Reprinted in Learning in Labour Markets, edited by Michael Waldman, Edward Elgar Publishing, Cheltenham, 2017.

Working Papers

  • Battling resistance: using machine prediction to improve antibiotic prescribing (with Michael Ribers) – paper draft available on request

Machine learning methods are increasingly providing economists with opportunities to design welfare improving policies when prediction problems are at their core. The alarming increase in antibiotic resistance is one such opportunity. In this paper we evaluate how machine prediction can reduce wasteful overprescribing without diminishing health outcomes. Specifically, we train a machine learning algorithm on administrative data from Denmark to predict bacterial caused urinary tract infections. The benchmark against which machine prediction must be evaluated is whether it can improve upon human decision making. Contrasting existing machine learning papers tackling prediction centered policy problems, we observe labels, namely patient test outcomes, independent of physician prescription choices. This allows us to directly evaluate health outcomes of prescription redistribution rules based on machine prediction. We find that redistribution rules based on a combination of machine prediction and physician autonomy can lower the overall amount of prescribing by up to 10 percent without reducing the number of correctly treated bacterial urinary tract infections. As Denmark is one of most conservative countries in terms of antibiotic prescribing, this result is likely a lower bound of what can be achieved elsewhere.

We analyze the effects of a hypothetical payment card fee regulation on bank profits, consumer welfare, and merchant welfare. We model consumers’ and merchants’ bank choices for debit card services, cardholders’ demand for card usage (conditional on bank choice), and how banks account for these in setting card fees to their customers. To estimate the model, we use bank-level data and survey data from the Norwegian debit card scheme, BankAxept. We conduct counterfactual exercises to analyze the effects of interchange fee regulations in the debit card scheme.

  • Antibiotic prescribing under uncertainty about resistance (with Michael Ribers)

The increasing level of antibiotic resistance constitutes a major worldwide health threat. Inappropriate antibiotic prescribing is considered one of the main drivers of increasing resistance. Hence, rational prescribing is an important policy objective. We develop a dynamic structural model of antibiotic prescribing for forward-looking general practitioners in the presence of uncertainty about antibiotics’ effectiveness. Our model endogenises information acquisition and features cross-patient learning from observed clinical microbiological testing. Reducing uncertainty is costly so that general practitioners have incentives to under-diagnose antibiotic resistance and prescribe inappropriately. We propose a framework for counterfactual simulations to evaluate policy measures such as mandatory resistance testing, a tax on (broad-spectrum) antibiotics, and the introduction of rapid testing technology.

Work in progress

  • Career concerns and managerial risk taking: evidence from the NFL (with Florian Schütt)
  • Governmental intervention on pharmaceutical markets (with Ulrich Kaiser)

My SSRN profile.
My Google Scholar profile.