Sources of productivity growth: Taiwan, 1966-1991

This research characterizes sectoral productivity growth of the Taiwanese economy for the period of 1966-1991. We adopt an expanded output concept, which includes labor, capital, and intermediate inputs as productive inputs, to calculate the Total Factor Productivity (TFP) growth. To construct quantity indices for the gross output, and for the labor, capital, and intermediate inputs, we employ modified fixed-base index methods. Within this framework, we can examine domestic interactions and international linkages which affect sectoral TFP growth. First, we apply cross-sectional analyses to sectoral TFP growth to identify TFP-influencing factors. This methodology allows us to analyze TFP growth in a comprehensive framework which is based on multiple characteristics of the production structure. Some interesting results from this approach include the following: positive but decreasing effect from the capital-labor ratio, conflicting effects from the domestic production capability, positive and sustaining effects from the imported intermediate inputs, and an insignificant effect from the exports. Second, we model the producer behavior by estimating the translog production function for each sector within an approximation context. Our procedures transform the constrained maximum-likelihood estimation into an unconstrained non-linear estimation so that the monotonicity and the concavity implied by the producer theory can be followed. These econometric procedures allow us to test hypotheses regarding the production and productivity growth: constancy of second-order characteristics, Hick\u27s neutrality of the productivity growth, and weak and strong separability of intermediate inputs. Also, we can examine the biases of productivity growth and substitution pattern among inputs of sectoral production processes. As our research adopts a methodology different from all preceding analyses of the Taiwanese miracle , we are able to generate results which are new to the existing literatures on the productivity growth of developing countries. Through the exact correspondence between the Tornqvist-Theil index and the translog function, we can integrate time-series and cross-sectional analyses which characterize the same economic process with different perspectives and assumptions