Automation and Top Income Inequality

Macro Lunch
PCPSE Room 101
United States
Abstract: One wellknown fact about the income distribution is that it can be approximated by a Pareto distribution. For almost 40 years, the inequality inside the top percentile of the income distribution has been increasing in the US, hence the estimated Pareto parameter for the top income distribution is decreasing. In this paper, we show that improvement in automation technology (number of tasks for which capital can be used) leads to increase in the top income inequality. We consider a model in which labor has a convex cost and capital has a linear cost. This leads to decreasing returns to scale profit function for entrepreneurs. As capital replaces labor in more and more tasks, severity of diseconomies of scale diminishes, hence the market share of top skilled entrepreneurs increases. If entrepreneurial skill is distributed with a Pareto distribution, then top income distribution can be approximated with a Pareto distribution. We show that the shape parameter of this distribution is inversely related to level of automation, i.e. as level of automation increases, the top income inequality increases. We rationalize convex cost of labor using one of the important differences between capital and labor: capital cannot shirk, whereas labor can. To prevent labor from shirking, an entrepreneur needs to spend additional resources for the labor they use, hence labor has a convex cost.