Trade, labour market rigidity, and aggregate productivity in OECD countries

Abstract

As the economy becomes more open to trade, aggregate productivity can increase by driving out the least productive firms (the selection effect). Since the selection effect reallocates resources toward the more productive firms, this process can be hindered by rigidity in domestic labour market institutions. Based on the selection effect by Melitz (2003), this article empirically examines how rigidity in labour market institutions affects the consequence of trade on aggregate productivity. Findings from panel dynamic ordinary least square (DOLS) estimators suggest that a high degree of labour market rigidity in an open economy reduces Total Factor Productivity (TFP). In particular, in the case of extremely high labour market rigidity but low foreign R&D capital stocks, openness to trade can cause a country to experience decreasing TFP.

Keywords:

• selection effect
• employment protection legislation
• panel cointegration

JEL Classification:

• F16
• C23

Acknowledgements

I am grateful to Robert McNown, Wolfgang Keller, Brian Cadena and Carol Shiue for helpful comments. The views expressed herein are those of the author and do not necessarily reflect the views of the Samsung Economic Research Institute.

Notes

¹ In this article, the selection effect implies the positive selection effect.

² Research on the impact of institutions on the foreign R&D spillover effect has been done by Coe et al. (2009). They show that well-established institutions, such as the ease of doing business, the quality of tertiary education, the strength of intellectual property rights and the origins of legal system, enhance the international spillover effect. However, they do not find significant evidence of effect owing to financial market development, labour market institutions, governance and ease of trade across borders.

³ Davidson et al. (2008) also explain the relationship between the selection effect and EPL. Their argument is that if there is little job creation in the incumbent and new exporting firms, the workers with high ability in nonexporting firms do not have much chance to transfer toward the exporting firms. Therefore, the inactive transfer of workers with high ability in nonexporting firms cannot accelerate the selection effect of trade.

⁴ On the basis of Helpman et al. (2010), Kang (2010) investigates the labour market conditions that determine the impact of trade on the average real wage. Unlike Melitz (2003) with the assumption of full employment, Helpman et al. (2010) have the labour market frictions due to emphasis on heterogeneity across both workers and firms; that is, workers below the ability threshold of a nonexporting firm are likely to be unemployed; workers between the ability threshold of a nonexporting firm and an exporting firm are employed only by nonexporting firms; and workers above the ability threshold of an exporting firm are employed by exporting firms.

⁵ Job creation occurs mainly in exporting firms, thus amplifying the positive impact of exports on the average wage. Job destruction occurs when high ability workers in nonexporting firms transfer to exporting firms and low ability workers in nonexporting firms are fired. This reduces the negative impact of imports on the average wage.

⁶ Since Bernard et al. (2007) assume homogeneous workers and heterogeneous firms, workers can affect aggregate productivity through changes in the common real wage of the entire economy. However, if heterogeneous abilities among workers are allowed, workers who do not meet the cut-off in nonexporting firms may affect the aggregate productivity by being unemployed.

⁷ Kambourov (2009) shows that high firing costs slow down the intersectional reallocation of labour after a trade reform.

⁸ As roughly suggested by Wooldridge (2002), when the number of time-series observations is equal to or greater than the number of cross-sectional observations, the assumption about the nature of the time dependence and time series analysis is required.

⁹ According to Potrafke (2010) using OECD dataset, globalization does not influence neither union density nor overall employment protection.

¹⁰ Some countries are excluded in this article due to missing data; Greece, Iceland, Israel, Portugal and South Korea.

¹¹ CEC are Austria, Belgium, Denmark, Finland, France, Germany, Italy, Netherlands, Norway, Spain, Sweden and Switzerland. ASC are Australia, Canada, Ireland, New Zealand, the United Kingdom and the United States.

¹² Eviews 7.0 reports the chi-square value based on MacKinnon et al. (1999) p-values for Johannsen’s cointegration trace test.

¹³ I use Eviews 7.0 in order to conduct the DOLS estimator applied to the ith country of the panel. Eviews 7.0 provides automatic selection for the conventional time-series DOLS lags and leads and for long-run variance whitening regression.

¹⁴ The coefficient of domestic R&D stock () by Coe et al. (2009) is 0.095, while that coefficient is about 0.22 in this article. This gap may come from excluding some countries and the human capital variable.

¹⁵ Union density is taken from OECD statistics.