Modern day bankers do borrow and lend and, often, do it rather well so that they do not lose much at all (one of Shakespeare's reasons for not lending). That's just as well because banks lend other peoples' money, not their own.
In the consumer market lending decisions are often made using credit analytics - creating an index of creditworthiness built from information and data about an individual both from public records, such as the voters' roll, the bankruptcy register and court judgment records and private information such as credit card usage, provided by banks and other lenders. Credit scoring, as it is known, is statistically consistent, reliable in that it delivers a lower level of default than old fashioned human judgement and can be controlled for different levels of risk, depending on the banker's risk appetite. At a micro level it can grade likelihood of default and can help to price risk (interest charged reflecting the likelihood of loss).
Couple this with the wealth of information that banks have on day to day transactions on our credit and debit cards whereby they can spot potential fraud by tracking odd payments behaviour and you have a relatively successful business model based on the analysis of available information, where losses are managed and high risks avoided.
But credit scoring is not without its flaws. It can only really capture available information and uses history to predict the future. It can commit both type 1 and type 2 errors and does , rather, reduce personal relationships to binary code, distancing bankers from their customers. This can add risk of a different kind as institutional trust built through long relationships is replaced with institutional distrust, bankers are replaced by robots, machines and computers - and all in the pursuit of profit for shareholders.
So, as I said. An excellent business model IF the working assumption is that the overriding aim is to improve profitability in the short term.
Now, is that an assumption that can be applied to British Universities?