Productivity and misallocation.

• Author: Hsieh, Chang-Tai
• Position: Research Summaries

The starting point of a large body of recent research on economic growth is the notion that differences in aggregate total factor productivity (TFP) may not be driven solely by technology but rather in part by allocative efficiency. The key building block of this literature is the idea that firms differ, and we do not necessarily want all the resources to be allocated to one firm. For example, suppose that there are a number of firms in a country and the output [Y.sub.i] of each firm is given by a standard production function, [Y.sub.i] = [A.sub.i]F([K.sub.i][L.sub.i]), where [K.sub.i] is the firm's capital stock (equipment and structures), [L.sub.i] is the firm's labor input (skill-weighted hours worked by its employees), F is the production function which combines capital and labor, and [A.sub.i] is residual firm productivity.

If each firm produces different products, we do not want all the inputs allocated to the firm with the highest [A.sub.i], as we value having access to a variety of differentiated products. Instead, what we want is for resources to be allocated across firms to equalize the revenue productivity of the firm, or [P.sub.i][A.sub.i]. Resources are misallocated when revenue productivity differs between firms. Reallocation increases aggregate TFP and generates growth when resources flow to firms with high revenue productivity.

[FIGURE 1 OMITTED]

Micro-data from manufacturing censuses suggest substantial gaps in revenue productivity across firms within India and China. (1) The gaps are also present in U.S. data, but are much smaller. Figure 1 plots the dispersion of revenue productivity in the three countries. In India and China, revenue productivity of firms in the 90th percentile exceeds that of firms in the 10th percentile by a factor of five. In the U.S., the equivalent gap in revenue productivity is a factor of three. These gaps in revenue productivity between firms may contribute to substantial gaps in aggregate TFP. In a standard model, aggregate TFP would increase by 43 percent in the U.S. in 1997, by 115 percent in China in 1998, and by 127 percent in India in 1994 if resources were to be reallocated to equalize revenue productivity across firms.

We now have a large body of evidence on gaps in revenue productivity at the microeconomic level, largely thanks to the detailed, firm-level data available for a growing number of countries. A project spearheaded by Santiago Levy at the Inter-American Development Bank provides detailed evidence on these gaps for a large number of countries in Latin America. (2) There is similar evidence from microeconomic data for a number of countries in Europe. (3) These studies find wide gaps in revenue productivity, consistent with substantial misallocation.

The literature has largely focused on measuring the static effects of firm-level gaps in revenue productivity, but the firm-level gaps are likely to also have important dynamic effects. If more-efficient establishments face larger distortions, it undermines firms' incentives to invest in better technology. Put differently, there are two effects of resource misallocation--the static effect and the dynamic effect of resource misallocation on growth in firm productivity. This has been highlighted in several case studies. (4) Evidence from firm-level censuses in India and Mexico is also consistent with the presence of dynamic misallocation. (5) Figure 2 shows that, by the age of 40, U.S. firms grow by a factor of eight while Mexican firms...