Introduction

Firms differ significantly in both pay and non-wage dimensions. While a substantial literature exists regarding firm-level wage heterogeneity and its implications for inequality, less research has historically focussed on how non-wage job characteristics vary across firms and how workers value them. Identifying the value of firm-provided amenities is empirically challenging because standard regressions often conflate preferences with sorting and unobserved firm characteristics. This paper addresses these challenges by using discrete choice experiments (DCEs) embedded in a large-scale survey of German workers to estimate firm-specific amenity valuations and examine their role in job choice, sorting, and inequality.

The researchers ask workers to rank hypothetical job offers from real firms that the workers themselves identify as likely application targets. By randomising wage offers across these firms, the study recovers money-metric valuations of firm amenities that are not confounded by equilibrium wage differences or assumptions about offer arrival rates. The survey is linked to administrative Social Security records and firm-level data from Kununu, a major employer review platform. This linkage allows for a detailed study of how amenity valuations relate to worker mobility, firm characteristics, and between-firm inequality.

Empirical setting and data

The analysis primarily utilises two sources: the CHH Worker Survey and the Integrated Employment Biographies (IEB) administrative records. The survey was fielded to full-time German workers aged between 25 and 50. The data includes over 2,000 workers who provided names of specific outside firms they would consider joining. These responses were standardised and matched to establishment identifiers in the German Social Security records. The IEB data provide complete employment histories, including wages and demographics, for the period between 2010 and 2019.

To complement these records, the researchers linked the survey data to employer reviews from Kununu. This platform allows current and former employees to evaluate workplaces across dimensions such as compensation, work-life balance, and organisational culture. The study by default weights firms by employment to ensure results reflect the experience of a typical worker. The resulting sample represents firms spanning all parts of the wage distribution and all major sectors.

Valuing firms’ amenities

Workers derive utility from both wages and non-wage amenities. The study models this utility as a function of log wages, amenity values in utility units, and idiosyncratic preferences. Amenity valuations are recovered through a two-step process. First, the researchers compute the share of workers who rank one firm over another to identify amenity values in utility units relative to a median reference firm. Second, these values are converted to money-metric units by dividing by the estimated marginal utility of log wages. This structural preference parameter is estimated from choices over randomised wage differences.

Discrete choice experiments are take-it-or-leave-it offers, allowing for estimation without assumptions regarding search frictions or endogenous wage-setting. The researchers validated the eigenvector approach used against a rank-ordered logit model, finding a correlation of 0.95. This confirms that the approach accurately recovers firm-level amenity values. The ex-ante value captured represents the total bundle of non-wage factors, including working conditions, career prospects, and job security.

Firm wages, amenities, and valuations

The study finds that amenity values are highly dispersed across firms. The spread between the highest- and lowest-valued firms is approximately one-third of a worker’s pay. Preferences over non-wage attributes significantly shape job choice; on average, 35% of workers selected the lower-wage offer in the discrete choice experiment. This implies meaningful trade-offs between pay and amenities even among firms workers consider as realistic application targets.

Between-group heterogeneity in valuations is substantial. The Pearson correlation in amenity valuations between men and women is only 0.266, and the correlation between college-educated and non-college workers is 0.231. These low correlations suggest that different demographic groups value firm attributes differently. However, a common amenity index remains a strong predictor of aggregate patterns. Conditional on wages, firms with higher amenity values are larger and receive more favourable reviews from employees on the Kununu platform.

Amenities and inequality

A key finding is that amenity values do not decline with firm wage premia. Across firms, amenity valuations are approximately orthogonal to employment-weighted firm wage premia and positively correlated when firms are weighted equally. This suggests that non-wage factors compound, rather than offset, inequality in firm wage premia. Between-firm inequality in total compensation is estimated to be roughly 20 percent larger in variance terms than in pay alone. Specifically, the variance of total compensation is 1.22 times the variance of pay premia when weighting firms equally and 1.20 times when weighting by employment.

Gender differences in amenity valuations also explain part of why women sort into lower-wage firms. While women work at firms that offer lower pay premia, they sort toward firms they value relatively more in terms of amenities. Female workers make steeper wage–amenity trade-offs when switching firms, with a fitted slope of -0.164 compared to -0.049 for men. Accounting for sex-specific valuations roughly offsets the gender wage gap in total firm-provided value under certain normalisations.

Conclusion

The research provides new evidence that non-wage attributes are critical to job choice and that amenity valuations are highly dispersed. Because high-wage firms do not systematically offer worse amenities, accounting for these non-wage factors widens the measured extent of between-firm inequality. In money-metric units, the employment-weighted signal variance of amenities is approximately one-third that of wage premia. These results suggest that standard models of the labour market should account for systematic firm-level variation in utility beyond simple monetary compensation.