Salary Matching and Pay Cut Reduction for Job Seekers with Loss Aversion

TL;DR

This paper models how loss aversion influences wage offers and acceptance, showing that employers strategically reduce pay cuts and that loss aversion significantly impacts salary distribution and policy effects.

Contribution

It introduces a behavioral search model incorporating loss aversion, empirically tests its predictions with Korean data, and demonstrates its importance for wage policy analysis.

Findings

Excess zero wage growth mass is 8.5 times larger with loss aversion.

Density immediately above zero wage growth is 8.8% higher than below.

Loss aversion improves model fit and increases the marginal value of additional pay.

Abstract

This paper examines how loss aversion affects wages offered by employers and accepted by job seekers. I introduce a behavioral search model with monopsonistic firms making wage offers to job seekers who experience steeper disutility from pay cuts than utility from equivalent pay raises. Employers strategically reduce pay cuts to avoid offer rejections, and they exactly match offers to current salaries due to corner solutions. Loss aversion makes three predictions on the distribution of salary growth for job switchers, which I empirically test and confirm with administrative data in Korea. First, excess mass at zero wage growth is 8.5 times larger than what is expected without loss aversion. Second, the density immediately above zero is 8.8% larger than the density immediately below it. Third, the slope of the density below zero is 6.5 times steeper than the slope above it. When estimating model parameters with minimum distance on salary growth bins, incorporating loss aversion substantially improves model fit, and the marginal value of additional pay is 12% higher for pay cuts than pay raises in the primary specification. For a hypothetical hiring subsidy that raises the value of labor to employers by half of a standard deviation, incorporating loss aversion lowers its pass-through to wages by 18% (relative to a standard model) due to higher elasticity for pay cuts and salary matches that constrain subsidized wage offers. Somewhat surprisingly, salary history bans do not mitigate these effects as long as employers can imperfectly observe current salaries with noise.