Research and policymaking on student loans has long been hampered by significant limitations in data availability and quality. University of Michigan economist Susan Dynarski provocatively compared this state of affairs to a large company that holds a huge portfolio of loans, many of them in default, but cannot get their loan division to cough up the data needed to assess the scale of the problem and what to do about it.

In the absence of better data from the US government, which makes the vast majority of student loans, researchers have been forced to rely on a variety of data sources, all of which have important limitations. Some data sources rely on surveys that may suffer from misreporting, whereas others are based on more accurate administrative data but are missing important information, such as borrowers’ income.

The wait for better data is finally over. A new paper by Adam Looney of the Department of the Treasury and Constantine Yannelis of Stanford University examines linked administrative records on a nationally representative group of federal student loan borrowers.

The key innovation is the linking of information on student lending maintained by the Department of Education with income data held by the Department of the Treasury. The only limitation is that the data do not include private loans, which make up less than 10 percent of student borrowing.

A recurring theme of this important new paper is that borrowers’ loan balances are a poor proxy for whether they are in trouble. The authors find that the recent troubling increase in defaults is driven largely by borrowers from for-profit and community colleges.

Almost half of for-profit borrowers default within five years of entering repayment, despite borrowing less than students at traditional four-year schools and graduate programs. Borrowers from these institutions have long struggled to repay their loans, but the increase in enrollment at for-profit schools and in borrowing levels at community colleges have pushed up the overall default rate.

The flip side of this coin is that many borrowers with significant amounts of debt have relatively high incomes. Looney and Yannelis report that borrowers in the top 20 percent of the income distribution, with a median income of about $96,000, hold 36 percent of all federal student loan debt. Many of the high-income, high-debt borrowers have graduate degrees: 68 percent of student borrowers with balances over $50,000 borrowed to attend graduate school.

Many of these findings are consistent with existing research on student loans. But they put to rest lingering questions related to data quality and allow for much more fine-grained analysis of an issue that continues to capture the public’s attention.

These new data will surely continue to produce new insights. Hopefully they will also refocus the public discussion on borrowers truly in financial distress—how to get them out of trouble and how to prevent future borrowers from finding themselves in a similar position.