Trade liberalisation and the composition of investment: theory and African application

While the effect of trade liberalisation on aggregate investment has been a major focus of study, its effect on the composition of investment has re¬ceived little attention. The effect on aggregate investment is somewhat con¬troversial but recent evidence suggest that it usually is positive. Sachs and Warner (1995) find that the openness dummy which they use significantly and substantially increases the investement/GDP ratio. Similarly, Levine and Renelt (1992) find that the result that the investment ratio is higher in liberalised economies is one of the few results of the empirical growth literature which passes the test of Extreme Bounds Analysis. However, there is some evidence that in Africa trade liberalisation has tended to reduce aggregate investment (Mosley et al, 1991), and in the theoretical literature several explanations have been proposed for a negative effect. One emphasises lack of credibility. When trade liberalisation is not considered fully credible it may be optimal for private agents to reduce fixed investment in favour of stocking imports (Calvo, 1987, 1988) or to remain liquid until policy uncertainty is resolved (Dixit, 1988). However, liberalisation can reduce aggregate investment even when fully credible. The first explanation for this relies on differences in factor proportions. If the protected import substitutes sector is capital intensive then trade liberalisation will reduce the return on investment (Buffie, 1992). The second explanation stresses that protection often applies only to consumer goods. Trade liberalisation then amounts to the removal of a subsidy on capital goods (Collier and Gunning, 1992, Buffie, 1992). In this paper we use variants of these two models to analyse how liberalisation might change the composition of investment. As we show, empirically episodes of trade liberalisation have sometimes been associated with very large changes in investment composition. When investment data are decomposed into tradable capital (equipment) and non-tradable capital (structures), the post-liberalisation slump in investment is found to be composed entirely of a fall in the former. Investment in structures, by contrast, appears to experience a boom. This phenomenon has implications both for analysis and measurement. Models which fail to distinguish between equipment and structures, or in which the two are used in fixed proportions, as in Buffie (1992), are evidently unable to explain the phenomenon. Substitutability between equipment and structures must be introduced either directly, or in the form of factor intensity differences between sectors. Measures of invest¬ment response which fail to make the disaggregation risk systematic bias. The purpose of this paper is to present this evidence and to develop simple models which can explain the conjunction. We present two models. The first one illustrates the equipment intensity effect: if the composition of investment differs sufficiently between sectors, with investment in the protected sector intensive in equipment and investment in the export sector in structures, then the conjunction occurs: equipment investment falls, but construction increases. In the second model there is an equipment subsidy effect. In this model equipment and structures are substitutes in the production of an aggregegate capital good which is used both for exports and importables. Hence there is no equipment intensity effect: the two sectors can differ in capital-labour ratios, but equipment-structures ratios are the same. However. in this model tariffs apply only to consumer goods so that there is an implicit subsidy on equipment investment. As a result trade reform raises the price of equipment relative to structures, the equipment subsidy effect. We show that if equipment and structures are good substi¬tutes and the protected sector is capital intensive while in the short run export supply is inelastic then the conjunction of an investment slump and a construction boom occurs. The equipment subsidy is necessary for this result. In the next section we present African evidence on changes in the composition of investment in the wake of trade reform. Section 3 presents two models which may explain this result, one illustrating the equipment intensity effect, the other one the equipment subsidy effect. Section 4 concludes.