How Have They Changed, and Why
By Ricardo Mejias Ph.D., Economist
June of 2013
Most of us have heard the frequently repeated statement that 50% of all marriages end up in divorce. This gets passed from one media “source” to another without anyone ever checking the original source of the information. So we decided to check with the ultimate authority on all things demographic: The Unite States Census Bureau.
Typical of statements often repeated in the media, the 50% turns out to be an oversimplification that does not begin to tell the important story about divorce rates. There are much more interesting numbers that tell us how the divorce rate has changed over the decades and suggest the reasons for their changes. But first, to understand the issues around divorce rates we need to answer this question:
What does it mean to say that some percent of marriages “end up” in divorce?
People can be married for many decades. Some get divorced at one year, five years, fifteen years or even sixty years after the marriage. And some die married. Therefore, we can only know for sure the rate at which marriages end up in divorce for people who were married far back enough in the past for all of them to have already died.
But we can also start with a more recent cohort of people who married on the same year and estimate the divorce rate of the remaining marriages on the last available year of their data. The more recent the cohort of marriages, the longer and less reliable is the estimated period.
Or we can state divorce rates as of a given wedding anniversary, such as “35% by the 25th anniversary“. This allows us to compare divorce rates between people who married on different years by the same standard.
A divorce rate alone, without: (1) stating the year of the marriages, (2) qualifying it by the anniversary when the divorce rate was calculated and (3) mentioning whether it is actual or estimated, is a meaningless number.\
It would be foolish to expect that divorce rates have been at the same 50% for many decades. Few things having to do with human behavior stays the same for very long. So we need to do our best to understand whether the divorce rate has been rising or falling during the last few decades.
The following article published by the Census Bureau sheds some light on the direction of the divorce rate:
Number, Timing, and Duration of Marriages and Divorces: 2009, Household Economic Studies, May 2011, by Rose M. Kreider and Renee Ellis, in Current Population Reports.
The data for the article is based on a Census Bureau survey of over 39,000 households given in 2009 to 55,597 adults that were married at some time in their lives. It is available on this website: http://www.census.gov/prod/2011pubs/p70-125.pdf
On the two graphs below, the solid lines labeled “Actual” are from Table 4 (on page 11) of this article. Each line represents a cohort of individuals married during one of the five-year periods shown on the table Years When Married on the upper right corner of the graphs.
The lines on the graphs show the percent surviving while married (on the vertical axis) at each wedding anniversary (on the horizontal axis). We can call these lines the marriage survival rates of each cohort. The divorce rate equals 100% minus the marriage survival rate.
At the 35th anniversary the marriage, the survival rate fell form 62.10% for the cohort of men married between 1960 and 64, to 57.90% for the 1965-69 cohort. At the 25th anniversary, that survival rate fell from 66.90% for the 1960-64 cohort to only 54.40% fort he 1975-79 cohort, a drop of about 12 percentage points. There was also a drop in the 10th anniversary survival rates of 10 percentage points between the same two cohorts.
After 1974 the marriage survival curves are too close together for the cohort-to-cohort changes to be significant. But, for men, the tenth anniversary survival rate gradually rose from the low of 73.40% for the 1975-79 cohort to 77.30% for the 1990-94 cohort.
In general, what we know from this study is that the men's marriage survival rates, of the 10th to 25th anniversaries plunged by about 12 percentage points between the 1960-64 and the 1975-79 cohorts. Then their survival rate at the 10th anniversary rose by about four percentage points between the 1975-79 and the 1990-94 cohort.
The women's marriage survival rates are generally lower and they are not as far apart between cohorts. But they tell the same story about when they fell and when they later partially recovered.
If you think about it, something looks wrong with the data when the marriage survival rates of men and women are different. As far as we know there were no same-sex marriages, polygamy or polyandry in the Unites States before the year 2009, so all marriages were one man to one woman. When they get divorced, both marriage partners count as divorced on the same year. Therefore, if a study were to follow through their lives a sample of couples who married on the same year, the marriage survival rates (and the divorce rates) have to be the same for men and women on each year after the marriage.
But such a study would be very long and expensive, which is why this study was based on interviews with 55,597 people in over 39,000 households in 2009. Since divorced people are very unlikely to be living in the same households, the former husbands of the divorced women interviewed (and the former wives of the divorced men) were very unlikely to be in the sample. Consequently, the divorce dates and total divorce rates of the women are unrelated to those of the men. This is why the men's and women's graphs above differ.
Another thing to keep in mind about this study is that it sampled people who were living in 2009, not on the year of they married. Therefore, both the married and divorced people who died before 2009 were not counted. Since married people live a little longer than single people, there could be a slight bias towards higher marriage survival rates, but nothing large enough to change any conclusions about how they changed.
Another way to compare long term marriage survival rates for the same cohorts of marriage years is to extrapolate the curves (shown as dashed lines on the graphs). The two earliest cohorts were extrapolated by a method that projects forward the changes in the slopes of the curve based on what they were like during the last three periods. All of the other curves were extrapolated by following the same slope as the actual and projected curve for the 65-69 cohort. This allows us to visually compare marriage survival rates (or Divorce Rate = 100% - Marriage Survival Rate) at all anniversaries up to the 60.
By this method, the post 1974 60th anniversary divorce rates are hovering around 56% to 62% for men and around 63% to 68% for women. Of course, in the total population, the men and women divorce rates should be the same. An estimate of that number would be an average of the men and women rates of this sample: 54.5% to 65.5%. So, as it turns out, the estimate of the more recent long term divorce rates is significantly higher than the much repeated 50%. But the extrapolation could have overestimated this divorce rate if the slope of the extrapolated marriage survival rate really flattens more towards the higher anniversaries than what is shown on the graph.
This is what I consider the most likely explanation:
The half-decade of 1965-69 was the start of an American cultural revolution. As part of this revolution, the “women's liberation movement”, in the broadest sense of this phrase, succeeded in opening up many professions for women (like those of doctors, lawyers, engineers and business executives) that had customarily remained almost completely closed to them. This encouraged many women to become educated in these professions and pursue them and it increased their labor force participation rate.
These changes took place with surprising speed, as this example from my personal experience illustrates: When I was a senior in high-school in 1960 there were no girls in my geometry and trigonometry class and I was the only boy in the typing class. When I started teaching Economics in college in 1966, I had only one or two women in each class and all were there because Economics was required of Home Economics majors. By 1969 women were about one third of the class and by the seventies they were half. More recently, women are more than half of all college students and recent graduates.
The rapidly increasing opportunities for education and employment in better-paid carriers made the generations of women entering their college years and marrying starting with the late sixties more economically independent of men. Their labor force participation rates and their income relative to men went up. Increasingly, more women who were later dissatisfied with their husbands did not find economic dependency to be an obstacle to divorce.
The graph below shows how the divorce rates of women rose together with their labor force participation rates between 1965 and 1990, particularly at the 10th year marriage anniversary.
This is supporting evidence of the theory that the freeing of many wives from being economically dependent on their husbands removed that dependence as an obstacle to divorce.
It should also be noted that in the nineteen seventies many states passed laws making divorce easier. It has been argued that this induced some married women to enter the labor force to be better prepared for the possibility of a divorce. This could have also increased the labor force participation rate of married women shown on the graph below.
The source of the labor force participation rates is: United States Census Bureau, 2012 Statistical Abstract, 597 - Labor Force Participation Rates by Marital Status, Sex, and Age on: http://www.census.gov/compendia/statab/2012/tables/12s0597.pdf
But by 1990 the 10-year divorce rate started dropping five years before the labor force participation rate stabilized. Yet it remained 8.3 percentage points above the 1970 rate.
Perhaps that gradual drop in the 10-year divorce rate is a sign that married couples have managed to improve how they cope with women working outside the home. This may be related to the postponement of the age of marriage, leading to marriages of more mature people. Or it could be happening because a smaller percent of the most likely people to divorce are marrying, since marriage rates have dropped.
In addition, we have to consider that we should expect the increase in the divorce rate due to rising female independence to stop at some point. This is because this increase in the divorce rate only affects the percent of wives who are dissatisfied with their husbands and their rising income removed the dependency obstacle to divorce. At some point, all wives dissatisfied with their husbands earn enough income to be able to get a divorce and the divorce rate stops rising.
Some would argue that it is good that women should not have to be tied to a husband they dislike any more than men should be tied to a wife they dislike. Women's improved employment opportunities removed the economic obstacle to break up an undesirable marriage, an obstacle that very few men did not have to face.
Others would object to anything that increases the divorce rate, particularly when the couple has children, due to the harm that divorce can do to them. Even with no children involved, divorce has a considerable traumatic effect on the divorcing couples. And some people have religious or moral objections to any type of divorce.
Whether the long term increase in the divorce rates is an acceptable price to pay for equal opportunities and higher employment rates for women is something we will let the readers decide for themselves.
This article does not have a copyright. You many copy it in part or in whole or make a reference to it. If you copy it, please include the following reference:
Divorce Rates: What They Are, How Have They Changed and Why,
by Ricardo Mejias, June, 2013, AAA Family Law
and link to this page on this website.