Sovereign Credit Ratings, Transparency, and Algorithms

Abstract Sovereigns sometimes borrow money from creditors. To demonstrate their creditworthiness, private credit rating agencies assess sovereigns and provide them with sovereign credit ratings. While the ethical issues associated with the credit scoring of individuals have received considerable attention, the ethical dimensions of sovereign credit ratings are underexplored. I argue that the current way in which sovereign credit ratings are calculated, which does not use opaque algorithms, is objectionable. This is because of concerns associated with opacity. The credit-rating process is opaque to actors on the outside, while in-principle transparent to credit rating agencies. Furthermore, some economists have proposed that opaque machine learning techniques should be used to calculate sovereign credit ratings. I argue that such a proposal gives rise to independent concerns about opacity. If such algorithms are used, the credit-rating process is opaque to actors on the outside and to the credit rating agencies employing them.