Misallocation and Manufacturing TFP in China and India

Large cross-country differences in output per worker are largely attributed to TFP, and most explanations locate the inefficiency within the representative firm. Following Restuccia and Rogerson (2008), this paper instead asks how much aggregate TFP is lost to misallocation across firms – the case where two firms with identical technologies face different effective input prices, so one has a low marginal product of capital and the other a high one, and output would rise if capital moved between them. The measurement device is a standard monopolistic-competition model with heterogeneous firms (Melitz (2003) without trade), in which the key result is that a plant’s revenue productivity should be equalised within an industry when there are no distortions: inputs flow to physically more productive plants until their extra output pushes their price down enough to equalise revenue per unit of input. Dispersion in revenue productivity therefore measures the wedges, and high revenue productivity marks a plant “held back” below its efficient size. Applying this to plant- and firm-level censuses – India’s Annual Survey of Industries for 1987-1994, the Chinese Annual Surveys of Industrial Production for 1998-2005, and the US Census of Manufactures for 1977, 1982, 1987, 1992 and 1997 – within four-digit industries, the dispersion of revenue productivity is much wider in China and India: the 90th-to-10th-percentile ratio in the latest year is 5.0 in India and 4.9 in China against 3.3 in the United States. Fully equalising revenue productivity within industries would raise manufacturing TFP by 86-115 percent in China and 100-128 percent in India – but also by 31-43 percent in the United States, which is why the headline counterfactual is the more modest move to US dispersion: 30-50 percent for China and 40-60 percent for India. Because average capital shares are about half in both countries, gains would be roughly squared if capital accumulated in response, so a 30 percent TFP gain in China implies a 67 percent long-run output gain and a 59 percent gain in India implies 153 percent. Over time, Chinese allocative efficiency improved about 2 percent a year from 1998 to 2005 – perhaps a third of measured Chinese industrial TFP growth – while India’s deteriorated about 1.8 percent a year from 1987 to 1994, which the authors call surprising given reforms begun in the late 1980s. The scope conditions are declared rather than buried: the calculation “heroically makes no allowance for measurement error or model misspecification,” the elasticity of substitution is set conservatively at 3 and the results are highly sensitive to it, and an extended set of checks on measurement error is reported as “inconclusive.”

O53 | Macro Paper Warehouse

Misallocation and Manufacturing TFP in China and India