Estimation of Factor Contribution to Growth under Structural Disequilibrium

SUPPOSE WE HAVE TIME series data on output capital and labor inputs and on factor shares in a rapidly industrializing economy, and we want to measure the contribution of various factors to growth. Would it be correct to weight capital and labor by their observed average income shares in order to assess the growth of productivity or technical progress over time?2 This would imply that we are assuming competitive equilibrium in both product and factor markets, constant returns to scale, and also a specific form of production function (usually Cobb-Douglas). Intuitively one feels that there are many cases where such simplifications might give us biased results. The use of the expression structural disequilibrium when talking of 'developing' economies has become a rather hackneyed phrase of which everyone is aware, yet rarely can the monster be forced into the straight jacket of quantitative theory and measurement.3 However, an attempt to tackle the empirical problem, at least partially, can be based on the modern approach to production function estimation which views production and distribution as a whole, by incorporating both the technical relations and the equilibrium side-conditions in one testable model.4 Arrow, Chenery, Minhas and Solow (ACMS) [2] have taught us that if one starts from an empirical relation between output per unit of labor VIL and the real wage w and assumes equilibrium in the labor and product markets, one can work out the implied general family of production functions (of any degree of homogeneity) which can then be put to an independent empirical test. Also, authors such as Hoch [5] and Mundlak [10] have shown systematic