Technological structure of trade and BOP-constrained growth in Turkey

ABSTRACT

This study investigates the empirical validity of the Multisectoral Thirwall's Law for Turkey over the last half century. Multisectoral Thirwall's Law facilitates the discussion of the effects of the sectoral composition of trade on the extent of the balance of payments (BOP) constraint and consequently on the long run growth prospects of an economy. In particular, structural change favoring sectors with Schumpeterian and Keynesian efficiencies is expected to improve these prospects. Lall's commodity classification on a technology basis is adequate for delineating such sectors. In this study, distinct export and import functions are estimated for primary production, low technology manufacturing, medium technology manufacturing and high technology manufacturing industries using the autoregressive distributed lag (ARDL) bounds test and Johansen approaches to cointegration. Resulting income elasticities are used to discuss the structural change in the technological content of Turkish trade and the validity of the Multisectoral Thirwall's Law. Results suggest that Turkey has come a long way in terms of improvements in Schumpeterian and Keynesian efficiency over the last 50 years and that the Multisectoral Thirwall's Law is empirically valid in the case of Turkey.

KEYWORDS:

• Balance of payments
• export-led growth
• trade structure

JEL CLASSIFICATION:

• E12
• F14
• F43

Disclosure statement

The author declares no conflict of interest.

Notes

1. Studies that try to model export and import functions for Turkey typically focus on the post-1980 era. This study covers both periods because the structural shift that occurred in early 1980s is exactly the kind of change this multisectoral framework will internalize.

2. Indicators presented in Figures 3 and 4 follow from ECLAC (2012) and facilitate a direct comparison of the Turkish experience to their discussion of the Latin American experience.

3. Alonso (1999) reformulates this test in terms of a test of cointegration between the actual and predicted growth rates, which completely overcomes the problem of a stochastic regressor. Since the Turkish actual and predicted growth series turn out to be stationary, using a cointegration test is not possible in this case.

4. In order to overcome the limitations of a single constant growth rate and a constant income elasticity of imports for the whole period in question, the test for the original Thirlwall's Law is usually carried out by analyzing overlapping sub-periods following Atesoglu (1993). In the MSTL framework, however, both the predicted growth rate and the elasticities are structurally dynamized in relation to the sectoral composition of trade, so analysis of overlapping sub-periods or rolling regressions to get changing elasticities and predicted growth rates are not necessary.

5. If the implied and estimated income elasticity of import demand series were non-stationary a cointegration test á la Alonso (1999) could have been possible. Turkish data at hand does not permit this.