product Diffemntiation, Journal of Policy Modeling 9(3):503-526. Gardiner, W.H., and Dixrt, P.M. (1987) Price Elasticity of Export Demand: Concept and Estimates. FAER ‘Jo. 228, U.S. Department of Agriculture, Washington, D.C. Johnson, P.R. (1977) The Elasticity of Foreign Demand for U.S. Agricultural Products, American 1 Economics 59(4):735-736. Journal of Agricu Johnson, S.R., Womack . A. WT.,Meyers, W.H., Young, R.E. II, and Brandt, J. (1985) Options for the 1985 Faru Bill: An Analysis and Evaluation. FAPRI Report No. l-85, Food and Agricultural Polic ;, Research Institute, University of Missouri-Columbia and Iowa State University. Meyers, W.H., and Helrudt, M.D. ( 1986) Trade Implications of the Food Security Act of 1985. Staff Report No. 86SX4, Center for Agricultural and Rural Development, Iowa Siate University. Meyers, W&I., Devados,, S., and Helmar, M.D. (1987) AgriculturalTrade Liberalisation: CrossCommodity and ,Jross-Country Impacts, Jolcnal of Poticy Modeling 9(3):455-484. Grcutt, G.H. (1950) Measlurementof Price Elasticities in International Trade, Review of Economics and Stat ‘,-rics32(2): 113- 132. Tyers, R. (1984) Agrict&ual Protection and MarketInsulation: Analysis of International Impacts by Stochastic Simulation. Research Paper No. 111, Australia-Japan Research Centre, Canberra. -. (1985) Interr&onal impacts of Protection: Model Structure and Results for EC Agricultural Policy Journal of Policy Modeling 7(2):219-251. Tyers, R., and Anders m, K. (1987) On Modelling the Effects of Agricultural Policies, T&kwhrifi voor Sociaai H.~st~nschappslijkonderzoek van de Ltmdbouw Jaargang 2, No. 2, pp. 150158. -. (1988a) “Li’&lising OECD Agricultural Policies in the Uruguay Round: Effects on Trade and V&!fae, Journal of Agricultural Economics 39(2):197-216. -. (1988b) ln-+&ect Price Transmission and Implied Trade Elasticities in a Multicommodity Wc.rldFcod Model. In Elasticities in International Agricultural Trade (C. Carter, Ed.) Boulder: Westview Press. -. (for&con+) Distortions in Werld Food Markets. Cambridge: Cambridge University Press. World Bank (1986) WSrld Development Report 1986. Oxford: Oxford University Press.
Rice Wheat CGrain sugar Dairy RMeat NRMeat
699 57772 6299 14164 118757 7520 14813
Rice 0.40
0.38 5.00 0.38 6.00 0.38 0.40
- 0.66
0.92 -0.10 -0.01 -0.01 -0.37
0.90
CGrain
-0.51 -0.10
Wheat
- 0.06 -0.05 0.50
Sugar
0.51 0.12
Dairy
- 0.03 1.02 -0.30
RMeat
- 0.48 1.14
NRMeat
0.02 0.25 - 0.90
NRMeat
0.02 -0.60 0.26
RMeat
-0.40 0.02 0.02
Dairy
Long-run elasticity of supply with respect to the price of:
0.05 -0.12
0.10 0.02 -0.20 0.01
0.25 945 -0.80 Rice -0.30 47850 0.01 Wheat 0.17 70195 CGrain 10533 sugar 107187 Dairy 7632 RMeat 14029 NRMeat Indirect demand parameters for coarse grain: Shares of livestock sectors grain-fed Grain use per unit of output
Reference production (kt)
Sugar
CGrain
Rice
Elasticity of Demand with Respect to the Price of: _.Wheat
Trend Consumption @t)
(a) THE EUROPEAN COMMUNITY
Table A.$: A Sampling of Key Demand, Supply, and Price Transmission Elasticities for Major Industrial Countries
(b) JAPAN
C6rain sugar Dairy RMeat NRMeat
wkak
Rice
le A.1:
-0.03
/,,
,,,, ,,
,,
0.01 0.14 - 0.40
Rice
0.04
0.10
t-2 t
0.12 -0.06
0.02
t-1
0.22 -0.14
-0.03
t-2
Ruminant meat
,,,,,,,, ,,,,,,,,,
- 0.05
Sugar
0.46 0.40
-0.80
Dairy
0.46 6.00
- 1.40 0.25
RMeat
Elasticity of Demand with Respect to the Price of:
0.07
CGrain
-0.01
0.10
- 0.02 - 0.02
t-1
Wheat
Reference Consumption (kt)
-0.22
-0.22 0.40 - 0.02
t
Dairy
Rice 10472 - 0.23 0.03 Wheat 633 1 0.24 -0.60 CGrain 19436 0.16 0.25 Sugar 285 1 O.@l Dairy 8113 RMeat 706 NRMeat 2904 Indirect demand parameters for coarse grain: Shares of livestock sectors grain-fed Grain use per unit of output
0.30 -0.22 -0.02
t-1
t-2
t-1
t
t-1
t-1
0.20
Sugar
coarse grak
Wheat
Short-run elasticity of supply with respect to the price of:
Rice
(Japan continued)
1.00 5.00
0.40 -1.00
NRMeat
t
0.76
t-1
t-2
-0.16
Nonruminant meat
t-1
0.08
0.30 -0.20
t-1
Rice Wheat CGrain
(c) THE UNfTED STATES -
Rice Wheat CGrain sugar Dairy RMeat NRMeat
Wheat
Rice Wheat CGrain Sllgti Dairy RMeat NRMeat
-0.15 0.30
t-1
Reference Consumption (kt) 2015 26958 155456
-0.01 0.01 -0.01 -0.03 -0.05
t t-2
-
Rice 0.20 0.60 -0.40
Wheat -0.30 0.60
CGrain Sugar
Dairy RMeat
Long-run elasticity of supply with respect to the price of:
Rice -0.20 0.01 0.01
0.10
t-1
Sugar
0.35
t-1
0.30 0.02
t-2
-0.10
t
-0.02 0.10 - 0.02
t-1
-0.02 0.40
t-2
Ruminant meat
Wheat 0.08 -0.12 0.08
CGrain 0.04 0.06 -0.20
0.07
Sugar
Dairy
RMeat
Elasticity of Demand with Respect to the Price of:
t
Dairy
0.50 - 0.06 0.80 -0.09 - 0.06 0.04 0.80 -0.23 - 0.06 Short-run elasticity of supply with respect to the price of:
- 0.02 -0.04
Coarse grain
Reference production (kt) 9375 675 399 853 6798 478 2619
0.33
t-1
t-2
- 0.05
Nonruminant meat
NRMeat
t
-0.10 0.99
NRMeat
t-1
-0.01
-0.15
-0.09 0.45
t-1
0.35 -0.02
-0.20
-0.01
t
-0.01 -0.02 -0.01
-0.30 0.40
t-1
0.02 - 0.50 0.20 0.67 6.00
-0.30 0.02 0.01 0.67 0.40
Wheat -0.20 0.80 -0.28 -0.53 0.75
CGrab
Sugar -0.04 Dairy
t-2
-
0.07
t-1 - 0.02
Sugar
0.07
t
0.02 0.01
t-1
Dairy
0.08
t-2
0.03 -0.20
t
- 0.20 0.72 -0.13
RMeat
t-2
t
-0.16 1.12
0.61
t-1
t-2
- 0.08
Nonruminant meaat
NRMeat
1.00 5.00
0.01 0.20 -0.80
A more complete list is also provided in
0.03 -0.10 0.24 0.32 - 0.05 - 0.02
t-1
Ruminant meat
0.28 -0.08 0.85 -0.24 0.03 -0.38 Short-run elasticity of supply with respect tenthe price of:
-0.04
Rice 0.75 -0.04
-0.10
Coarse grain
(W 4713 72301 211494 5321 61807 10578 13991
-0.20
!L@q-run elasticity of supply with respect to the price of:
0.05
Swrce: The parameters of the model are presented as detailed estimates in Tyers and Anderson (forthcoming). Tyers and Anderson ( 1988b).
CGrain sugar Dairy RMeat NRMeat
Rice Wheat
Wheat
Rice
Rice Wheat CGrain Sugar Dairy RMeat NRMeat
Reference production
8693 Dairy 60503 RMeat 11890 NRMeat 13825 Indirect demand parameters for coarse grain: Shares of livestock sectors grain-fed Grain use per unit of output
sugar
USSR
United States
Japan
EFTA
EC-10
Canada
Australia
SR LR SR LR SR LR SR LR SR LR SR LR SR LR
0.78 1.00 0.68 1.00 0.09 0.20 0.11 0.79 0.20 1.00 1.00 1.00 0.05 0.45
0.11 0.63 0.68 1.00 0.08 0.11 0.11 0.79 0.06 0.25 1.00 1.00 .os 0.45 0.69 0.96 1.00 1.00 0.24 0.58 0.15 1.00 0.20 1.00 1.00 1.00 0.02 0.17
P 0.69 0.96 1.00 1.00 0.13 0.26 0.15 1.00 0.02 0.12 1.00 1.00 ox!2 0.17
c
P
Wheat c
Coarse grain
0.62 .84 0.90 0.90 0.11 0.46 1.00 1.00 0.06 0.55 0.82 1.00 0.06 0.30
P
C 0.23 1.00 0.90 0.90 0.11 0.22 0.30 0.30 0.03 0.12 0.71 1.00 0.06 0.30
Rice
0.73 1.00 0.27 0.46 0.24 0.45 0.01 0.04 0.10 0.24 0.60 0.61 0.05 0.20
P 1.00 1.00 0.08 0.40 0.14 0.45 0.01 0.04 0.10 0.24 0.21 0.53 0.05 0.20
c
Ruminant meat
0.46 0.52 0.08 0.40 0.12 0.76 0.13 0.68 0.49 0.63 1.00 1.00 0.05 0.20
P
0.25 0.34 0.83 0.85 0.62 0.76 0.16 0.16 0.47 0.86 1.00 1.00 0.05 0.20
C
Nonruminant meat
Table Q-2.: A Sampling of Key Elasticities of Transmission of International Price Changes to Domestic Prices”
0.40 0.45 0.96 0.40 0.08 0.30 0.06 0.19 0.03 0.08 0.07 0.36 0.05 0.13
P
0.13 0.39 0.96 0.40 0.08 0.30 0.06 0.19 0.03 0.08 0.06 0.18 0.05 0.13
c
Dairy products
0.49 0.54 0.07 0.25 0.00 0.00 G.%I 0.00 0.00 0.00 0.10 0.48 0.02 0.04
P
0.00 0.00 0.12 0.60 0.00 0.00 o.GG 0.00 0.00 0.00 0.10 0.48 0.02 0.04
c
Sugar
SR LR SR LR SR LR SR LR SR LR SR LR 0.05
0.60 0.15 0.90 0.09 1.00 0.40 0.60 0.80 1.00 0.42 0.79
0.44
0.60 0.15 0.90 0.09 1.00 0.40 0.60 0.80 1.00 0.42 0.79
0.54 0.87 0.14 0.80 0.47 0.94 0.85 1.00 0.70 0.80 0.57 1.00
0.05 0.70 0.14 0.80 0.46 1.00 0.85 1.00 0.70 0.80 0.35 0.42
c
P
P
c
COtUW? grain
Wheat
0.35 0.58 0.17 0.26 0.20 0.60 0.49 0.74 0.56 0.56 0.16 0.46
P
.,,,,,
,,
,,
/,.,
,,,,,,
,,,
,//
consumer prices, respectively.
88,
,,
3,
,,,,
,,/
3,
,,,,
8,
,,,,
c 0.48 0.66 0.15 0.40 0.05 0.40 0.17 0.30 0.58 0.63 0.44 0.60
P 0.05 0.50 0.15 0.40 0.05 0.40 0.17 0.30 0.77 0.90 0.44 0.60
c
Ruminant meat
0.17 0.25 0.15 0.60 0.05 0.40 0.18 0.50 0.43 0.46 0.72 0.77
P
0.05 0.22 0.15 0.60 0.20 0.40 0.18 0.50 0.66 0.80 0.72 0.77
C
Nonruminant meat
0.10 0.16 0.15 0.25 0.05 0.20 0.01 0.20 0.34 0.35 0.54 0.51,
P
0.05 0.12 0.15 0.25 0.02 0.20 0.01 0.20 0.34 0.35 0.54 0.54
c
Dairy products
0.19 0.23 0.09 0.20 0.02 0.20 0.24 1.00 0.00 0.00 0.24 0.90
P
0.05 0.20 0.09 0.20 0.02 0.20 0.24 1.00 0.00 0.00 0.24 0.90
c
Sugar
,,,
./,,,,,m,
,,,,
geometric lag structure connecting them); P and C refer to domestic producer and
0.05 0.40 0.17 0.26 0.05 O.?O 0.31 0.58 0.56 0.56 0.26 0.32
Rice
“SR and LR refer to short-run and long-run elasticities (with a Nerlovian
Brazil
Argentina
Thailand
Indonesia
India
China
Table A.2. (continued)
PRICE ELASTICITIES
IN INTERNATIONAL
FOOD TRADE
341
APPENDIX HI: DERIVATION OF IMPLIED TRADE ELASTICITIES FROM THE GLS MODEL The model treats each of its 30 countries or country groups as “large,” in that changes in the trade volumes of each induce nonzero changes in international prices. Of special interest are the excess supply or demand elasticities (of “the rest of the world”) that face each country. Where these are comparatively small in magnitude for particular countries, those countries have the capacity to increase domestic welfare at the expense of the rest of the world by imposing trade taxes. Consider a world comprising N countries (K = 1, . . . , N) and several commodities. Let mf be the excess demand in the rest of the world for commodity i produced by country K. This excess demand might be either positive or negative, the latter indicating that country K is a net importer. The excess demand elasticity es is then the proportional adjustment that occurs in rnfin response to a given proportional change in the price at which country K trades commodity j (Pi, where j and i may or may not refer to the same commodity): dm% dP. et =2/I mf Pjm
(A.11
We summarize here an analytical approximation to the elements of lc. structural equations and parameters of the mauix [et], based on tPa the model (a more detailed account is provided in TyerS atid Anderson 1988b and forthcoming). First, we seek an expression for the adjusment in excess demand in the rest of t:rc world, L%v$~ in terms of changes in prices. This adjusment is simply the sum of the adjustments made in the countries, k, of the rest of the world group: dmf = Z (t&z + dcz - &a + dSia. k#k
(A.2)
first term is the adjustment in direct human demand, &, the second is that in livestock feed demand, cc, the third is that in production, qjk, and the final term is the adjustment in stock levels~ s,. It is convenient to express each of these terms as functions of domestic prices in country k (#A). The direct human consumption adjustment is readily derived from the differentiation of the direct consumption equation to give
The
(A.3)
where aijkis the elasticiity of direct human demand for commodity i with respect to the consumer price of j and & is the consumer price of j, both in country k. The corresponding adjustments in feed demand, production, and
RodTyersandKymAnderson
342
stock levels depend on the length of run. Here we consider three extreme cases discussed in the paper: adjustment in the same year, after one year, and in the long run. tintheSameYear The relationship between feed demand and livestock production is based in the model on fixed input-output coefficients that link that demand to an index of steady state output for each livestock product. Taking derivatives of the terms in the feed demand equations, which depend on prices in the curre year, an summing over the feedconsuming sectors, n, we obtain following expression: (A.41
where 0~~~is the fixed input-output coefficient linking feed demand to onB is a coefficient indicating the deviation of the steady-state output and +T same-year production response from that in the long run (steady state), as explained in detail in Tyers and Anderson (1988b). ISnR is the partial adjustment elasticity for the production of livestock product n in country k (if Sti is unity, the entire response is immediate). The product of ankand bhjk is then the same-year output response elasticity for livestock product n with respect to the price of c ity j in country k. Because livestock producers respond as consumers of feed grains, the.prices in this equation are consumer prices where commodity j is a livestock feed and producer prices where it is a livestock product. The corresponding production adjustment is derived in the same way: (A.3
where the same xule applies regard ng the prices in the final term of equation (AS) as applies for (A.4). The stock level adjustment dsikis a complex function of the changes in both production and consumer prices, dqikand dpz (derived in Tyers and Anderson 1988b). When it is combined withequations (A.3), (A.4), and (AS), one obtains the following expression for the adjustment in the overall excess demand of country k in terms of price changes: (A.6)
are constants in any given year. de elasticities via e
ion (A.
1) 9
expressions relating
PRICE ELASTICITIES IN INTERNATIONAL FOOD TRADE
343
domestic prices, &’ andp$, to international prices, Pi, are also lzeded. These are given by (A.71
where +iM and +F are the same-year international-to-domestic price transmission elasticities for consumers and producers, respectively. The adjustment in the excess demand of “the rest of the world” for the exports of focus country, K, is then (A-8)
where =he sum over countries k excludes the focus country, K. From equation (A. l), the excess demand elasticity facing country K is then
Adjustment after One Year In this case the terms in equation (A.2) depend on the sum of the same-year and the one-year lagged adjustments in consumption, production, and stocks. It is convenient to begin with the production adjustment, where the one year production response is (A. 10)
where b,, is the one-year component of the long-run elasticity of production response of commodity i with respect to the price of commodity j in country k. Once again, the prices in this equation may be either consumer or producer prices, as in equations (A.4) and (AS). The response of feed demand after one year then depends on the adjustment of steady-state livestock production to changes in price. In this case the feed consumption adjustment is
where ai& is the same adjusted input-output coefficient as in equation are coefficients indicating the deviation fo the same(A-4), TOnk and +rlnk year and one-year response in the production of n from that in the long run (steady state), d& is the same-year production adjustment given in equation (AS), and && is the one-year adjustment given in equation (A. 10).
Rod Tyers and Kym Anderson
344
The adjustment in country k’s overallexcess demand can then be formulated in terms of changes in domestic producer and consumer prices, as in equation (A.@. It remains to link these domestic price changes to changes in international prices. This is readily achieved by using the formulation of the one-year price transmission elasticity in terms of the corresponding same-year and long-run elasticities, which are both parameters in the model. Because the expression is identical for domestic consumer sand producer prices, we drop the P and C superscripts: (A. 12)
where $y is the long-run price transmission elasticity. An expression similar to equation (A.@ readily follows. The one+ies can then be calculated from a further equation year trade @ g Run
Adjustment in t
In this case, the global excess demand elasticities embody all longrun quantity adjustments to price changes. Again, equations (A. 1) and (A.2) apply, except that changes in stock levels in the long run do not effect annual excess demand. Accordingly, the third term in equation (A.l) is dropped. The changes in direct and feed consumption are now (A. 13)
dcf, =
xn %t
qnk
2 @tijk+ btnjk+ bad i
dP*
F-
(A. 14)
Jr:
Note that the three components of long-run supply elasticities are summed in the second term of equation (A. 14). This same total longrun supply elasticity is the basis for the change in production: dq,&= qi&Z (bwk + b,, + b& 91s J
(A. 15)
Pjk
Substituting equations (A.13), (A.14), and (A.15) into (Al), and setting dsikto zero, we derive and expression like equation (A.@. The expression for the long-run trade elasticity then differs from equation (A.9) only in that the constants Ajk and Bjkdiffer and the price transmission elasticities that appear are the long-run values +i?tnd +s?