Secondary products and the measurement of productivity growth

The paper considers alternative treatments of secondary products in input-output systems and analyzes their implications for the measurement of productivity growth at both the sectoral and overall level. Two standard models of secondary products are used: (1) the commodity technology model and (2) the industry technology model. It is argued that the first model correctly relates sectoral and overall levels of productivity growth; the second model, though more conventional, aggregates sectoral levels to a biased estimate of overall productivity growth. Estimates of the two measures are provided using U.S. 85-sector input-output data for 1967, 1972, and 1977. The empirical results indicate that the alternative assumptions do not lead to significantly different estimates of commodity-level and industry-level productivity growth over this period for the full economy but do for several sectors. Moreover, changes in secondary production did not contribute significantly to the decline in productivity growth over this period but secondary production was found to have a much lower rate of productivity growth than primary production.