Methodological Rift Over California's $20 Fast-Food Wage: Job Losses vs. Null Employment Effects

Methodological Rift Over California's $20 Fast-Food Wage: Job Losses vs. Null Employment Effects

The academic and policy debate over the employment impacts of California’s $20 fast-food minimum wage (AB 1228, effective April 1, 2024) is anchored in a deep methodological rift. Three landmark studies published in early 2026 expose a critical distinction between studies that directly measure employment using administrative or granular location data, and those that infer job losses by modeling consumer price elasticities.

Direct Administrative Measurement vs. Conventional Differencing

A central challenge in minimum wage literature is controlling for pre-treatment macroeconomic trends. In NBER Working Paper 35171 (May 2026), Arindrajit Dube uses official administrative Quarterly Census of Employment and Wages (QCEW) data through 2025Q3 to show how choice of econometric methodology alters employment conclusions. A conventional Difference-in-Differences (DiD) model yields a negative employment own-wage elasticity (OWE) of -0.19 (implying modest job losses). However, because California’s labor market was on a unique trajectory prior to the wage hike, conventional DiD fails to establish a proper parallel trend. When employing Synthetic Difference-in-Differences (SDID)—which reweights control areas to match California's pre-treatment path—the OWE shrinks to -0.04, which is statistically and economically indistinguishable from zero.

Granular Mobile Device Location Tracking

Corroborating the null-employment findings, Michael Reich and Denis Sosinskiy (UC Berkeley Center on Wage and Employment Dynamics, March 2026) introduce a highly innovative dataset: mobile device location data tracking actual worker presence at individual fast-food establishments across the United States. This study, covering two full years of data, find that the policy increased average weekly wages for covered fast-food workers by 11% without reducing employment. By observing worker presence at the establishment level, they can distinguish policy-specific effects from broader industry-wide shifts.

The Monopsonistic Quit-Reduction Channel

One of the most significant theoretical developments in the 2026 literature is the identification of a clear mechanism explaining how restaurants absorb higher wages without cutting headcounts. Dube’s analysis of Quarterly Workforce Indicators (QWI) data reveals a massive drop in employee separation rates, yielding an own-wage elasticity of -1.7 to -4.2. This indicates that a $20 wage floor significantly reduces costly employee turnover (quits), allowing restaurants to capture substantial operational savings that directly offset the increased cost of labor.¹ This provides strong empirical support for a monopsonistic labor market model, which is further analyzed in Fast-Food Price Hikes and the 50-100% Cost Pass-Through: Who Pays for the $20 Wage?.

Inferred Job Losses via Price-Demand Modeling

In contrast to direct employment measurements, Jeffrey Clemens, Olivia Edwards, Jonathan Meer, and Joshua D. Nguyen published NBER Working Paper 34990 (2026), focusing on price pass-through. They find that California fast-food prices rose by 4.9% to 5.1% relative to other MSAs. Crucially, Clemens et al. do not directly measure employment in this paper. Instead, they use a consumer demand model to infer that these price increases must have reduced fast-food demand by 3.9% to 4.1%, asserting that these quantity reductions "align with employment reductions estimated in prior research."

This reveals the core of the methodological weaponization: studies claiming substantial job losses often rely on theoretical price-elasticity inferences or uncorrected DiD models, whereas studies directly tracking administrative payrolls (QCEW) or establishment-level physical presence (mobile data) find that employment levels remained stable.