Last week Joseph Baines and I published a new research brief with the Roosevelt Institute on corporate tax and concentration. The report can be found here. In this post I won’t try to summarize the arguments and findings – those can be found in this Twitter thread. Instead I want to discuss some of the trials and tribulations of using firm-level tax data to study concentration.

In our earlier work Joseph and I compared the effective tax rates of large and smaller corporations in the United States. Our original plan with the Roosevelt brief was to update these findings and to extend these analysis of ETRs to different sectors. What we found in our initial draft for Roosevelt largely mirrored findings from our previous research: namely that the ETRs of the largest corporations are significantly and persistently lower than their smaller counterparts.

But an astute referee commissioned by the Roosevelt Institute had misgivings with our methodology. Because we included loss-making firms, the ETRs that we reported bore no resemblance to the statutory tax rate inscribed in law. For some time periods, the ETR for smaller corporations (when loss-making companies are included) are above 50 percent, even when using an aggregate method to smooth out the data. Once we started to break things down by sector, the ETR climbs even higher. Reporting these ETRs may be fine for an academic study. But they are misleading and can be easily misinterpreted in a public-facing facing publication like the one we did for Roosevelt.

At this point we faced a dilemma. Including loss-making firms is necessary in the study of concentration, otherwise we get a skewed view of the state of competition in the economy. But including loss-making in the study of tax gives us these wonky ETRs that are difficult to interpret. My brilliant co-author Joseph found a way out of this dilemma. Rather than ETRs, he proposed that we follow studies of personal income inequality and examine the pre- and post-tax profit shares of large corporations. This allows us to retain loss-making companies and avoid the issue of difficult-to-interpret ETRs. By comparing the pre- and post-tax profit share we get a sense of the role of the tax structure in redistributing income between corporations of different sizes.

The new method is not perfect. Including loss-making companies means that technically the profit share of a cohort could be over 100% (for the period from 2019-2022 the post-tax profit share of large corporations was 99%). But in our view, these problems are insurmountable. We need to retain loss-making companies in our analysis if we want to say anything meaningful about concentration and monopoly power in the corporate sector.

This is an instance where valid criticism forced us to re-think our research and led us to a new way of looking at tax and concentration. What peer review was meant for!