Low marks for the Department of Education

In what can charitably called “gross incompetence”, the Department of Education has recklessly and massively underestimated the cost of key new income-based repayment student loan programs.  The earliest projections seem to have been off by at least a factor of 2 ($25 billion compared to $53 billion), and many think that number will grow more in the future.

The best case scenario is that taxpayers are out billions of dollars, with flaws in the programs’ designs being allowed to persist because no one knew there was a problem.  At worst, the future of programs such as Public Service Loan Forgiveness (which forgives the remaining loan balances of individuals who work in certain occupations with a public service component) or Graduate PLUS (the program which provides loans to attend graduate school) could be in jeopardy. 

{mosads}These are generally good programs which need certain tweaks to make them sustainable, but I worry that the sheer ineptitude highlighted in a recent report from the Government Accountability Office report could land them on the chopping block.  To be clear, this is not the result of a small statistical quirk or a well-intentioned assumption that turned out to be false.  This is statistical malpractice, and it was entirely preventable.

To be clear, forecasting the borrowing and repayment behavior of all students is not an easy task.  On top of detailed statistical and institutional knowledge, it requires making many assumptions about the future, some of which will turn out to be incorrect.  And even a very unexpected result does not necessarily invalidate a model.   

But the careful approach taken by most other government agencies stands in stark contrast to that of the Department of Education.  To name just a few of the most egregious errors, the models did not (1) account for inflation, (2) account for the fact that more people might enter a given repayment program while at the same time actively trying to recruit 2 million  additional borrowers, and (3) account for basic differences in the terms of loans that students have.  To call these rookie mistakes would be offensive to rookies.

I say this as someone who makes a living analyzing data (and therefore I am sympathetic to the difficulties of statistical modeling), and also as someone who has previously defended the Department of Education.  This is not defensible.  

So where do we go from here?  Some have (perhaps rightly) called for the budgetary projection function to be taken away from the Department of Education and placed in the hands of an agency with a better track record.  There is a faction on the right who want to eliminate the department entirely, a position held by at least half of the GOP primary, most notably President-elect Trump.  

The reform I would personally like to see come from this mess is an increase in the statistical transparency with which the Department of Education (and for that matter all other agencies) operates.  There is fundamentally no reason that the computer code which generates these important projections should not be made freely and publically available.  

While I, and the many other researchers who analyze education policy, always want more detailed data, I recognize that there are confidentiality constraints that are in place for a reason.  There is nothing confidential nor proprietary in the computer code.  I doubt very much that the Department of Education would be in such an embarrassing position today if they had been forced to publish their programs online years ago.  This sort of transparency forces people to be more careful in their work (my code is certainly cleaner and better documented when I know someone else will be looking at it), and also allows those writing the code to benefit from the knowledge and expertise of researchers around the world.  One need only look at the resounding success of open source computer software to see the benefits of such a policy.  

Given that the stakes of these calculations are high, an unexpected shortfall will necessitate cuts to other areas of government, there is a compelling public interest in doing the best job possible forecasting the impact of any policy.  In short, #FreetheCode. 

Douglas Webber is an Assistant Professor in the Department of Economics at Temple University.  He studies labor economics and the economics of higher education. Follow him on Twitter @dougwebberecon.

The views expressed by authors are their own and not the views of The Hill.