Export premium, productivity, trade openness and wage inequality in China : empirical evidence from firm-level data.

Abstract

This thesis uses Chinese firm-level data to investigate the relationships between the export premium, firm productivity and wage inequality. Using Chinese annual survey data for all state-owned firms and other non-state-owned firms with sales on mainland China over 5 million RMB, the author finds that there is a series of premiums for exporters compared with non-exporters. On average, exporters pay higher wages, produce more, sell more, add more value, employ more labour, have higher capital intensity, and have higher productivity (based on 1999-2003 data). Firms with relatively high export values will also be relatively more productive. Quantile results show that the premium decreases with the increase of the quantile. In addition, the export premium declines over time and across the industries, provinces and ownership types, and the higher the export intensity, the lower the export premium. The thesis further investigates the question: what determines the export premium – the selection effect or learning-by-exporting effect? First, the author uses the Olley and Pakes (1996) method to control both selection and simultaneity bias to estimate the reliable firm productivity. Then the author tests the self-selection and learning-by-exporting effects both parametrically and non-parametrically. The author finds both strong self-selection and learning-by-exporting effects at the aggregate level. The higher the productivity the firm has today, the easier for the firm to export tomorrow. The learning-by-exporting effect is the most significant in the second yearafter exporting. However, at the more disaggregated level, no significant learning effect is found within sectors and within middle and western provinces. A significant learning effect is found in eastern provinces. The learning-by-exporting effect across different ownership types is not robust to different testing methods. In addition, the author uses Chinese privately-owned firm-level survey data to investigate the heterogeneous export premium associated with different levels of trade. Firms engaged in international trade have higher premiums than firms which trade only across province borders. Firms which trade across province borders have higher premiums than firms that only trade within their province. Furthermore, export premium deviation between international trade and interprovincial trade is much smaller compared with the export premium deviation between interprovincial trade and inter-county trade. This finding implies that compared with the inter-county premium, the premium at interprovincial level is similar to the premium at the international level (though the former is actually less than the latter). The export premium caused by the self-selection effect can reflect the trade cost and it tells that trading goods across provincial borders within China is as onerous as crossing national borders. The next question to consider is whether engaging in international trade causes the wage inequality between firms to increase? To find out the answer, the author adopts a two-stage estimation strategy to study the effect of international trade on wage inequality. The first stage uses the Chinese annual survey firm-level data to calculate the wage inequality indexes—Gini and Theil of each province; as well as two dimensions of trade openness—intensive margin and extensive margin of each province. The second stage uses the panel data to study the impacts of trade margins on wage inequality between provinces. The results show that the variation of trade openness itself can explain nearly 70 percent of variation of wage inequality across China’s provinces and the extensive margin has a larger impact on increasing wage inequality than the intensive margin. Instrumental variable (IV) regression results imply that with one unit of increase in trade openness, the intensive margin increases wage inequality by nearly one unit and the extensive margin increases wage inequality by 1.2 to 1.3 units.