Nobel Prize Worthy Lending

Nobel Prize Worthy Lending

In the wake of Nobel prize winners Al Roth and Lloyd Shapley, it’s worth analyzing the matching algorithms to see their application to banking.

I’ll walk through the basics of the matching algorithm here, but Alex Tabarrok provides a much better and more thorough primer here.

The Basics

The Nobel prize winning algorithm deals with matching. For example, men and women to get married or transplant recipients. Essentially, the algorithm allows one group to select its ideal match, allow the other group to reject or retain the offer, then rejected first group members make their offer to their second choice, which in turn rejects or retains the offer. This continues until there are no more offers.

The reason the algorithm is good is because it “converges.” That means there is actually a solution in the vast majority of cases. But, it also has this great property that there are no pairs that would rather be together but are not. There may be disappointed first parties and disappointed second parties, but no mutually disappointed pairs.

In my case, I want to discuss the algorithm for lenders and borrowers.

The big problem we have in both consumer and commercial lending right now is a mismatch. Banks want to lend to those businesses that need the lending least of all. But of course, banks want to lend to someone, because if they don’t, they don’t make money. On the flip side, borrowers would like the best choice – and given the fear that banks may be left out – borrowers may benefit from bank offers that otherwise wouldn’t be made.

Of course, banks don’t just make one loan. Fortunately, Roth has extended the original algorithm for more complicated matching like banks that want to make multiple loans.

So perhaps the algorithm goes like this: in some virtual marketplace a set of lenders and borrowers come together to exchange information. Borrowers make “public” enough information for banks to draft an offer.

  1. Banks make their offers to their best potential borrowers.
  2. Borrowers reject the offer or retain it.
  3. Rejected banks make new offers to additional borrowers to replace those that were rejected.
  4. New borrowers reject the offer or retain it.
  5. Continue until no more offers are made.

The advantage of taking this approach is that banks will be able to lend out according to their goals to their optimal portfolio. Borrowers, in turn, will get to choose offers and reject those that don’t meet their needs. Banks would obviously have to choose their offers and portfolio of targeted borrowers based on a complex set of profitability and risk metrics. But given an optimization-based pricing and selection approach, they could always guarantee an optimal portfolio – given that not all their offers will be accepted.

You might argue that banks wouldn’t participate in such a virtual marketplace, but I assert they already do. Consider such platforms as Lending Tree. Also, if the benefits of such a virtual marketplace outpace those of the traditional marketplace for borrowers, alternative players such as and LendingClub may come in to fill the gap.

In particular, other marketplaces, such as assigning medical residents to hospitals, have shown that the “greedy” approach of making early offers and forcing quick decisions leads to suboptimal results, including backing out and unhappy pairings. Banks see this when borrowers finance early to get a better rate. It leads to poor profitability and low customer satisfaction.

The algorithm is certainly something to think about. It very well may lead to better financing availability and a much more profitable lending relationship.