While browsing Reddit yesterday, I happened upon a post about how “millennial home ownership shrinks as student debt grows”. When I first saw it, the post was near the top of Reddit’s front page. When I took the screen cap below, it had 13,700 upvotes and nearly 3600 comments. I wouldn’t be surprised if it’s over 15,000 upvotes by now.
The source article — written by “Fred” at Financeography.com — makes the following argument:
At the beginning of 2016, the home ownership rate for those 30 and under sat at about 27.7%, the lowest it has been in decades. On the other hand, student loan debt rose to $1.2 trillion, though it has already surpassed the $1.3 trillion mark earlier this year. In a development which should surprise nobody, there seems to be a pretty clear correlation between the growing student loan debt Americans hold and the under 30 home ownership rate.
The article is accompanied by this graph:
The implication is that there’s a strong negative correlation between student loans and the ability for young adults to purchase a home. (A negative correlation means that as one of these variable increases, the other decreases. They move in the opposite directions.) Or, as the Reddit post claimed: “Millennial home ownership shrinks as student debt grows.”
This is a bullshit article using bullshit stats to reach a bullshit conclusion. It’s the latest in a long line of “millennials have it rough” stories that drive me nuts because:
- Their conclusions are false.
- They give tacit permission for young folks not become boss of their lives. (If things are so shitty, then it’s not their fault they can’t get ahead. Right?)
This article is crap, and it’s crazy that so many people find it compelling. Although I know I should let it slide — “OMG SOMEONE IS WRONG ON THE INTERWEBS!” — I devoted most of my Sunday to researching and writing a rebuttal.
Here’s my thesis: There’s no connection between student loans and homeownership rates. There’s certainly no negative correlation between the two. In fact, as we’ll see, there’s a moderate positive correlation. (But that correlation is meaningless.) Average student debt has actually plateaued and is declining, and homeownership for young adults is moving lockstep with most of the rest of the United States.
In short, Reddit and Financeography are dead wrong. Let’s look at why.
Homeownership Rates for Millennials
It took some digging, but I think I found the homeownership data used by Financeography.
On the U.S. Census Bureau Vacancies and Homeownership page, you can download a spreadsheet containing annual estimates of the housing inventory by age of householder (from 1982 to present) [37K Excel document]. Financeography grabbed a subset of the data. Their graph only shows homeowners under age 30, which is fine. But the graph is misleading because it only shows numbers since 2005, which is when homeownership was peaking for young adults (and everybody else).
Using the Census Bureau spreadsheet, I manually collated the data for people under 30 — but I went back to 1982, which is the first year for which the government has numbers. Here’s the base table, which includes the number of households headed by somebody under 30 (in 1000s), the number of homeowners under 30 (in 1000s), and the derived homeownership rate:
And here’s the full graph of the homeownership rate for young adults, from 1982 to present:
Looks a bit different, doesn’t it? When you’re not trying to lie with your statistics, you get a different picture. Yes, fewer folks under 30 own homes than they did ten years ago, but can you think of another reason this might be the case? A reason that doesn’t involve bullshit connections to student loans?
For added context, here’s a U.S. Census Bureau graph of homeownership rates in the U.S. for the same time period (divided by age group):
Homeownership rates have declined for everyone — not just millennials. In fact, young adults might not be the group that’s been hit hardest. (It looks like 45-54 year olds had a decline that’s at least as steep, although I didn’t calculate percentages to be certain.)
The decline in homeownership for millennials appears to be part of a natural cycle. And it’s certainly tied to homeownership in the country as a whole.
But what about student loans?
Average Student Loan Debt Per Person
Long-term historical data on student loan debt was almost impossible to find. The Federal Reserve Bank of New York has a couple of publications available online with info back to 2003. (That’s the data Financeology used to create its graph.) That’s because, according to a footnote in this Fed report, student loan reporting to credit bureaus before 2003 was unreliable.
So, we’re going to use a stand-in dataset. (Which is okay, I think. “Total student loan debt” doesn’t seem like a meaningful statistic. It’s like talking about home prices while citing the total value of the real estate market. It makes no sense.)
For my analysis, I’ve used data from the College Board, the non-profit group best known for administering the SAT exam. The organization also has a wealth of data related to student borrowing patterns since 1970. (In fact, it’s the only such dataset I was able to locate.)
For instance, here’s a recent table showing average aid per student over time (with data running from 1972 to present).
Once again, I downloaded the Excel spreadsheet with raw numbers in order to create my own table. Here’s the average amount of loans (from all sources) per full-time equivalent student (FTE) in 2015 dollars:
If you read background behind these numbers, student loans increase when the government makes new loan programs available. In 1994, for example, a new direct lending program was introduced. Borrowing jumped. (It doubled in two years!)
Today, the average full-time student does take roughly $7500 in loans every year. That’s a total of about $30,000 during a four-year college career. That’s a lot, no doubt. Whether that much education debt is good or bad is a debate for another day.
Here’s a full graph of this data:
I hope you’ll agree that it makes much more sense to look at the average loan amount per student than the total outstanding debt for the entire country.
When you examine student loans per peson, you get a completely different picture than if you focus on the total student loan debt in the United States. While the latter may be increasing — perhaps because more kids are going to college? — the former has experienced a modest decline after a longer plateau.
The Bottom Line
Now let’s get geeky.
In statistics, correlation coefficients measure the degree to which two variables are related. A correlation coefficient can range from -1.0 to 1.0.
- If its value is 1.0, that means there’s a perfect positive relationship between the two variables. As one moves up, the other moves up with it.
- If the value is -1.0, there’s a perfect negative relationship between the variables.
- As one moves up, the other moves down. A correlation coefficient of zero means there’s no relationship between the two variables.
The graph from Financeography shows a strong negative correlation between total student loan debt and homeownership rates for young adults, but only for the years between 2005 and 2015.
But this graph is telling a story that doesn’t exist. Or, more precisely, a story that doesn’t matter. Worse, it’s only providing a single chapter from that story.
If you look at the bigger picture — data from 34 years instead of 11 — and you choose to look at numbers that matter, the results are much different. Here’s a graph I made that combines two variables from 1982-2015: the homeownership rate for people under 30 and average loans per FTE.
When you calculate the correlation coefficient for the actual numbers in my graph, you get a modest positive correlation: 0.46. This isn’t strong by any means, but it is positive. As homeownership rates rise, so too do student loans. As student loans decline, so does the homeownership rate.
That’s totally different than what Financeography would have you believe, isn’t it?
Here’s the bottom line: Just because somebody cites some stats, that doesn’t mean they’re accurate — or that the correlations they’re trying to draw are actually meaningful. (Or that correlation implies causation.)
And just because something makes it to the front page of Reddit, that doesn’t mean it’s true. It just means redditors wish it were true.
I spent six hours researching and writing this article. There’s a lot of material that didn’t make the final edit, and some of you money nerds might find it interesting. For example, the U.S. Census Bureau maintains a website with all sorts of info on homeownership. They recently released their latest quarterly report on residential vacancies and homeownership [PDF]. At the website for the Federal Reserve Bank of St. Louis, you can explore tons of data related to housing. For instance, you can create (and customize) your own graph of the U.S. homeownership rate from 1965 to today. Lastly, don’t forget my April article about the history of the U.S. housing market, in which I shared overall homeownership rates going back to 1890.