The micro drivers of aggregate employment

In my PhD thesis, I study how the aggregate employment growth is affected by secular trends or exogenous shocks using quantitative models of firm dynamics and machine learning algorithms in the counterfactual analysis. To this end, I leverage the comprehensive firm-level data for a small open economy, Belgium. It is quarterly data on firm-level employment and wage bill from the National Social Security Office (NSSO) of Belgium on all firms paying social security contributions during the period from 2003Q1 to 2020Q2 period. In the first chapter, I study the firm growth in Belgium using Markov transition matrices and the associated job flows. I observe that firms, conditional on employment growth, adjust their employment only slightly by tracking their quarterly movements within the size distribution. While some firms grow more rapidly, their contribution to the overall job creation or job destruction is insignificant. Therefore, it is not the magnitude but the frequency of employment adjustment that determines firm growth. Using this mechanical observation, I approximate firm dynamics as the birth-death process, type of a Markov process. This specification implies that all firms grow or shrink in size by one employee at a time and allows to link the net job creation rate by size to overall firm size distribution. In a counterfactual experiment, I show that a hypothetical dampening of the net job creation rate among small firms leads to a decrease in the large firm share. In the second chapter, I analyze how the secular decline in firm entry affects the aggregate employment of Belgium using a dynamic decomposition of aggregate employment by age and sector. I allow for the firm survival rate and the conditional employment growth rate to differ by age and sector. This dynamic framework captures well the trend evolution of aggregate employment, and I use it to generate the counterfactual evolution of aggregate employment under different patterns of firm entry. To this end, I disentangle the entry margin into two channels along which it affects the aggregate employment, the overall employment of new firms at the entry stage, or the start-up employment, and the share of start-up employment by sector, or the sectoral composition of start-ups. The simulation results suggest that the decline in start-up employment slowed down the growth rate of aggregate employment by 26% over the 2009Q2-2017Q1 period by shifting the age distribution of firms toward older firms. The sectoral composition of start-ups accelerated the decline in the manufacturing sector and prevented the distribution sector from a potential decline while leaving the aggregate employment unchanged. This chapter has been published in Small Business Economics. In the third chapter, which is my job market paper, I evaluate the impact of wage reduction on employment using a policy to suspend the Belgian automatic wage indexation system in 2015 as a natural experiment. Since the policy affected all firms in Belgium and resulted in an economy-wide real wage decrease of 2 percent, no suitable control could be found. While predicting firm behavior ex-ante is a tall order in general, predicting firm behavior ex-post in response to a hypothetical wage indexation, that used to happen frequently in the past is an ideal setting to use machine learning for policy evaluation. Therefore, I use a machine learning based approach to construct a counterfactual control group where I predict the firm-level employment for the scenario without the policy, where the wages get indexed as before. To verify the validity of the counterfactual group, I use the out-of-sample cross-validation technique on the data before the suspension. Later I use the same cross-validation results to correct for the prediction bias of the algorithm in a difference-in-differences setting. Therefore, using the actual treatment and counterfactual control groups, I identify the impact of wage indexation on employment. I find a positive impact on employment of 0.5 percent, which corresponds to a labor demand elasticity of -0.25. This effect is more pronounced for manufacturing firms, where the elasticity can reach -1. In my final chapter, I explore the impact of the COVID-19 lockdown on aggregate employment in Belgium. It is already clear that the pandemic has a massive effect on employment worldwide, but it is not clear how long it will last and what are the implications of lockdowns in terms of jobs lost in ten years. To this end, I augment the dynamic decomposition of aggregate employment from the first chapter by allowing the life-cycle dynamics of firms, the survival rate and the conditional growth rate, vary with the business cycle. In particular, I apply a machine learning based approach to predict the counterfactual evolution of life-cycle dynamics. Using this dynamic framework augmented with machine learning, I forecast the evolution of aggregate employment over the 2020-2030 period under various economic scenarios. In doing so, I distinguish between start-ups and incumbent firms with both short- and long-term effects on employment. In the short-term, I expect to see significant losses of employment coming mainly from mature incumbent firms. In the long-term, the missing generation of start-ups during the lockdown is expected to have a significant and growing effect of slowing down the employment growth even a decade after the lockdown.

Publication year:2021

Accessibility:Closed