The Payroll-Tax Puzzle: More Jobs, Flat Wages, Higher Profits
This series studies Brazil’s Plano Brasil Maior (PBM), a reform that sharply reduced employer payroll taxes in selected sectors between 2011 and 2014. Its central question is not simply whether a tax cut can create jobs. It is when the same kind of relief expands employment, when it mainly raises firms’ profits, and how local labor-market competition determines the difference. The evidence gives readers a way to judge a payroll-tax policy without confusing a lower cost for employers with an automatic gain for workers.
The puzzle is straightforward to state. In the sectors examined by the paper, employment increased by about 5 percent after the reform. The available measure of new-hire wages did not change significantly. Firms’ profits rose. Those results do not fit the quick version of the idea that a payroll tax is merely shifted between employers and workers until the market reaches a new wage. They also do not show that tax cuts are a costless route to employment. They point to a more specific question about how employers compete for labor.
A tax cut with three different outcomes
Payroll taxes are paid by employers, but their economic effects can reach several places. A firm facing a lower tax on each employee may hire more people, increase pay, reduce prices, invest, or keep the saving as profit. Which response dominates depends on the firm’s production opportunities, the supply of workers, and the alternatives available to those workers. A market with many employers competing for similar workers is not economically equivalent to one in which a few employers dominate local hiring.
The PBM reform makes that distinction visible because it did not reduce labor costs for every employer at once. Under Brazil’s standard tax regime, eligible firms were released from a flat 20 percent payroll contribution and instead paid a 1–2 percent contribution based on gross revenue, net of exports. The reform was introduced for selected sectors at different times. Firms in the separate SIMPLES regime were not directly affected, and the paper’s main sample excludes sectors whose eligibility depended on product codes that the researchers could not measure reliably at the firm level.
That design produced a selective change in relative costs. A treated firm could become cheaper to operate than some nearby competitors while those competitors remained under the old rules. The size of the reduction also differed across sectors because a revenue-based contribution affects labor-intensive and less labor-intensive businesses differently. The paper estimates an average reduction of roughly half of the original payroll-tax burden, not one identical subsidy for every treated firm.
The immediate result was more formal employment. But “more formal employment” is a measured outcome, not yet a complete account of what happened to work. Existing firms may have expanded. Firms may have entered the observed standard-regime population. Some informal jobs may have become registered. Firms may also have moved from SIMPLES into the standard regime to take advantage of the reform. These possibilities can coexist, and they carry different meanings for a policy whose stated purpose was job creation.
The wage result is equally important for interpreting the reform. The paper uses contractual hourly wages for new hires in the administrative data. Those wages show small and statistically insignificant changes. That is evidence against a large pass-through to this measured wage margin; it is not proof that every worker’s total compensation, hours, benefits, or bargaining position stayed exactly the same. The profit result is also qualified: the source reports a substantial increase using a structural approach, so the magnitude depends on the model and its assumptions even though the direction fits the broader pattern.
Together, the outcomes force a more disciplined question. If firms received lower labor costs, why did they hire more in some places without visibly raising new-hire wages, and why did the same relief produce less employment where local hiring was more concentrated? The answer cannot be read from the tax rate alone.
Why the three outcomes can coexist
A competitive labor-market story gives the reader a useful starting point, but not necessarily the final explanation. If workers can move easily among employers and firms compete strongly for them, a change in an employer-side tax can be reflected in wages. The tax saving may be competed away as firms bid for labor. In that case, employment might change little even though the distribution of the burden between firms and workers changes.
The Brazilian evidence points toward a different arrangement. Employers can have some power over wages even when there is more than one employer. Workers may face costs of moving between firms, places, or formal and informal jobs. A firm that offers a lower wage can save on labor costs, but it may also make it harder to attract workers or expand its workforce. The firm therefore faces a trade-off: it can retain some surplus through lower pay, yet it may need a larger workforce or a more competitive position to grow.
This is the intuition behind an oligopsony, a market in which several employers compete for workers but do not behave as perfect wage takers. The term matters here only because it names a relationship visible in the PBM setting: the number and relative size of employers competing for labor can shape how a common-looking policy benefit is used. It is not the claim that every Brazilian labor market is a single-employer monopoly, nor that concentration by itself measures every aspect of worker power.
A selective tax cut changes relative labor costs. In a less concentrated local market, a treated firm has more nearby competitors that can attract the same workers. It may have less room to turn the tax saving into a lower relative wage without losing labor. Expansion in employment becomes a more important response. In a more concentrated market, the dominant employers may be able to retain more of the saving because workers have fewer comparable alternatives. The tax cut can then increase profits without producing the same hiring response.
This mechanism explains why employment and profits can rise together without requiring a large measured wage increase. It also explains why the policy’s average effect can conceal sharply different local outcomes. A national statistic may say that employment rose five percent, while the more useful policy question is where that increase occurred and where the fiscal relief was absorbed without much additional hiring.
The contrast also keeps the evidence from being overinterpreted. The paper’s model predicts that a labor subsidy can raise both employment and wages in general, with the employment effect larger for most plausible parameter values. The observed wage estimates are small and imprecise, so the data are more informative about the employment gradient than about a precise wage gradient. The model rationalizes the joint pattern; it does not prove that every other explanation, including wage-setting frictions, is absent.
Nor is a payroll-tax cut necessarily the same policy when it applies to every employer in a market. If all firms face the same change, their relative labor costs may not shift in the same way. The incidence could then look different, with wages absorbing more of the change. The PBM’s selective eligibility is therefore not an administrative detail at the edge of the story. It is part of the mechanism that makes local competition relevant.
The question the evidence must answer
The source is an IZA discussion paper, explicitly presented as preliminary research, so its findings should be read as a carefully developed case rather than a universal formula for job creation. Its main evidence comes from RAIS, a matched employer–employee administrative census of formal employment. The researchers use admission and dismissal dates to reconstruct monthly sector outcomes from 2009 through 2014, allowing them to compare sectors that entered the PBM at different times with sectors that were never treated in the main sample.
That comparison matters because a raw before-and-after change could simply reflect a sector’s existing trajectory. The event-study estimates show no differential employment trend before implementation, then an increase after the tax change. The response is stronger in sectors expected to receive a larger reduction. These features make the labor-cost explanation more credible, while not turning the policy into a randomized experiment. Treatment timing correlates with some initial sector characteristics, the sample covers selected sectors rather than the whole economy, and later policy reversals and Brazil’s severe downturn are outside the main period.
The rest of the evidence must therefore keep several distinctions in view. The 5 percent figure is a sector-level formal-employment estimate, not a claim that every treated firm hired five percent more people. The wage result concerns new hires’ contractual hourly wages, not every component of worker welfare. The employment increase includes firm growth and observed entry, but the data cannot cleanly separate new firms from formalization at the entry margin. A household-survey exercise suggests that formalization may account for a meaningful share of the increase, but its confidence interval is consistent with a much wider range of possibilities.
The central relationship remains local competition. The paper measures concentration across 558 Brazilian micro-regions using a wage-bill Herfindahl–Hirschman Index. Employment responds mainly in less concentrated markets; it does not appear to respond in the more concentrated group. This pattern is the bridge between the reform’s institutional design and its policy meaning. Before deciding whether a payroll-tax cut “worked,” one must ask whether the local employers receiving the relief actually compete hard enough for workers to turn the saving into jobs.
That is why the next step is not to debate tax cuts in the abstract. It is to examine exactly how the PBM changed employers’ obligations, how the comparison was built, and which firms were inside or outside the direct treatment. Only then can the employment, wage, firm, and profit results be read at the level of precision they deserve.