When a Payroll-Tax Cut Creates Jobs—and When It Mainly Creates Profits

The Brazilian PBM reform gives a conditional answer to the question in this title. A large, selective reduction in employer payroll taxes increased formal employment, but the response was concentrated in less-concentrated labor markets. Wages did not show a significant response in the available new-hire measure, firms’ profits rose, and the fiscal cost was substantial. The final judgment is therefore not that payroll-tax cuts work or fail in general. It is that their employment value depends on who is exposed, how employers compete for workers, what “employment” includes, and what the policy is meant to achieve.

The causal chain is conditional from the start

The PBM did not simply remove a tax from every Brazilian employer. It replaced a 20 percent payroll contribution with a 1–2 percent gross-revenue contribution for eligible standard-regime firms in selected sectors, while firms in SIMPLES were not directly affected. The reform reached sectors at different times between 2011 and 2014, and the effective burden reduction varied with each sector’s labor intensity. On average, the reduction was about 51 percent of the former burden, but the sector-level change ranged from a small increase to an 84 percent reduction.

That design matters because it changed relative labor costs. A treated firm could face a lower cost per employee than some employers competing for the same workers. The policy therefore created the conditions for a labor-demand response, but it did not dictate the margin through which firms would adjust. They could expand an existing workforce, enter or become visible in the standard regime, formalize workers, change tax status, or retain some of the saving as profit.

The event-study evidence supports a meaningful employment response. Formal sector employment begins to rise after implementation, intensifies over subsequent quarters, and stabilizes at about 5 percent above the comparison path. Sectors receiving larger expected burden reductions respond more strongly: the paper reports roughly 7.3 percent employment growth above the median intensity and 2.2 percent below it. The pre-treatment path is flat, and alternative estimators designed for staggered implementation produce results consistent with the main estimates. Those features make the labor-cost explanation credible without turning the PBM into a randomized experiment.

The wage result changes the interpretation of that response. Contractual hourly wages for new hires show small and statistically insignificant changes. The evidence therefore points to quantity adjustment rather than a large measured price adjustment: firms used more formal labor without visibly bidding up the recorded hourly wage of new entrants. That is a statement about the observed wage margin, not proof that no incumbent pay, benefits, hours, or job quality changed.

The firm evidence fills in part of the chain. Existing firms’ average employment rises by about 3 percent, while net entry rises by about 2 percent. The paper summarizes the sectoral response as roughly 60 percent from growth of existing firms and 40 percent from entry. But the entry label has an administrative boundary. Firms can move from SIMPLES into the standard regime, and the share making that move rises by about 0.7–0.9 percentage points in treated sectors. When all tax regimes are included, the employment effect falls to about 3.6 percent. The directly treated standard-regime estimate and the broader all-regime estimate are not rivals from which one must choose a single “real” number; they answer different questions about exposure and selection.

Formalization adds another layer. RAIS records formal work, so its increase cannot distinguish a newly created formal position from an informal position that became registered.