The S-curve in emerging markets.

Bahmani-Oskooee, Mohsen ; Kutan, Ali ; Ratha, Artatrana 等

INTRODUCTION

The relationship between trade balance and terms of trade or trade balance and real exchange rate continues to be of substantial interest in trade literature and by policy makers. There are three distinct methods of investigating the link between trade balance and real exchange rate or terms of trade, which includes testing the Marshall-Lerner condition, the J-curve, and the S-curve.

The Marshall-Lerner condition is an indirect method of assessing the effectiveness of devaluation in improving trade balance. It asserts that as long as the import and export demand elasticities add up to more than unity, a decline in terms of trade or a real devaluation will have a favourable impact on trade balance. While a seminal work by Houthakker and Magee (1969) used standard econometric methods in estimating the import and export demand elasticities, more recent studies by Rose (1990), Bahmani-Oskooee and Niroomand (1998), and Bahmani-Oskooee and Kara (2005) have relied upon cointegration analysis and estimated the elasticities for different countries (developed as well as less developed).

The above studies have shown that the Marshall-Lerner condition is satisfied in most countries. However, there have been circumstances under which trade balance has continued to deteriorate despite the fact that the Marshall-Lerner condition is satisfied. See Magee (1973), Junz and Rhomberg (1973), Meade (1988), Bahmani-Oskooee (1985), and Rose and Yellen (1989) for examples. Given that the Marshall-Lerner condition is a long-run condition, this occurrence shifted the emphasis to the short-run dynamics that traced the post-devaluation time path of trade balance to the widely known concept of the J-curve. The argument put forward by researchers is that there are adjustment lags between the time that a currency is devalued and the response of trade balance. Thus, a short-run deterioration is consistent with a long-run improvement in trade balance and seems like a necessary feature of devaluation. For an extensive review of the literature, see Bahmani-Oskooee and Ratha (2004).

More recently, another body in the literature has emerged that concentrates on analysing the cross-correlation function between trade balance and terms of trade. Using the cross-correlation function, Backus et al. (1994) were the first to show that trade balance is negatively correlated with current and future movements in terms of trade but positively correlated with past movements. They labelled this finding for OECD countries as the S-curve because of the asymmetric shape of the underlying cross-correlation function. Following the same logic and analysis, Senhadji (1998) investigated the pattern for 30 less developed countries and recovered the same empirical regularity in most cases. An important question that we address in this paper is, given the evidence for developed and less developed countries, whether a similar S-pattern could be observed for emerging economies as well. Evidence supporting such a pattern would suggest that similar dynamic, general equilibrium models could also be applied to these economies. Besides the general interest on the topic among policy circles, studying the link between terms of trade and balance of trade for emerging economies is quite important for several other reasons as well.

First, these new emerging markets of Europe have had a huge surge in their trade flows since the beginning of market reforms starting in the early 1990s. At the beginning, real exchange rates initially appreciated in all these economies. During the initial years of the transition, the key economic goal was to achieve a stable inflation environment, instead of competitiveness, because the elimination of government restrictions in general, and trade liberalisation measures in particular, increased aggregate demand, putting pressure on prices. Bahmani-Oskooee and Kutan (2007) find that over time, competitiveness, measured by terms of trade fluctuations, has played a much more important role as inflation was controlled and price convergence with the European Union (EU) is achieved. Some countries undertook a series of official devaluations, while others have switched to more flexible exchange rate policies. (1) Over the years, there has been a significant increase in trade flows, particularly to the EU. An important question then is whether there has been a significant link between movements in the terms of trade and trade balance during this time period.

Second, these economies are relatively small open economies, relying heavily on export revenues in their catching-up efforts with the EU. On May 2004, among our sample countries, Cyprus, the Czech Republic, Hungary, Poland, and the Slovak Republic joined the EU. Bulgaria, Croatia, Romania, and Turkey are the candidate countries. Russia has also been transforming her economy significantly and may apply for an EU membership in the future. Thus, real economic convergence towards the EU standards is the ultimate objective of their economic integration decision with the EU. Fluctuations in export revenues due to adverse terms of trade changes may delay the real convergence process, putting stronger pressure on the real sector of the economy, hurting competitiveness and ongoing restructuring efforts.

The rest of the paper is organised as follows: A review of the S-curve literature is provided in the next section. The data, definitions, and cross-correlation functions between terms of trade and trade balance are presented in the subsequent section, followed by concluding remarks and suggestions for future research in the final section.

REVIEW OF RELATED LITERATURE

Backus et al. (1994) develop a two-country stochastic growth model to study the properties of the short-term movements in trade balance and terms of trade in some OECD countries. They propose a dynamic general equilibrium model in which both trade balance and terms of trade are endogenous. Their model produces an interesting pattern, called the S-curve, describing the relationship between trade balance and terms of trade in the short run. It is called the S-curve because it resembles the letter S. According to the S-curve model, a negative correlation exists between trade balance and current and future movements in terms of trade but a positive correlation with past movements. This S-curve link between the two variables depends heavily on two key sources of economic fluctuations, namely, productivity and government purchases shocks. For example, a one-time positive productivity shock, which is assumed to be persistent, results in an increase in output and improvement in terms of trade. This productivity shock raises consumption spending, but less than output.

However, investment spending rises temporarily by more than consumption as capital is moved to its most productive location. Consumption and investment together increase more than output; as a result, a deficit in the trade balance occurs initially. Over time, however, the temporary boom in investment spending slows down and the trade deficit hence turns into a surplus. This response produces a pattern, resembling an S-curve, tracing the short-run dynamics of trade balance and terms of trade. Backus et al. (1994) test their model for 11 OECD countries and find that the trade balance is negatively correlated with current and future movements in terms of trade but positively correlated with past movements. However, they find relatively weak evidence for the S-curve for Austria, and Canada and the United States.

Given the evidence for the OECD countries, a natural question to ask is whether this S-curve may characterise the short-run movements between trade balance and terms of trade for less developing economies. Senhadji (1998) takes up this issue for 30 less developed countries. He develops an international real business cycle theory model that departs from a small country assumption by assuming a downward-sloping export demand function. His model builds upon a stochastic growth model of an open economy which faces productivity shocks as well as shocks to the trade partner's income. The theoretical model is calibrated to investigate the link between terms of trade and balance of trade for less developed countries under different parameter values for productivity and income shocks. The model is shown to replicate the S-curve for less developed countries as well. As in Backus et al. (1994), in Senhadji's (1998) model productivity shocks are shown to be critical in producing the S-curve.

In a more recent paper, Bahmani-Oskooee and Ratha (2007a) wonder whether the rather weak S-curve evidence found for the United States and other countries in Backus et al. (1994) is related to the aggregation of data. They argue that aggregation could not allow us to observe all the fluctuations observed in bilateral trade flow variables because of the smoothing-out effects as a country's terms of trade or trade balance could be weakening with one trading partner but becoming stronger with another. To deal with this issue, they estimate S-curves between the United States and her 24 trading partners. Initially, they present S-curves based on aggregate trade data of the United States. The results confirm the evidence reported in Backus et al. (1994), which suggests weak, at best, evidence for S-curve. However, data based on disaggregated or bilateral trade variables show much better support. This disaggregation approach has received further support by Bahmani-Oskooee and Ratha (2007b) when they carry out the S-curve analysis between Japan and her trading partners.

Not many studies investigate the link between terms of trade and trade balance for the emerging economies of Europe. To our best knowledge, there have been only three, but indirect, studies. Kemme and Teng (2000) provide evidence on the impact of real exchange rate misalignments on export performance. They find that the misalignments tend to reduce export growth in Poland. Egert and Morales-Zumaquero (2005) investigate the impact of nominal exchange rate volatility on export flows in 10 Central and Eastern European transition economies. They find that higher exchange rate volatility hurts export performance, and the negative effects of exchange rate volatility vary across countries. Bahmani-Oskooee and Kutan (2007) test the J-curve hypothesis for 11 European emerging economies and find support for the J-curve hypothesis in only three of the 11 countries, that is, Bulgaria, Croatia, and Russia. While in many of the countries real depreciation had a short-run effect, the short-run effects did not last into the long run.

DATA, DEFINITIONS, AND CROSS-CORRELATION FUNCTIONS

Monthly data, spanning from 1990M1 to 2005M6 in most cases, are used to compute the cross-correlation functions. The sample comprises Bulgaria (1995M1: 2004M6), Croatia (1992M1: 2005M6), Cyprus (1990M1: 2005M6), Czech Republic (1993M1: 2005M6), Hungary (1990M1: 2005M6), Poland (1990M1: 2005M6), Romania (1991M1: 2005M6), Russia (1993M11: 2005M6), Slovac (1993M1: 2005M6), and Turkey (1990M1: 2005M6). All data are collected from the International Financial Statistics of the International Monetary Fund (CD-ROM).

Following the practice in the literature, terms of trade is defined as the price of imports relative to the price of exports. For emerging markets no such data are available on a monthly basis. Therefore, following Bahmani-Oskooee and Ratha (2007a) the inverse of the real effective exchange rate is used as a proxy. Backus et al. (1994) define trade balance as net exports relative to Gross Domestic Product (GDP). Again, in the absence of monthly data on GDP, real trade balance is defined as net exports deflated by the consumer price index. The series are all detrended using the Hodrick-Prescott filter. Figure 1 reports the results. (2)

[FIGURE 1 OMITTED]

A graphical inspection of the cross-correlation functions in Figure 1 indicates that support for the S-curve is strong in cases of Bulgaria, Croatia, Poland, and Slovac, and weak in cases of the Czech Republic, Hungary, and Turkey. There seems to be no support in cases of Cyprus, Romania, and Russia. Certainly all these economies in our sample have experienced significant productivity shocks due to restructuring, including the Balassa-Samuelson type effects, so that the lack of productivity shocks cannot be the explanation for not detecting the S-curve. (3) A more plausible explanation is aggregation bias. Despite this, however, evidence indicates that, in the majority of cases, we have strong or marginal evidence for the S-curve, suggesting that these models developed either by Backus et al. (1994) for developed countries or by Senhadji (1998) for less developed countries may also be applicable to many emerging economies. More importantly, terms of trade fluctuations in seven countries seem to have a predictable pattern with respect to movements in trade balance. Policymakers in emerging countries may hence learn useful lessons from the experience of developed and less developed countries where a similar dynamic link between terms of trade and trade balance has been taking place.

CONCLUDING REMARKS

Backus et al. (1994) found that the cross-correlation function between terms of trade and trade balance resembles a horizontal S-curve and labelled this finding the S-curve. Senhadji (1998) confirms this pattern for a set of developing countries. We contribute to this literature by adding further support for this pattern from the emerging market economies. In particular, we find strong evidence of an S-curve in cases of Bulgaria, Croatia, Poland, and Slovac, weak evidence in cases of the Czech Republic, Hungary, and Turkey, and no support in cases of Cyprus, Romania, and Russia.

It is worth noting that empirical regularities based on aggregate trade data may be biased since aggregation can potentially suppress some of the patterns observed in trade at bilateral levels. Indeed, by employing bilateral trade data, Bahmani-Oskooee and Ratha (2007a, 2007b) came up with excellent support for the S-curve for the United States and Japan. A similar exercise in cases of emerging market economies is likely to bring in further support for the posited pattern. Of course, this claim cannot be substantiated until the trade data between each emerging economy and her trading partners become available. In addition, future research should test the applicability of the dynamic models developed by Backus et al. (1994) and Senhadji (1998) to emerging markets in a more formal way.

REFERENCES

Backus, DK, Kahoe, PJ and Kydland, FE. 1994: Dynamics of the trade balance and the terms of trade: The J-curve? American Economic Review 84 (1): 84-103.

Bahmani-Oskooee, M. 1985: Devaluation and the J-curve: Some evidence from LDCs. The Review of Economics and Statistics 67: 500-504.

Bahmani-Oskooee, M and Kara, O. 2005: Income and price elasticities of trade: Some new estimates. The International Trade Journal 19: 165-178.

Bahmani-Oskooee, M and Kutan, AM. 2007: The J-curve in the emerging economies of Eastern Europe. Applied Economics (forthcoming).

Bahmani-Oskooee, M and Niroomand, F. 1998: Long-run price elasticities and the Marshall-Lerner condition revisited. Economics Letters 61: 101-109.

Bahmani-Oskooee, M and Ratha, A. 2004: The J-curve: A literature review. Applied Economics 36: 1377-1398.

Bahmani-Oskooee, M and Ratha, A. 2007a: The S-curve dynamics of US bilateral trade. Review of International Economics 15: 430-439.

Bahmani-Oskooee, M and Ratha, A. 2007b: Bilateral S-curve between Japan and her trading partners. Japan and World Economy 19: 483-489.

Dobrinsky, R. 2003: Convergence in per capita income levels, productivity dynamics and real exchange rates in the EU acceding countries. Empirica 30(3): 305-334.

Egert, B, Drine, I, Lommatzsch, K and Rault, C. 2002: The Balassa-Samuelson effect in central and Eastern Europe: Myth or reality. Journal of Comparative Economics 31(3): 552-572.

Egert, B and Morales-Zumaquero, A. 2005: Exchange rate regimes, foreign exchange volatility and export performance in central and Eastern Europe: Just another blur project?. William Davidson Institute Working paper no. 782.

Houthakker, HS and Magee, SP. 1969: Income and price elasticities in world trade. Review of Economics and Statistics 54: 325-347.

Junz, HB and Rhomberg, RR. 1973: Price competitiveness in export trade among industrial countries. American Economic Review 63: 412-418.

Kemme, D and Teng, W. 2000: Determinants of the real exchange rate, misalignment and implications for growth in Poland. Economic Systems 24: 171-205.

Kocenda, E and Valachy, J. 2005: Exchange rate volatility before and after regime switch: A new measure and new evidence from transition. CERGE-EI Working paper.

Magee, SP. 1973: Currency contracts, pass-through, and devaluation. Brookings Papers of Economic Activity 1: 303-323.

Meade, EE. 1988: Exchange rates, adjustment, and the J-curve. Federal Reserve Bulletin 74(10): 633-644.

Rose, AK. 1990: Exchange rates and the trade balance: Some evidence from developing countries. Economics Letters 34: 271-275.

Rose, AK and Yellen, JL. 1989: Is there a J-curve? Journal of Monetary Economics 24: 53-68. Senhadji, AS. 1998: Dynamics of the trade balance and the terms of trade in LDCs: The S-surve. Journal of International Economics 46: 105-131.

(1) For a comprehensive review of the development in exchange rate policies in these economies, see Kocenda and Valachy (2005).

(2) For a detailed method of producing cross-correlation functions and the graphs, see Bahmani-Oskooee and Ratha (2007b).

(3) For a discussion of productivity and related issues in these economies, see, among others, Egert et al. (2002) and Dobrinsky (2003).

MOHSEN BAHMANI-OSKOOEE (1), ALI KUTAN (2) & ARTATRANA RATHA (3)

(1) Economics Department, University of Wisconsin-Milwaukee, MiLwaukee, WI 53201, USA. E-mail: bahmani@uwrn.edu

(2)Department of Economics and Finance, Southern Illinois University Edwardsville, Edwardsvilie, IL 62026, USA

(3) Economics Department, St Cloud State University, St Cloud, MN 56301, USA