Change in American Families: Favoring Cohabitation over Marriage

Change in American Families: Favoring Cohabitation over Marriage

Demographics

Change in American Families: Favoring Cohabitation over Marriage

Jagadeesh Gokhale · · 22 min read
Change in American Families: Favoring Cohabitation over Marriage

Cohabitation rates have increased significantly during the last two decades. Cohabiting individuals appear to have weaker workforce engagement and earnings. With changing U.S. demographics, the trend toward favoring cohabitation over marriage appears likely to continue.


Key Points

  • The share of married Americans has declined secularly since the mid-1990s, and the share of cohabiting couples among dual-headed families has increased.

  • Across education groups, high school dropouts and those with a high school diploma have the lowest marriage prevalence rates and highest cohabitation rates.

  • Changes in the demographic composition of the U.S. population portend steeper declines in marriage and increased prevalence of cohabitation in the United States.

1. Introduction

Family formation and dissolution patterns emerge from complex mate selection processes and decisions to cohabit or marry in the context of education acquisition, financial security, career development, decisions to procreate, and other factors. The economic benefits of two-parent families have been well documented (Kearney, 2023). This brief, however, is limited to describing one aspect of family formation where trends are clear and patterns consistent over time: increased cohabitation rates among dual-headed families. A family with a married head and spouse is associated with providing a more trusting and stable environment for raising children than cohabitation alone.

The information presented below is compiled from the Annual Social and Economic (ASEC) supplement of the Census Bureau’s Current Population Survey (CPS). Marriage rates in the United States have declined secularly since the mid-1990s. As Figure 1 shows, the overall share of married individuals has trended downward — from 55.9 percent in 1996 to 46.4 percent in 2023. And the share of cohabiting couples has increased from 3.7 percent in 1996 to 9.1 percent in 2023.

Figure 1: Population shares of married and cohabiting individuals.

Data from the Annual Social and Economic (ASEC) supplement to the Current Population Survey.

2. Trends in Marriage and Cohabitation rates by Demographic Attributes:
A. Education

Figure 2’s Panel A shows that marriage rates declined for all education (e) groups. Individuals with less than a high school education (e<HS) have the lowest marriage share. In contrast, those with more education — high school, some college (C), and college-or-more education (e=C and e>C) — have higher marriage rates. The correlation of marriage shares (levels) with education also suggests that Americans are postponing marriage while acquiring more education.

Figure 2’s Panel B shows that cohabitation rates have increased rapidly among all education groups. Moreover, those with lower education have higher cohabitation rates, again suggesting that cohabitation may be an intermediate stage during education acquisition and career development along the transition toward marriage.

Figure 2.

e=educational attainment; HS=High School; C=College.
Data from the Annual Social and Economic (ASEC) supplement to the Current Population Survey.

B. Race

Figure 3’s Panel A shows that marriage rates have declined secularly among all races, with the lowest marriage prevalence rates among black and mixed-race individuals. Panel B shows that cohabitation rates have trended upward across all race categories.

Figure 3.

Data from the Annual Social and Economic (ASEC) supplement to the Current Population Survey.

C. Age Group

Figure 4’s Panel A shows marriage rates declining secularly among all age groups. Young individuals appear to be postponing marriage, and older individuals’ marriages attrit from divorce and mortality. Among the middle three age groups shown in Figure 4, those in the 55-64 age group have the highest marriage rates, but those rates have declined rapidly.

Figure 4’s Panel B shows that cohabitation rates have trended upward for all age groups with the fastest increase occurring for the 65+ age group.

Figure 4.

Data from the Annual Social and Economic (ASEC) supplement to the Current Population Survey.

D. Under-age children in the family

Figure 5’s Panel A shows the lowest marriage prevalence rates for individuals with no children and higher marriage rates for those with children, especially those with 2 or more children. Panel B of Figure 5 shows, however, that cohabitation rates are increasing rapidly, even among families with 2 or more under-age children.

Figure 5.

Data from the Annual Social and Economic (ASEC) supplement to the Current Population Survey.

3. Marriage, cohabitation, work hours, and annual earnings

Whatever the reasons for the secular historical decline in married individuals’ shares and increase in cohabiters’ shares, survey micro-data suggest that cohabiting correlates with lower labor force engagement and lower economic productivity as indicated by annual wage earnings.

Figure 6 shows calculations of average hours worked and average annual indexed wages from ASEC micro-data information for 1996-2023. To focus the analysis on working-aged individuals, only singles aged 18-59 and married individuals with the spouse also aged within 18-59 years are included in the sample1 Panel A of the figure shows that, among males, average hours worked for married males are highest (dashed blue line), followed by those of cohabiter males (dashed red line), and followed by those of single males (dashed yellow line).

Figure 6.

Data from the Annual Social and Economic (ASEC) supplement to the Current Population Survey.

In Panel A, the wider gap in hours worked between married males and females (dashed and unbroken blue lines, respectively) compared to cohabiting males and females (dashed and unbroken red lines, respectively) suggests that the former engage in greater division of labor between working in the labor market and at home compared to the latter.

Panel B of Figure 6 shows that married and cohabiting females’ wages (blue and red unbroken lines) are comparable. Still, the wages of cohabiting males (dashed red line) are much lower than those of married males (dashed blue line). This supports the conjecture that financial insufficiency may be a reason for cohabiting couples to remain unmarried.3

The information summarized in Figure 6 on average annual hours worked and average indexed earnings is decomposed by just two factors: Marital (/married /cohabiting /single) status and sex. But many other concomitant factors could be triggering differential decisions on hours and earnings by family type and sex. These include differences in education, race, health status, the presence of children, and others.

A regression of annual (indexed) wages against several control variables is implemented to more carefully isolate the effect of marital/cohabiter/single status on annual work hours and wage earnings,4 The controls include year-fixed effects, sex, age group, race, education, spouse’s race and education, health status, single/cohabiting/married status, number of children, employment status (self-employed or wage worker), and immigrant status. The controls also include interactions with marital status and sex. The results suggest that cohabiting individuals of working age work fewer hours and earn lower wages than married and single individuals.5 The regression’s predicted wage earnings (/hours worked) are calculated by year, sex, and marital status and displayed in Figure 7.

Figure 7.

Data from the Annual Social and Economic (ASEC) supplement to the Current Population Survey.

Panel A of Figure 7 shows that after accounting for the concomitant influence of other demographic and economic attributes, predicted hours worked by married and cohabiting individuals are closer to each other (compare blue and red dashed lines with those in Panel A of Figure 6). However, Figure 7’s Panel B shows that the difference between married and cohabiting males’ wage earnings (blue and red dashed lines, respectively) remains substantial (although it is slightly smaller compared to Figure 6’s Panel B). In contrast, the difference between married and cohabiting females’ wage earnings (blue and red unbroken lines, respectively) remains similar to that of Panel B of Figure 6.

After controlling for many other potential influences, cohabiters work slightly fewer hours than married couples but earn significantly less – suggesting a correlation between cohabiting and employment in low-value-added occupations and industries.

4. Projections of U.S. marriage and cohabitation rates

As in most micro-data surveys, the PWBM microsimulation’s projections combine cohabiters with married individuals in the “married” category. Knowing historical trends in the shares of cohabiters among all couples by detailed demographic attributes enables making projections of the shares of cohabiters among all couples. ASEC information on the share of cohabiting individuals by gender, race, and education is applied to PWBM projections of the share of all individuals who are “married” (married plus cohabiters).

As indicated earlier, the historical shares of cohabiters among all “married” individuals vary by race, gender, and education, as do the growth rates of those shares. Combining cohabiter shares (including differential growth trends of those shares) with changing future demographic composition of the population could either reverse or extend historical growth in cohabiter shares. For example, more educated individuals have lower cohabiting rates and their shares in the future population are projected to increase and push overall cohabiter shares downward. On the other hand, cohabitation rates are higher among Hispanic and mixed-race individuals, and their populations are projected to grow larger in the future, pushing cohabitation shares upward.

Figure 8: Historical and projected shares of cohabiting and married individuals.

Source: Author’s calculations.
Projections based on combining historical cohabitation rates from ASEC microdata and projections of married plus cohabiting individuals from the PWBM microsimulation.

Figure 8 shows the results of applying historical cohabiter rates and trend growth in those rates to the projected population generated by the PWBM microsimulation. As before, only singles aged 18-59 and married individuals with at least one spouse aged 18-59 are included in the sample.

Cohabiter rates are projected to increase from just under 10 percent today to over 16 percent by 2040. Correspondingly, the share of legally married individuals is projected to decline from just over 46 percent today to below 40 percent by 2040. Because cohabiting men and women appear to have weaker workforce engagement, earnings, and productivity, it is important to distinguish between legally married and cohabiting individuals when estimating future economic trends, especially as their respective shares are projected to shift significantly.



This analysis was produced by Jagadeesh Gokhale under the direction of the faculty director, Kent Smetters. Mariko Paulson prepared the brief for the website.

Kearney, Melissa S. “The two-parent privilege: How Americans stopped getting married and started falling behind.” In The Two-Parent Privilege. University of Chicago Press, 2023.

Menasce-Horowitz, Juliana, Nikki Graf, and Gretchen Livingston, (2019). “Marriage and Cohabitation in the U.S.” Report, Pew Research Center: Views on Marriage and Cohabitation in the U.S. | Pew Research Center.

U.S. Census Bureau, (2018). “Living with an Unmarried Partner is now Common for Young Adults”: For Young Adults, Cohabitation Is Up, Marriage Is Down (census.gov).

  1. The calculations are weighted by the survey’s person weight (“ASECWT”).  ↩

  2. ASEC’s individual annual earnings (of the previous year compared to each survey year) are indexed using the Social Security Administration’s annual average wage index series available at: https://www.ssa.gov/oact/cola/AWI.html#Series. The indexing neutralizes the effects of inflation and productivity growth in different years to put all earnings of sample individuals on par with each other before executing statistical analyses.  ↩

  3. The lack of financial readiness is cited by cohabiting survey respondents among key reasons for not being engaged or married (op. cit.).  ↩

  4. The regression framework used is follows standard “Mincer” approach where the logarithm of wages is regressed against variables such as education, age (a proxy for experience) and other control variables besides marital status and sex. It assumes that the error of the dependent variable (either log hours or log earnings) is normally distributed.  ↩

  5. ASEC cross-section weights are used in the regression so that each year’s data sample representative of the worker population aged 18-59.  ↩