An essential issue in economics is understanding why some countries are rich and others poor. A consensus view has emerged in the literature whereby productivity is at the core of the differences in income across nations. Over the last fifteen years, progress has been made in our understanding of cross-country income differences in part by the increasing recognition of the importance of production heterogeneity—firms—and the allocation of factors of production across them for aggregate outcomes. The progress has been enhanced by the much wider availability of microeconomic data sets of firms across a growing number of countries. There has been a productive interplay of macro- and microeconomic approaches to development. Two papers in this issue of Economía reflect very well the synergy that is growing across subfields, and both papers provide valuable insights for the overall role of firms and productivity on development. Although both papers offer insights and implications that are broad across many fields in economics, I focus my discussion in the context of the macroeconomic development literature and, in particular, the connection of misallocation and aggregate productivity.
In “Firm Dynamics and Productivity: TFPQ, TFPR, and Demand-Side Factors,” John Haltiwanger discusses relevant measurement issues surrounding estimates of firm-level productivity and the implied inference of distortions, misallocation, and aggregate productivity derived from a variety of microeconomic data sets of firms.
What is typically meant by misallocation? The concept of misallocation is tightly related to a particular economic structure. To focus the discussion, consider a simple one-period economy where a single homogeneous good is produced by heterogeneous establishments that differ only in their productivity. That is, more productive establishments produce more output for a given set of inputs. Assume for illustration that there are decreasing returns to scale in inputs at the establishment level, although this assumption is not essential. Incidentally, the aggregate production function implied by this setting features [End Page 51] constant returns to scale with the number of firms as an input, consistent with standard practice in macroeconomics. In this simple context, the efficient allocation of inputs across establishments is one that maximizes total output given an aggregate set of inputs. The efficient allocation involves assigning more inputs to more productive establishments and the same amount of inputs to equally productive establishments, so that all establishments feature the same physical marginal product of factors. More productive establishments are larger than less productive ones precisely to achieve equalization of marginal products. Misallocation occurs when an allocation deviates from the efficient benchmark allocation, although it may well be that the efficient allocation can never be achieved in practice. Moreover, the aggregate cost of misallocation can be expressed as the ratio of actual aggregate output to efficient output as determined by the efficient allocation. With this setup in mind, it is easy to see how market and centralized economies can generate allocations that differ from the first best. Examples include any tax or subsidy to a subset of establishments, even if well intended to promote productive enterprises or sectors; any allocation of inputs not guided by productivity, regardless of whether this is done via price distortions or centralized arrangements with limitations to reallocation; and any market friction that prevents the allocation of resources to the best uses. It is also easy to see that both the measurement of productivity at the microeconomic level, in this case the establishment, and the measure of misallocation critically depend on the specific structure/abstraction, such as one good, one sector, no dynamics. As a result, any relevant misspecification will translate into a biased picture of productivity and misallocation. The abstraction focuses on a subset of potential sources of misallocation, but in practice misallocation can occur in many other settings, such as across producers in different sectors, across occupations, across space, across ethnic or racial groups, or across time; and the importance of these sources can differ across time and space.
As Haltiwanger points out, in most productivity studies, microeconomic data provide the value of outputs (or more generally revenue) for a given establishment or firm. Hence, the issue discussed at length is how to recover physical measures of productivity at the firm level and...