Marie Gram Pietraszek defends her PhD thesis at the Department of Economics

Candidate:

Marie Gram Pietraszek, Department of Economics, University of Copenhagen

Title:

Empirical Essays on Labor Economics: Coworker Networks, Application Success, and Graduate Trajectories

Supervisor:

• Nikolaj Arpe Harmon, Associate Professor, Department of Economics, University of Copenhagen

Assessment Committee:

• Morten Bennedsen, Professor, Department of Economics, University of Copenhagen
• Giovanni Mellace, Professor, Department of Economics, University of Southern Denmark
• Oskar Nordström Skans, Professor, Department of Economics, Uppsala University

Summary:

The labor market is a cornerstone of any economy, playing a pivotal role in shaping the well-being of individuals and the economy. Unemployment, a key facet of the labor market, carries significant implications. High levels of unemployment not only result in financial hardship for individuals but also contribute to social challenges, including increased inequality and diminished quality of life. Understanding the complexities of the labor market is instrumental for policymakers aiming to design interventions that alleviate unemployment and enhance job search strategies. This PhD dissertation consists of three self-contained chapters that study empirical aspects of the labor market and job search.

The first chapter (co-authored with Nikolaj Harmon and Jonas Maibom) examines the mechanisms behind the effect of coworker networks on job search outcomes. We combine linked Danish administrative data on job applications made by the universe of UI recipients with a theoretical job search framework. We employ a quasi-experimental identification strategy that generates variation in coworker networks from the timing of past job transitions to identify the effect of social connections. This enables us to quantify the relative importance of the different mechanisms driving the effect of coworker connections on job search outcomes. We find that the effect of having a coworker connection at a firm increases the likelihood of an application resulting in a hire at the given firm by 133 percent for Danish UI recipients. One-tenth of this overall effect arises because social connections increase the likelihood that an application results. The remainder of the effect stems from changes in application behavior because it is more attractive for the UI recipient to apply to a connected firm. In contrast, we find that direct information effects - in the sense of being more likely to notice a given employment opportunity - appear unimportant in our setting.

The second chapter employs machine learning models to predict job application success probabilities. I combine novel data on job applications with standard administrative data sources to predict application success. I find that the histogram gradient-based boosting model exhibits better calibration and lower mean squared error in out-of-sample testing. I present descriptive statistics on the predictions, revealing an average success probability of 0.98 percent, with the variation in success probabilities driven largely by differences between UI recipients and to a lesser extent variation within the UI recipient. I present a descriptive analysis of the changing success probabilities throughout the first year of the UI spell, showcasing the practical applicability of the predicted success probabilities for research. The average success probability declines over time, which is attributed to the effect of UI duration, dynamic selection, and changes in application behavior. I present evidence suggesting that approximately two-thirds of the decline is linked to the effect of UI duration, while the remaining third stems from dynamic selection. UI recipients adjust such that they apply for jobs with a slightly smaller success probability over time. Lastly, I repeat the analysis split by gender. The conclusion is consistent, but I conclude that the effect of duration is more pronounced for men.

The third chapter (co-authored with Lykke Sterll Christensen) investigates the causal effect of fluctuations in the labor market on employment and social outcomes of graduates. We combine Danish registry data and unemployment data for the period 2007-2019. We construct a monthly and occupation-specific measure of labor market conditions, which we employ as shifters in a shift-share design. We determine exposure to the shifters from a previous cohort. We find that people graduating under worse labor market conditions are more unemployed, and those in employment earn less and work less. Additionally, we find that worse labor market conditions during graduation increase singlehood and have a lasting negative impact on fertility. We repeat the analysis for men and women separately and find that the effect of fertility is largely driven by women. To investigate the drivers behind the decline, we examine responses for both graduates in a couple and in employment and find a reduction in fertility for both groups. Lastly, we find women graduating at an unfortunate time tend to work more in the private sector.

An electronic copy of the thesis can be requested here: lema@econ.ku.dk