BU Today

In the World

The Dollars and Sense of Marriage (Continued)

Student Spotlight: An Essay by Fergus Hodgson (CAS’06)

A Touch of Theory

The relevant theoretical gains from marriage shall be referred to within the context of each explanation for marital decline, but in order to clarify the framework for this analysis the benefits of marriage deserve a short mention before we begin. The gains to an individual from a marital commitment are many, (and from a social standpoint also; hence the desire to promote marriage within the 1996 welfare reform), but economic theory reduces them down to three of primary importance:

• Specialization/division-of-labor based on comparative advantage and increased returns to sector-specific human capital investment.
• Income/wealth pooling, a gain from household economies of scale, a form of insurance, and a support network in the case of illness or unemployment.
• The sharing of children, a semi-collective/public good. The sharing of a child is able to increase or hold constant rather than decrease the satisfaction that a parent gains from that child, and within marriage, a parent is not excluded from a child, as might occur otherwise. Additionally, a child might usually present a financial vulnerability and hinder vocational specialization, however marriage alleviates such concerns.

The economic data to be outlined is not necessarily of isolated exogenous variables, part of a one-way causal sequence to marriage rates. Although the economic analysis supports a causal role, the statistical evidence is only correlational data, consistent with the hypothesis. The direction of causality need not be one-way. For example, greater economic opportunity for women is one accepted explanation for reduced marriage rates. However, lower marriage rates are likely to have women anticipating the need for self-sufficiency, preparing themselves better for employment, and seeking out higher-paying jobs. Hence, rising wages for women can be both a cause and symptom of falling marriage rates. But this does not necessarily present a problem; both causal relationships can comfortably occur simultaneously. However, teasing out the proportional direction of causality among such relationships without experimental data is largely impossible.

Legalization of Contraception and Abortion

Theoretically, contraception and abortion can explain declining marriage rates through a variety of routes: shifting of perceived responsibility for a child’s welfare away from the father to the mother, changing sexual habits away from commitment and marital exclusivity, and reduced birth rates among married and unmarried women. Oral contraceptives, far more effective than prior methods, were approved by the FDA in 1960, but in many states remained illegal to unmarried people until 1970, when restrictions on contraception were judged to be unconstitutional. The liberalizing of abortion laws began to accelerate in the early 1970s, but the Roe v. Wade decision of 1973 made all abortions (even those at later stages in pregnancy) approved by the Supreme Court and more widely available to women. With legalization complete, the utilization of oral contraceptives and abortion was rapid, closely preceding and correlating well with the greatest period of marital decline (1970 to 1990). In the early 1970s, when starting sexual intercourse, couples’ use of “the pill” almost tripled in the early 1970s, and 80 percent of US women born after 1945 are thought to have used oral contraceptives at some time. Similarly, the number of abortions in the US grew every year from 1970 through 1990, multiplying by an astounding factor of fifteen, and in
1980, when the rate of abortion increase began to slow down, 59 percent of extramarital pregnancies were terminated by abortion.

George Akerlof, Janet L. Yellen, and Michael L. Katz (AYK) have investigated changes to perceived responsibility for children among men as a consequence of contraception and abortion; the logic being that if a man has no power over whether a baby is aborted, and he can not be certain whether the women is correctly using her oral contraceptives, how can he be responsible for taking care of the child (usually manifested by marriage with the mother-to-be)? This argument may be appealing, but it provides little in the way of testable predictions, and even AYK despondently admit that the survey evidence supporting a relationship is weak at best. A man’s perception of responsibility is rather difficult to gauge, particularly when his provision of financial support is so heavily influenced by child-support legislation. Additionally, the question of responsibility for the care of a child is one of ethics, not of economics. That does not mean that the introduction of new technology doesn’t sway public opinion, but rather that we are not going to resolve the deeper ethical question or delve into people’s minds with economic tools alone.

Yet all is not lost. We can examine the correlation between the decision to marry in the event of an illegitimate pregnancy with the legalization of abortion and contraception. At first glance, rising illegitimacy as a result of contraception and abortion may appear counterintuitive. After all, aren’t these the very tools used to prevent such occurrences? Perhaps if all other factors, such as sexual habits, remained constant, these conflicts would be avoided. However, a decline in “shot-gun” weddings, as mentioned in the introduction, certainly did occur after contraception and abortion were legalized.

To clear up any confusion as to why such a trend occurred and to avoid a post hoc fallacy, an argument to explain causality is needed. AYK, in a more verifiable manner, place the sex/relationship market into the supply-and-demand framework, and theorize that since women no longer faced the risk of pregnancy they were generally more willing to engage in extramarital intercourse without a commitment from men to marry in the event of pregnancy. AYK’s rather dehumanizing supply-and-demand analysis assumes that men are the demanders of premarital sex, favoring a lower priceless or no commitment to marry in the event of pregnancy,—and women, the suppliers, will take the highest price—as much commitment or chivalry as the competitive relationship market will allow. As contraception was introduced, and the risk of pregnancy was all but removed, the cost of premarital intercourse for women was lowered, and the equilibrium in the market was changed by an outward shift of supply. Even though most women would have preferred to have a commitment to marriage in the event of pregnancy, and many women resisted use of contraception and abortion on a moral basis, they were competing for men in a market that no longer afforded such luxury.

The demand-supply model accurately predicts greater proportional illegitimacy, but it also predicts increased extramarital sexual activity and the reduced ability of society and women to command and enforce a commitment to marriage from men (as manifested by the aforementioned decline in marriage upon illegitimate pregnancy). Consistent with an increase in the quantity of extramarital sexual activity, the proportion 18 year old females who were sexually active grew from 29 percent in 1950 to 63 percent in 1980. In the long-run, as the market settles at a new equilibrium, and extramarital sexual activity becomes the norm, a loss of the social stigma surrounding illegitimate births is also to be predicted.

The final piece to the marriage puzzle related to abortion and contraception may be the overall decline in birth rates, but the assertion that birth control caused the decline is debatable. Since the gain from a marital relationship is increased by the expectancy and presence of children, the ability to prevent children, or at least lower the number for each woman, would cause fewer people to seek marriage and would facilitate more people leaving marriage because they now face less financial vulnerability and/or concern over causing harm to children. However, the relationship between birth control and lower birth rates is more tenuous than expected. Although the birth rate fell by more than a third between 1950 and 2000—the decline being most concentrated among married women—the birth rate of unmarried women more than doubled.

To obscure the picture even more, specific demographics, such as teenagers and Medicaid recipients, have shown positive relationships between abortion availability and birthrates. Gary Becker argues that reduced birthrates are better explained by greater employment opportunities for women, highlighting the course of events in Japan. Japan legalized abortion in 1948, but, unlike other industrialized nations, oral contraception was banned right up until 1999. Even without the birth-control technology shock of the 1970s, Japan has still experienced a birth-rate decline of 40 percent and a rise in divorce since 1950, similar to that of the US. On the other hand, the US and Japan do share a common trend of more women in the workforce, to be discussed in the next section.

The sequence of events in Japan serves to keep us on our toes when we might otherwise give too much credence to effects of abortion and contraception. The legalization of abortion and contraception correlates well with marital decline, and it provides a good explanation for changed sexual-relationships and the decline of shot-gun weddings; however, the causal role with birth rates is ambiguous. And of course, we must keep in mind, only a small group of people marry due to the stimulus of an illegitimate pregnancy. There are other reasons for entering marriage, and birth control does less to change these reasons. Abortion and contraception do provide part of the explanation for declining marriage rates, but plenty is left to be resolved.

Greater Employment Opportunities for Women, and Growing Income Inequality Among Men

I have grouped these two factors together because, as shall be explained, each goes hand-in-hand when considering their effect on marriage and divorce. Rising wages for potential marital care-laborers (traditionally women) mean less dependence on men and a higher opportunity cost of child-birth and home-based labor (consistent with the lower birth rates of Japan and the USA). Unless the wages of potential private-market “providers” (traditionally men) rise comparably, the provision and security that men can offer will be relatively less lucrative, and men will be less able to compensate women for lost wages and induce them into marriage. Additionally, as women face greater market opportunities and are educated in the same manner as men, the historic gains from specialized human-capital investments and trade based on comparative advantage within marriage will decline.

When male wages become more unequal, and a larger proportion of men go without stable employment, women seeking “provider” labor within marriage will deem a smaller proportion of males to be marriageable. Fittingly, male employment and income levels are positively associated with marriage incidence. While wealthier men face less competition with rising inequality, and could potentially marry more rapidly, the poorest of men are left in a weak position. When combined with rising wages for women, increased inequality among men brings a double-edged sword for the poorest of men. Not only is the lower-income sector of men earning less and struggling to fulfill a provider role, what little they do have is less desirable to women who are now earning more on the private market. While rising incomes of women would theoretically be expected to lower marriage incidence, this effect is only to be predicted from a declining disparity between men and women—men’s incomes either being held constant or growing at a slower rate than women’s. If men’s and women’s incomes were to grow at the same rate, no change in marriage rates would be predicted. Any increase to the opportunity cost of marriage that a woman might face could be compensated for with a larger marriage premium transferred from the higher income of the man. Additionally, maintained income-disparities would allow the continuation of gains from gender specialization between the home and private market. Supporting the importance of income-disparities, divorce is positively correlated with a wife’s income relative to her husband’s, but not with her absolute income.

Before discussing the observed rising incomes of women, a point made in the introduction must be revisited. The proportional direction of causality between marriage rates and female incomes in the long run is difficult, if not impossible, to ascertain statistically. The trends must be approached with caution, keeping in mind that some of the growth of incomes for females can be attributed to the decline in marriage itself. On the other hand, the positively reinforcing nature of each variable may magnify the effect of income changes on marriage. Wages for women have risen relative to those of men, along with an expected increase in labor force participation, and this does correlate with the decline of marriage, but the picture is complicated by variation across income and education demographics. Over the last fifty years, human capital (education and experience) has become more important than physical strength in the labor market. Hence, one advantage that men possessed has been eroded. A declining income-disparity is all but inevitable given this context, and any ideological resistance from employers towards women entering the workforce was outlawed by the Civil Rights Act of 1964. In real terms, hourly earnings of women over the age of fourteen grew by almost 50 percent between 1950 and 1978 while general wages, consistent with a declining disparity, grew by only 20 percent. Fitting nicely as a correlation, the most rapid period of growth in wages for women occurred in the 1960s, a few years prior to the heaviest declines in marriage incidence and the greatest growth in divorce.

The shift towards greater female labor force participation, particularly among married women with children (which tripled between 1950 and 1977), reduced the proportion of married couples specializing between the home and private market. The gain from specialization within marriage, at least in monetary terms, had declined, and by the year 2000, only 16 percent of families were made up of an employed father, home-maker mother, and children, compared with 42 percent in 1970. Despite what appears to be a clear relationship between rising relative wages of women and a lower married population, when we look at specific demographics this conclusion becomes less assured. The marital decline has been greatest among poorer communities, and yet lower-income women have always tended to seek work. Women without a high school diploma who worked fewer hours in 1980 than in 1940 and received the most modest earnings gains defy my analysis because their marriage rate declined from 78 to 56 percent between 1970 and 1995, a greater drop than any other education demographic. The apparent counterexample of relatively uneducated women leads to an acknowledgement that growing inequality among men is also needed in order account for this potentially contradictory data. Different demographics of women face different potential mates, so relative rather than absolute wage changes will vary depending on the advancement or recession of the relevant potential mates. For example, positive assortative mating on the basis of education means that half of all United States married couples have the same level of schooling. The earnings gains for uneducated women were small, but in the face of growing male inequality, their gains relative to their potential mates, uneducated and poorer men, were higher than that of college-educated women.

A growth in male wage inequality certainly has occurred, consistent with the relative wage analysis and reducing the proportion of marriageable men, but it does not clearly match up with the marital  decline. Males with little education have seen their wages stagnate or fall, but this did not begin until the 1970s, slightly after the incidence of marriage had already begun to decline. The bottom tenth percentile of fulltime male workers earned less in inflation-adjusted terms in 2003 than in 1969. The slight statistical mismatch between marriage rates and male income inequality suggests that the rising opportunities of women were initially responsible for the male-to-female wage disparity; but neither male inequality, nor female advances alone are sufficient to account for marital decline. We must consider both the relative incomes and the spreads across genders.

With comparatively lower wages for men at the bottom end of the income distribution, and fewer women willing to take these men on as mates, lower marriage incidence was to be expected among low-income men. However, the population-wide effect was compounded by declines in the marriage of high-income men as well. Although the wealthiest men could now marry more rapidly if they chose to, low ratios of eligible men to women reduce the power of women to command a marital commitment in return for sex and care-labor (as was the case after the introduction of birth control). Low ratios of eligible males to females are associated with lower rather than higher rates of marriage among wealthy men, but the effect, although theoretically and empirically viable, is not as large as that of the changes to relative wages and marriageable proportions. Rising market opportunities for women relative to men appear to have played an important, if not dominant, role in the decline of marriage by reducing the demand for children, raising the opportunity cost of care-labor, and reducing the economic gain from specialization between home and private market work. Divorce also became more viable with fewer children and higher wages reducing a women’s financial dependence on her husband. Growing inequality among men has doubly compounded the change by reducing the proportion of men considered marriage worthy and limiting the power of women in the relationship market. Although the long-term trend of declining income-disparities fits declining marriage rates, apparent statistical inconsistencies highlight the importance of considering income inequalities among potential mates and relative rather than absolute gains for women.

Increased Government Provision of Welfare and Social Services

The provision of government transfers (more commonly known as welfare payments) and the enforcement of child support payments reduce the financial vulnerability of women, especially those with children. Consequently, the wealth and risk pooling gain from marriage is eroded, making single-parenthood, either by way of divorce or simply marriage avoidance, more economically viable. The criteria for welfare payments may also inflame this situation by providing a perverse incentive to remain single in the event of a pregnancy, at least officially, because married couples are less likely to be eligible. Government social services such as day-care, pre-school, elementary education, retiree welfare, and healthcare make the role of traditional care-labor increasingly redundant and less valuable when the government alternative, even if of lower quality, is subsidized or free. Hence, there is a decline in the demand for care-labor that has traditionally provided by women within the marital setting and the ability of both males and females to remain unmarried and fully employed in the private sector increases, even in the presence of children.

Changing social values have also been attributed to welfare provision: women receiving welfare payments for getting pregnant and keeping babies born out of wedlock hardly send the message that such actions are morally reprehensible, as was assumed in the past, particularly when married women do not receive the payment. However, this effect is difficult to monitor, and it requires some plausible but debatable assumptions in order to fit the data. Because the generosity of welfare benefits declined in the late 1970s, but illegitimacy rates did not, we must assume that either changing social attitudes are able to shift in one direction more rapidly than the other (once attitudes have changed, turning them around is more difficult) or social attitudes are irresponsive to small incremental changes, such as inflationary devaluation, and require a lengthy period of time to respond. In addition to the observational struggles, welfare and social services are as likely to be a symptom of changing social values as they are a cause.

Therefore, changing social values should not be used as an explanation, but instead the decline of female financial vulnerability and the reduced demand for spousal care-labor should be of primary concern. The connection between the growth of social spending and welfare with illegitimacy and fewer marriages is well founded theoretically, but not necessarily clear empirically. Federally mandated transfer programs such as unemployment compensation, food stamps, and social security, grew in real terms by almost a factor of five between 1950 and 1976, and the value and number of recipients of Aid to Families with Dependent Children (AFDC) had been growing since inception in 1936, rising rapidly during the 1960s. (AFDC is especially relevant because the program had eligibility criteria that provided a disincentive for marriage.) However, despite the growth of AFDC and other transfers that coincided well with the decline of marriage, particularly in the occurrence of an extramarital pregnancy, the decline or stagnation of AFDC payments during the late 1970s and the 1980s did not see a reversal of the illegitimacy trend.

A variety of explanations can account for the apparent break from a simple welfare-illegitimacy relationship, but one could reasonably assert that the relationship is weaker than theory implies. One explanation is that only a small proportion of the population is poor enough to be eligible for welfare when making decisions about living arrangements. Therefore, unless we believe that this small group can change social values, we must accept that welfare policy, even if it does distort marriage behavior, is not applicable as an explanation for the general population. An alternative theory—inferring a greater destructive impact of welfare—is that society takes a long time to adjust to incremental changes. AFDC payments had been growing in real terms from 1936 right up to the late 1970s, and the effects of welfare may have arisen over many years, even decades. While AFDC payments did see a gradual decline in the late 1970s and early 1980s, the adjustment would then not be expected until the mid to late 1980s, by which time divorce (and to a limited extent, illegitimacy) had plateaued, divorce peaking around 1980 and illegitimacy in the early 1990s. The nature of welfare is something that can be experimented with, and these experiments often occur across state lines. However, such data should be approached with caution because it has limited ability to explain a population-wide welfare-marriage relationship. Natural experiments have non-random selection, and those selected for voluntary welfare experimentation are usually current welfare recipients, so we can draw scant conclusions regarding effects on potential recipients and the general population.

The Minnesota Family Investment Program (MFIP) is one of few welfare experiments to have been attempted. The two-year trial, which started in 1994, possessed attributes that were later adopted in the 1996 reform, so it is particularly relevant. Half of all applicants were kept on the AFDC program while those remaining were placed on MFIP. MFIP reduced disadvantages to working by using a more gradual tapering of benefits and by tailoring the eligibility criteria to increase benefits to married couples. After the two years, the MFIP group had an almost 50 percent higher proportion of recipients married and less than half the proportion divorced, with a confidence level of greater than 99 percent. The MFIP finding bodes well for the ability of welfare reform to have a positive impact, and it is testament to the distortion of welfare on marriage of recipients. However, it does not tell us the impact on potential recipients, nor does it gauge the effects on wider society. Therefore the welfare-illegitimacy relationship and the magnitude of the impact on general marriage rates are still inconclusive.30

What About Cohabitation?

The reader may be thinking that this discussion and the public concern over declining rates of marriage is a little alarmist, and that the lower frequency of marriage is purely a consequence of more couples living together in a stable manner but merely choosing to remain unmarried. While cohabitation fails to provide many economic benefits attributed to marriage, an observation of increased cohabitation could leave doubts lingering regarding the credibility of this analysis and the importance of changing marriage rates.
Therefore, leaving out a reference to cohabitation would do a disservice to the discussion. While a growth in cohabitation has occurred, the rise is not close to bridging the gap created by so many fewer marriages. Cohabitation is by nature informal, so relevant statistics are scarce, but the census bureau does publish rates of households shared by two unmarried adults over the age of 16. Granted, many inaccuracies enter into the bureau’s broad definition, tending to overstate cohabitation by approximately a quarter, but this statistic still provides an indication of the growth of cohabitation within the US. In general, one third of the decline in married relationships is accounted for by increased cohabitation. Even by combining marriage and cohabitation (forgetting the differences) and ignoring the cohabitation overestimation in the measurements, the proportion of women living with either a husband or partner still declined by almost 10 percent between 1960 and 2000.


Economic, theoretical, and empirical analysis has provided great insight into why we have observed such a sharp decline in the married proportion of our population. The analysis points to the legalization of abortion and oral contraception, the decline of male-female wage disparity, and expanding welfare and social services. Alone, none of these explanations are adequate. However, taken collectively, as occurred between 1965 and 1980, they can account for the dramatic decline of marriage. In the case of marital decline, we are dealing with multiple coinciding explanations rather than an either-or scenario. With so many interplaying factors, gauging the relative contribution of each in terms of percentages, however tempting, is beyond the depth of this article. And without experimental data, which is limited to a few welfare trials, attempts to do so would only provide a degree of correlation, not causality.

I propose further research into the role of incarceration rates, religious participation, and divorce law changes. Mindful of the role of the aforementioned impact of male inequality on marriage rates, an increased prison population leading to even fewer men within the poorer demographics would exasperate the lack of marital opportunity for poor women. Since joint religious participation is a gain from marriage, and many religions promote the traditional family model, if declining religious participation, has occurred, may also have played a role. However, one would then need to explain in more economic terms why religious participation has declined. Liberalized divorce laws may be a symptom of marital decline than a cause of it; but a reduced transaction cost of marriage termination would justify a causal relationship, so as an ongoing legislative question, divorce laws deserve attention.