Does real exchange rate volatility affect sectoral trade flows?

• Author: Caglayan, Mustafa
• Position: Statistical data

1. Introduction

   A review of the empirical and the theoretical literature that spans the period after the breakdown of the Bretton Woods agreement reveals that there is no consensus on the impact of exchange rate volatility on trade flows. Several theoretical studies arrive at the conclusion that exchange rate volatility can have a negative impact on trade flows. (1) Equally, several others conclude that the effect is uncertain or positive. (2) Interestingly, one cannot reach a firm conclusion from empirical studies, either. Results are conflicting and sensitive to various factors. (3)

   When we focus on the recent empirical literature, we come across several possible reasons why researchers have reached conflicting conclusions. Early empirical research, which concentrated on aggregate U.S. or (37 data, suggests that exchange rate uncertainty may have a positive or negative effect on trade flows. (4) Recent research that focuses on bilateral rather than aggregate trade data of advanced countries concludes that exchange rate volatility has no or little effect on trade flows. (5) In this study, we utilize a broader data set that contains both advanced and emerging top trade partners of the United States. Hence, we avoid the narrow focus on the United States and the advanced country data that have characterized much of the literature.

   We should point out that the inclusion of advanced and emerging countries in our investigation is important, as recent research suggests that exchange rate volatility has a significant negative impact on trade flows of emerging countries. For instance, Grier and Smallwood (2007) conclude that while real exchange rate volatility has a significant negative impact on international trade for emerging countries, there is no such effect for the advanced economies. Several other researchers also report similar findings for different sets of emerging countries on the linkages between exchange rate volatility and trade flows. (6) Although one can claim that the presence of a significant relationship may be due to the lack of proper financial tools in emerging countries that firms can use to hedge against exchange rate fluctuations, Wei (1999) cannot find empirical evidence to that end. In this article, we utilize data from nine advanced and five emerging countries.

   Although the use of country-specific bilateral trade data is an improvement over aggregate trade data, sectoral trade data can help us further disentangle the linkages between exchange rate volatility and trade flows that may exist across sectors but not in bilateral data. However, there are only a handful of articles that use sectoral data to investigate the impact of exchange rate uncertainty on sectoral trade flows. Also, the early literature that used sectoral data summarizes the impact of exchange rate volatility on sectoral trade flows in one coefficient as researchers implement panel data methodologies. In contrast, we focus on country sector-specific bilateral relationships and investigate dozens of models. (7) Our data are organized with respect to bilateral sectoral trade flows between the United States and its top 13 trading countries. Our 14-country data set includes the United States, Japan, Germany, the United Kingdom, France, Italy, the Netherlands, Ireland, Canada, South Korea, Singapore, Malaysia, China, and Brazil and covers the period between 1996 and 2007 on a monthly basis.

   Another important factor that may affect the results in this literature is the method that one uses to generate a proxy for real exchange rate volatility. (8) Generally, the early research has used a moving average standard deviation of the past monthly exchange rates or variants of ARCH methodology to generate a proxy for exchange rate volatility. We utilize daily spot exchange rates to proxy for exchange rate volatility employing a method proposed by Merton (1980). This method, also used by researchers including Baum et al. (2004) and Klaassen (2004) in similar contexts, exploits daily exchange rate movements to proxy for monthly exchange rate volatility. Furthermore, both studies indicate that this approach yields a more representative measure of volatility avoiding problems associated with proxies derived from ARCH methodology or moving standard deviations. In particular, the Merton (1980) methodology avoids potential problems, including high persistence of shocks when moving average representations are used or low correlation in volatility when ARCH/GARCH models are applied.

   Last but not least, our empirical model takes the form of a simple distributed lag model where we allow each variable to affect trade flows up to six lags, which is shown to be adequate to capture the explanatory variables' impact. We keep those models that yield a stable dynamic relationship and discard the remaining models, which are dynamically unstable. In total, we scrutinize over 200 models where we discuss the impact of volatility measures across sectors and countries. To address an interesting suggestion raised by Baum et al. (2004), we also allow for income volatility and an interaction term between income and exchange rate volatilities in our model. They suggest that higher volatility of foreign income may signal greater profit opportunities inducing entry into the market or delaying exit from the market. Also, the interaction term between foreign income and exchange rate volatilities may help capture indirect effects emanating from any of these variables, which may capture the impact of the expansion or retention of trade flows as foreign income and the exchange rate fluctuate while addressing the presence of nonlinearities in the model.

   Our results provide evidence that exchange rate uncertainty has little effect on sectoral trade flows. We find that the impact of real exchange rate volatility on trade flows is significant in about only 6% of the models at the 5% significance level, where the effect is positive. Furthermore, although this relationship is slightly stronger for the emerging countries, our findings do not support earlier findings that exchange rate volatility plays an important role for emerging country trade flows. Overall, our results show that there is little effect of exchange rate volatility on sectoral trade flows of advanced and emerging economies.

   When we investigate the effects of income volatility and the interaction term between exchange rate volatility and income volatility on trade flows, we come across some interesting observations. It turns out that the interaction term is significant in almost all cases when exchange rate volatility plays a significant role in the model. Furthermore, it takes the opposite sign to that of exchange rate volatility, reversing the impact of exchange rate volatility on trade flows. From this perspective, it is apparent that omitting the interaction term from the analysis would lead to wrong policy prescriptions. When we observe the role of income uncertainty, we see that this variable significantly affects trade flows in 6% of the models at the 5% level, while its sign is generally the same with that of exchange rate volatility. This variable seems to play a more important role when we concentrate on exports of the United States to its trading partners. This is not surprising, as the income of the trading partners over the period under investigation was much more volatile than that of the United States.

   We finally check for the robustness of our findings by implementing a semirestricted model to test those effects arising from exchange rate and income volatilities and their interaction. Our investigation provides support for our earlier conclusion that exchange rate uncertainty has negligible impact on trade flows.

   The reminder of this article is organized as follows. Section 2 outlines the model, discusses our volatility measures, and provides information on the data. Section 3 reports the empirical results, and section 4 concludes.

2. Model Specification

   Most of the early research that concentrated on the impact of exchange rate volatility on trade flows used country-level aggregate or bilateral trade flow data. However, as Bini-Smaghi (1991) indicates, because sectoral data do not constrain income and price elasticities across sectors, one should employ sector-specific data when exploring the linkages between trade flows and exchange rate movements. Yet there are only a handful of studies that utilize sectoral data. (9) These studies follow an Armington (1969) approach and estimate both price and output elasticities. In particular, to capture export flows from country i to j, the model takes the form

   [X.sub.ijt], = f([P.sub.ijt], [Y.sub.jt], [[sigma].sub.ijt]), (1)

   where [Y.sub.jt], [X.sub.ijt], [P.sub.ijt], and [[sigma].sub.ijt] denote income of country j and exports, relative price, and exchange rate volatility from country i to country j, respectively. The price and output elasticities (coefficients associated with relative prices and output) are estimated in a panel context using sectoral trade flow data for each sector. Naturally, this approach yields a single sector-specific price and output elasticity along with the impact of exchange rate volatility, which is then compared across sectors.

   Our approach differs from the above specification, as we model the impact of exchange rate volatility for each sector- and country-specific trade flow separately. Given that we have 14 countries where data are ordered with respect to i) sectoral exports of 13 countries to the United States and ii) sectoral exports of the United States to the same set of countries, the maximum number of models that we can estimate is 260. However, because of a lack of data on exports from Ireland to the United States for sectors 4 and 5, we estimate 258 models. Of these 258 cases, we discard 28 models, as they fail the dynamic stability conditions, rendering us with 230 models to scrutinize. Our model...