Wage Dispersion and Search Behavior: The Importance of Non-Wage Job Values [report]

Robert Hall, Andreas Mueller
2015 unpublished
We use a rich new body of data on the experiences of unemployed jobseekers to determine the sources of wage dispersion and to create a search model consistent with the acceptance decisions the jobseekers made. Heterogeneity in non-wage job values or amenities among jobseekers and jobs is a central feature of our model. From the data and the model, we identify the distributions of four key variables: offered wages, offered non-wage job values, the value of the jobseeker's non-work alternative,
more » ... d the jobseeker's personal productivity. We find that, conditional on personal productivity, the standard deviation of offered log-wages is moderate, at 0.24, whereas the dispersion of the non-wage component of offered job values is substantially larger, at 0.34. The resulting dispersion of offered job values is 0.38. We also find high dispersion of personal productivity, at 0.43. People looking for jobs seek to earn money. But holding a job involves an opportunity cost in terms of less time for other activities. A job also has a non-wage dimension-it can be vexatious, fulfilling, or both. We call this dimension the non-wage value of a job. Job-seekers consider all three dimensions in deciding whether to accept an offer. They also judge job offers against the probability that they will be offered a nicer job or higher wage if they decline an offer and incur the cost of continuing to search. We develop a model of the search process and explore its implications for the probability distributions of key variables. Our model embodies the now-standard view that employed people search along with the unemployed. Searchers consider the job value of an offer, modeled as the product of the wage and the non-wage value. A member of either group forms a reservation job value and accepts a job that offers at least that value. Wages and non-wage values vary across workers, because some workers are more productive than others, because workers have different opportunity costs of being employed, because there is variation across workers in the non-wage values they receive from a given job, and because of the randomness of job offers. Our goal is to decompose the variation across individuals into probability distributions of four variables: (1) personal productivity, (2) the opportunity cost of holding a job, (3) the offered wage, and (4) the offered non-wage job value. Our model of the search process makes strong enough assumptions to identify our four probability distributions. We introduce and defend these assumptions shortly. We use data from a novel panel survey of job-seekers, carried out by Alan Krueger and Andreas Mueller (the KM survey). Respondents in the survey were selected at random from individuals who were drawing unemployment benefits in New Jersey in the fall of 2009. The survey reports reservation wages, wages of job offers, and the acceptance of an offer. We also use administrative data on wages linked to the survey respondents. The new survey permits more refined measures of dispersion than earlier data sources. The survey identifies the dispersion of non-wage values of jobs in the following way: Many job-seekers accept jobs that pay less than their previously reported reservation wages. This outcome reveals that the accepted job has an unusually high non-wage value. A smaller fraction reject jobs that pay more than the reservation wage, a sign of a low non-wage value. Accordingly, we can measure the distribution of the non-wage job value directly from the acceptance frequency stated as a function of the ratio of the offered wage to the reservation wage. The survey data also identifies the dispersion of personal productivity. Our idea is that reservation wages and offered wages both depend on personal productivity, so the covariance of the two variables equals the variance in personal productivity across job-seekers. Our measure of dispersion is the standard deviation of the log of a variable. Our estimates imply that the overall dispersion of the log of the offered wage is 0.52. The dispersion of
doi:10.3386/w21764 fatcat:4s6p2js6qzfzjgonsgymcnxdu4