Some Thoughts on the use of the Survey of Consumer Finances (SCF) to study Student Debt
A recent Federal Reserve Bulletin article provided some interesting statistics on student debt. This post summarizes the numbers in the FRB article and discusses the limitations of the use of SCF to study issues related to student loans.
Summary on Student Loan Analysis in the FRB article:
The analysis in the FRB article on student debt pertains to “young” households defined as households where the head of household was younger than 40 at the time of the survey.
The main student debt results reported in this study are as follows:
- Percent of young households with student debt went from 22.4 in 2001 to 38.8 percent in 2013.
- Conditional student debt mean went from $16,900 to $29,800.
- Conditional median student debt went from $10,500 to $16,800
The Distribution of Student Debt Balances for the Young Household Population is as follows:
|Distribution of Student Debt Balances Reported in FRB Article|
|2001 SCF||2013 SCF||Diff.|
|Less than $25,000||78.2||64.2||-14|
|Between $25,000 and $50,000||15.6||17.1||1.5|
|Between $50,000 and $100,000||5.6||13.2||7.6|
Numbers in table above are taken form pie charts in Figure B of page 26 of the FRB study.
- Share of student debt held by young households went up for households making less than $30,000 from around 12 percent to somewhere in the mid 20 percent. The share of student debt held by people making $30,000 to $60,000 per year went down from around 40 percent to around 28 percent. The share of student debt held by individuals with households making more than $60,000 went up slightly to approach 50 percent.
- Around 70 percent of families with student debt had one member of the household with a bachelors degree or higher.
- Young families with student debt pay around 3.8 percent of their household on student loan repayments. One reason why this number is so low is that some households can defer payment on student loans for a variety of reasons.
Comment One: Other studies which examined large databases of recent graduates have reported much larger student debt burdens the FRB study.
Stats in Brief U.S Department of Education
- This paper found that the percent of BA degree recipients in public 4-year institutions with a repayment burden greater than 12 percent of income rose from 14 percent in 2001 to 26 percent in 2009.
- In private non-profit institutions this number went from 25 percent to 39 percent.
Amazingly, this increase in payment burdens occurred during a period of modest interest rates.
Comment Two: The under-40-household population is too broad to be useful to study the impact of the rapid increase in student debt mentioned in comment one. A person who was 40 has likely been repaying their undergraduate loans for 18 years and their graduate loans for at least 10. These older “young” households were not affected by the recent increase in student debt.
Perhaps this issue could be addressed with the chart below but the sample size of the SCF is a significant limitation.
|Student Debt Burdens|
|Age||2001 SCF||2013 SCF|
|31 to 39|
Comment Three: The sample size of groups of households on the SCF defined by different levels of education, age, income and other issue is very small. The SCF covers the entire population with 4568 cases from a general survey and 1458 cases from an IRS sample of high-income households. Since higher income households tend to be older I suspect that the IRS sample under sampled the young households that are the focus of this study. (I can’t answer this issue authoritatively. Readers interested in this issue need to contact the FRB staff.)
If 30% of the SCF households are under 40 the incidence of student debt numbers are based on a sample of around 1,800. Around 38.8 percent of the population has a student loan the mean and median loan figures are based on a sample size of around 700 or less. Around 28 percent of people over 25 have a BA degree. But the number of households with a BA might be a bit lower because of a marriage effect. Hence, there may be around 200 households in the sample with student debt and a BA. Less than half of these households have a household head under age 30.
Around 20 percent of students go to the more expensive private non-profit institutions. Hence I expect the SCF sample contains at most 40 households with people who went to some private school. It is likely around 10 people in this survey have student debt over $100,000.
The SCF sample sizes are small. The results are likely very imprecise. I suspect micro data on student loans would provide a more accurate view of student debt in America.
Comment Four: The Survey of Consumer Finances contains a lot of information on household assets, debts and decisions. It would be useful to categorize young households on the basis of their amount of student debt and look at the impact of student debt on financial decisions and outcomes. One categorization would involve households with no student debt, <$30,000, $30,001 to $75,0000, and greater than $75,000. I am interested in how these household differ and on the changes in these groups in several variables including — home ownership, credit card interest and consumer debt totals interest rates on different types of loans, contributions to 401(k) plans and IRAs.
Student debt is likely to impede the growth of wealth for households through two channels — delayed home purchases and reduced 401(k) contributions.
Again, the sample size of the SCF limits the amount of information, which might be obtained from this exercise.
Comment Five: The report states that “highly educated families continue to hold the majority of education debt in 2013. Nearly three fourths of education debt is held by families where one member has a BA degree or higher.”
It is a misnomer to call a person with a BA degree highly educated. A BA degree is now required for a lot of entry level jobs that did not formerly require a college education. Also, the unemployment rate and underemployment rate for current College grads are around 8.5 percent and 16.8 percent respectively.
Concluding Thoughts:. The SCF is not the most useful source of evidence on this topic. Some additional tabulation based on SCF data might prove interesting but researchers interested in student debt issues should consider other sources of data.