2 minute read

Annex 2A. Estimating Productivity, Marginal Cost, and Markups

Next Article
Concluding Remarks

Concluding Remarks

housing, while clarifying property rights, land markets, and zoning, and addressing local environmental externalities. While it may be appealing to consider place-based interventions that directly improve transportation and housing supply, care must be taken to address the underlying land markets, zoning, and property rights, and enact a complementary set of policies to make such interventions more valuable to local businesses and local residents. Chapter 7 provides an assessment of transportation and housing programs using the framework outlined in the overview and expanded upon in chapter 6.

Second, the productivity benefits will only arise with the necessary transformation of the economy. Hence, a broader reform agenda is necessary to make that happen. This relates as much to cultivating shares of value chains in manufacturing as in promoting higher-end services and, for the least developed countries, including transformation within agriculture. Chapter 5 discusses these steps in the context of lagging regions. Chapter 7 focuses on special economic zones. But in the long term, the agenda must be national.

Annex 2A. Estimating Productivity, Marginal Cost, and Markups

To estimate agglomeration elasticities, Grover and Maloney (2021) use a standard specification:

yit = γ denl(i)t + θ Xit + FEjt + εit (2A.1)

where yit represents the set of firm-level productivity measure: wages, wit; revenue total factor productivity TFPR, tfprit; and physical (or quantity) total factor productivity TFPQ, tfpqit or other related outcomes of interest: price, pit; marginal cost, mcit; and markup, μit The population density in location (l) is represented by denl(i)t, where firm (i) is located at time (t). Xit includes firm-level controls, such as firm age and size (measured by the number of workers lit, included in wage and price regressions only). All variables, except firm age, are in natural logarithm. FEjt are pair-wise four-digitindustry-year fixed effects to help factor in any sectoral or time trend. Base estimations apply an ordinary least squares technique, and standard errors are clustered at the level of the spatial unit.

Firm-level average wages (total wages/number of workers) and prices are directly observed in the data. The firm-level price index, Pit, is the aggregated index computed using a weighted average of a firm’s product-level price data.15 TFPRit is the Hicksneutral measure of firms’ efficiency, commonly used in the literature to capture the residual output variation after controlling for tangible inputs of production. TFPRit is unobserved to the econometrician but known to the firm when making its periodic input decisions. This induces a correlation between productivity and inputs resulting in the well-known “simultaneity” or “transmission bias” (Marschak and Andrews 1944). To solve for this endogeneity, the production function for each two-digit industry is estimated following the structural approach in Gandhi, Navarro, and Rivers (2020).16

This article is from: