The Distribution of Total Education Debt among U.S. Households

How big are U.S. households’ education loans?

The topic of education debt and its impact on U.S. households has attracted more interest with Senator Elizabeth Warren’s recent proposals for a publicly-subsidized debt relief plan. This memo is the first in a series examining the impact of education debt on household balance sheets and cash flows, using data from the Survey of Consumer Finances1.

In this memo, I examine the distribution of household education debts. I examine the prevalence of education debt, and the overall size of education debts held by households.

Findings

Prevalence of Education Debts

Many U.S. households carry education debt, although most do not. The data suggest that about 22% of US households owe any money on educational loans. Typically, these debts helped cover the household heads’ education. Slightly more than 19% owe money on their own loans, and just under 4% owe money for their childrens’ education. In other words, just over one-fifth of American households stand to benefit from an education debt reduction scheme.

Distribution of Education Debts

The median educational debt-bearing household owes $19,000. One quarter owe less than $8,310, and another quarter owe more than $41,600. The top ten percent of debtors carried more than $80,000 in student debt The figure below depicts our estimates of total education debts among debt-bearing households:

Distribution of education debt loans among education debtor households, United States, 2016. Source: Federal Reserve

Note that the x-axis has a custom scale. Had we used evenly-spaced categories, the figure would have a strong right skew. Six-figure education debts are quite rare among educational debtors, and are very rare in the population at large. A proposal to eliminate up to, say, $50,000 in student debt would benefit a large majority of houseohlds. Keep in mind that these are houseohld figures, which would consolidate the debts of two married people would enjoy even more relief if debt relief is provided on a per person basis, as is the case with Senator Warren’s most recent proposal.2

When interpreting these figures, remember that this is the prevalence of debt among the 22% of households with education debts. So, for example, the 10% of education debtor households who owe more than $80,000 corresponds to 2.2% of the overall country. Maintaining a cognizance of this point is important, because it means that people with more than, say $40 thousand, is in fact a rarity in society-at-large.

First Impressions

I’m sure more proposals for educational debt relief will emerge over the course of the Democratic primary. There might be quibbles over who would qualify for it. Proposals will differ over whether to offer relief in the thousands or tens of thousands, and how much maximum relief to offer. There are limits to which conclusions we can draw.

Many Will Be Affected

The data do make it clear that education debt relief is a program with a decently-wide franchise, in the sense that it affects about as many households as a major entitlement program. Presumably, the program would cost less than a major entitlement, because debt relief is generally a one-shot event, as opposed to a continuing expense. If this debt relief program comes without structural changes to education pricing or lending, then there are some prospects for the student debt problem to just start up again, which might cause subsequent generations to ask for this precedent to be repeated. Without concurrent structural adjustments to the industry, then one might ask whether debt forgiveness is just an economic payout to a targeted constituency.

Most Debtors Seem to have Manageable Debts

It seems like most people have relatively modest education debts, around the size of the cost of a new car. I would assume that these kinds of debts could easily be managed by an employed, college-educated household over a five- or ten-year term. Of course, not everyone who takes out debts graduates with a degree, and not all degree-holders maintain employment. These are all questions that can be probed further with this data, if there is interest.

First Reactions After Seeing Data

My first exposuree to these data have led me to pause in supporting higher education debt forgiveness. My initial reflex is to want these kinds of programs, because I believe that unaffordable higher education is a problem. However, I am not convinced that relieving past debts, especially debts this large, would address my reasons for wanting more government investment in education access. I have a lingering suspicion that proposals like Warren’s would benefit a lot of wealthy people who purchased very expensive schooling. I am less concerned with this “problem” – I see this behavior as an already-rich person’s attempt to purchase their way up the ladder. I’m more concerned with the kind of students we serve at CUNY, for whom a $7,000 tuition is a heavy lift because they lack the parental support of their more privileged counterparts.

In a world of limitless resources, I’d say great — let’s do it all. However, political and in turn fiscal realities mean that certain efforts at progressive economic reform will have to hit the chopping block. My first reaction is to think that this is good candidate for early cuts.


  1. Federal Reserve Bank (2017) Survey of Consumer Finances, 2016. Online database. https://www.federalreserve.gov/econres/scfindex.htm
  2. It is possible to make more precise estimates from the data, but I will bracket the development of precise evidence here. I can follow up if anyone else is interested.

For More

You can download the R Markdown file used to generate these results from Open Science Framework. The data used in this analysis is available for download here.

I Bought My Book’s Digital Distribution Rights

I used my royalties to buy my book’s digital distribution rights, and am now selling it for $2.99 on Amazon Kindle.

The biggest mistake that I made in negotiating my first solo book contract was not thinking about selling price. When you are an Assistant Professor, you are so focused on getting tenure that you don’t think about book prices. You need peer acceptance (read: letters), and it is possible to get a free copy to the main players in that process. I just wasn’t focused on my book’s list price.

After I got tenure, my book’s price became a sore spot for me. The book was listed at $60 a copy. I’m sure that, in a strict economic sense, that price renders the maximum total book revenue. It hits price insensitive parts of the book market, like libraries and mandated student purchases, and milks each individual buyer as if they were six people buying $9.99 Kindle editions. A book that’s mostly dry-ish household finance data analysis is only going to sell so many copies, so the revenue maximizer tries to get as much out of each potential customer.

The problem is that profit maximization and audience reach are two different things. Who was going to pay that much for an academic book on household finance? Not me. I could not bring myself to encourage people to buy a book that expensive, and there was no chance that I’d ever require my students to pay that kind of money. You can encourage people to ask their librarians to order the book, but you are going to lose tons of readers going through libraries. So many intentions to read die on the inter-library loan vine.

So I used some of my book royalties to buy digital distribution rights from my publisher. I am selling an Kindle edition of my book for the minimum allowable price: $2.99, and will be distributing free copies through my web site in the fall. Along with free electronic copies of my book, I will distribute single chapter PDFs, high quality media for use in classroom slides, and chapter summaries.

The book is a study of financial insecurity in the U.S., and its relationship with the high cost of wellbeing-essential products due to America’s neglected public services. Here an interview about the book:

The book’s chapters can be used as a basis for a range of interesting class discussions, like:

  • Are people really worse off today?
  • Exactly how have people’s money situations gotten worse over the past few decades?
  • Who counts as “rich”, and how much money do they really have?
  • Are poor people really so poor?
  • How many of us are ultimately reliant on government aid?
  • Why can’t people just tighten their belts in response to financial problems?

And many more. Check it out!

Zero Sales Businesses

A substantial proportion of entrepreneurs aren’t actually selling anything.

This analysis is interested in the prevalence of paper entrepreneurs, people who self-identify as self-employed or operating their own business but whose business registers no sales. An example of this kind of entrepreneur might be an unemployed person who works as a “consultant” (and lists themselves on LinkedIn as such), but do not have any clients. Another might be someone who claims to have an actively-managed business in order to construe and write off household expenses as business expenses in order to maximize tax deductions.

I expected to see a secular rise in the prevalence of no sales businesses for at least three reasons. First, I was under the impression that America’s tax system and business enviornment evolved to encourage small businesses formation. Many “zero sales” businesses might be startups at their very earliest stages of development. Given overall rates of entrepreneurship seem to have remained unchanged1, I expected the pool of small businesses to be have proportionally more startups than in the past.

Second, I presumed that the evolution of taxes would create incentives for households to declare business ownership and push household transactions through those businesses as a tax management strategy. I thought that either new regulations or laws created new opportunities to save money in this way, or that new technologies made it easier for people to recognize and capitalize on business deductions (e.g., Turbotax may alert people to the potential for business deductions, and give people a means to capitalize on them).

Third, the title of “entrepreneur” is more socially desirable than that of a retiree or unemployed, and I expected more people to assume these roles when creating public identities on web sites like LinkedIn or on their resumes. Given that job precarity and the forced “retirement” of older workers are widely-discussed as major trends, I thought there was a potential for this dynamic to raise the proportion of “zero sales” entrepreneurs.

Data and Methods

We use data from the Survey of Consumer Finances2 This analysis only considers entrepreneurs who own actively-managed businesses. The other group traditionally counted among entreprneurs – those who self-identify as self-employed – are not included because their business transacting data is not in the data set. We are interested in the percentage of households whose actively-managed businesses collectivelygenerate no sales. If a household has one business with sales, then they are not countes as zero sales entrepreneurs.

Findings

The figure below depicts the estimated percentage of business owners who make no sales. Since 1992, the proportion of actively-managed businesses that registered no sales appeared to bounce within the range of 8.8% (in 2001) and 11.8% (1995). I am not strongly confident that the large change from 1989 to 1992 represents a real change, or whether it is a byproduct of any differences in the survey or sampling mechanisms in these surveys.

Discussion

To some degree, this was an unanticipated finding to me. I found many reasons to expect a rise in these kinds of businesses. These failed expectations could be the result of faulty presumptions, including:

  • Zero sales businesses may not signal business startups
  • Tax policies or economic regulations did not change the ease of forming new businesses
  • Tax policies or technology did not change in a way that sufficiently encouraged enough people to form businesses for the purpose of tax management.
  • Comparatively few people usurp the role of “entrepreneur” to cover up unemployability

For More

You can download the R Markdown file used to generate these results from Open Science Framework. The data used in this analysis is available for download here.


  1. Cohen, J. N. (2019, March 1). Prevalence of Entpreneurship among U.S. Households, 1989 – 2016. Retrieved from <osf.io/q5xmu>
  2. Federal Reserve “Survey of Consumer Finances” Data from triennial survey, 1989 to 2016. Available for download at https://www.federalreserve.gov/econres/scfindex.htm

Personal Incomes from Entrepreneurship, 2016

The distribution of personal incomes from entrepreneurship, from 2016 Survey of Consumer Finances

This analysis examines the proceeds that entrepreneurs earn from their business enterprises using data from the Survey of Consumer Finances.

Data and Methods

We use data from the Survey of Consumer Finances1. We confine our attention to the subset of households that meet one of two conventional criteria for identifying entrepreneurs in survey data: (1) a head who reports a “self-employed” labor market status or (2) reported ownership of an activley-managed business. If either condition is met, the household is included in our analysis.

Entrepreneurial income is captured by two financial statement metrics:

  1. Self-employed income. Questions that ask self-employed respondents “how much [do you] earn before taxes in your main job”2
  2. Business Profit Taken. From questions asking “(In addition to salary,) how much [was] personally received from the business before taxes?”3

Results

Figure 1 (below) presents a bar chart depicting the distribution of entrepreneurship-related income across all households who have a self-employed head or an actively-managed business. Recall that our sub-sample is limited to entrepreneurs only. Some observations of these findings:

  • About 29% of American entrepreneurs earn no money from their businesses
  • The vast majority of American entrepreneurs (59%) did not earn enough from their businesses to sustain a livelihood above the poverty line (above $15,000).
  • About 15% earn more than a roughly median household income of $50,000.
  • A smaller minority (2%) earned more than $200 thousand in 2016.

Discussion

Entrepreneurs are often described as a comparatively wealthy group, and entrepreneurship is often portrayed as a vehicle of wealth. Only about one-seventh of U.S. households participate in entrepreneurship4 The vast majority of these businesses could only sustain the most modest livelihood, if at all. Only a small proportion of this narrow subset of American society earns enough to be counted among high income households. Overall, I read these data as suggesting that few people succeed in generating high income from their businesses.

For More

You can download the R Markdown file used to generate these results from Open Science Framework. The data used in this analysis is available for download here.


  1. Federal Reserve “Survey of Consumer Finances” Data from triennial survey, 1989 to 2016. Available for download at https://www.federalreserve.gov/econres/scfindex.htm
  2. SCF variable x4112 where Head #1 is self-employed, and x4712 where Head #2 is self-employed.
  3. Variables x4131, x4731, x3337
  4. Cohen, Joseph N. 2019. “Prevalence of Entpreneurship among U.S. Households, 1989 – 2016.” OSF. March 1. osf.io/q5xmu.