Does the US Tax Code Favor Automation?

Abstract

We argue that the US tax system is biased against labor and in favor of capital and has become more so in recent years. As a consequence, it has promoted inefficiently high levels of automation. Moving from the US tax system in the 2010s to optimal taxation of capital and labor would raise employment by 4.02% and the labor share by 0.78 percentage points, and restore the optimal level of automation. If moving to optimal taxes is infeasible, more modest reforms can still increase employment by 1.14-1.96%, but in this case efficiency can be increased by imposing an additional automation tax to reduce the equilibrium level of automation. This is because marginal automated tasks do not bring much productivity gains but displace workers, reducing employment below its socially optimal level. We additionally show that reducing labor taxes or combining lower capital taxes with automation taxes can increase employment much more than the uniform reductions in capital taxes enacted between 2000 and 2018.

AI and Shared Prosperity

“…At the time of writing, autonomous delivery is not widely commercially available, but a few startup companies are piloting them in a limited set of locations. 1 For the purposes of this discussion, we will examine autonomous grocery delivery, while noting that autonomous delivery vehicles could be used to deliver other types of goods as well.…”

“…Most of those jobs will require a college or advanced degree and will likely be geographically concentrated. They will also likely be fewer in number compared to the number of displaced customer 1 See, for example: https://venturebeat.com/2021/01/27/starship-raises-17-million-to-send-autonomous-delivery-robots-to-new-campuses/ and https://medium.com/nuro/california-dmv-grants-nuro-first-ever-av-deploymentpermit-ca424ebd2 service associates identified in step 1, because one chatbot development company can service many corporate clients.…”

“…However, aside from voluntary action on the part of AI developers to consider the economic interests of workers in the AI development process, government can also take steps to ensure that the regulatory environment does not provide incentives for excessive automation and ever greater concentration of earning opportunities. For example, the incentives for innovators are affected by tax policies -current tax regimes that favor capital over labor [1] and policies that limit labor mobility [13] create strong incentives to develop AI applications that focus disproprotionately on labor-saving use cases. If these policies remain in place, AI advancement might bring about levels of automation well above what is socially optimal, to the disproportionate detriment of the economically vulnerable workers with limited access to retraining opportunities.…”

Automation and Taxation

“…Existing studies on the nexus of automation and taxation mostly take an optimal taxation perspective. For example, Acemoglu et al [2020] argue that the US tax system is biased in favor of capital which leads to a sub-optimal reduction of the labor share for "marginally automated jobs". Applying the optimal taxation framework by Diamond and Mirrlees [1971] to a task-based model calibrated on US tax rates, the authors show how a tax reform could raise the labor share.…”

“…When ATs diffuse and replace labor at a large scale, the tax basis might be significantly undermined. This argument is put forward to support that a robot tax is needed to ensure the sustainability of public finances [Kovacev, 2020, Süssmuth et al, 2020, Acemoglu et al, 2020, Rebelo et al, 2019. However, the impact of automation is complex, including many second-order effects.…”