I already outlined a few interesting data tidbits that are in Federal Student Aid’s annual report (large PDF), but the report also has some information that is worth taking a closer look at in its own post. One of these areas is student loan default rates .

Loan default rates are one of the few federal performance metrics that give any sense of whether students are struggling with too much to repay in debt. This measure is usually presented in what is known as a cohort default rate. This is calculated by establishing a cohort for each year, which is defined as the number of individuals that took out student loans and left their institution (through graduation or dropping out) in that year. The default rate is then calculated by dividing the number of students from that cohort who default on their loans by the end of the next federal fiscal year by the total number of borrowers from that group. In other words, student loan borrowers who left their institution in 2007 are tracked until Sept. 30, 2009, the end of the federal fiscal year, but not further.

While this measure provides some insight into student loan repayment success, it also has numerous flaws. Chief among them is that cohort default rates say nothing about long-term repayment prospects, making it impossible to see if failing to pay back loans is something largely confined to the first few years after leaving an institution, or if it grows over time. (This 2007 Education Sector report talks more about cohort default rates.)

The annual report released yesterday by the Federal Student Aid Office, however, provides some interesting longer-term measures known as the Cumulative Lifetime Default Rate (CLTDR). In contrast to the cohort measure, which treats all borrowers the same regardless of how many loans they took out, the CLTDR calculates the percentage of loans that entered repayment in a given year and then sees what percentage of them had defaulted by different points in time. In other words, the CLTDR for 2003 loans as reported in 2009 shows the percentage of loans that entered in repayment in 2003 that had defaulted by the end of the 2009 fiscal year. The annual report also has data for the same group of loans over several years, making it possible to see how much the percentage of defaulted loans increases over time.

The table below shows the CLTDR for the 1999 through 2007 loan cohorts by the number of years after that group entered repayment.* The black shaded part represents data points that have yet to be collected because insufficient time has elapsed.

Not surprisingly, the data show that longer measurement windows results in higher default rates. Only 5.5 percent of loans that entered repayment in 2003 defaulted within two years, but by 2009 that rate nearly doubled to 10.9 percent. With the exception of the 2007 cohort, this same pattern seems to hold for the other loan groups as well. But this substantial increase over time raises the question of whether this is the result of a rapid increase in defaults, followed by a leveling off, or does it reflect a more prolonged and steady expansion?

To answer that question, the table below shows the  year-to-year percentage point changes in the default rate. The 1999 and 2007 loan groups have been excluded because neither has multiple data points.

As the data show, the most pronounced percentage point increases in the CLTDR do occur in the first few years after entering repayment, with most groups showing annual increases of around 2 percentage points in their default rate. This then slows to a change of about 1 percentage point in the fifth and sixth years. But even though the rate of change does slow down over time, it is still around 1 percentage point a year, even six years into repayment. This suggests that student loan default is not a short-term worry, but something that persists for a long period of time.

At the same time, the high default rates displayed here should be cause for concern, especially among the most recent group. For starters, it’s likely that looking at loans rather than borrowers in terms of defaulting actually undercounts the problem. Consider the following hypothetical group of two borrowers with eight loans total, four each. Under the CLTDR method, if a single student defaults on one of his or her loans, then the default rate is one out of eight, or 12.5 percent. By contrast, if one student defaulted on any of their loans, the cohort default rate for that population would be 1 out of 2, or 50 percent. This explains why the Department’s estimates for cohort default rates measured under a three- or four-year window are somewhat higher than CLTDR examined for the same period of time.

Just like all other default measures, the CLTDR is not without its flaws. But until the Department provides better longer-term cohort default rate data, it is one of the few longer-term metrics available.