Cryptography and the Economics of Supervisory Information: Balancing Transparency and Confidentiality

Abstract

We elucidate the tradeoffs between transparency and confi dentiality in the context of fi nancial regulation. The structure of information in fi nancial contexts creates incentives with a pervasive effect on fi nancial institutions and their relationships. This includes supervisory institutions, which must balance the opposing forces of confi dentiality and transparency that arise from their examination and disclosure duties. Prudential supervision can expose confi dential information to examiners who have a duty to protect it. Disclosure policies work to reduce information asymmetries, empowering investors and fostering market discipline. The resulting confi dentiality/transparency dichotomy tends to push supervisory information policies to one extreme or the other. We argue that there are important intermediate cases in which limited information sharing would be welfare-improving, and that this can be achieved with careful use of new techniques from the fi elds of secure computation and statistical data privacy. We provide a broad overview of these new technologies. We also describe three specifi c usage scenarios where such benefi cial solutions might be implemented.

DStress

“…Our results show that these models could be evaluated on the entire U.S. banking system in less than five hours on commodity hardware, using about 750 MB of traffic per bank. In our technical report [57], we also show that the use of differential privacy (which was already suggested by the OFR working paper [30]) does not significantly diminish the utility of the systemic risk measure. In summary, this paper makes the following three contributions:…”

“…This provides some basic security, but it is not sufficient to discover complex interdependencies, which would require looking at data from all the banks. This is why, in a recent working paper [30], the Office of Financial Research (OFR) has started investigating ways to perform system-wide stress tests while protecting confidentiality.…”

Sensitivity and computational complexity in financial networks

“…Flood et al (2013) proposed using tools from differential privacy (Dwork, 2006;Dwork et al, 2006) to provide a means of computing global network characteristics while preserving the privacy of each individual institution's holdings. A high sensitivity implies a worse tradeoff between privacy and accuracy when calculating network statistics.…”

Privacy and mechanism design

“…How should policy be designed given that the summary used to guide it is privacy-preserving and therefore necessarily coarse? [FKOS13] propose and study these concerns in the context of the financial industry, and provide an excellent overview of the trade-offs involved.…”