A law passed in June 2004 banned the use of gender preferences in job recruiting in Austria. At the time over 40% of openings on the nation’s largest job-board specified a preferred gender. We use data on filled vacancies, merged to employer records, to study how the legal prohibition of gender preferences affected hiring and job outcomes. Prior to the ban, most vacancies with a stated preference signalled stereotypical preferences (e.g., a preference for females at a majority female workplace), but a minority stated preferences to recruit against stereotypes - a subset we call “non-stereotypical” vacancies. Vacancies with a gender preference were very likely (>90%) to be filled by someone of the preferred gender. We develop models based on pre-ban vacancies to predict the probability of specifying a preference for female, males, or neither gender. We then conduct event studies of the effect of the ban on different predicted preference groups. We find that the ban led to a rise in the fraction of women hired for jobs that were likely to be targeted to men (and vice versa), reducing the degree of gender segregation across firms. Partially offsetting this effect, we find a reduction in the success of non-stereotypical vacancies in recruiting workers that would diversify the gender mix of the workplace, and a rise in filling times for these vacancies. For the larger set of stereotypical vacancies, however, vacancy filling times, wages and job durations were largely unaffected by the ban, suggesting that the law had at most small consequences on job match efficiency.

This paper examines the extent to which changes in international market relative prices lead to shifts in firms’ skill requirements through trade, in the context of an exogenous currency appreciation. On January 15, 2015 the Swiss National Bank unexpectedly abandoned the exchange rate floor with the Euro, causing a 15% increase of the value of the Swiss franc, which remained relatively stable in the subsequent years. This unforeseen appreciation immediately impacted the relative price of trade, creating new incentives and opportunities for import, while simultaneously reducing expected profits for firms exposed to foreign competition. I study how this sharp change in trade conditions affected skill requirements in Switzerland using novel data on trade and labor demand. Specifically, I merge trade data containing information on each single import or export transaction made by Swiss firms to firm-specific job postings data. I find that in the two years after the shock firms with a workforce more exposed to offshorability and automation, increased imports and posted more job ads for highly skilled workers. For these firms, a 10 percent increase in monthly import translates into a 2.1 percent reduction in the routine intensity of their labor demand.

Analyses of spatial or network data are now very common. Nevertheless, statistical inference is challenging since unobserved heterogeneity can be correlated across neighboring observational units. We develop an estimator for the variance-covariance matrix (VCV) of OLS and 2SLS that allows for arbitrary dependence of the errors across observations in space or network structure and across time periods. As a proof of concept, we conduct Monte Carlo simulations in a geospatial setting based on U.S. metropolitan areas. Tests based on our estimator of the VCV asymptotically correctly reject the null hypothesis, whereas conventional inference methods, e.g., those without clusters or with clusters based on administrative units, reject the null hypothesis too often. We also provide simulations in a network setting based on the IDEAS structure of coauthorship and real-life data on scientific performance. The Monte Carlo results again show that our estimator yields inference at the correct significance level even in moderately sized samples and that it dominates other commonly used approaches to inference in networks. We provide guidance to the applied researcher with respect to (i) whether or not to include potentially correlated regressors and (ii) the choice of cluster bandwidth. Finally, we provide a companion statistical package (acreg) enabling users to adjust the OLS and 2SLS coefficient’s standard errors to account for arbitrary dependence.

This paper investigates the effect of supporters on the performance of soccer players by skin color using objective player performance data and an automated skin color recognition algorithm. Identification comes from an exceptional change in access to stadiums: due to the COVID-19 restrictions, one third of the games of the highest Italian soccer league 2019/2020 season were played in closed stadiums. I identify a significant increase in the performance of non-white players, relative to white players, when supporters are banned from the stadium. The effect does not differ between home and away games, and players playing in top versus minor teams, while weaker players are impacted more than others. These results suggest that discrimination faced by non-white individuals affects performance.