Payroll Tax Cuts

Lobel (2024) investigates the implications of large payroll tax cuts in the context of Brazil, which implemented a payroll tax reform in 2012. Due to arbitrary sector-specific legal requirements, tax rates were reduced by 20 percentage points for a small subset of firms. The political process that determined eligibility often assigned remarkably similar sectors to different eligibility statuses, as illustrated by the cases of hotels and motels. Eligible and ineligible firms are not only similar in levels but, most importantly, in pre-reform trends. This resemblance between groups allows the paper to implement in a difference-in-differences specification. The paper uses novel, anonymized administrative tax microdata, which enables the tracking of firms and workers over time, both before and after the reform. The paper estimates that consumers bear 65% of the incidence of payroll taxes, with firm owners bearing 23% and workers bearing 12%.

The paper incorporates these results into an estimate of the MVPF for payroll tax cuts.

MVPF SUMMARY

Lobel (2024) calculates the MVPF of payroll tax cuts as

where dB is the estimated worker incidence (0.12), d\pi is the estimated firm owner incidence (0.23), and dB is the estimated consumer incidence (0.65).

dT is the net fiscal cost to the government, scaled by the welfare variation (the policy-induced change in the benefits to consumers, firm owners, and workers). The net fiscal cost to the government is:

where \Delta \tau is the change in the payroll tax (-0.2), \tau_0 is the payroll tax rate in the pre-reform period (0.3), \beta_L is the estimated effect of the Brazilian tax reform on employment (0.12), and , \epsilon is the estimated labor supply elasticity (4.15).

Putting these pieces together yields a net fiscal cost of \left(-0.2 + 0.3 \cdot 0.12 \cdot \frac{4.15 + 1}{4.15}\right)=-0.15.

To put the net cost on the same scale as the worker, firm owner, and consumer incidence, Lobel (2024) Scales this value by the welfare variation to generate the governmental incidence: dT = -0.6. This yields: