Monday, July 07, 2014


Cherry Picking

You know the problem with serving the needy?  They’re just so...needy.

Several Silicon Valley startups have decided to cherry pick the least needy students and make a buck from them.  They’re designed around identifying the recent graduates most likely to repay their loans, and offering them preferred financing.  (Hat-tip to @tressiemcphd for catching this one.)

You’d think such a thing wouldn’t be necessary.  In any rational universe, students would be able to refinance their loans without much trouble.  (Taken farther, in a rational universe, they wouldn’t need loans in the first place.)  But that isn’t always possible, which leads to a market opportunity.

That’s not to say that the startups have necessarily done their homework.  Data-analytics types will feel their eyes roll back in their heads when they read quotes like these:

“Intuitively, it just makes sense,” said Paul Gu, co-founder of Upstart. “The same sort of person who is going to care about their GPA in college is later going to care about their FICO score. That same person tends to be somebody who is very organized and is going to be disciplined with their money.”

Winners are just so much more efficient than losers.  

In a way, I’m grateful to the startups for demonstrating the basic flaw in treating loans as aid.

The point of “aid” is to help the people who are least able, financially, to attend college.  Cherry picking the people with the highest likelihood of repayment should, all else being equal, make the remaining public program look that much worse through a process of adverse selection.  Take away the easy cases, and you’re left with the hard cases.  

In other words, a sort of short-term rationality would lead to a greater long-term irrationality, because aid and cherry-picking operate by different logics.  Aid is based on helping those who need it.  Profit, here, is based on finding the ones who don’t really need it.  If the startup succeeds -- a big “if” -- then I predict we’ll start seeing ideologically-based victory dances celebrating the superior efficiency of the private sector, and ignoring its roots in cherry picking.  

I’m all for efficiency in the service of a coherent mission.  But efficiency for its own sake -- in econ 101, I remember learning that profit derives from identifying inefficiencies -- can easily send you in directions contrary to your mission.  The insidious part of that is that it happens step by step, and each step seems obvious or inevitable at the time.

From the perspective of one of the lucky few students to whom the startup reaches out, the appeal is obvious.  You get a better interest rate, and you replace a loan that can’t be discharged in bankruptcy with one that (presumably) can.  You may even get better customer service.

But as a substitute for a meaningful financial aid policy, this is nuts.  

In a way, it’s reminiscent of Sebastian Thrun’s statement that students from non-elite universities just weren’t capable of handling the demands of MOOCs.  A disruption that was supposed to save higher education turned out to work only with the students who least needed the help.  As for everyone else, well, on to the next thing.  Only the worthy shall be disrupted.  The rest can continue to flounder in public institutions whose funding pales next to what any self-respecting venture capitalist could blow on the next variation on instagram.

I can’t exactly condemn the startups for offering a service that should be freely available anyway, even if their methods seem a little, uh, undercooked.  But the sheer audacity of the business plan should give the rest of us pause.  What if we were to direct that kind of creativity to helping the people who actually need help?  What if we decided to disrupt economic polarization, instead of universal programs?

Yes, it’s harder.  Those of us on the ground have known that for a very long time.  Cherry picking our best successes, and then blaming us for what’s left, just makes it that much harder.

<< Home

This page is powered by Blogger. Isn't yours?