What if I told you that getting a college degree provides no long-term financial benefits unless you get married? What if I told you that the typical single business owner is wealthier than most married couples who don’t own businesses?
Though both of these statements might seem far-fetched, statistically, both are true.
For the past few years I’ve been obsessed with understanding how wealth is built in the United States. I even published a book about it. Despite this, my quest hasn’t stopped. I recently ran a bunch of regressions on my favorite wealth dataset (the Survey of Consumer Finances) and I was a bit surprised at what I found.
The punchline is that the two most common paths to wealth in the United States are:
- Marry a college graduate (or get married as a college graduate)
- Start a business
Individuals in these two groups have more wealth than all the other groups I examined. But before I explain why this is true, let’s start by discussing what predicts wealth in the first place.
What Predicts Wealth?
When it comes to predicting who has wealth and who doesn’t, the first thing I considered was age. After all, age is highly correlated with wealth accumulation because older individuals have had more time to work, save, and invest their money.
For my first model, I simply regressed U.S. household net worth against different age cohorts (<35, 35-44, 45-54, etc.) in the 2022 data to see how well age predicted wealth. Thankfully, the results fit my expectation and were all statistically significant.
Model 1: Age
As you can see in the table below, the model predicted that the median net worth for someone under 35 (the baseline) was $2,228.03. If we wanted to find the median net worth prediction for those aged 35-44, we would simply multiply this baseline value by the multiplier on that row:
So, for those age 35-44, the model would predict a median net worth of $29,610.52 [$2,228.03 * 13.29]. For those age 65-74, the model would predict a median net worth of $231,314 [$2,228.03 * 103.82]. And so forth.
Were these predictions close to the actual values? Sometimes. For those 65-74, the predicted median net worth was $231,314 while the actual median net worth was around $266,000. For those 35-44 the prediction was $29,611 while the actual was closer to $91,000. So hits and misses.
Thankfully, we can improve our predictions by adding more variables to the model.
Model 2: Education
The next thing I decided to add was educational attainment. After all, we’ve been told for decades that going to college is a financial home run. Well, is it? According to the data, yes.
Those who received a college degree had a predicted median net worth that was 2.92x higher than those with only a high school education (the baseline). More importantly, this was statistically significant at the 1% level:
The predicted median net worth of all U.S. households under 35 with a high school education (the baseline) was $1,961.85. For those with a college degree, that number shot up to $5,728.60 [2.92* $1,961.85].
While having more education can help you on the upside, having less of it can definitely harm you on the downside. As you can see in the table above, individuals with no high school education only had 13% (0.13x) as much wealth as their high school graduate counterparts. This was true regardless of age, since we already controlled for age in our regression.
But we can improve this model even further by considering its structure. You have to remember that this is household wealth data. Because of this, we know that our results will be skewed if there is more than one person in a household. After all, a married couple is likely to have two earners compared to only one for a single member household.
To adjust for this, let’s add marital status.
Model 3: Marriage
When we add marital status to the model, we see a few big things. First, the multiplier on marriage is 27.15! This means that a married couple has about 27x more wealth than their single counterparts with the same age and education status. Second, because of this huge marriage multiplier, the baseline predicted median net worth (for single high school graduates under 35) is now only $356.63:
Unfortunately, marriage explains a lot of the net worth variation in the data, and it’s statistically significant.
In addition, the multiplier on a college degree dropped from 2.92 (in Model 2) to 2.21, though it’s still statistically significant. This got me thinking, what if we adjust for the interaction of marriage and educational status. This would let me know whether it’s just a college degree that impacts wealth, whether it’s just marriage, or whether it’s a combination of the two.
Model 4: Education * Marriage
When we include the interaction effects of education and marriage in the model, something shocking happens—the wealth multiplier on having a college degree declines below 1 and is no longer statistically significant! I’ve highlighted both of these changes in the table below. As you can see, the multiplier on a college degree dropped from 2.21 (in Model 3) to 0.89 while losing all statistical significance:
This means that, statistically, the wealth benefit of having a college degree is near zero…unless you get married. Think about that. Someone with a college degree who isn’t married has the same predicted median net worth as a single high school graduate.
We know that college degrees are correlated with higher incomes, but those gains are typically offset by debt and a four-year delay into the workforce. While high school grads can start building wealth at 18, many college grads don’t start until 22 and are already a bit behind (i.e., in debt).
This suggests that one of the primary benefits of college isn’t actually the skills, but the fact that you get to meet lots of people and (hopefully) pair off with one of them at some point in the future. This dual high-income household formation is what helps to create this wealth premium.
This isn’t just conjecture either. There’s academic research showing how assortative mating can lead to higher income inequality (which also contributes to rising wealth inequality). One writer even argued that this dynamic creates an informal American caste system:
The primary caste division that I’m pointing to is often called “the Professional Managerial Class (PMC),” and that’s a good enough label for the central division.
Essentially, we see roughly ~70-80% of the population that largely don’t have undergrad degrees, are fatter, shorter, have fewer marriages, work fewer hours per week, and own less real estate.
And then we see the PMC, which has undergrad and grad degrees, high marriage rates, high home ownership, taller and fitter men, longer work weeks, and skinnier and more educated women.
What’s crazy is how extreme this effect is in the data. If you are a single college graduate, your net worth is not statistically different from a single high school graduate. But get married and it’s likely to be 35.6x higher! We get to that value by multiplying the college degree multiplier (0.89) by the marriage multiplier (8.19) by the college degree*marriage interaction multiplier (4.89) [0.89*8.19*4.89 = 35.6].
In other words, if a high school graduate gets married, their predicted wealth increases by 8.19x (the marriage premium). But if a college graduate gets married, their predicted wealth increases by 35.6x, over 4x more.
I know what you might be thinking. How much of this college * marriage effect is just rich kids marrying other rich kids? Well, we have a solution for that, and it’s called adjusting for inheritance.
Model 5: Inheritance
When controlling for households that had any inheritance, the model suggests that predicted wealth increases by 4.98x for those who had one (highlighted below). Unsurprisingly, inherited wealth is a predictor of current wealth:
But, even after adjusting for inheritance, our total college marriage multiplier still exists at 28.3x. Once again we get to this by multiplying the college degree multiplier (0.72) by the marriage multiplier (7.00) by the college*marriage interaction multiplier (5.62) [0.72*7.0*5.62 = 28.3]. Though this overall multiplier declined from 35.6x (in Model 4), it demonstrates that there’s still a real benefit to being married to a college graduate (or being a married college graduate yourself).
But, marrying a college graduate is just one proven path to wealth. The other is business ownership.
Model 6: Business Ownership
When we include whether an individual owns a private business in the model, a new (statistically significant) high wealth contender emerges (highlighted below):
This new model suggests that a single business owner would have about 9.9x the wealth of a single high school graduate who doesn’t own a business. Once again, this is after controlling for age, inheritance, and education.
Additionally, after adding the effect of business ownership, our model now only predicts a 19.1x wealth increase for being a married college graduate [0.60*5.76*5.53 = 19.1] relative to a single high school graduate (the baseline).
This is a big decrease from 28.3x (in Model 5) because that model was including the wealth impact of business ownership itself in that college educated marriage premium. By adjusting for this, this premium now shows the wealth impact of being married and college-educated for those that do not own businesses.
But there’s one final adjustment we should make to improve the model further—looking at the interaction between business ownership and marriage itself.
Model 7: Business Ownership * Marriage
When we add the interaction of marriage and business ownership, a few big things change. First, the premium on being a single business owner sky rockets to 36.5x from 9.9x in Model 6 (highlighted below). Second, the interaction between being a business owner and married is less than one, which means that the predicted wealth boost of getting married overwhelming goes to non-business owners. Lastly, the college degree premium has decreased further (and remains statistically insignificant):
Putting it all together, this model (which is our most accurate one yet) highlights the two primary paths to wealth—being married to a college graduate (or being a married college graduate yourself) and business ownership.
In this final model, the overall married college graduate premium is 22.2x [0.54*6.44*6.39 = 22.2x]. So while a single college graduate shows no statistically significant difference in wealth (compared to a single high school graduate), a married college graduate is expected to have 22.2x more net worth!
The other high wealth path is business ownership. Single business owners tend to have 36.5x more wealth than their single, non-business owner counterparts (from the multiplier in the table above).
How does marriage impact this? It increases predicted wealth, but only by an additional 16% (1.16x). This 1.16 comes from the marriage premium (6.44) multiplied by the business owner*marriage interaction multiplier (0.18) [6.44 * 0.18 = 1.16]. Taken together, married business owners are predicted to have 42.3x more overall wealth [36.5*6.44*0.18 = 42.3] than single, non-business owners.
This is an intriguing result because it suggests that while getting married does provide some wealth boost to business owners (1.16x), it’s nowhere near the wealth boost we see for college educated, non-business owners (22.2x) relative to their single counterparts. Business ownership is such a powerful wealth generator on its own that the marriage bonus doesn’t provide as much benefit as it does to a college graduate.
So, if you aren’t going to be a business owner, the next best bet for getting wealthy is marrying a college graduate (or becoming a college graduate and getting married).
Of course, getting married by itself isn’t necessarily the cause of the increased wealth. Though being married can lead to cost savings (e.g, shared expenses, tax benefits, etc.), what seems more likely is that the type of people who get married have other characteristics that are also positively correlated with building wealth.
For example, if you have a high income, you will be more attractive on the dating scene and more likely to get married, all else equal. So what’s important isn’t just getting married, but being the type of person who gets married. Because these types of people tend to be wealthier than their non-married counterparts for a variety of reasons.
But is there more going on with this story than meets the eye? Let’s see as we wrap things up.
Is It All Survivorship Bias?
While I would love to tell you that getting married to a college graduate or becoming a business owner is your guaranteed path to wealth, unfortunately, no such foolproof plan exists. Why? Because of survivorship bias.
When we see that married college graduates and business owners are wealthy, we are only looking at the successful marriages and successful businesses in the data. What we have overlooked are all the divorced college graduates and failed business owners who should bring the median values downward, but don’t.
The mechanism of failure causes this because as soon as you fail you leave the desired group. Divorce makes you single and, therefore, only impacts the overall wealth statistics of single people. A failed business turns you into a “non-business owner” and only impacts the overall wealth statistics of non-business owners. Do you see the problem?
But it’s not all survivorship bias. How do I know? Because if it was, we would see some of it in the data. There would be lots of struggling business owners who failed to build wealth. There would be lots of married people who failed at it as well. But because the effect sizes are so strong, it suggests that something real is going on here that helps build long-term wealth.
This doesn’t mean that you should get married (as a college grad) or that you need to start a business. But making choices that are more like what those people do is likely to benefit your finances for decades to come.
Thank you for reading!
If you liked this post, consider signing up for my newsletter.
This is post 491. Any code I have related to this post can be found here with the same numbering: https://github.com/nmaggiulli/of-dollars-and-data