Marriage Now Belongs to the Well-Off and Educated
Why are lower income, less educated people, especially women, marrying less and divorcing more often?
Hey, everyone! My name is Jordan, and I have a Ph.D. in Economics from the University of Pennsylvania. I want to ask important questions and answer them meaningfully using the hard work economists have put into their research. This is my first Substack post doing just that. If you learn anything interesting from this, please like and subscribe!
Since the 1980s, marriage in America has undergone a pivotal transformation. While college-educated adults have had stable marriage rates and seen their divorce rates decline, those without a college degree have experienced the opposite: decreasing marriage rates and rising divorce rates. These trends emerged in the aftermath of the 1970s, a decade marked by a dramatic surge in divorce across the board. But by the early 2000s, a stark “marriage divide” and “divorce divide” had taken hold, especially for women, revealing that higher education now plays a central role in shaping who marries, who stays married, and who doesn’t marry at all.
Consider two women: one working a low-wage job with no college degree, the other a college-educated professional with a stable salary and benefits. For the poorer woman, the economic gains from marriage, which would include shared housing and pooled income, offer economies of scale that could substantially improve her standard of living. The additional household income from a partner might mean the difference between financial strain and basic stability. In contrast, for the wealthier woman, those same economic gains are less important; she already enjoys a level of comfort and independence that makes the gain from a spouse’s income relatively small.
Yet, it’s the poorer woman who is now less likely to ever marry and much more likely to divorce. With increasing income for those with a college degree, why would this ever be the case? This question has been at the forefront of many economic and sociology research papers for the past 20 years, with one paper documenting and crossing out some key potential explanations. In fact, this has been happening for other countries as well.
We have seen an increase in divorces for college-educated men married to non-college-educated women and the opposite occur for husbands with less education than their wives. Education serves as a general “proxy” (substitute) for lifetime income, but one paper also considers a wife’s percentage of a total couple’s earnings and finds that it no longer has a positive correlation with divorce. So, the divorce divide happening for women is primarily being driven by what I call “mixed education marriages.”
This begs the following question: what has economics and sociology research proposed as potential reasons for the marital and divorce divide? In this post, I am going to only focus on divorce, as marriage decisions could be based on the risk of divorce itself and warrant an entirely different post; we have an issue of “reverse causality” here, which means marriage decisions may depend on divorce trends and the other way around. However, since divorce is conditional on being married, it is easier to determine what exactly is influencing divorce trends, even if they may be partially influenced by how many people in an area are single (think along the lines of cheating or trying to find a new spouse).
I will go over a list of the research and logic behind a few popular proposed reasons for the divide, and I will show why some of these reasons do not entirely address the growing divide. Note that as someone who has contributed to this research, I am going to save my proposed idea for last, as I argue it is a large contributor.
Age at first marriage is lower for less educated people who don’t graduate from college, which means they make “worse” decisions
College-educated people want to invest more into their children’s success, and they haven’t seen a strong decrease in children born within marriage like non-college-educated people
Women have become increasingly more educated compared to men, which changes the kinds of marriages happening
Alcohol and drug consumption differs across income and education levels, leading to more disagreements and abuse
Unemployment is higher among lower income and education groups, and income volatility has increased over time — these cause more arguments
Divorce laws began changing in the late 1970s and have allowed poorer women married to richer men the ability to get a divorce while simultaneously disincentivizing richer women married to poorer men from getting a divorce
Age at First Marriage:
People often point out that adults without a college degree usually tie the knot younger than college-graduates. They think, “Well, isn’t it the case that people who don’t go to college get married at a younger age than people who do go to college? So they’re not as careful about who they are marrying?” It’s true that non-college couples still marry earlier (and this may explain a large portion of the current divorce divide we have seen since 2005), but that educational gap hasn’t really shifted since 1980. What has changed? Everyone, no matter their schooling, is waiting longer to get married. Therefore, we should have seen the divorce rates decrease across all groups, but this is not what happened.
If people today were still marrying at the same young ages they did in 1980, overall divorce rates would actually be even higher for all groups. In the graph below, I show what happens if people continued to marry at the same ages as they did in 1980 (as represented by the dashed lines), meaning the percentage of first marriages happening in each age group stays the same.
This shows us that had women married at the same ages since the 1980’s, divorce rates would have increased even more for the non-college group and almost remained the same for the college group. These results are consistent with what we see in an economics paper that shows that the increase in age at first marriage can explain around 60% of the decrease in divorce hazard. Why has the first marriage age increased over time? There are a few hypotheses; one paper argues that an increase in income volatility, or the amount of changes in income, over time is responsible, while other papers consider the impact of contraception (i.e. birth control pills, IUDs, etc.) and increasing gains for a woman being in the labor force. More on these ideas at a later time because there is a lot to unpack here.
We now see that the age at which people marry doesn’t explain the trend.
Investment in Children:
One economics paper hypothesizes the following: 1. Parents with college degrees pour more time and money into their kids than parents without degrees. 2. Because of that extra investment, college-educated parents are more likely to stay married for their children’s sake. The evidence for the second point is strong: college-educated couples do seem to let concerns about their kids weigh heavily when deciding whether to divorce; I find in the PSID that college-educated women with children are over 3 times less likely to divorce than those without children and that non-college-educated women are 2 times less likely.. The first point was partially addressed in another paper that found childcare investment increased among the college-educated since the 1990’s, potentially due to increased competition in college attendance. However, according the the PSID data, there isn't evidence that people's behavior toward divorce has changed over time with regard to children. Instead, we only see a decrease in divorce rates among college-educated women for both those who had children during their marriages and those who didn't.
Meanwhile, family-planning patterns have been changing in opposite ways for the two education groups. Among women without college degrees, births inside marriage have fallen sharply and births before marriage have become much more common relative to women with college degrees. While research has found that having children within a marriage decreases divorce rates, other research has found that bringing children into a marriage might actually increase divorce rates… but “causal inference” from these points are both hard to prove. Why? Because couples can make decisions on having children in anticipation of potential divorce and may also decide to marry because they had a child out-of-wedlock; these marriages can be fundamentally different.
I can only show how much the change in birth trends over time could potentially account for the divorce trends below, but I cannot provide any “causal evidence” (i.e. I can’t say these birth trends “cause” anything) for the change in divorce trends. The following graphs were meant to be used as guidelines during my own research, just as the age at first marriage graph was used above.
What we can see from the graphs above is that, yes, these changes in children born within and outside of marriages could potentially account for some of the changing divorce trend for women, but they cannot explain most of the change. Divorce rates would have been slightly lower for non-college-educated women in all instances, but divorce rates would have only slightly increased for college-educated women.
We now see that investment in children doesn’t explain the trend.
Women’s Increasing Education Rates
Since the 1980s, many more women have gone to college, and that shift makes us ask whether today’s female college graduates are the same kind of people as the smaller group who earned degrees decades ago. If the new crop of graduates has different strengths or personalities, the mix of qualities inside each education group along with the couples formed within those groups could look very different now. One possibility is that marriages in which neither spouse finished college now bring together partners with fewer “soft skills,” such as patience or conflict-resolution, which might lead to more fights. However, this hypothesis assumes that the average “skill” level of college-educated women has fallen as college attendance rose; if that were true, we would expect higher divorce rates for everyone, which contradicts the data.
In another sense, cognitive and social skills would have closer matches between the men and women in each group except for the non-college-educated women with college-educated men. We need more research in this area to truly determine the argument’s validity and determine if “closer matches” are occurring for the college-educated women marrying non-college-educated men. In other words, this argument is inconclusive. One study, however, looks at how people with different education levels pair up, finding that individuals who marry someone with less education ("marrying down") tend to have lower cognitive abilities compared to those who marry someone with equal education.
So, women’s increasing education rates most likely do not explain the trend — or at least it’s questionable.
Alcohol and Drug Consumption
Alcohol and opioid addiction have placed a heavy burden on low-income adults — especially men — over the past few decades. Overall drinking has fallen for every education group, yet opioid misuse has climbed among men without college degrees. In a survey I implemented for a current working paper, alcohol abuse showed up most often in marriages where neither spouse finished college, and it’s reasonable to think opioid problems hit this group hard as well. While the widening divorce gap is still driven mainly by mixed education marriages, this potential explanation is a contender for the purely non-college, low income marriages.
Now we see that alcohol and opioids are not responsible for the trend.
Unemployment, Income Volatility, and Wages
Sudden swings in a family’s earnings make divorce more likely, and there is some debate about whether these ups and downs have become more common over time. What has not changed is the chance of losing a job altogether; unemployment rates have stayed roughly flat and different among education groups. That tells us most of the new instability comes from “how much” people earn in a job, not from whether they have one. One study even suggests that shakier paychecks are one reason couples now wait longer before getting married; the same paycheck instability is less likely to influence divorce rates as it would already have encouraged many couples to delay marriage in the first place. Some research has not found an obvious increase in income stability over this time period. Income instability, if it is actually growing, would still fail to explain the increase in divorce rates among the college-educated men married to non-college-educated women, but it could explain the lower income marriages and lower education.
There is this idea that men today are “less marriageable” due to decreases in labor force participation (not the same as unemployment, where men would actively be looking for work) and income. In which case, we might see effects on the lower income marriages with this trend as well. The lack of labor force participation may be more attributable to initial marriage decisions, as 1. men who are married have not experienced quite as strong of a decrease in labor force participation over time and 2. men generally increase labor force participation and work hours soon after marrying.
Contrary to popular belief, unemployment does not explain the trend.
Changes in Divorce Laws
In the late 1960s the U.S. began shifting from a “fault-based” divorce, where one spouse had to prove adultery, abuse, or some other wrongdoing, to “no-fault” divorce. California led the charge in 1969, when Governor Ronald Reagan signed a law letting either spouse end a marriage simply by citing “irreconcilable differences” and scrapping fault rules for dividing property and setting alimony; Alaska, Kansas, and Oklahoma had already allowed one-sided (“unilateral”) filings, but California’s reform was broader and soon copied by other states.
Under the older system, both partners had to agree to split and one of them had to be officially “at fault,” so couples who privately agreed to divorce often staged phony allegations to satisfy the court. No-fault laws ended that charade: spouses no longer had to point fingers, though in some states they still both had to say the marriage couldn’t be saved. This is where we draw the line between “unilateral” no fault divorce and “mutual consent” no fault divorce. Under the former (and what California implemented), only one person has to file for divorce. Under the latter, both spouses must agree to “irreconcilable differences.” 30% of states ended with the latter mutual consent laws.
Below, I show what happens with the percentage of marriages that end in divorce when they begin under a fault divorce law or a unilateral, no-fault divorce law. I go further into this in my actual paper, but this serves as the motivation. I found that, on average, this divorce divide between mixed education marriages closes when unilateral, no fault divorce is introduced.
These unilateral no-fault divorce laws allowed non-college-educated, poorer women married to college-educated, well-off men who wanted a divorce to actually get a divorce. First, under fault-based divorce, a judge would grant a divorce only on the basis of adultery, cruelty, or desertion. If the husband didn’t privately agree to the divorce, the courtroom fight cost time and money, so it was mainly an option for people with resources, such as college-educated women. Second, if you did prove your partner was at fault, the judge could award you a larger share of the couple’s property, making the legal battle even more appealing to those who could afford it, like the college-educated women.
We don’t have precise data on how often these fault-based cases occurred, but literature hypothesizes they were more common than today, where most divorces are settled quickly and without a courtroom showdown. While the no-fault divorce laws reduced barriers for non-college-educated women, they slightly decreased divorce incentives for college-educated women. The model in my paper finds that these changes in laws can account for 37% of the changes in mixed education marriages.
Researchers still don’t agree on whether unilateral, no-fault divorce improves or harms people’s well-being. Only one paper has tried to weigh all the gains and losses of unilateral, no-fault rules against “mutual-consent” rules that require both partners to agree. It finds that wealthier men and women generally favor unilateral, no-fault divorce, while lower-income men and women prefer mutual consent. The model does really well with real-life patterns in how much couples work and earn over their lives along with lifelong divorce rates, but it leaves out two potential factors: (1) in no-fault settings, a spouse might negotiate concessions to stop a divorce, and (2) in mutual-consent settings, couples can face steep legal costs to prove fault. We need more studies to measure the welfare implications of these laws using different assumptions to get a consensus.
There are other studies that look into specific effects from the implementation of unilateral no fault divorce laws. One study found that unilateral divorce in the U.S. substantially decreased domestic violence and female suicides. Yet, another study found that the children exposed to unilateral divorce were more likely to be less educated and also divorce in adulthood.
Conclusion
Since 1980, changes in divorce laws have shaped much of the rise and fall in divorce rates, especially by making it easier for lower-income women to leave unhealthy marriages; so, we can argue that it explains the trend until 2005. But divorce is now falling for everyone, so the case for changing the laws back to requiring fault-based divorces isn’t as strong as it used to be in, say, the 1990’s. Either way, research has not come to a conclusion on whether no-fault divorce laws are good or bad policies, but the tradeoffs seem to suggest they are strongly beneficial to low income women.
Some state legislators have introduced bills to bring laws closer to fault-based divorce once again — Texas is currently considering a bill that would get rid of its no-fault divorce reason known as “insupportability,” Tennessee passed a bill that makes divorce fault-based for covenant marriages, and Oklahoma has a bill striking no-fault divorce that is stalled but still in consideration.