Price and exchange rate transmission in Russian meat markets.

Osborne, Stefan R. ; Liefert, William M.

INTRODUCTION

A major goal of Russia's economic reform that began in the early 1990s has been integration into the world economy. Integration involves liberalising both foreign trade and investment. The benefits of integration include growth in the gains from trade according to comparative advantage, increased foreign investment, and access to the world's best technology and management practices.

Trade liberalisation involves ending the state's monopoly on foreign trade, as well as reducing domestic price controls and barriers to trade. Lessening these controls results in world market prices becoming the dominant factor in determining domestic prices. The movement in relative domestic prices (price ratios) to relative world prices becomes the very means by which countries gain from trade according to comparative advantage. Countries benefit from producing for export those goods whose domestic prices (reflecting costs) initially are less than world prices (at existing exchange rates), and from producing less and importing more of those goods whose domestic prices are initially above world prices. Russia has strengthened its policy of trade liberalisation by moving to (generally) floating exchange rates, such that rates are determined mainly by flows in trade and capital investment.

This paper examines the transmission between changes in both world trade prices and Russian exchange rates and changes in Russian consumer retail prices for meat. The empirical focus of the paper is estimation of price and exchange rate transmission elasticities (TEs) for Russia during 1994-1999 for beef and pork.

The estimated TEs are useful for two reasons. The first is as indicators of Russia's progress toward its reform goal of integrating its agricultural and food economy into the world agricultural economy. As implied earlier, the degree to which world prices determine domestic prices is a good indicator of an economy's integration into world markets. The transmission of changes in world prices to domestic prices is in turn a strong indicator that world prices are largely determining domestic prices (an example of good transmission being that a 10 percent rise in the world price for wheat raises Russia's domestic prices for wheat by 10 percent). If transmission for Russia is poor, domestic agriculture and food prices will likely deviate from world prices, such that the country's commodity volumes and mix of trade will be suboptimal (not at the levels that maximise the gains from trade).

The second reason TE estimates are useful is for forecasting. Predicting changes in Russian agricultural and food production, consumption, and trade, as well as how such changes would affect world agricultural prices and trade volumes, requires knowledge of price and exchange rate transmission. Commodity forecasting models for Russian agriculture, such as those of the Economic Research Service of the US Department of Agriculture and the Organization for Economic Cooperation and Development, explicitly require values for price and exchange rate transmission elasticities.

A number of studies (Gardner and Brooks, 1994; DeMasi and Koen, 1996; Berkowitz et al., 1998; Goodwin et al., 1999; Loy and Wehrheim, 1999; Berkowitz and DeJong, 2001) have examined price integration for foodstuffs within Russia--that is, how well domestic food markets work to eliminate price differences between regions. These studies are related to the issue of price and exchange rate transmission, in that the factors that segment domestic regional markets are also likely to weaken transmission between world and domestic prices.

The studies, in general, find that Russian domestic price integration is far from complete--that is, that regional agricultural and food markets within Russia are segmented from each other. Gardner and Brooks, DeMasi and Koen, and Loy and Wehrheim find that price cointegration between regions has been poor--that is, substantial variation in food prices has existed between regions that cannot be explained by distance and transport costs. The results of Goodwin et al. are more mixed, as prices in retail stores are more integrated than those in farmers' markets. Berkowitz et al. finds the greatest degree of price integration, though it is still not strong.

Although these studies can help in gauging the general magnitude of price and exchange rate transmission in Russian agriculture, they do not specifically measure transmission. Also, with the exception of the work by Berkowitz and DeJong, which covers the second half of the 1990s, these works encompass only the early transition years (not going beyond 1995). Our study, on the other hand, covers 1994-99. Lastly, out study estimates TE for 31 cities within Russia, and thereby allows us to determine which cities (or regions) have stronger transmission, reflecting better policies and infrastructure for linking the domestic agricultural economy to world agricultural markets.

The next section examines why price and exchange rate transmission in Russian agriculture might be weak. The following section discusses the method and data used in estimating price and exchange rate TEs. The subsequent section examines the results, and the last section presents the paper's main conclusions.

CONTEXT

In the Soviet Union, foreign trade was a state monopoly, and state planners determined the mix and volume of imports and exports. (1) The state also set domestic producer and consumer prices for almost all goods, (2) including those imported and exported (Bornstein, 1987). Although world prices might have influenced the state price setters to some degree, no formal relationship existed between world market and domestic prices. For agricultural goods, large differences existed between world market prices and domestic producer and consumer prices (see Organization for Economic Cooperation and Development, 1998). The state also set the official exchange rates between the ruble and foreign currencies. Ruble exchange rates were changed in response to movements in foreign rates only to keep cross exchange rates involving the ruble equal. (3) Given the strong state control over both domestic prices and exchange rates, transmission was negligible between changes in both world market prices for agricultural and food products (as well as all other goods) and exchange rates and domestic prices.

The economic reform that began in Russia in 1992 after the dissolution of the USSR ended the state monopoly on foreign trade, generally freed domestic prices (though this did not happen overnight and some price controls still remain, such as for energy), and created generally floating exchange rates. Since 1994 (the beginning of the estimation period in this paper), formal controls on agricultural and food trade at the federal level have not been overly restrictive. Tariffs on most agricultural imports have ranged from 5 to 20 percent, though some have been as high as 30 percent (such as for poultry and sugar). Quotas and other quantitative restrictions have been rare, sugar imports being a major exception. State trading does not formally exist, though some of the agencies that administered the country's foreign trade during the Soviet period and have now been privatized retain links to the state. Nonetheless, at the federal level agricultural trade has not been too strongly controlled. (4) Consequently, one might suspect that price transmission between the world and domestic markets is fairly high.

There are countervailing conditions, however, that diminish price and exchange rate transmission for foodstuffs in Russia, or more generally, that weaken Russia's integration into world agricultural markets. (5) One condition is that the domestic infrastructure for moving agricultural goods is deficient (which also segments domestic regional markets). Although storage is also inadequate, the main weakness in physical infrastructure is transportation, particularly the poor road system. The cost of shipping agricultural and food products between regions can exceed producer prices. Weak transportation and storage also increase the risk of spoilage for perishables such as meat.

Commercial and institutional infrastructure is also weak. Wehrheim et al. (2000) argue that undeveloped institutional infrastructure is the main problem facing Russian agriculture. Producers above all need a system for the quick and inexpensive dissemination of market information. Without knowledge of trade opportunities and prices, concerning not only foreign but also internal trade, regional producers are segmented from each other as well as cut off from the world market. Producers and traders also need a financial system that allows fast and affordable access to capital and a strong system of commercial law that protects property and enforces contracts. The absence of such market infrastructure increases the costs and risks of producing and, in particular, selling output--that is, it raises the transaction costs of doing business.

Another condition contributing to low transmission is market power held by suppliers throughout the food distribution system, involving both domestically produced and imported products (Interfax, Food and Agriculture Report, Moscow, twice monthly). Market power is likely, given that the state agencies responsible for food distribution during the Soviet period have been privatized, but often facing little or no competition. The exercise of market power would further separate domestic prices from trade prices.

An additional cause of low transmission is that regional governments throughout the country have controlled to varying degrees prices and profit margins for local producers. These restrictions create price differences between regions, as well as between regional markets and the world market, thereby further segmenting internal agricultural markets and isolating them from the world market (Interfax, Food and Agriculture Report, Moscow, twice monthly). Although these regional controls have diminished in recent years, this could largely be the result of growing crop harvests over 1999-2002, caused mainly by favourable weather, that reduced governments' perceived need to intervene in agricultural markets. A drop in output of major crops such as grain that renewed fear of inadequate local food supplies could revive such restrictions. Nonetheless, for the purpose of analysing the results in this paper, these controls were stronger during out period of estimation than in recent years. Organised crime also exacerbates market segmentation, through such actions as blocking entry and extorting rents.

The importance of price and exchange rate transmission in Russian agriculture for world agricultural trade depends in part on the magnitude of Russian agricultural and food trade. The Soviet Union was a large importer of grain, soybeans, and soybean meal, used mainly as feed for its growing livestock herds. During the 1980s, the country accounted for about one-sixth of world grain imports. During transition, however, the livestock sector in Russia and most other countries of the former USSR has contracted by about half (both inventories and output; Cochrane et al., 2002). In response to this severe downsizing, grain and oilseed imports have fallen substantially. Rather than importing feed to maintain large livestock herds, Russia has become a major importer of meat--beef, pork, and poultry (Table 1). During the second half of the 1990s, Russia took 10 percent of total world imports of meat, and almost 17 percent of world poultry imports. The country's share in world imports of all agricultural and food products over this period was about 3 percent. (6) (The figures in Table 1 exclude intra-EU imports from world imports.)

Russia's imports of agricultural and food products are important not only to world markets but also to the country's domestic economy. Before the financial crisis of 1998, imports accounted for about 20 percent of all food consumed in Russia (Liefert and Liefert, 1999). In the wake of the extreme depreciation of the ruble following the economic crisis of August 1998, imports' share in food consumption fell substantially, though it has since rebounded. During the last 5 years, imports have supplied close to a third of the meat and vegetable oil consumed by the country. During 1996-2000, the share of agricultural and food products in Russia's total imports was about 25 percent (Russian Federation State Customs Committee Tamozhennaia Statistika Vneshnei Torgovli Rossiiskoi Federatsii (Customs Statistics for Foreign Trade of the Russian Federation), Moscow, annual).

Russia is a small agricultural exporter, with agricultural and food products accounting over 1996-2000 for only about two percent of the country's total exports. The value of agricultural and food exports over this time was 13 percent of the value of agricultural imports. Sunflowerseed is the dominant export, with average annual sales during 1996-2000 of 1.1 million metric tons.

The importance of Russian price and exchange rate transmission, for both the Russian agriculture and food economy and world agricultural markets, also depends on how stable are world agricultural prices and Russian exchange rates. The more unstable these are, the more significant transmission is in determining the effects of changes in world prices and exchange rates on trade volumes. During the transition period, both world agricultural prices and exchange rates have fluctuated considerably. Generally speaking, world agricultural prices rose substantially from 1993 to 1997, and then plunged. For example, US export prices for a bushel of wheat (fob Gulf ports) in 1994, 1996, and 1999 equaled $4.09, $5.63, and $3.04, respectively. Unit values for US exports of beef and pork fell, respectively, from $1.52 and $1.26 a pound in 1995, to $1.17 and $0.98 in 1999 (Economic Research Service, US. Dept. of Agriculture, Agricultural Outlook, Washington, DC, monthly through 2002).

From 1992 to 1996, the Russian ruble depreciated severely in nominal terms (Table 2). In real terms, the ruble depreciated by almost 90 percent in 1992, but then from 1993 to 1996 appreciated in real terms. The reason for the real appreciation is that, although the ruble's rate of nominal depreciation was itself high, the rate of domestic inflation exceeded the nominal depreciation rate. In response to the economic crisis that hit in August 1998, the rune depreciated again in both nominal and real terms. The drop in Russian agricultural imports in 1998 and 1999 (Table 1) indicates the exchange rate's importance in determining the volume of Russia's agricultural trade, as ruble depreciation substantially increased domestic ruble prices for imported foodstuffs. In 2000, the ruble began to appreciate again in real terms, rising in real terms in 2000 and 2001 by 5 and 15 percent, respectively.

METHODOLOGY AND DATA

The TE for a good (as estimated in this paper) equals the percent change in the Russian consumer retail price for the product divided by the percent change in the border price (or exchange rate). (7) A value of 1 gives perfect transmission, a value of 0 no transmission.

For the period 1994-99, we estimate price and exchange rate TEs for beef and pork. These products are chosen because imports constitute a relatively large share of Russian domestic consumption. In 1997, the share of imports in the total consumption of beef and pork was 24 and 22 percent, respectively (Liefert and Liefert, 1999).

The livestock product for which imports provide the largest share of domestic consumption is poultry. During the last 5 years, imports have provided about half of all poultry consumed (and as much as 66 percent in 1997 before the crisis of 1998), with the bulk of imports coming from the United States. (8) Yet, we regrettably do not include this commodity in our study, for the simple reason that Russia began reporting prices for poultry by city only in October 1998.

TEs are computed separately for 31 of the largest cities in Russia. We make no estimates of national aggregate TEs for specific commodities. The city-based results will show large differences between cities. In light of this variation, national aggregate estimates would be misleading and difficult to interpret.

As mentioned earlier, we compute TEs between changes in Russian consumer retail prices for beef and pork and changes in border prices/ exchange rates. For foodstuffs, TEs are more typically computed for producer prices rather than consumer retail prices (for example, see Mundlak and Larson, 1992; Quiroz and Soto, 1995). The main reason is that most traded foodstuffs are unprocessed products, such that imports usually correspond in degree of processing to farm gate output. A close correlation, and potentially equivalence, can therefore exist between domestic producer prices and border prices. Retail prices for foodstuffs, however, cover costs from processing, distribution, and retail sale, in addition to the cost of primary agricultural production, such that retail prices exceed border prices, and are not as closely correlated.

TEs can nonetheless be estimated between border prices/exchange rates and domestic retail prices. Many studies have estimated TEs between farm gate (producer), wholesale, and retail prices within a country (see Hahn, 1990; Goodwin and Holt, 1999; Goodwin and Harper, 2000, all of which involve US beef and pork). Calculating TEs between border prices and retail prices is as conceptually and empirically valid as computing TEs between producer and retail prices. The TE between the border and retail price should equal the TE between the border price and producer price, times the share of the producer price in the retail price. Appendix A provides a demonstration under general conditions for this result. We define 'full transmission' between the border and retail price as that involving perfect transmission (100 percent) between the border and producer price. With full transmission, the TE between the border and retail price would equal the share of the producer price in the retail price of the foodstuff (as equation (A.6) in Appendix A shows).

Table 3 gives the structure of the retail value of Russian beef and pork in 1995 and 1999. Based on the table's figures for the share of producer prices in retail prices, TEs between border prices/exchange rates and retail prices for both beef and pork that correspond to 'full transmission' would appear to be in the range of 50-60 percent.

Yet, full transmission between border and retail prices for Russian beef and pork during our period of estimation (1994-99) should probably be higher than this range, for two reasons. The first reason involves the method by which the Russians determined the 'downstream' costs and markups that contributed to retail prices--processing and retail costs, taxes, and profit markups. During the Soviet period, the processors' and retailers' profit, taxes, and retail cost were all calculated as percentage markups, with the percentages set by state authorities. This means that a change in the cost of primary production for a foodstuff of x percent came close to changing the retail price by x percent. If all downstream cost and value elements were determined during our estimation period as percentage markups, full transmission between border prices/exchange rates and domestic retail prices would result in border to retail price TEs equal to 100 percent.

As part of market-oriented reform during transition, retail and profit markups were generally freed from state control, to be established by processors and retailers themselves. Yet, this change in price-setting policy was made gradually and non-uniformly, such that in many regions during our period of estimation, downstream margins continued to be determined (or influenced) by state-set percentage markups (Interfax, Food and Agriculture Report, Moscow, twice monthly). The continued use of fixed percentage markups in pricing increases the expected values for border to retail price TEs.

The second reason the border to retail price TEs should be higher is that imported beef and pork can range from being wholly unprocessed to fully processed and ready for retail sale. The smaller the gap in processing between imports and retail product, the higher transmission should be between border prices and domestic retail prices. Although most beef and pork imported by Russia during the estimation period was largely unprocessed, some was processed. (9) For both this and the other reason examined involving percentage markups, 'full transmission' between border and retail prices should yield TE values for border to retail prices more in the range of 60-70 percent than 50-60 percent.

In computing the price transmission elasticities, real as opposed to nominal values are used for both domestic and trade prices (which involves deflating nominal prices by domestic and foreign CPIs, examined in more detail later). Real prices are appropriate because changes in consumer and producer behaviour are driven largely by changes in real, as opposed to nominal, prices.

Likewise in calculating the TEs for the exchange rate, the real as opposed to the nominal rate is used. The real exchange rate equals the nominal rate adjusted for changes in both domestic and trade prices (by multiplying the nominal rate by the ratio of the change in trade prices to the change in domestic prices). Given that we use real values for domestic and trade prices, it would be inconsistent in the TE estimations to use nominal rather than real exchange rates. (10)

The data needed to estimate the TEs are domestic consumer prices, border prices, the domestic and foreign CPI, and exchange rates. The Russian Federation Ministry of Agriculture (Sbornik Informatsionnikh Materialov dlia Territorii Rossiiskoi Federatsii (Collection of Information Materials for the Territories of the Russian Federation), Moscow, quarterly) has collected domestic prices for beef and pork for markets in 80 Russian cities (mainly capitals of the various 88 oblasts, republics, and autonomous districts) quarterly from January 1994 to December 1999. All 31 cities for which we make calculations are included in the data set.

The border prices for goods are calculated as unit values for imports into Russia, computed from the import volume and value data in Tamozhennaia Statistika, the quarterly foreign trade publication of the Russian State Customs Committee. The publication reports all import (as well as export) values in US dollars. Russian Federation State Committee for Statistics (b) (Rossiiskii Statisticheskii Ezhegodnik (Russian Statistical Yearbook), Moscow, annual) provides the Russian CPI and rune exchange rates. For the foreign CPI, we use the CPI of the United States (US Bureau of Labor Statistics, CPI-All Urban Consumers, www.bls.gov). (11)

We index the nominal domestic prices by the Russian CPI and then average them into quarterly prices so that they can be compared with the quarterly border prices. The real exchange rate is calculated by multiplying the ruble/dollar exchange rate by the ratio of the US to Russian CPI.

The technique that we choose to estimate the TEs depends on whether the data used in the estimation are stationary or non-stationary. The data series for many economic variables, such as prices, often have the characteristic that the value of the variable at any point in the time series is correlated with past values (or observations) in the series. If the correlation among values is absolute, the data series is non-stationary. Data series that are not strongly correlated are stationary. For whatever specific technique we use to estimate the TEs, the independent variables will be the border prices and (real) exchange rate, while the dependent variables will be the domestic prices. In order for a relationship to exist between dependent and independent variables, the data for both variables must either be stationary or non-stationary. This is because no statistically testable relationship can exist between the levels of a stationary and a non-stationary variable. (12) In the case of our estimation exercise, stationarity incompatibility would mean that the TE involving the border prices/exchange rate and domestic prices must be zero.

The standard test for determining whether a data series is stationary or non-stationary is the augmented Dickey-Fuller (ADF) test (Dickey and Fuller, 1979). We use the ADF test to determine the stationarity status of the data for the three key variables in our TE estimation--domestic and border prices and the real exchange rate. We find that the data series for both of our independent variables--the border prices and exchange rates--are nonstationary. Recall that we wish to compute price and exchange rate elasticities separately for 31 Russian cities for two different products. The ADF test shows that the data for domestic beef prices are non-stationary for only 12 of these cries, while the data for domestic pork prices are non-stationary for 16 cities (Table 4). Given that TEs between stationary and non-stationary data must be zero, these results by themselves indicate that transmission in Russia for beef and pork is generally poor.

The ADF unit root test has low power, so it is important to discuss the implications of using this test in out analysis. The ADF test has low power because it is difficult to distinguish between a stationary data series with a high degree of autocorrelation and a data series that is truly non-stationary. Since the null hypothesis of the standard ADF unit root test is nonstationarity, the unit root test is biased toward finding that data are non-stationary. The implication is that we might find that city prices are non-stationary when they are actually stationary. Given, however, that the border price and exchange rate data have both been round to be nonstationary, our test is biased in favour of finding too many compatible cases. The consequence is that our results are biased toward finding a stronger rather than weaker relationship between border prices/exchange rates and domestic prices. We find weak price transmission in Russia in spite of this bias. (13)

TEs can be estimated for only those city-product pairings that pass the stationarity compatibility test (28 of the 62 possible pairings). The simplest way to do so would be to use ordinary least squares (OLS), employing the following equation:

(1) In ([P.sup.d]/[CP[I.sup.d]) = [[beta].sub.1] ln([P.sup.f]/CP[I.sup.f]) + [[beta].sub.2] ln(E) + [epsilon].

[P.sup.d] and [P.sup.f] are the domestic consumer retail price and foreign trade (border) price of the good, respectively, and CP[I.sup.d] and CP[I.sup.f] are the domestic and foreign consumer price indices. The real exchange rate E equals the product of the nominal exchange rate (rubles per US dollar) and the ratio of the foreign to domestic CPI. [[beta].sub.1] would be the estimated TE for the border price, and [[beta].sub.2] the estimated TE for the exchange rate.

Given that all the data we are using in the estimations are non-stationary, the problem arises that for non-stationary data, the standard deviations calculated from OLS regression have non-standard distributions. The consequence is that one cannot compute standard confidence intervals or do significance tests. Wald-like tests developed from the Johansen and Juselius cointegration test, however, have asymptotically standard normal distributions. We use these Wald-like tests to provide confidence intervals and significance tests (Johansen and Juselius, 1990). As we have only 23 data points (quarterly data over 1994-99) with which to do the cointegration tests, the critical values of the JJ test are adjusted to take into account the small sample size. (14) Appendix B provides more detailed explanation of how we estimate the TEs using the Johansen-Juselius (JJ) cointegration approach. (15)

RESULTS

Table 5 presents the TE estimates using the JJ method. For those city-product pairings that pass the stationarity compatibility test, results are given only for those for which the estimates are significant (at the 10 percent level). In the table, 'S' means that the domestic price data for the city-product pairing in question are stationary. As discussed in the previous section, because the border price and exchange rate data used in out estimations are nonstationary, the stationarity compatibility test for the data is not met. We therefore do not calculate TEs for these pairings.

The estimates indicate that price and exchange rate transmission for foodstuffs in Russia is weak. Transmission is particularly poor for beef, with significant results for only four cities. Most of the significant beef TE estimates are greater than one, though in only two cases (Orenburg and Vladivostok) does the confidence interval lay wholly above one (excluding that for Irkutsk, whose estimates are not significant). The most likely reason for this overshooting is interference by regional governments in price-setting, such that market conditions play a secondary role in determining prices.

Moscow beef has almost perfect transmission. Moscow relies on imports for more than half of its total food consumption (Interfax, Food and Agriculture Report, Moscow, twice monthly). Thus, it has strong incentive both to minimize policy impediments to food imports and to improve the physical and institutional infrastructure for bringing them in. We argued earlier that because most of Russia's imported meat was unprocessed, full transmission between border and retail prices would involve border to retail price TEs in the neighbourhood of 60-70 percent, not 100 percent. Yet, although there could be some overshooting, the beef results for Moscow suggest not only that the city has good transmission, but that much of its imported beef is already processed, behaviour consistent with its affluence relative to the rest of the country.

The estimates, however, indicate much lower transmission for Moscow pork compared to beef. St. Petersburg also imports over half of its food, and has the advantage in terms of location and infrastructure of being a port. Yet, for neither beef nor pork do we get significant results.

Transmission appears to be somewhat stronger for pork in general, given that we get significance for eight cities. Only a few of the pork TE estimates, however, are close to the range of 60-70 percent, values which, as we argued earlier, would indicate strong transmission.

One might hypothesize a relationship between transmission and cities' per capita wealth, in that rich cities might be more likely to be heavier consumers of foreign (especially Western) foodstuffs than poor ones, as well as import more processed as opposed to unprocessed foods. Yet, no strong correlation appears to exist. Moscow is the richest city (see the income index in Table 4), and we get significance for Moscow beef and pork, with high TE estimates for beef. Yet, Tables 4 and 5 (using the income index) reveal little overall correlation in the country between wealth and either data stationarity compatibility or statistical significance.

The evidence of generally poor transmission is threefold: (1) of the 62 city-product pairings, less than half (28) pass the stationarity compatibility test; (2) of these 28 pairings, for less than hall (12) do we get significance at the 10 percent level; and (3) of these significant pairings, for less than half of the TE estimates do we get results between 0.5 and 1.2 (excluding those estimates that show strong overshooting).

The results serve as an indicator of Russia's progress toward integrating its agriculture into the world economy, specifically with respect to the test that world market prices should largely determine domestic prices for tradable goods. The low TE estimates show that changes in world prices and exchange rates are not being transmitted well to changes in domestic prices, the consequence being that deviations will exist between domestic and world prices. The economic cost to Russia is that its agricultural production and trade are not at the optimal volumes and mix that would maximize the gains from trade according to comparative advantage.

Although our estimates show Russian price transmission to be weak in an absolute sense, the results are consistent with the findings of most other empirical work on agricultural price transmission, which also reveal generally low transmission. Tyers and Anderson (1992) compute both producer and consumer price elasticities for grain, sugar, meat, and dairy for all major countries over the period 1961-83. The elasticities for most country-commodity pairings are less than 50 percent, and for many pairings less than 25 percent. The weighted average price transmission elasticities for China in the short and long run are 0.19 and 0.48, Japan 0.24 and 0.47, EC-10 (at that time) 0.17 and 0.38, and the United States 0.7 and 0.78.

Tyers and Anderson calculate consumer price transmission elasticities for two groups of meat: ruminant (beef, mutton) and non-ruminant (pork, poultry) meat. The results again show generally low transmission, especially for ruminant meat. For example, the short and long run elasticities for ruminant meat for the EC-10, Japan, and United States are 0.02 and 0.04, 0.10 and 0.24, and 0.21 and 0.53, respectively, while the same elasticities for these countries for non-ruminant meat are 0.62 and 0.76, 0.47 and 0.86, and 1.0 and 1.0. For countries such as Argentina, Australia, and New Zealand that are major exporters of (ruminant) beef and mutton and therefore lack import controls that could 'separate' domestic and world prices, the elasticities are higher for ruminant than non-ruminant meat. For example, Australia's short- and long-run consumer price elasticities for ruminant meat are 1.0 and 1.0. The estimate of 1.0 for Australian ruminant and US non-ruminant meat show that near-perfect transmission is possible.

Unlike Tyers and Anderson who examine transmission for both producer and consumer prices, most empirical work on transmission covers only producer prices. Yet, this work also supports the conclusion that transmission throughout the world is generally low. Quiroz and Soto (1995) calculate aggregate producer price transmission elasticities for agricultural goods for 78 countries over the period 1966-91, and for most countries find little or no transmission. For the more recent period of 1990-99, Sharma (2002) computes producer price transmission elasticities for wheat, maize, and rice for eight Asian countries (including India, Pakistan, and Indonesia). The simple average of his significant results for the short run is 0.27, and for the long run 0.65. In a study covering 1968-78 that computes aggregate producer price and exchange rate transmission elasticities for agricultural goods, Mundlak and Larson (1992) find that transmission for most countries is much higher than shown in other work. For 49 out of 57 countries, price transmission lies between 0.85 and 1.07. Yet, Quiroz and Soto argue that the high transmission results of Mundlak and Larson stem mainly from a serious problem of positive autocorrelation, a problem that the former avoid in their own study by using a dynamic error correction model.

Russia's performance with respect to transmission therefore looks less vulnerable to criticism when compared to that of the test of the world. Russia's performance appears generally on a par with that of most developing countries. Its transmission is clearly worse than that in the developed market economies that are large agricultural exporters, such as the United States, Australia, and Canada, which have less incentive to impose agricultural import controls which could separate domestic prices from world prices. Also, the main causes of low transmission in most countries (whether developed or developing) are identifiable price and trade policies at the national level. (16) Russia, on the other hand, has relatively mild price and trade controls at the national level; (17) rather, its main transmission-impeding causes are regional state policies and controls and weak physical and commercial infrastructure. Given the less transparent and more structural nature of these obstacles, Russia faces a more difficult task than most other countries in identifying and correcting its impediments to price and exchange rate transmission.

Russia is currently negotiating entry into the World Trade Organization (its accession bid formally began in 1993), which might raise hope that WTO membership would reduce market-intrusive government policies that hurt price transmission and integration into world markets. Yet, as just mentioned, state transmission-impeding policies are mainly at the regional rather than federal level, and often involve actions, such as restricting outflows of foodstuffs, that the Russian federal government itself has been trying for years to suppress. Enforcement of WTO-membership rules that would improve transmission and transparency within Russia's many far-flung regions could prove difficult.

Price and exchange rate transmission is important for forecasting Russian agricultural production and trade. Forecasting models for Russian agriculture, such as that of the Economic Research Service (ERS) of the US Department of Agriculture and Organization for Economic Cooperation and Development, in fact explicitly require values for the parameters of price and exchange rate transmission elasticities. Our TE estimates indicate that fairly low values should be chosen for these model parameters. Given that the task of improving Russia's physical and institutional infrastructure for agriculture is an inherently slow process, low values should be used not only for short run predictions (the next 1-3 years), but for longer forecasts as well.

As discussed earlier, during the transition period Russia has been a big meat importer. If transmission were greater, would Russian meat imports be higher? The general answer is yes, though the answer must be qualified. In the short run, Russian meat imports will rise when either world prices rail or the ruble appreciates in real terms. The greater the degree of price and exchange rate transmission, the greater will be the rise in imports. During our period of transmission estimation, however, both world trade prices for meat and the ruble's exchange rate fluctuated considerably, moving in both directions (as examined earlier).

The explanation as to why greater transmission would have resulted in higher meat imports involves a more general argument. In a study on Russia's comparative advantage in agriculture in the mid- to late-1990s, Liefert (2002) finds that the ratios of Russian ruble-denominated domestic prices to US dollar-denominated Russian border prices for the meats lay above such price ratios for grain and agricultural inputs (such as fertilizer, fuel, and feed). This indicates that Russian relative prices for meat lay above world market relative prices. (18) The causes of disparities between Russian and world prices are the same as the causes of poor transmission between world and Russian prices--market intervention by regional governments and weak physical and institutional infrastructure. Both of these factors work to isolate domestic regional markets from the world market. Strictly speaking, poor transmission creates disparities between Russian domestic and world prices only when world prices or exchange rates change (as just argued). However, transmission can also serve as an indicator of the strength of the factors that push domestic prices toward world prices, or alternatively, that keep domestic and world prices apart. Because Russian domestic relative prices for meat during the transition period have lain above world relative prices, if the factors that contributed both to poor price and exchange rate transmission and to long-term disparities between domestic and world prices had been weaker, Russian meat imports would have been higher.

CONCLUSION

This paper finds that transmission between changes in (a) world trade prices and Russian exchange rates and (b) Russian domestic consumer prices for meat is low. Transmission elasticities are estimated for both border prices and exchange rates for 31 Russian cities. The commodities covered are beef and pork, which are among Russia's most heavily imported foodstuffs. For most city-product pairings, we find evidence of either weak, or no, transmission.

The results indicate that Russian integration into world agricultural and food markets is poor, specifically with respect to the test that world prices should determine domestic prices for tradable goods. Because transmission between world prices and exchange rates is weak, deviations will exist between domestic and world prices. Consequently, Russian agricultural production and trade will deviate from the optimal volumes and mix that would maximize the gains from trade according to comparative advantage. The main ways transmission could be strengthened would be to improve the country's weak physical and institutional infrastructure for agriculture and end market intervention by regional governments, both of which cut regional markets off from the world market (as well as segment regional markets from each other).

The results are important for model-based forecasting of Russian agricultural production and trade, which requires values for the parameters of price and exchange rate transmission elasticities. Our transmission elasticity estimates indicate that fairly low values should be chosen for these model parameters.

APPENDIX A. TRANSMISSION BETWEEN BORDER AND RETAIL PRICES

We begin by defining the following variables:

[P.sub.c.sup.d] the domestic consumer retail price for a foodstuff;

[V.sub.a] the value of the foodstuff from domestic primary agricultural production, which equals the domestic producer (farm gate) price;

[P.sub.a.sup.f] the border price for the primary agricultural good;

[V.sub.p] the value of the foodstuff from domestic processing, distribution, and retail sale.

We wish to derive the TE between [P.sub.a.sup.f] and [P.sub.c.sup.d].

(A.1) [P.sup.d.sub.c] = [V.sub.a] + [V.sub.p]

Let e be the price transmission elasticity between [P.sub.a.sup.f] and [V.sub.a], such that:

(A.2) [V.sub.a] = [([P.sup.f.sub.a]).sup.e]

(A.3) [P.sup.d.sub.c] = [([P.sup.f.sub.a]).sup.e] + [V.sub.p]

Assuming that no relationship exists between [P.sub.a.sup.f] and [V.sub.p], we get:

(A.4) [partial derivative][P.sup.d.sub.c]/[partial derivative][P.sup.f.sub.a] = e[([P.sup.f.sub.a]).sup.e-1]

We now divide both sides of the equation (A.4) by [P.sub.c.sup.d]/[P.sub.a.sup.f]. Given that [P.sub.c.sup.d]= [([P.sub.a.sup.f]).sup.e] + [V.sub.p], we get:

(A.5) %[DELTA][P.sup.d.sub.c]/%[DELTA][P.sup.f.sub.a] = e[([P.sup.f.sub.a]).sup.e-1] [P.sup.f.sub.a]/[([P.sup.f.sub.a]).sup.e] + [V.sub.p]

(A.6) %[DELTA][P.sup.d.sub.c]/%[DELTA][P.sup.f.sub.a] = e[([P.sup.f.sub.a]).sup.e]/[([P.sup.f.sub.a]).sup.e] + [V.sub.p]

This shows that the TE between the border price for the primary product and the retail price equals the TE between the border price and the producer price (e), times the share of the producer price in the food's retail price. (Recall from equation (A.2) that the producer price equals [([P.sub.a.sup.f].sup.e]). The analysis for the TE between the real exchange rate and retail price is similar.

APPENDIX B. ESTIMATION OF TEs USING JJ COINTEGRATION TECHNIQUES

Estimation of transmission elasticities using JJ cointegration techniques involves running the following error-correction model (with all the variables defined as in Equation (1)):

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

[PI] and [DELTA] are matrices of parameters to be estimated. The estimate of [PI] will be a 3 x 3 vector. If the JJ test finds that there is a cointegrating vector between the three variables, that means there exists at least one 3 x 1 matrix [alpha] and one 3 x 1 matrix [beta] such that [PI] = [alpha][beta]', [alpha] is the adjustment vector (to shocks), while [beta] is the cointegrating vector. Let [[beta].sub.1], [[beta].sub.2], and [[beta].sub.3] be the elements of the cointegrating vector corresponding to ln([P.sup.d.sub.t-1]), ln([P.sup.f.sub.t-1]), and ln([E.sub.t-1]), so that the estimated cointegrating relationship is

(B.2) [[beta].sub.1] ln([P.sup.d.sub.t-1]) + [[beta].sub.2] ln([P.sup.f.sub.t-1]) + [[beta].sub.3] ln([E.sub.t-1] = 0

The estimated price and exchange rate TEs are -[[beta].sub.2]/[[beta].sub.1] and -[[beta].sub.3]/[[beta].sub.1] respectively.

The standard deviations associated with the estimated elasticities are calculated as follows. The JJ test in most statistical software packages reports the estimates of the eigenvalues of matrix [PI] from largest to smallest, and the corresponding eigenvectors. The first eigenvector is [beta], the cointegrating vector. Let [[lambda].sub.1] be the corresponding eigenvalue. Further, let [[lambda].sub.2] and [[lambda].sub.3] represent the other two eigenvalues, and [v.sub.2] and [v.sub.3] their corresponding eigenvectors. Johansen and Juselius find that when there is just one cointegrating vector, and the linear restriction K holds such that K'[beta] = 0, then the following quantity is asymptotically standard normal:

(B.3) [T.sup.1/2]K'[beta]/[{(1/[[lambda].sub.1] - 1) (([(K'[v.sub.2]).sup.2] + [(K'[v.sub.3].sup.2)}.sup.1/2]

In other words, the asymptotic standard deviation is

(B.4) [square root of ((1/[[lambda].sub.1] - 1) ([(K'[v.sub.2]).sup.2] + [(K'[v.sub.3]).sup.2])/T)

Table 1: Russian agricultural and food imports

 1996 1997

 Volume Value Volume Value

Meat 2,183 2,141 3,283 3,118
 Beef 600 789 993 1,189
 Pork 450 586 820 884
 Poultry 1,116 753 1,444 1,026
Grain 3,607 713 3,371 601
Oilseeds (b) 18 19 23 18
Vegetable oil 383 320 765 456
Sugar 3,149 1,275 3,485 1,148
Vegetables and potatoes 753 336 1,173 389
Fruit 1,583 859 2,301 920
Other 5,103 5,598

Total agriculture and food 10,765 12,249

 1998 1999

 Volume Value Volume Value

Meat 2,612 2,322 2,701 2,196
 Beef 672 772 752 772
 Pork 725 719 800 698
 Poultry 1,165 805 1,080 705
Grain 1,714 248 6,852 632
Oilseeds (b) 28 16 237 69
Vegetable oil 622 346 940 425
Sugar 4,060 1,210 5,901 1,168
Vegetables and potatoes 1,201 343 1,724 361
Fruit 1,809 687 1,222 425
Other 4,453 2,462

Total agriculture and food 9,625 7,737

 2000 Share in total
 world imports
 Volume Value (1996-2000) (a)
 (%)

Meat 2,135 1,576 10.13
 Beef 496 564 8.44
 Pork 470 386 11.30
 Poultry 1,151 613 17.01
Grain 4,672 551 2.00
Oilseeds (b) 60 22 0.22
Vegetable oil 696 287 2.85
Sugar 4,817 765 14.10
Vegetables and potatoes 1,356 313 2.73
Fruit 1,690 625 3.49
Other 3,060 2.01

Total agriculture and food 7,199 2.95

(a) In value terms. Intro-EU imports are not considered part of
world imports.

(b) Includes no meal or oil from crushing.

Note: Volume data are in thousands of metric tons; value data in
millions of US dollars.

Source: USDA (Foreign Agricultural Trade of the United State
(FATUS), www.ers.usda.gov/df/FATUS/) for the meats; United Nations
(UN Trade Statistics, www.intranetapps.fas.usda.gov/untrade/) for
all other commodities

Table 2: Russian exchange rates

 Nominal exchange Change in real
Year rate (a) exchange rate (b) (%)

1992 222 -88
1993 928 126
1994 2,190 75
1995 4,571 40
1996 5,124 29
1997 5,784 -1
1998 10 (c) -27
1999 25 -24
2000 28 5
2001 29 15

(a) The rates give rubles per US dollar, and are averages of monthly
rates.

(b) Positive value means appreciation; negative value depreciation.

(c) In January 1998, a currency reform rebased the entire monetary,
price, and exchange rate system by dividing all values by 1,000. To
compare pre- and post-monetary reform values, multiply all post-reform
rates by 1,000.

Source: PlanEcon, Review and outlook for the Farmer Soviet Republics,
Washington, DC, annual

Table 3: Structure of retail value (percent) of Russian beef and pork

 1995 1999
Element of value
 Beef Pork Beef Pork

Cost of primary product
 (producer price) 56.1 52.6 58.7 56.7
Cost of processing 10.6 10.6 14.5 14.7
Profit of processor 6.7 6.7 5.2 5.3
Taxes (a) 8.9 9.1 6.7 7.8
Cost of distribution and retail
 sale 13.8 17.2 11.7 12.2
Profit from distribution and
 retail sale 3.9 3.8 3.2 3.3
Total 100 100 100 100

(a) Taxes on Russian foodstuffs are assessed after processing.

Source: Russian Federation State Committee for Statistics. Tseni v
Rossii (Prices in Russia), Moscow, 1996 and 2000

Table 4: Stationarity of price and exchange rate data

 Income index Beef Pork

Border price NS NS
Exchange rate NS NS

Cities (a)

Moscow 3.23 NS NS
Tyumen 1.66 NS
Samara 1.19
St. Petersburg 1.05 NS NS
Perm 1.04
Irkutsk 1.01 NS NS
Kazan 1.00 NS
Kemerovo 0.94 NS
Krasnoyarsk 0.92
Rostov on Don 0.87 NS NS
Ufa 0.86 NS
Khabarovsk 0.85 NS
Ekaterinburg 0.84 NS
Chelyabinsk 0.83
Yaroslavl 0.81
Ulyanovsk 0.77 NS NS
Krasnodar 0.76 NS NS
Orenburg 0.74 NS NS
Nizhniy-Novgorod 0.74
Omsk 0.72 NS
Saratov 0.72
Novosibirsk 0.71 NS NS
Astrakhan 0.70
Voronezh 0.69
Izhevsk 0.69 NS
Tula 0.67 NS NS
Barnaul 0.66
Vladivostok 0.65 NS
Volgograd 0.61
Ryazan 0.57
Penza 0.53 NS

(a) The cities are ordered according to an income index, calculated by
dividing the per capita income of each city by the cost of the
recommended food consumption basket computed for each city by the
Rusian Federation State Committee for Statistics, Regioni Rossii
(Russia's Regions), Moscow, annual.

Note: NS means non-stationary. A blank entry means that the data are
stationary.

Source: Own estimates

Table 5: Estimates of transmission elasticities

City Income Beef
 index (a)
 Trace
 PTE ERTE statistic (b)

Moscow 3.23 1.00 0.85 45.25 **
 [1] (0.15) (0.14)

Tyumen 1.66 S
 [2]

Irkutsk 1.01 2.23 1.21 35.25 (c)
 [6] (0.57) (0.41)

Kazan 1.00 S
 [7]

Ufa 0.86 S
 [11]

Ekaterinburg 0.84 S
 [13]

Orenburg 0.74 2.67 1.31 38.77 *
 [18] (0.61) (0.68)

Novosibirsk 0.71 1.05 0.71 43.24 **
 [22] (0.32) (0.27)

Izhevsk 0.69 S
 [25]

Vladivostok 0.65 2.11 1.22 42.47 **
 [28] (0.38) (0.35)

City Pork

 Trace
 PTE ERIE statistic

Moscow 0.37 0.24 37.34 *
 (0.15) (0.13)

Tyumen 0.61 0.36 60.22 **
 (0.14) (0.08)

Irkutsk 0.75 0.40 40.60 **
 (0.19) (0.16)

Kazan 0.27 0.03 40.81 **
 (0.13) (0.11)

Ufa 0.42 0.42 41.53 **
 (0.16) (0.16)

Ekaterinburg 0.46 0.30 40.98 **
 (0.27) (0.22)

Orenburg 0.14 -0.05 61.90 **
 (0.08) (0.06)

Novosibirsk 0.64 0.32 26.85 (c)
 (0.20) (0.19)

Izhevsk 0.24 0.07 53.90 **
 (0.08) (0.08)

Vladivostok S

(a) Brackets in this column give the city's per capita income rank
(out of 31).

(b) The trace statistic tests the significance of the cointegrating
relationship in the equation Linking domestic prices and border prices
and the exchange rate.

(c) Although the TE estimates are not significant, they are given
simply because the TE estimates for this city for the other meat are
significant.

* Significant at 10 percent level.

** Significant at 5 percent Level.

Note: PTE is price transmission elasticity; ERIE is exchange rate
transmission elasticity. S means domestic price data are stationary.
Standard errors are in parentheses.

Source: Own estimates

Acknowledgements

We thank Carlos Arnade, Mary Bohman, Michael Trueblood, and Thomas Vollrath for helpful comments. Any remaining errors are our own. The views expressed are the authors' alone and do not in any way represent official USDA views or policies.

(1) For more information as to how foreign trade in the Soviet Union was planned and managed, see Gregory and Stuart (1986).

(2) The main exceptions were agricultural goods produced on the private plots of state and collective farm workers and sold in farmers' markets.

(3) For example, if the US dollar rose by 10 percent vis-a-vis the Japanese yen, the Russian official exchange rates involving the dollar and yen would be adjusted such that one dollar now bought 10 percent more rubles than did one yen.

(4) For more information concerning agricultural trade restrictions during the transition period, see Organization for Economic Cooperation and Development (1998).

(5) These conditions are also discussed in the studies cited earlier as to why Russian domestic price integration for foodstuffs is poor.

(6) Russia's most heavily imported agricultural good in terms of world import share has been sugar, though much of it comes from neighboring Ukraine.

(7) The border price for an imported good equals the world trade price plus transport costs (the import's cif [cost, insurance, freight] value), while the border price for an exported good equals the world trade price (the export's fob [free on board] value). As the products for which transmission elasticities are calculated in this paper are imported rather than exported by Russia, the border prices used in this paper are cif values.

(8) During 1996-2000, Russia took about 30-40 percent of all US poultry exports (US Department of Agriculture ((USDA), Foreign Agricultural Trade of the United States (FATUS), www.ers.usda.gov/ db/FATUS). The reason a more precise figure is difficult to give is that during these years, much of the US poultry shipped to Russia went through Baltic ports, and was identified in the official trade data as exports to the Baltic countries rather than to Russia.

(9) This information was obtained directly from Russian agricultural specialists and meat traders.

(10) An example of this inconsistency is the following. Assume that Russia experiences a major depreciation in the ruble in nominal terms combined with high inflation (the depreciation being a likely contributor to the inflation). Assume also that in estimating the TE for the exchange rate, we use real consumer prices for foodstuffs, but the nominal as opposed to real exchange rate. We would get a large percent change in the exchange rate, but small percent changes in the real inflation-adjusted consumer prices. The calculated TEs would therefore be small. Yet, in real terms, the exchange rate TEs would be much higher. The main reason for the small calculated TE is that the rise in domestic food prices was adjusted for inflation (deflated), but the nominal exchange rate was not inflation-adjusted. If we used the real exchange rate in the TE estimations, we would be adjusting the exchange rate as well as domestic consumer prices for inflation. The change in the exchange rate would then also probably be small. Coupling the small change in the real exchange rate with the small change in the real consumer price would yield a larger TE.

(11) Although the US inflation rate during our period of calculation was not identical to inflation rates in other countries exporting to Russia, the other major exporters were also developed market economies. Inflation rates in the United States and other developed market economies over our calculation period were relatively low, typically less than 5 percent a year. Thus, the choice of which countries' CPI to use in the computations would have little effect on the results.

(12) A qualification is that a relationship can exist between stationary and non-stationary variables if all the non-stationary variables are cointegrated with each other. In out case, this would mean that if Russian trade prices and exchange rates were cointegrated, a relationship would exist between domestic prices and both the trade prices and exchange rates. In out study, however, trade prices and exchange rates are not cointegrated. The technique we use to test for this cointegration (Johansen and Juselius, 1990) is similar to that we employ to estimate our TEs. It therefore is examined later in the paper.

(13) We do two additional tests to show that stationarity exists among out domestic price data. The Levin-Lin-Chu technique tests whether all our domestic price data are non stationary. This hypothesis is rejected at the 1 percent level of significance (with a t-statistic of -123). The ImPesaran-Shin technique tests whether any of out domestic prices are stationary. At the 1 percent significance level (with a t statistic of--2.5), we can reject the hypothesis that there are no stationary data.

(14) If the cointegration test has n endogenous variables, the time series is T periods long, and the cointegration test uses k lags, multiplying the asymptotic critical values of the cointegration test by T/(T-nk) gives the approximate critical values for small samples (Ahn and Reinsel, 1990; Cheung and Lai, 1993).

(15) A JJ cointegration technique is also used to test whether the trade prices and exchange rates in our study are cointegrated, as discussed in footnote 12.

(16) For information concerning the agricultural trade policies of countries throughout the world, see the country trade policy reviews of the World Trade Organization. Parts of the reviews are downloadable at the WTO website www.wto.org.

(17) For example, Russia's tariffs for most agricultural imports in 2002 ranged from 5 to 20 percent, while the average "bound" agricultural tariff for the world in 2000 was 62 percent (Gibson et al., 2001). A "bound" tariff is the maximum tariff allowed by a country's membership in the World Trade Organization. Although actual tariffs for some country-commodity pairings are below bound levels, the world value-weighted average-bound tariff for agricultural products lies well above Russia's average tariff.

(18) Since high relative prices (reflecting costs) indicate comparative disadvantage, from these results Liefert concludes that Russia has an apparent comparative disadvantage in producing meat compared with grain and agricultural inputs.

REFERENCES

Ahn, SK and Reinsel, GC. 1990: Estimation for partially nonstationary multivariate autoregressive models. Journal of the American Statistical Association 85: 813-823.

Berkowitz, D and DeJong, DN. 2001: The evolution of market integration in Russia. Economics of Transition 9(1): 87-104.

Berkowitz, D, DeJong, DN and Husted, S. 1998: Quantifying price liberalization in Russia. Journal of Comparative Economics 26(4): 735-760.

Bornstein, M. 1987: Soviet price policies. Soviet Economy 3(2): 96-134.

Cheung, YW and Lai, KS. 1993: Finite-sample sizes of Johansen's likelihood ratio tests for cointegration. Oxford Bulletin of Economics and Statistics 55 (3): 313-328.

Cochrane, N, Bjornlund, B, Haley, M, Hoskin, R, Liefert, O, and Paarlberg, P. 2002: Livestock sectors in the economies of Eastern Europe and the former Soviet Union. Agricultural Economic Report no. 798, Economic Research Service, US Dept. of Agriculture, Washington, DC, February.

DeMasi, P and Koen, V. 1996: Relative price convergence in Rassia. IMF Staff Papers 43(1), 97-122.

Dickey. D. and Fuller, WA. 1979: Distribution of the estimates for autoregressive time series with a unit root. Journal of the American Statistical Association 74: 427-431.

Gardner, B and Brooks, K. 1994: Food prices and market integration in Russia: 1992 93. American Journal of Agricultural Economics 76(3): 641-646.

Gibson, P, Wainio, J, Whitley, D, and Bohman, M. 2001: Profiles of tariffs in global agricultural markets. Agricultural Economic Report no. 796, Economic Research Service, US Dept. of Agriculture, Washington, DC, January.

Goodwin, BK, Grennes, TJ, and McCurdy, C. 1999: Spatial price dynamics and integration in Russian food markets. Policy Reform 3(2): 157-193.

Goodwin, BK and Harper, DC. 2000: Price transmission, threshold behavior, and symetric adjustment in the US pork sector. Journal of Agricultural and Applied Economics 32(3): 543-553.

Goodwin, BK and Holt. MT. 1999: Price transmission and asymmetric behavior in the US beef sector. American Journal of Agricultural Economics 81 (3): 630-637.

Gregory, P and Stuart, R. 1986: Soviet economic structure and performance, 3rd edn. Harper Row: New York.

Hahn, WF. 1990: Price transmission asymmetry in pork and beef markets. Journal of Agricultural Economics Research 42(4): 21-30.

Johansen, S and Juselius, K. 1990: Maximum likelihood estimation and inference on cointegration--with applications to the demand for money. Oxford Bulletin of Economics and Statistics 52(2): 169-210.

Liefert, W. 2002: Comparative (dis?)advantage in Russian agriculture. American Journal of Agricultural Economics 84(3): 762-767.

Liefert, W and Liefert, O. 1999: Russia's economic crisis: Effects on agriculture. Agricultural Outlook, Economic Research Service, US Dept. of Agriculture, Washington, DC, June, pp. 15-18.

Loy, JP and Wehrheim, P. 1999: Spatial food market integration in Russia. In: Peters, GH and von Bravn, J (eds). Food Security, Diversification and Resource Management: Refocusing the Role of Agriculture, Proceedings of the 23rd International Conference of Agricultural Economists, Ashgate: Aldershot, UK. pp. 421-431.

Mundlak, Y and Larson, DE 1992: On the transmission of world agricultural prices. The World Bank Economic Review 6(3): 399-422.

Organization for Economic Cooperation and Development (OECD). 1998: Russian federation: Review of agricultural policies. OECD:Paris.

Quiroz, J and Soto, R. 1995: International price signals in agricultural prices: Do governments care? GERENS and ILADES/Georgetown University.

Sharma, R. 2003: The transmission of world price signals: the concept, issues, and some evidence from Asian cereal markets. In: Agricultural Trade and Poverty. OECO: Paris.

Tyers, R and Anderson, K. 1992: Disarray in world food markets: A quantitative assessment. Cambridge University Press: Cambridge, UK.

Wehrheim, P, Frohberg, K, Serova, E and von Braun, J. (eds). 2000: Russia's agro-food sector: Towards truly functioning markets. Kluwer Academic Publishers: Dordrecht, Netherlands.

STEFAN R. OSBORNE & WILLIAM M. LIEFERT

Economic Research Service, US Department of Agriculture, 1800 M St, NW, Room 5062, Washington, DC 20036-5831, USA. E-mail: sosborne@ers.usda.gov