اثر انعطاف پذیری کار بر نتایج بازار کار زنان: برآورد از مدل های جستجو و چانه زنی

In this article, we develop a search model of the labor market in which jobs are characterized by work hours’ flexibility. Workers value flexibility, which is costly for employers to provide. We estimate the model on a sample of women extracted from the CPS. The model parameters are empirically identified because the accepted wage distributions of flexible and non-flexible jobs are directly related to the preference for flexibility parameters. Results show that more than one-third of women place a small, positive value on flexibility. Women with a college degree value flexibility more than women with only a high school degree. Counterfactual experiments show that flexibility has a substantial impact on the wage distribution but a negligible impact on the unemployment rate. These results suggest that wage and schooling differences between males and females may be importantly related to flexibility.

Anecdotal and descriptive evidence suggests that work hours’ flexibility, such as the possibility of working part-time or choosing when to work during the day, is a job amenity women particularly favored when interviewed about job conditions.1 On average, women spend more time than men in home production and child-rearing and less in the labor market.2 In this paper, we measure women’s preference for job flexibility and its effects on labor market outcomes by estimating the parameters of a search and matching model of the labor market with wage bargaining. We show how preferences for flexibility affect labor market outcomes and the shape of the accepted wage distribution. Finally, we assess the welfare and labor-market implications of policies favoring job flexibility. We describe the model in Section 3. Jobs can be flexible or not, and flexible jobs are more expensive to provide.3 Workers have preferences for wages and flexibility and meet with firms to bargain over these dimensions. Wage heterogeneity arises from the bargaining process as a result of idiosyncratic match-specific productivity and heterogeneity in preferences for flexibility. We show that because of search frictions, the wage differential between flexible and non-flexible jobs is not a pure compensating differential.4 In Section 4 we discuss the identification of the model parameters with data on wages and job flexibility. The model predicts wage distributions for flexible and non-flexible jobs. To provide intuition for the parameter identification, we show that if all the workers have the same preferences for flexibility then the accepted wage distributions of flexible and non-flexible jobs have non-overlapping supports. The size of the gap is measured by the monetary value of the preference for flexibility, which is equivalent to the compensating wage differential paid to the worker that marginally rejects a flexible job over a non-flexible job. Preferences for flexibility also imply a wider support of the wage distribution of flexible jobs. The firms’ cost of providing flexibility is also identified because a higher cost implies fewer flexible jobs in equilibrium. We describe the data in Section 5. Working hours’ flexibility includes both the possibility of working fewer hours and the option of organizing the working hours in a flexible way. Some papers focus on the first type of flexibility by studying part-time work and hours–wage trade-offs.5 Data limitations make it difficult to study the second type of flexibility. Although our model and estimation method apply to a general definition of flexibility, our data force us to approximate flexible jobs as part-time jobs. In the empirical implementation, we define a job as flexible when the worker provides less than 35 h of work per week. Section 6 describes our estimation strategy. Our estimation approach uses a simulated method of moments to minimize a loss function that includes several moments of the wage distributions of flexible and non-flexible jobs and moments of the distribution of unemployment durations. The parameter estimates show that approximately 37% of college-educated women have a positive preference for flexible jobs, valued between 1 and 10 cents per hour, but only about 20% of them choose such jobs in equilibrium. The value of flexibility for women with at most a high school degree is estimated to be equal to or less than 2.5 cents per hour. The structural-model estimates allow us to evaluate policy interventions, which we present in Section 8. We assess the welfare effects of the flexibility option by comparing our estimated model with an environment in which flexibility is not available. Next, we analyze a policy that reduces the cost of providing flexibility. Because of equilibrium effects, if flexibility is more costly or not available, some individuals observed in flexible jobs might decide to work in non-flexible jobs, whereas others might decide to remain unemployed. These workers preferences and productivities are relevant to assess each policy’s overall labor market effect. Search frictions and preferences over job amenities also imply that policy intervention may improve welfare because the compensating differentials mechanism is only partially at work.6 These counterfactual experiments suggest that flexibility has a large impact on the accepted wage distribution. However, the impact on overall welfare and unemployment is very limited. This implies that if men had significantly lower preferences for flexibility–as some anecdotal evidence seems to indicate–then these policies would have the potential to reduce the gender wage gap without significantly affecting overall welfare.

In this paper, we estimate the parameters of a dynamic search model of the labor market in which workers and firms bargain over wages and the provision of flexibility. We maintain the narrow definition of flexibility most frequently found in the literature–flexibility as the availability of part-time work–and show that women value this amenity. In our estimates, we also find that college graduates and high school graduates value flexibility differently. College graduates place higher value on having flexible jobs. Moreover, we find that jobs requiring a college education can provide flexibility at a lower cost. Because women might also choose schooling to accommodate their preference for job flexibility, we speculate that this might explain some of the observed differences in schooling achievements between men and women. The counterfactual experiments reveal that the impact of flexibility is quite substantial on some labor market outcomes (i.e., wages and hazard rates, the distribution of employment between flexible and non-flexible jobs) but negligible on others (notably, unemployment). For example, without flexibility, for college-educated women that value flexibility the most, the average wage would be 74% lower, while the unemployment rate would be at most 0.06% higher than in the benchmark environment with flexibility. We infer from our experiments that policies reducing the cost of flexibility provision could be very effective in changing the realized wage distribution at little cost in terms of employment. Our approach presents four main limitations. First, in the empirical application of our model, we define flexible jobs using part-time jobs. A more appropriate definition should also capture the option of organizing work time in a flexible way. Second, we estimate the model by schooling groups and we find significant differences between them. We did not, however, integrate a schooling decision in the model and in the estimation procedure. We think devoting future work to filling this gap is particularly promising to test our conjecture that expectations on future job amenities, such as flexibility, are important components of the schooling choices of women. Third, employers and workers in our model are very stylized. In particular, we assume a homogenous cost of providing flexibility, and workers are heterogeneous only in education and preference for flexibility. Estimating heterogeneous costs and correlations between costs and industries could help explain why we observe different preferences across different skill levels and could deepen our understanding of the feedback of the labor market on schooling choices. Just as we find that different levels of schooling are correlated with preferences for flexibility, we could find that different types of jobs or schooling (e.g., college majors) are correlated with preferences for flexibility because they increase the likelihood of working in jobs and industries that provide flexibility at low cost. Fourth, we have found a strong impact of flexibility on wages and a significant correlation between preference for flexibility and level of schooling. A large amount of literature on gender differentials on both variables currently points to a puzzle. Recent data on US workers show women earning lower wages than men, despite having a positive schooling differential. Results from the counterfactuals reveal that changes in the provision of flexibility (e.g., through a subsidization of its cost) have a large differential impact for people that value flexibility differently. Therefore, a higher preference for flexibility for women with respect to men (if proved) could potentially explain a large portion of the gender-wage differential. Exploring this conjecture requires estimating the model for a sample of men. Our sample does not include enough males working in flexible jobs to provide reliable estimates of their preferences for flexibility. We hope that a more complete data set providing a better definition of flexibility (and possibly more detailed schooling and firms information) will also generate enough data variation to estimate the model on a sample of men.