The not so distant effect of distance within a time zone

Abstract

This article extends the distance and time zone trade literature by examining the impact on trade of larger distance within a time zone between a country pair. Although countries cannot control the physical distance between them and other countries, they do have some control over the time zone difference. We find the quartile for distance within each time zone difference and use these to create conditional distance quartiles which measure the overall impact of falling in the 25th–90th quartile for distance, given the time zone difference. We find that a larger distance within a time zone comes at an additional cost. For trade purposes, countries that could move from the 90th to the 25th conditional quartile for distance by increasing their time zone difference by two hours are better off doing so.

Keywords:

• exports
• time zone effects
• distance measures

JEL Classification:

• F10
• F13
• F14

Notes

¹ For example, this negative distance effect is still present for services and financial capital trade (Portes et al. (2001)) which have no transportation costs.

² Evidence of this negative effect has been found empirically by Stein and Daude (2007) and Hattari and Rajan (2012).

³ Theoretical support was provided by Marjit (2007) and Kikuchi and Long (2010) and found empirically by Tomasik (2013).

⁴ The sample of countries in this study is similar to the one used by Stein and Daude (2007), with any differences caused by data availability. List of countries in the sample available upon request.

⁵ The gravity model takes the following form: =, where is the total exports from country i to country j, and are country i and j’s GDP, respectively, and is the distance between the countries. Silva and Tenreyro (2006) recommend estimation of the gravity model in this original, multiplicative form.

⁶ The test for the residuals produced a test statistic of 6.6e+05 and a p-value of 0. Hence, there was overwhelming evidence suggesting heteroscedasticity in the data.

⁷ A test of difference in coefficients across regressions was conducted using the Suest command.

⁸ To calculate the marginal effect using PPML estimation, the formula used is . Marginal effects are reported in this section.

⁹ Following Stein and Daude (2007), for the United States, the more centrally located Chicago was used over Washington D.C.

¹⁰ For the United States, only time zones in the contiguous 48 states were considered.

¹¹ These results are available from the authors upon request.