Trade and Labor Standards in the European Union: A Gravity Model Approach

Abstract

Using a gravity model, we examine whether labor standards are important determinants of bilateral export performance for EU-15 countries over the period 1988–2001. We assess the conventional wisdom that countries with low labor standards and less stringent regulations have performed better in terms of trade performance, and we use a panel data set in a triple-indexed gravity model to conduct our empirical investigation. We find several instances where improvements in labor standards are related to improvements in export performance, suggesting that productivity improvements as a result of better standards outweigh the costs of these standards themselves. These results are robust to the choice of dependent variable, specification tests, and robustness checks. The standard variables used in gravity equations conform to theoretical expectations and are highly significant.

Keywords:

• international trade
• labor standards
• gravity equation

Notes

¹This dominance of the conventional wisdom is the result of several factors, including the growing backlash against globalization through trade and investment (see Chapter 1 in Bhagwati, 2004), a genuine concern for workers' rights, and a form of disguised protectionism by developed countries that are trying to link standards with trade.

²A look at data (for example unionization rates, social protection as a percentage of GDP or hours worked) for EU-15 countries reveals that there are important differences even though these countries have similar political systems (Van Beers, 1998). If one considers the number of ILO conventions ratified by EU-15 countries, which is a rather crude indicator of labor standards, there is quite a range in the data with Spain having ratified 129 and Austria 53 (out of a possible 185 Conventions).

³See also Dehejia and Samy (2008) for a brief history of labor standards in the European Union.

⁴According to estimates from UNCTAD, intra-EU trade as a percentage of total exports for EU and accession countries was more than 60% in 2004.

⁵ Kanbur and Chau (2002) have in fact confirmed empirically that ratification of ILO Conventions is indeed costly.

⁶More generally, we can think of the cost due to the labor standard as part of the total trade costs (direct and indirect).

⁷The specification of the gravity model in log-linear terms allows one to interpret the different coefficients as elasticities, even though the model itself is initially expressed in multiplicative form.

⁸We have also examined what happens when individual exporter and importer effects are considered and the significance of the results did not change significantly. The point estimates for the labor standards varied by 0.1 to 0.2 in absolute value and there was no systematic evidence that our estimates were biased upwards or downwards.

⁹Please refer to the Appendix for data sources.

¹⁰Since we do not have separate data for Belgium and Luxembourg, these two countries had to be removed from the EU-15 countries, leaving us with only 13 countries. We also did not have a lot of data on labor standards, including occupational injuries for these two countries.

¹¹Results with the other indicators are provided in the appendix, and additional results (discussed but not included in the paper) are available upon request.

¹²This is also carried out in Rodrik (1996) and Dehejia and Samy (2004), for example.

¹³We thank an anonymous reviewer for pointing this out.

¹⁴An LM test for the residuals of the static model (column 3 of Table 2) did not show any evidence of autocorrelation in the residuals.

¹⁵However, they align reasonably well with the cross-sectional findings for developing countries (Rodrik, 1996; Dehejia & Samy, 2004, 2009).