Working Paper Series no. 803: Heterogeneous Policy Distortions and the Labor Share

Government policies that impose unequal constraints across firms are widespread. Be it among policymakers or academics, these policies are suspected of having large damaging macroeconomic effects, particularly in terms of aggregate productivity and output. In this paper, I investigate the impact of heterogeneous policy distortions on a another major macroeconomic outcome: the labor share of income.

The concept of heterogeneous policy distortions spans a wide range of policies, from trade policy to fiscal policy or labor market regulation. My analysis of the link between this type of distortions and the labor share rests upon a simple empirical regularity: bigger firms tend to have a smaller labor share. Consequently, any policy limiting the weight of bigger firms in the economy potentially causes a redistribution of national income towards labor. The main contribution of this paper is to show that this composition effect can be substantial, which contributes to our general understanding of the determinants of the labor share.

In order to assess the quantitative importance of heterogeneous policy distortions for the labor share, I propose an extension of the neoclassical growth model with two-way heterogeneity across firms. In the steady-state, firm-level production functions vary both in terms of total factor productivity (TFP) and of factor elasticity. I calibrate the model using French production data for the period 1995-2007. In particular, I set the level of correlation between the two idiosyncratic technological parameters so as to match the observed correlation between size and the labor share across firms. Then, I study the impact of policies which create dispersion across firms in terms of output price or factor price. By distorting the allocation of productive resources, such policies can impact the weight of capital-intensive firms in the economy. This results in a distortion of the aggregate relative demand for capital and labor which, ultimately, impacts the labor share of national income.

My results show that considering labor share differences across firms to quantify the impact of heterogeneous distortions is primordial in the case of output price distortions. Taxing the output of the 50% most productive firms results in an increase of the national labor share ranging from 1 percentage point (hereafter “p.p.”) to 2 p.p., for tax rates from 10% to 50%. These effects double when I take into account selection, that is the fact that taxing productive firms allows some unproductive firms with high labor share to survive in the economy. To put these numbers in perspective, the labor share in France was 5 p.p. larger than in the US in 2005. 5 p.p. also corresponds to the global drop in the labor share over the past 30 years, as estimated by Karabarbounis and Neiman (2014). My simulations suggest that the removal of productivity-based output distortions could explain from 20% to 75% of this evolution.

The mechanism presented in this paper is important for understanding what movements in the labor share can tell us about economic efficiency. There is evidence that the global decline in the labor share observed over the last decades results from an increase in average markups (De Loecker and Eeckhout, 2017, 2018), and thus potentially a loss of economic efficiency. On the contrary, my results suggest that this decline is consistent, both qualitatively and quantitatively, with a better allocation of resources towards large firms over time.