Exchange rate uncertainty and US bilateral fresh fruit and fresh vegetable trade: an application of the gravity model

Abstract

In order to analyse the effect of exchange rate uncertainty, we apply an empirical gravity equation to two sets of US bilateral trade data: fresh fruit over the period 1976–1999 for a panel of 26 countries; and fresh vegetables over the period 1976–2006 for a panel of nine countries. Based on panel estimation methods, and using both a moving SD measure and the Perée and Steinherr (1989) measure of exchange rate uncertainty, the results show that US bilateral fresh fruit trade has been negatively affected by exchange rate uncertainty. We also find some evidence that the exchange rate between the US dollar and the currencies of Latin American trading partners accounts for most of the negative impact of exchange rate uncertainty on bilateral trade flows in fresh fruit. In contrast, when using panel estimation methods and both measures of exchange rate uncertainty, we find no statistically significant evidence for any negative effect of exchange rate uncertainty on US bilateral fresh vegetable trade. However, we do find a statistically significant negative effect for exchange rate uncertainty when we estimate a US export gravity equation for fresh vegetables using the same panel of countries.

Keywords:

• exchange rates
• uncertainty
• fresh fruit and fresh vegetables
• panel data sets

JEL Classification::

• F14
• F31
• C33

Acknowledgements

Funding for this research was provided through Cooperative Agreements No. 43-3AEK-3-80068 and No. 58-3000-7-0127.

Notes

¹ Earlier work on agricultural trade and exchange rates, focused on the impact of, changes in the level of the real exchange rate, and agricultural exports. Examples include Batten and Belongia (1986), and Bessler and Babula (1987). This research has been summarized and reviewed by Kristinek and Anderson (2002).

² The recent paper by Karemera et al. (2011) extends the earlier work of Cho et al. (2002), by examining the impact of exchange rate uncertainty on vegetable trade for an OECD panel data set over the period 1996–2002.

³ The banana-exporting countries include Colombia, Costa Rica, Côte d’Ivoire, Ecuador, Guatemala, Honduras and Panama, while Southern-Hemisphere countries consist of Argentina, Australia, Brazil, Chile, New Zealand, Peru and South Africa.

⁴ See Sheldon (2006) for an extensive discussion of the theoretical foundations for the gravity model.

⁵ More broadly, distance may capture other barriers to trade (Anderson and van Wincoop, 2004).

⁶ While the expected sign is negative, De Grauwe (1999) shows that, if the utility function is sufficiently convex, firms may export more rather than less when uncertainty arises. Bachetta and van Wincoop (2000) have forwarded similar arguments.

⁷ See Egger and Pfaffermayr (2003) for a useful discussion of the econometric specification of the gravity equation.

⁸ Note that the fresh fruit trade sample is truncated at 1999 due to introduction of the Euro.

⁹ Correlation between real GDP per capita and trade was observed to be higher than the correlation between real GDP and trade. Furthermore, real GDP and population were found to be highly positively correlated with each other.

¹⁰ http://faculty.haas.berkeley.edu/arose/RecRes.htm.

¹¹ Information about common membership in free trade agreements was obtained from the website: http://www.dfait-maeci.gc.ca/nafta-alena/menu-en.asp.

¹² In constructing the exchange rate uncertainty measure there is a certain amount of arbitrariness involved concerning the choice of the measure and the time period covered. In this article, n = 2, 3 or 5 years. The results are generally robust to the time period covered by the measure, the most important issue being the form the measure takes.

¹³ The data can be found at the website: http://www.ers.usda.gov/data/exchangerates/

¹⁴ In both the case of fresh fruit and vegetable trade, an alternative to estimating Equation 6 by fixed effects is to estimate it via random effects. In doing the latter, while the results for exchange rate uncertainty were found to be robust across the fixed and random effects estimates, application of the Hausman test concluded in favour of the fixed effects model, so we do not report the random effects results in the article. See Egger (2000) for a discussion of fixed versus random effects estimation of the gravity equation.

¹⁵ In the case of the moving SD measure of uncertainty, , when adjusting for group-wise heteroscedasticity in the results, while still finding that the estimated parameter on region r = 2, Latin America, was negative for n = 2, 3 and 5 years it was statistically insignificant, so we do not report those results here.

¹⁶ Based on Bahmani-Oskooee and Hegerty's (2007) survey, we chose to estimate an export rather than an import gravity equation, based on the fact that this has been by far the most common approach in the empirical literature on exchange rate volatility.

¹⁷ We also estimated the export equation using a random effects specification. Application of the Hausman test could not discriminate between a fixed or random effects specification. We chose to report only the fixed effects results, but the random effects results are available from the authors upon request.