Why Competition Determines Where the Jobs Appear

The PBM reform changed the cost of labor, but its employment effect was not evenly distributed across Brazil’s local labor markets. This chapter explains how the paper measures that difference and why a lower payroll-tax burden produced more hiring in less concentrated markets while doing less for employment in more concentrated ones. The result gives the employment–wage–profit pattern a mechanism: the relevant question is not only how much relief a firm receives, but how many credible employers compete for the same workers.

Concentration is a market condition, not a firm label

The paper measures local labor-market concentration with a wage-bill Herfindahl–Hirschman Index, or HHI, across 558 Brazilian micro-regions. A micro-region groups neighboring municipalities with economic ties and a shared structure of production. The wage-bill measure asks how much of the local labor market’s total wage bill is associated with each employer, then summarizes whether that total is spread across many employers or dominated by a few. The index is higher when a small number of firms account for larger shares.

In the paper’s sample, the HHI ranges from 0.003 to 0.433, with a median of 0.029. Those numbers are not universal thresholds for a competitive or uncompetitive labor market. They describe the distribution of the proxy in the Brazilian micro-regions used for the analysis. The authors divide markets above and below the sample median to make the contrast legible, then use a more granular exercise with the index as a continuous interaction. The median split is a way to compare settings, not a policy rule that says a market becomes concentrated at exactly 0.029.

The contrast is sharp. Firms in less concentrated labor markets account for virtually all of the estimated employment response to the PBM reform. Employment in more concentrated markets does not appear to respond to the tax change in the same way. Wage effects remain small and statistically insignificant in both groups. The result is therefore not simply that concentrated markets produce a different wage adjustment. In the estimates, they show much less of the employment adjustment that appears in lower-concentration markets.

A continuous-HHI analysis points in the same direction. Moving from a less concentrated to a more concentrated micro-region is associated with a smaller post-implementation employment response, while the pre-treatment interactions are flat. The estimates using a Poisson approach are qualitatively similar but noisier because a more disaggregated sector–micro-region panel contains many cells with zero employment. That loss of precision matters: the granular exercise is a robustness check that supports the pattern, not a reason to claim an exact response for every locality.

The concentration result is also not a disguised firm-size result. The paper reports broadly similar employment effects for small and large firms within sectors. It finds similar responses across workers with and without a high-school degree and across regions with high and low levels of informality. Those comparisons suggest that the central division is the organization of the local hiring market rather than a simple distinction between large firms, skilled workers, or formal regions.

That conclusion still needs to be kept at the level the evidence supports.