Back to the basics: Revisiting the development accounting methodology

The standard baseline estimate in development accounting is imprecise because of a mismatch between the estimate of physical capital and the estimate of physical capital's share, the fraction of total income accruing to physical capital. I adjust for this mismatch, and in so doing, incorporate natural capital. I also treat factor shares as variables, not constant parameters. To accommodate these adjustments, I carry out a development accounting analysis using translog multilateral indices of outputs, inputs and productivity. Results reveal that the correction for the mismatch between physical capital and its share, which is the weight assigned to the physical capital input in development accounting, reduces the variation in output per worker explained by observables by as much as 15 percentage points relative to the standard baseline. Most of this reduction is due to a decline in the explanatory power of physical capital per worker. Natural capital per worker, which is usually ignored, is found to explain up to 7.2% of the variation in cross-country output per worker. Variation in factor shares, also omitted from most studies, explains up to 6.3% of the variation in cross-country output per worker, which is nearly half as much as all observables together explain.