Measurement of technical change : a case study of manufacturing industries in Singapore

Singapore's economy has been growing rapidly for the last two decades. Contrary to the conventional belief that productivity growth and technical progress are the main forces behind the rapid growth, many recent research findings point out that the rapid growth of output in Singapore has been accompanied by very little productivity growth measured by total factor productivity growth (TFPG). Studies on productivity measurement in Singapore mainly employ the non-parametric productivity accounting approach. Firms are assumed to be operating in perfectly competitive long-run equilibrium, which is characterized by the conditions of perfect competition, full utilization of capacity and constant returns to scale. If these conditions are not met, the conventionally calculated TFPG is different from the primal and dual estimates of technical change. We derive an expression that gives the TFPG bias when the conditions of perfectly competitive long-run equilibrium are not met. The TFPG bias can be positive or negative, depending on whether the adjusted cost share is greater than or equal to the corresponding factor payment share. Specifically, if factor payment shares are equal to cost shares, there is no bias to TFPG unless there is imperfect competition and the shares add to less than one. In this case the bias is positive and increases with the growth of inputs and with the degree of imperfect competition. However, if cost shares deviate from the factor payment shares and the latter sum to one, the direction of the bias in TFPG is ambiguous. Our second derivation shows the relationship between the primal and dual rate of technical change. It shows that the dual rate of technical change is exactly equal to the primal rate if the conditions of perfectly competitive long-run equilibrium are met. Otherwise, the two measures of technical change are different. The results of our calculation of the TFPG and the estimate of the rate of primal technical change show that the differences between TFPG and the primal technical change in individual industries are generally small and vary in sign. There is no clear difference between TFPG and `T`, as calculated for Singapore manufacturing, and on average the values are approximately equal. In the parametric approach, we estimate a system of factor demand equations derived from a generalized Leontief cost function with equations that reflect the market demand and market equilibrium conditions. This system of simultaneous equations enables us to generate estimates for the dual rate of technical change, the degree of economies of scale and a conjectural measure of competition. Different demand specifications are fitted in the equation system to compare the results for the corresponding industry. The main findings show that the estimation results can vary considerably from one demand specification to another, in particular this applies to the largest and fastest growing industry, the electronic products and components industry. It is perhaps difficult to make any generalizations given the estimation results are not robust. However, some industries whose results are more robust than those of other industries. It seems to illustrate that assuming a particular demand specification for all industries may lead to estimation problems and anomalies and it is well worth the time for researchers to pay more attention to the specification issue. Despite the difficulty mentioned above, a crude conclusion could be drawn: economies of scale and perfect competition appear to be prevalent in most industries with mixed results for technical change. An attempt is made to explain inter-industry technical change. Using an econometric model that employs a host of variables that describe market structure and industrial characteristics, we find some evidence that market concentration, direct foreign investment and direct exports contribute positively to technical change. However, these relationships are not statistically significant to allow us to draw any firm conclusion. Furthermore, we find a significant inverse relationship between technical change and returns to scale, reflecting the peculiar nature of Singapore development or mis-specification error in the equation system.