From employer wage-setting power to the consequences of pay transparency and the role of benefits in compensation inequality, many core questions about labour markets hinge on how job seekers respond to wage and benefit differences across job ads. Combining a field experiment on over 100 Swiss job boards with a survey, this column shows that workers are imperfectly informed about wage and benefit differences across jobs. Posting wages redirects search towards better-paying jobs. Workers also value certain benefits highly – notably company cars and the option to work from home. Since higher-paying jobs tend to offer more benefits, inequality in job value exceeds wage inequality across jobs.

Pay-transparency reforms in Europe and the US are changing what workers can learn about prospective employers (Cullen 2024). Whether these reforms improve workers’ options depends on how job seekers use information about pay and benefits. A central question is therefore how strongly job seekers respond to wage and benefit differences across job ads. This response depends on three channels: how much they value wages and benefits, how much they already know about wage and benefit differences across jobs, and how much they update their beliefs when new information is disclosed. Distinguishing these channels matters for understanding employer wage-setting power, workers’ willingness to pay for non-wage benefits, and inequality in total compensation.

Two complementary experiments

To answer these questions, in Beerli et al. (2026) we ran a pre-registered (AEARCTR-0011023) field experiment on the 112 job platforms of Jobchannel, one of the largest private job-board networks in Switzerland, covering roughly 20% of web traffic in the online vacancy market. For three months in 2023, around 265,000 users were randomly assigned to a control group or to treatment arms in which job ads displayed wage and benefit information sourced from kununu, the market-leading employer-review platform in the German-speaking region. During the experiment, job seekers made over 12 million consequential decisions over nearly 200,000 real vacancies. We track whether users open an ad and whether they take further actions – such as clicking “apply now”, navigating to the employer’s posting, or saving the ad. To our knowledge, this is the first experiment to provide revealed-preference estimates of how both posted wages and job benefits affect job seekers’ interest in real vacancies across a broad range of occupations.

Behavioural responses alone, however, cannot reveal how much job seekers value wages and benefits. Even those who place a high value on pay may respond little to posted wages if they are already aware of pay differences across jobs or discount the information provided entirely. We therefore complement the field experiment with a belief-elicitation survey of roughly 1,500 job seekers recruited from the same platforms. Respondents saw eight real job ads matched to their usual search occupation and region, with or without the same wage and benefit information as in the field experiment. They then reported their expected pay and benefits for each job. The control group’s answers reveal how well-informed job seekers are. The comparison with treated respondents reveals how much they update their beliefs.

Posting wages redirects job search

Consistent with earlier experiments (e.g. Belot et al. 2022; see also Kircher et al. 2018), we find that treated job seekers direct more attention to ads offering higher wages. Figure 1 reports our estimates: in our experiment, a 1% increase in the posted wage raises the likelihood of engaging with a job ad by about 0.4%. Extrapolating this elasticity implies that displaying wage information on all vacancies would shift ad views towards vacancies in the top quintile of the wage distribution by approximately 30% relative to those in the bottom quintile. Importantly, the effects are nearly identical for consequential actions such as clicking “apply now”, printing, and saving the ad, suggesting that the response goes beyond curiosity clicks.

Figure 1 Posted wage elasticity of ad views  

Note: The figure reports point estimates and 95% confidence intervals of the posted-wage elasticity of ad views, for all occupations and separately for occupations with low (below 35%), medium (35–70%) and high (above 70%) shares of female workers.

The response is unevenly distributed. It is higher in occupations with a high share of female workers (0.61 where more than 70% of workers are women, compared with 0.33 where fewer than 35% are; Figure 1). The elasticity is also roughly twice as large in the lowest-paying occupation tercile as in the highest-paying tercile. These patterns suggest greater scope for pay transparency to redirect job search in lower-paid and female-dominated occupations – a timely finding as EU member states implement the Pay Transparency Directive, which aims to narrow gender pay gaps (De Poli et al. 2026).

Job seekers know half the picture – and update only partially

However, responses to posted wages reflect not only how much job seekers value pay, but also how much they already knew about pay differences and how far they adjust their beliefs when shown new information: even job seekers who care strongly about wages will barely react to a posted wage if they had already anticipated it – or if they do not incorporate this information into their beliefs, for example, because their wage expectations are anchored to their current pay, because they doubt they could obtain the posted wage themselves, or because they do not trust the information. This is where our second experiment comes in: by eliciting the wage expectations of survey respondents who saw real job ads from their own search scope – with and without wage information – we can directly measure both what job seekers already know and how much they revise their beliefs. 

Figure 2 illustrates the results. It plots, for each job ranked by its benchmark wage (x-axis), the wage respondents expect to be paid, regardless of whether they received the benchmark wage information for that job. The wage beliefs of respondents who received no wage information are shown in blue (line and dots) and the corresponding beliefs of respondents who were shown the benchmark wage in red. These patterns reveal that job seekers perceive roughly half of the wage differences between jobs within their search scope: when one job pays 10% more according to the employer-review platform, job seekers in the control group expect only about 5% more. Even when wage information is displayed directly on the ad, job seekers close only about 29% of the gap between their prior beliefs and the benchmark wage. This estimate lies towards the upper end of belief-updating estimates in the recent literature, which range from 15% to 35% (Altmann et al. 2026, Andresen et al. 2026, Bhatiya et al. 2026, Christensen et al. 2026), suggesting that job seekers do not heavily discount information from an employer-review platform.

Figure 2 Wage beliefs versus kununu wages 

Note: The Figure shows a binned scatter plot of own-wage beliefs against benchmark wages, both in logs, for the control group (blue) and the wage treatment groups (red), with linear fits. The grey line is the 45-degree line.

Adjusting the posted-wage elasticity for prior knowledge and incomplete belief updating reveals that workers value wages more than their reaction to posted wages suggests: a 1% increase in a job seeker’s expected pay raises interest in a job by around 2.5–2.9%. This own-wage elasticity is several times larger than the posted-wage elasticity of 0.4 – implying that studies relying on posted wages alone may substantially underestimate how much workers care about pay. 

What are job benefits worth?

Interestingly, job seekers know even less about benefits than about wages: their beliefs capture at most 15% of the actual differences across firms in the availability of flexible working hours, work from home and firm-sponsored childcare, compared with roughly half for wages. Accounting for these differences in prior knowledge is therefore essential when comparing how much workers value wages and benefits. Figure 3 reports our willingness-to-pay estimates for each of the twelve benefits studied, obtained by combining the field-experimental responses with the survey-based corrections for prior knowledge and updating. Four benefits stand out: job seekers are willing to forgo 10.4% of their wage for a company car, 8.3% for the option to work from home, 6.1% for firm-sponsored childcare, and 4.4% for a parking spot. The valuations for company cars and childcare align closely with objective cost benchmarks. Most other benefits have small, statistically insignificant valuations, and one, the company doctor, appears to be a disamenity, perhaps because it signals hazardous working conditions.

Figure 3 Job seekers’ willingness to pay for job benefits

Note: The figure reports point estimates and 95% confidence intervals of the willingness to pay (WTP) for each of the 12 benefits, expressed as a percentage of the wage.

Benefits amplify inequality

These valuations matter for how we measure inequality. Among vacancies with matched wage information, the average job offers benefits with a monetised value of approximately 11% of wages, or CHF 741 per month. Because higher-paying jobs also offer more benefits – the lowest-paying decile of firms offers around six benefits on average, versus eight in the highest-paying decile – inequality in total job value exceeds wage inequality. Accounting for non-wage benefits in our sample thus amplifies, rather than compresses, compensation inequality across jobs. This finding is consistent with US evidence that differences in working conditions can reinforce wage inequality (Maestas et al. 2023).

Policy implications

Three lessons emerge. First, workers are imperfectly informed about wage and benefit differences across jobs, giving room for transparency initiatives to improve job search outcomes. Second, pay transparency works through job search: posting wages visibly redirects job seekers’ interest towards better-paying jobs, especially in low-paid and female-dominated occupations. Third, job seekers value some benefits highly, and know far less about benefit differences across jobs than about pay.

Source : VOXeu