A causal relationship between exports and productivity at the plant level: The case of Turkey.

摘要

This dissertation first examines the performance of Turkey's apparel, textile, and motor vehicle and parts industries using plant-level data from 1987--1997. The interesting results include the significant impact of foreign ownership, technology purchase, subcontracting, location, plant size, and skilled labor on the productivity of plants. Furthermore, most of the changes in industry-level productivity result from the reallocation of resources from less to more productive plants. Finally, exporting plants are more productive than their domestically-oriented counterparts.;Next, the dissertation investigates the causal relationship between exports and productivity using plant-level panel data from the Turkish apparel and motor vehicle and parts industries from 1990--1996. Three different models are utilized to accomplish this goal. The first model examines the relationship between the past export status of plants with the productivity of plants today using the Ordinary Least Square (OLS) estimator with and without controlling for the export history of plants. The results show that the plants' prior market experience is a factor in their current productivity.;The OLS model, however, neither controls for the endogeneity nor the unobserved plant-specific effects that persist over time. Moreover, it does not distinguish between the long- and short-run relationship between exports and productivity. Thus, a second model, the Generalized Method of Moment with an Error-Correction specification, is utilized to incorporate these three issues in the examination of the causal relationship between exports and productivity. The empirical answers provided with this model indicate that there is a bidirectional causal relationship between exports and productivity both in the long- and short-run.;The assumption that each plant within an industry shares a common production function relating input and output was the basis for the measurement of productivity used in the previous two models. Thus, a Semi-Parametric Model is utilized to incorporate the plant-level heterogeneity across plants in the measure of productivity. After utilizing this third model, we find statistically significant results in the apparel industry, however, not the motor vehicle and parts industry in regards to the effect of exporting on productivity. With respect to the effect of productivity on exporting, however, significant results were seen in both industries.;On the policy front, suggestions in regards to the causal relationship between exports and productivity following our findings must consider that industries have different structures and dynamics, thus they learn by exporting to different extents. However, the effects of productivity on exports are realized regardless of the industry. Thus, exports can be expanded by improving the productivity of plants.