Doubly Bound: The Cost of Credit Ratings

Conclusion

Conclusion

THE METHODOLOGY OUTLINED ABOVE is intended to show what is possible and invite further discussion. I recognize that other researchers and practitioners, working with different data sets and analytical approaches, will select different indicators and choose different weights, ranges, or mappings.

More ambitious changes to the modeling methodology are possible. For example, all the indicators need not be economic or accounting values. Automated textual analysis of a government’s web site, legislative proceedings, and news coverage may yield qualitative indicators that could be coded and then included in a rating index.

More important than the methodology chosen is my general assertion that model-based ratings—calibrated to assign a greater proportion of AAA municipal bond ratings yet sensitive enough to detect at-risk entities—can provide a viable alternative to the contemporary municipal bond rating system. A publicly available model that is easy to implement would be virtually cost-free. If ratings consumers accepted such a system, it could produce savings for government bond issuers— and thus taxpayers—of roughly $2 billion per year.