This article in many ways, is the “debt” counterpart to my 1993 Journal of Financial Economics piece and offers an even broader critique of the reflexive belief that more complete financial markets are always better.
I conclude thus:
Lemon problems do not stop the sale of well over a million used cars in the U.S. each year, but they do prevent the operation of a market in which buyers place sight-unseen bids for used cars offered by unknown sellers. In fact, anonymous markets for physical goods are restricted mainly to metals or agricultural commodities. Most goods—including new or secondhand cars, shoes and homes—are purchased from identifiable sellers.
Buyers also prefer to examine specific items—test-driving cars or trying on shoes, for instance—before they make a purchase.
Outside finance, revolutionary technological advances have not turned many goods or services into anonymously traded commodities. Rather, the advances have reduced the cost of communicating and using detailed information, mitigating information asymmetries, and helping buyers select items that match their preferences. And technology has reduced anonymity: in contrast to the street-hailing of taxis, users of ride-hailing apps can screen drivers based on their ratings. Similarly, consumers can review the ratings of plumbers on the web instead of randomly picking one from the telephone directory.
Why, then, did anonymous debt markets—markets that require investors to forgo information they could secure in private transactions—experience such remarkable growth in the U.S.? Until the 1980s, creditors in the U.S. were willing to give up information to get tradability mainly in the case of bonds issued by governments or blue-chip companies. The subsequent securitization of trillions of dollars of loans extended to unknown individuals in the U.S. has required investors, loan originators, and government housing finance agencies to rely on generic credit bureau scores. But while reliance on scores that loan originators cannot manipulate reduces the information asymmetry problems that investors would otherwise face, it also makes estimates of default risks noisier, increasing the unwarranted extension and denial of credit.
Technological advances did not preordain the revolutionary “completion” of anonymous credit markets in the U.S. U.S. banks could have used technology to improve decentralized, case-by-case lending. Or, like many European lenders, U.S. banks could have developed proprietary credit-screening algorithms that incorporate a wide range of data about applicants. U.S. government-sponsored agencies also could have developed rule-centered Artificial Intelligence systems to automate case-by-case mortgage underwriting. Instead, fair lending and credit reporting laws and government-sponsored housing finance agencies favored tradability over information by promoting—to the point of virtually requiring—reliance on bureau scoring. Policy choices fostered the expansion of anonymous credit markets to an extent that few would have thought prudent or possible.