Minimum Wage Effects on Employment

Pioneered the use of natural experiments to study minimum wage effects, comparing fast-food employment in NJ (treatment) and PA (control) after NJ's 1992 minimum wage increase.

No significant negative employment effect; point estimate of +0.59 FTEs per restaurant (SE 1.20), contradicting the competitive labor market prediction.

Foundational paper for the modern minimum wage literature; our identification strategy builds directly on their cross-border comparison design.

Generalized the border-discontinuity approach to all contiguous US county pairs straddling state lines, addressing spatial heterogeneity bias in national-level studies.

Employment elasticity of -0.01 to 0.02 (not statistically distinguishable from zero) for restaurant workers, with earnings elasticity of 0.20-0.25.

Our empirical strategy follows their contiguous county-pair design; we extend it with staggered treatment timing methods.

Introduced a bunching estimator that examines the entire wage distribution around the minimum wage, avoiding reliance on a single comparison group.

Missing jobs below the new minimum are offset by excess jobs above it; net employment effect near zero with an elasticity of -0.036 (SE 0.042).

Their distributional approach complements our design; we use their bunching estimates as a benchmark for our ATT estimates.

Exploited a large (60%) minimum wage increase in Hungary to study firm-level adjustment margins, including employment, prices, profitability, and capital-labor substitution.

Employment fell by 10% at affected firms, but most adjustment came through reduced profits (-13%) and higher prices (+3%), not layoffs.

Informs our heterogeneity analysis by firm size and profit margin; different context but same causal question.

Studied Germany's 2015 national minimum wage introduction using administrative employer-employee data, showing reallocation from small to large firms.

Minimum wage reduced employment at small firms by 2.3% while increasing earnings by 4.8%; workers reallocated to higher-paying, more productive establishments.

Uses the same DiD framework in a different institutional context; their reallocation findings motivate our firm-size heterogeneity analysis.

Used administrative data from Washington State to track individual workers before and after Seattle's minimum wage increase, separating incumbent from new worker effects.

Hours worked by low-wage workers fell 6-7%, partially offsetting hourly wage gains; experienced workers saw net earnings increases while new entrants faced reduced opportunities.

Their individual-level tracking approach motivates our robustness check using worker-level (rather than county-level) outcomes.

Challenged the contiguous county-pair design by arguing that spatial controls absorb real variation, and that traditional panel methods with state and time fixed effects recover negative employment effects.

Employment elasticities of -0.15 to -0.20 for teens and -0.10 to -0.15 for restaurant workers when using longer panels and state-level variation.

Primary counterpoint to our identification strategy; we address their critique by implementing the Callaway-Sant'Anna estimator that handles staggered timing transparently.

Developed and estimated a structural model of restaurant industry dynamics with putty-clay technology, where minimum wage effects depend on the vintage of capital in place.

Short-run employment elasticity of -0.04 but long-run elasticity of -0.18, as firms gradually exit and adjust technology; welfare effects depend on discount rates and adjustment costs.

Their structural estimates bracket the reduced-form effects we identify; we compare our ATT to their model-implied treatment effects.

Provided the modern theoretical framework for labor market monopsony, showing that search frictions generate employer wage-setting power even in markets with many employers.

When labor supply to individual firms is imperfectly elastic, moderate minimum wage increases can raise both wages and employment, rationalizing the Card-Krueger findings.

Core theoretical motivation for our paper; the monopsony model predicts the near-zero employment effects we find empirically.

Developed an equilibrium search model generating a non-degenerate wage distribution even among homogeneous workers, where larger firms pay higher wages to reduce turnover.

The model predicts that minimum wages compress the lower tail of the wage distribution and can reduce frictional unemployment by raising the opportunity cost of non-employment.

Provides the structural search-theoretic underpinning for the monopsony interpretation of our reduced-form results.

Developed group-time average treatment effects ATT(g,t) for staggered DiD designs that are robust to treatment effect heterogeneity, unlike TWFE estimators.

TWFE can be severely biased with heterogeneous treatment effects; their estimator recovers interpretable causal parameters under parallel trends conditional on covariates.

Our primary estimation method; we implement their R package `did` for all main specifications and event study plots.

Comprehensive review of recent advances in DiD methodology, covering staggered adoption, pre-testing, sensitivity analysis, and alternative identification strategies.

Recommend the Rambachan-Roth sensitivity analysis for assessing robustness of parallel trends assumptions; document widespread misuse of pre-trend tests as validation.

We follow their recommended best practices: Callaway-Sant'Anna for estimation, Rambachan-Roth for sensitivity, and honest confidence intervals for pre-trends.