network.GIF (5000 bytes)line.gif (198 bytes)

MEMBERS.gif (1390 bytes)

WPAPERS.gif (2205 bytes)

INTERNET.gif (2600 bytes)


contact info |more about the macarthur3.gif (1180 bytes)


The Feminization of Poverty: Past and Future

Sara S. McLanahan
Bendheim-Thoman Center for Research on Child Wellbeing
Princeton University

Erin L. Kelly
Department of Sociology
Princeton University


To reach first author:
Office of Population Research
21 Prospect Avenue
Princeton, NJ 08544-2091
Phone: 609-258-5944
Fax: 609-258-5804



In 1978, Diana Pearce, a visiting researcher at the University of Wisconsin, published a paper noting that poverty was becoming "feminized" in the United States. According to Pearce, almost two-thirds of the poor over age 16 were women. Women's economic status had declined from 1950 to the mid-1970s, Pearce claimed, even though more women had entered the labor force in those years. Female-headed households in particular formed a larger and larger percentage of the poor. Pearce blamed the feminization of poverty on the lack of government support for divorced and single women. She argued that "for many the price of that independence has been their pauperization and dependence on welfare" (Pearce, 1978, p. 28).

Following on Pearce's observation, Sara McLanahan and her colleagues used data from the U.S. Census to examine trends in men's and women's poverty rates between 1950 and 1980 (McLanahan, Sorenson, and Watson, 1989).1 While providing empirical support for Pearce's claim that poverty was becoming feminized, they also showed that both men’s and women’s poverty rates had fallen dramatically during this period. Noting that the feminization of poverty was due to a relative rather than an absolute decline in women’s economic status, McLanahan and her colleagues blamed the feminization of poverty on changes in the family which had uncovered women's latent economic vulnerability. Among working age adults, the growth of single parent families was the crucial factor; among the elderly, it was declines in mortality and increases in the propensity to live alone.

Recently, Paula England (1997) noted that the feminization of poverty has stopped going up and may even be reversing among some groups. England attributes these changes to improvements in women's earnings, which have risen both absolutely and relative to men's during the 1980s and 1990s. She argues that since 1980, the gains in women's employment and earnings have been large enough to offset the changes in family structure and that these gains are beginning to close the poverty gap between men and women.

In this chapter, we examine the trends in men’s and women’s absolute and relative poverty rates between 1950 and 1996. We discuss and assess three explanations for the feminization of poverty, including Pearce's claims that the meager welfare benefits led to the feminization of poverty, McLanahan's argument that changing family structures were the principle culprit, and England's assertion that gains in women’s earnings are beginning to close the gap. We end by reviewing cross-national evidence of the feminization of poverty. These comparisons show how family structure, earnings, and government transfer programs have allowed some countries to avoid the feminization of poverty.

1.1 Defining the Terms

Before we begin our story, we need to say something about how poverty is defined and how the feminization of poverty is measured. In the United States, poverty is defined as not having enough income to pay for basic needs, such as food, clothing and shelter. Poverty is a family attribute. In other words, if a family is classified as poor, all the members of that family are also poor.2 To determine whether or not a family is poor, the Census Bureau sums the income of all the members of the family and divides by the "need standard" or "poverty threshhold" for a family of that size.

The "needs standard" or "poverty threshold" was developed in 1964 by Mollie Orshansky, an economist at the Social Security Administration (Ruggles, 1994). To determine a family’s "basic needs," Orshansky used data from an expenditure survey conducted in the mid 1950s which showed that the typical American family spent about a third of it’s income on food. From this study, Orshansky reasoned that in order to live above the poverty line, the family’s income should be about three times the cost of the "basic food basket," defined as the minimal nutritional requirements for a healthy diet. To determine the latter, Orshansky used information from the Department of Agriculture. She then estimated the cost of the food basket and multiplied by a factor of three to get the official "poverty threshold" for the average American family. Ultimately, a set of thresholds was created to take account of economies of scale associated with families of different sizes and compositions. Each year the poverty thresholds are adjusted for changes in the cost of living. Otherwise they remain very similar to the ones created by Orshansky in the early 1960s.

The official poverty line used by the U.S. government has been criticized on numerous accounts (Ruggles, 1994; Citro and Michael, 1996). First, the poverty line is based on gross earnings, before taxes. Families may be unable to meet their basic needs with the income they bring home, but they will not be counted as poor if their before-tax income is above the poverty threshhold. In this way, the current U.S. poverty rates may underestimate poverty among the working poor.

Second, government transfers, like the Earned Income Tax Credit (EITC), food stamps, public housing, and MedicAid, are not counted as income for the purposes of determining poverty.3 Some people may be counted as poor because their earned income is very low, but they may be able to meet their basic needs through using other government programs. In this way, the current U.S. poverty rates may overestimate poverty among people who receive benefits from certain government programs.

Third, the poverty measure implicitly assumes that money is equally shared among members of a family. This assumption is not necessarily correct. It is certainly plausible that the person who earns the money may have more control over its use, or a man might control the money as part of his breadwinner identity, or there might be some other unequal arrangement.

A fourth criticism of the poverty line measure is that it is based on an absolute rather than a relative standard. A poor family today is much worse off, relative to the average family, than a poor family was in the 1960s (McLanahan 1995). This is because the living standards of the median family have increased dramatically during the past 35 years, while the poverty thresholds still assume that families need only three times their basic food needs. Some scholars have suggested setting the poverty line at a given proportion of the median income, rather than having it reflect a fixed amount of money. For example, families with incomes below 50% of the median family income (adjusted for family size) would be classified as poor. With a relative measure, poverty measures economic standing in the society, rather than marking economic destitution or the inability to feed, clothe, or house one's family.

Many other countries use a relative standard for poverty (Smeeding, Higgins, and Rainwater, 1990). As we will see in the cross-national comparisons below, if the U.S. adopted a relative standard, poverty rates would be significantly higher than they are under the official poverty measure. The difference reflects different ideas of what poverty is -- relative economic disadvantage or the inability to cover basic needs.

The term feminization of poverty focuses our attention on sex differences in poverty rates and the fact that they have grown in the last half-century. Feminization describes both the unequal state of men's and women's poverty rates and the processes by which women's risk of poverty has increasingly exceeded that of men's. In Pearce's articles (1978, 1984), she focused on the proportion of the poor who were female. Other scholars, including England and McLanahan and her colleagues, have focused on the ratio of women’s poverty rates to men’s rates (Casper, McLanahan and Garfinkel, 1994; McLanahan, Casper and Sorensen, 1995; McLanahan et al, 1989). A sex-poverty ratio greater than one means that women have a higher poverty rate than men. For example, a poverty rate of 1.50, which is close to the number today for all adults, means that women's poverty rate is about 50 percent higher than men's.

Sex differences in poverty rates are a function of the differences in poverty rates among single men and single women weighted by the proportion of adults who are single (i.e. not married). Since poverty is measured at the family level, sex differences in poverty can only exist among unmarried men and women. If all adults were married, and if boys and girls were evenly distributed across households, the sex-poverty ratio would be 1.0. Of course, some families would still be poor, but there would be no sex differences in poverty because each poor man would be linked to a poor woman.

The sex-poverty ratio, our measure of the feminization of poverty, can change when either the percentage of single adults changes or when the poverty rates of single men and women change. As we shall see below, both these factors have changed during the past four decades.

1.2 Trends in Poverty Rates and Poverty Ratios

Figure 1 reports poverty rates for men and women of all ages from 1950 to 1996. Several points about figure 1 are worth noting. First, blacks have much higher poverty rates than whites, regardless of their sex; and women have much higher rates than men, regardless of their race.4 Second, poverty rates fell dramatically between 1950 and 1970, and have leveled off since then. The trends are quite similar for whites and blacks and for men and women: large declines during the 1950s and 1960s, followed by very small changes during the 1980s and 1990s.

The leveling off in poverty rates after 1970 actually masks two quite distinct trends among young adults and the elderly. After 1970 (as shown in Figures 2A and 2B), elderly adults continued to experience falling poverty rates, while young men and women saw their rates go up. The disparity between the two age groups is especially prominent among men. In 1950, elderly males had the next to highest poverty rate; in 1996, they had the lowest rate. Over the same period, young women went from having the next to lowest poverty rate to having the highest rate.

Table 1 reports the trends in the sex poverty ratios between 1950 and 1996. The first set of numbers is taken from the McLanahan et al. paper (1989) which was based on the 1950, 1960 and 1970 decenial censuses. The second set of numbers is taken directly from the Current Population Surveys. Notice that the two data sets yield different ratios for young adults in 1970. This is because a different set of age cutoffs was used in the two analyses. To examine trends between 1950 and 1970, McLanahan and her colleagues used 18-24 as their age cutoff. To examine trends between 1970 and 1996, we used 18-30. Thus, we would not expect the two sets of figures to be identical in 1970. The sex poverty ratios for middle aged, elderly, and the total population also differ slightly, although the discrepancy is much smaller. In this instance, the analyses used the same age cutoffs and the difference is due to differences in the data themselves.

In presenting these numbers, our point is not to provide precise estimates of the sex poverty ratio for any particular group or period. Rather it is to document the trends in the data. For this purpose, we are confident that our numbers tell a consistent story.

To sum up briefly, a rather dramatic feminization of poverty occurred between 1950 and 1970, among all age groups and among both whites and blacks. Interestingly, it occurred at a time when poverty rates for both men and women were falling rapidly. Because men's poverty rate declined faster than women's rates, the sex-poverty ratio increased.

After 1970, the story becomes a bit more complicated. For the population as a whole, the feminization of poverty peaked in 1970 among whites and in 1980 among blacks. Then it leveled off. The latter trend masks two rather different patterns. Among the elderly, the feminization of poverty has continued to rise unabated. Between 1980 and 1996, the sex-poverty ratio grew by 69 percent among whites and by 65 percent among blacks. Whereas, in 1950, elderly women were only slightly more likely to be poor than elderly men, today they are more than twice as likely to be poor. Overall, poverty rates for elderly men and women decreased significantly, as in the earlier decades, but again, men's rates have fallen much more quickly than women's.

Among working aged adults, the feminization of poverty stopped increasing some time after 1970, and may even be reversing. Among white middle aged adults, a "de-feminization of poverty" occurred in both the 1970s and the 1980s.


Why has the sex poverty ratio grown so rapidly over the last fifty years? Why were women only slightly more likely than men to be poor in 1950, whereas today they about 50 percent more likely? What accounts for the leveling off, and the possible reversal, of the feminization of poverty among some age groups after 1980? In this section, we try to answer these questions. We focus on three changes that have profoundly altered the social and economic conditions of American men and women during the past forty years: changes in the family, changes in the economy, and changes in the welfare state. As described below, some of these changes have clearly benefited men more than women. Others have benefited women more than men, and still others have had different effects at different times.

2.1 Changes in the Family

Starting with changes in the family, several demographic trends are important. The first of these is the delay in the age of first marriage. Throughout the 1950s, the typical young woman married when she was about 20 years old and the typical young man married when he was about 23. This situation prevailed throughout the 1950s. By 1990, however, the median age at first marriage-the age at which half of the population has married for the first time - was 24 for women and 26 for men. (McLanahan and Casper, 1994).

A second major change is the rise in divorce (Figure 3). Whereas, in 1950, most people married and remained married until they (or their spouses) died, today over half of all couples end their marriages voluntarily. The divorce rate -- the number of divorces each year per 1,000 married women -- rose steadily during the first half of the twentieth century and increased dramatically after 1960. Over half of all marriages contracted in the mid-1980s were projected to end in divorce (Castro Martin and Bumpass, 1989). The divorce rate leveled off during the 1980s, but this was not necessarily a sign of greater stability. We would have expected such a leveling off, given the increase in the average age at first marriage, and given the fact that the large baby boom cohorts have reached middle age and passed through the period of their lives when they were most likely to divorce (Bumpass and Sweet, 1989).

The increase in divorce, coupled with the decline in marriage, meant that an increasing proportion of adult women were living separately from men and relying on themselves for economic support. Since women generally earn less than men for various reasons (see below), single women have a higher risk of being poor than single men. In short, if nothing else changes, declines in marriage will lead to increases in the sex poverty ratio.

We should point out that American women are more independent (socially, psychologically and economically) today than they were in the 1950s, and more of them are opting out of traditional family arrangements in spite of the economic costs. Since poverty only measures one aspect of wellbeing, we cannot say that women who live alone are "worse off" overall than men or than women living with men. Indeed, the fact that so many women are choosing to remain single suggests that the benefits of independence may outweighs the costs associated with a lower standard of living.

A third trend affecting family arrangements is the increase in children born outside marriage (Figure 4). In 1960 about 6 percent of all births were to unmarried couples whereas by 1996 over a third fell into this category. The increase is even more dramatic among black families-from 22 percent in 1960 to nearly 70 percent in 1996. These changes in fertility are due to several factors, including the decline in marriage which increases the pool of women at risk for having an out-of-wedlock birth, the decline in the birth rate of married couples (which was especially steep during the 1960s and early 1970s), and the increase in the birth rate of single women (which has risen since 1980).

The net result of the changes in fertility and the increases in divorce among couples with children has been the growth of single mother families (Garfinkel and McLanahan, 1986). In 1960, only 8 percent of families were headed by a single mother; today the number is about 27 percent. Blacks are much more likely than whites to fall into this category, with over half of all black families now consisting of a single mother and her dependent children. However, the trends are similar for all race and ethnic groups.

Single parenthood has affected women's poverty rates much more than it has men's. Whereas in principle, responsibility for children could be evenly distributed between mothers and fathers after a divorce or a nonmarital birth, in practice, women almost always have custody of the child (about 90 percent of single parents are women). Thus, not only do single women earn less money than single men, their needs, as defined by the poverty threshold, are greater than men’s because they have more dependents. Again, we should point out that most mothers, when faced with a divorce or a nonmarital birth, prefer to have custody of their child, despite the economic hardships associated with single motherhood. However, as noted above, these women pay a high financial price for their independence.

A fourth demographic trend affecting the family life of men and women is the increase in "nonfamily" households, particularly, the increase in one-person households. In 1940, only 10 percent of all households were classified as "nonfamily" households, whereas in the early 1990s, over 30% were of this type (Sweet and Bumpass, 1987). Until recently, nonfamily households were more common among blacks than among whites. Today, however, the proportions are nearly the same for both races. Hispanics, in contrast, have always had lower rates of nonfamily households.

Young adults and the elderly are especially prone to living in nonfamily households. For young people, this change is associated with the decline in marriage and the increase in cohabitation. Interestingly, couples who are cohabiting are not counted as families by the Census Bureau, even though they may be pooling their incomes and sharing expenses. Thus the poverty rates of many of these couples are likely to be overstated. Since young women are somewhat more likely to cohabit than young men, treating cohabitators as nonfamily households leads to an overestimate of women's poverty rates relative to men's (McLanahan and Casper, 1994).

The elderly are also much more likely to live independently today than they were fifty years ago (Treas, 1995). This change is generally viewed as a sign of progress, and most elderly people report that they prefer their independence. Whereas in the past, a majority of elderly people lived with their adult children once they were widowed, today, more and more widows and widowers live alone. The increase in the proportion of elderly living alone is due in part to increases in Social Security income.

A final trend affecting the sex differential in poverty rates is the increase in life expectancy. More specifically, women are likely to live longer than men and so their retirement income has to stretch over more years. The differences in men's and women's life expectancies have grown over these years. In 1940, men at age 65 had a life expectancy only 17 months shorter than that of women aged 65. In 1993, men at age 65 had a life expectancy 47 months -- almost four years -- shorter than that of women aged 65 (Advisory Council on Social Security, Volume 2, 1996, p. 261).

2.2 Changes in the Economy

Changes in family composition reflect changes in the needs of the family -- the denominator of the poverty function. There have also been important changes in men’s and women earnings and income experiences -- the numerator of the poverty function -- during this period. Women's labor force participation has increased over most of the twentieth century (Bianchi and Spain, 1996; Blau, 1997). The changes have been larger in recent decades than in the early 1900s, with many American women seeking and finding employment since 1960.

The increase has been most dramatic for married women and for women with children. In the early 1950s, only about 30 percent of married mothers with school aged children were working outside the home. By 1990, this number had risen to over 73 percent. The figures for mothers with preschool children (under age 6) are even more dramatic. In 1960, only 19 percent of married mothers with preschool children were in the labor force, whereas, by 1990, 59 percent were employed. Women’s employment increased over this whole period, but especially in the 1970s and 1980s. These years also saw some decreases in men’s employment, with the reductions occurring mostly among younger and older workers (Wetzel, 1995). Young men, especially those with little education and few marketable skills, withdrew from the labor force. Older men also left work, many of them taking early retirement because their jobs were being downsized or their plants closed (Wetzel, 1995).

The increase in women’s employment means that women are gaining more experience in the labor market, which, all else being equal, should increase their ability to support themselves and their families. Equally important, women are also gaining on men with respect to their hourly wages (Spain and Bianchi, 1996; Blau, 1997). From 1960 to 1980, full-time women workers averaged about 60% of full-time male workers' earnings. Since 1981, however, the ratio of women's earnings to men's earnings has increased to about 70%. During the 1980s, highly-educated workers did well while those without a college education saw falling wages (Wetzel, 1995). This division was true of both men and women, but highly-educated women’s wages increased more than men’s while less-educated women’s wages fell less for women. Overall, women's earnings have increased relative to men's, raising the female-to-male wage ratio.

England’s (1997) argues that women's earnings have helped reduce the sex-poverty ratio is supported by the timing of the relative wage increases and the sex-poverty ratio decreases. The declining sex gap in wages is caused by both women's increased earnings (due to greater labor force participation and higher wage rates) and men's stagnant wages since the late 1970s. In fact, men's economic status has declined in recent years as women's earnings have improved slightly. For whites, the period in which women's relative wages began climbing corresponds to the period in which the sex-poverty ratio fell. This finding suggests that women's wages have led to the reversal in trends.

The exception to the de-feminization trend also offer some support for England's argument about wage effects. If changes in women's earnings (relative to men's) are largely responsible for the reversal in the sex poverty ratio, we would not expect to observe a reversal among the age groups who are less likely to be working, namely among retired persons. From 1970 to 1996, the sex-poverty ratio did not decline for women over 65 years. Since elderly women are less likely to be employed, and quite likely to be living without men, they did not benefit from the improvements in employment and earnings that accrued to women in the 1970s and 1980s.

2.3 Changes in Public Benefits

Besides relying on their earnings, many Americans receive income from government transfers. Government transfers serve as a buffer against the risk of poverty and economic insecurity. If a family's income from earnings does not meet its needs, transfer income may provide the extra income needed to lift the family out of poverty. However, government transfer programs differ in how generous they are, and thus how successfully they keep households out of poverty. The three main government transfer programs in the US are Aid to Families with Dependent Children (now TANF), which is a program mostly benefiting poor single-mothers and their children , Supplementary Security Income, a program benefiting disabled, blind, and poor elderly adults, and Old Age Survivors, Disability and Unemployment Insurance, a set of programs for disabled, unemployed and retired workers.

The first two programs, TANF and SSI, provide "income-tested" benefits which means that a family must be poor in order to qualify for the benefit. The income test creates a "catch 22" with respect to poverty. If a family that qualifies for the benefit manages to increase its income (from other sources) enough to lift it above the poverty line, the family no longer qualifies for the benefit. In fact, eligibility thresholds are generally set below the poverty line, so a family can be cut off from these programs even though their income is well below the poverty line. Thus income-tested programs, by design, do not lift many families out of poverty, although they reduce the depth of their poverty. Social insurance programs, in contrast, are guaranteed to all adults who qualify, regardless of their economic status. These payments (which include most "Social Security" benefits) are also much larger, per person, than TANF or SSI. As a consequence, their poverty reduction effects are very large.

We should point out that several other government programs providing support to low-income families were initiated during the 1960s, 1970s, and 1980s, including food stamps, public housing and housing vouchers, Medicaid, and the Earned Income Tax Credit (EITC). The EITC, which supplements the earnings of the household head in working poor families, is one of the most generous and most effective anti-poverty programs in the U.S.. However, since benefits are provided through the tax system, they are not counted as income in calculating the official poverty rates. Neither food stamps nor housing subsidies is counted as income, although both of these programs cover ‘basic needs" and are similar to cash transfers in many respects.

If government transfer programs played a large role in the feminization of poverty and the recent reversal of the trend, then changes in these programs should correspond to changes in poverty rates and sex-poverty ratios. To some extent, the trend in benefits corresponds to the trends in poverty rates. During the years in which feminization first occurred—1950 to 1970—both AFDC and Social Security were expanding rapidly, not only in terms of the number of people covered but also in terms of the value of the benefits themselves. AFDC rose during the 1950s and 1960s, as the federal government waged a "War on Poverty." Similarly, Social Security pension coverage was expanded during these two decades and the value of the benefit was increased several times.

As shown in Figure 1, poverty rates fell dramatically during this time. Social security benefits undoubtedly played a role in reducing poverty among elderly men and women, and welfare did the same for single mothers. Recall that these were the years in which family structure was changing quickly, and the increased availability of AFDC probably helped to offset some of the negative consequences of these changes on women’s economic status.

Many conservatives would disagree with the statement that women’s poverty rates would have been higher without welfare benefits. Instead, they would argue that the growing availability of welfare benefits during this period actually increased women’s risk of poverty by fueling the growth of single mother families (Murray, 1984). While, theoretically, it makes sense to think that AFDC benefits might encourage some poor women to leave a bad marriage or to keep a child who was conceived out-of wedlock, the empirical evidence on this issue indicates that the size of the effect is very small (Moffitt, 1997). This finding should not surprise anyone who is familiar with the demographic trends described above (delays in marriage, declines in marital fertility, increases in one-person households) since these trends have been occurring among women of all social class backgrounds, including women who would never been forced to rely on welfare. Nevertheless, the conservatives raise an important point and draw our attention to the fact that the relationships between family change, labor market change, and changes in public transfers are probably reciprocal.

After 1970, the trends in AFDC and social insurance diverged sharply. Social Security pensions for the elderly were raised again and indexed to inflation in the early 1970s, and the value of these benefits continued to rise throughout the 1970s, 1980s and 1990s. In contrast, welfare benefits, which were not linked to inflation, peaked in the mid-1970s and then began to fall. Initially, the decline in AFDC was due to high inflation in the second half of the 1970s. After 1980, however, declines were due to direct cuts in benefits as well as to inflation.

The decline in government transfers for women of childbearing age after the mid-1970s is consistent with the leveling off of poverty rates among women between the ages of 18 and 64 during the 1970s and 1980. That is, poverty rates stopped declining about the time that welfare benefits stopped increasing. Similarly, the rise in social security benefits for retired adults after 1970 is consistent with the continuing decline in poverty rates among the elderly.

Given that social security has played an important role in driving down the poverty rates of the elderly, why have women’s poverty rates been slower to decline than men’s? The feminization of poverty among the elderly is a result of several factors. First, pensions for never married women are lower, on average, than pensions for never married men since women earn less than men while they are working. Second, retirement benefits are set when a person reaches age 65 (or retires), and despite that fact that they are indexed to inflation, the original benefit is not adjusted for increases that post-date retirement. In particular, people who retired before the early 1970s have substantially lower pensions than people who retired after 1970. Since women live longer than men, they are more likely to fall into this group than men. Finally, elderly women are more likely to live alone than elderly men, and thus they must live on one pension rather than two. For all these reasons, poverty rates among elderly women have not declined as fast as poverty rates among elderly men.

2.4 Decomposing the Trends

To try and gain a better idea of the forces behind the leveling off of the sex poverty ratio between 1970 and 1996 and the relative importance of changes in the family and changes in labor market opportunities, we conducted a simple thought experiment. We asked, what if only family structure had changed between 1970 and 1996 and everything else had stayed the same, what would the sex-poverty ratio have been? Next we asked a similar question about changes in employment and earnings. If only men’s and women’s personal earnings had changed between 1970, what would the poverty ratio have been? To answer these questions we estimated multivariate regression equations that treated poverty in 1970 as a function of family characteristics, personal earnings and a set of other variables measured in 1970. Family structure was divided into five categories: married couple (with or without children), single parent with minor children, one-person households, other households with children, and other households without children. We ran separate equations for men and women, blacks and whites, and three age groups: 18-30, 31-64, and 65+.

Next, we substituted the 1996 means for the family structure variables into the 1970 equations and obtained predicted poverty rates for men and women, holding all the other values constant. Using the predicted poverty rates, we then calculated a new set of sex-poverty ratios for each sex, race, and age group. We repeated the process, this time substituting 1996 personal earnings for 1970 earnings and holding family structure constant. The results are reported in Table 2.

According to Table 2, if only family composition had changed between 1970 and 1996, the sex poverty ratio for young whites would have been 1.47, which is 14 percentage points higher than the observed ratio in 1996. In contrast, if only men’s and women’s earnings had changed during this period, the sex poverty ratio would have been less than one. These simulations are only illustrative. However, they demonstrate that employment changes strongly favored women during this period and offset some of the negative effects associated with changes in family composition. The patterns for middle aged white women and for both young and middle aged black women are similar. If only family composition had changed between 1970 and 1996, sex poverty ratios would have been even higher than they actually were. If only earnings had changed, the ratios would have been much closer to parity.


Another approach to understanding the feminization of poverty is to ask why it does not occur in some settings. Casper et al (1994) examined sex differences in poverty in eight countries and asked how some countries have managed to avoid the feminization of poverty (Figure 5). They considered how factors related to income (i.e. age, education, and employment status) and needs (i.e. marital status, parental status) led to different poverty rates and sex-poverty ratios among these eight countries. These researchers found that large gender differentials in poverty rates have been avoided in three countries, for three different reasons. The important factors are distinctive family structures in Italy, distinctive employment patterns in Sweden, and distinctive government transfers in the Netherlands. The Netherlands findings are useful because government transfers have not played a large role in reducing poverty or preventing the feminization of poverty in the U.S.. This example shows that government programs can and do reduce poverty rates and close the gap between men's and women's poverty if they are generous enough.

Italy has very high marriage rates and so almost no difference in men's and women's likelihood of living in poverty. Because almost every adult man is paired to an adult woman, there are poor Italians but very little difference between men's and women's likelihood of living in poverty. If U.S. adults had marital patterns equal to Italian adults (but otherwise had the same characteristics that they actually do have), the sex-poverty ratio in the U.S. would be reduced by 71 percent (Casper et al, 1994).

In Sweden, men are actually slightly more likely than women to live in poverty. This unusual situation is explained by women's high labor force participation. Since Swedish women are very likely to be employed and there is only a small wage gap between men's and women's average earnings, women are not more likely than men to live in poverty. If U.S. women had the same labor force participation rates as Swedish women, the sex-poverty ratio would be reduced by 60 percent. In some other Western countries with relatively low levels of women's employment, female labor force participation rates similar to Sweden's would entirely eliminate the poverty gap (Casper et al, 1994, p. 600).

While Sweden illustrates the potential of women's earnings for reducing the sex-poverty ratio, it is important to realize that Swedish government policy facilitates women's employment and women's high relative earnings in several ways. Government policies that encourage women's employment include a tax system that taxes persons on their own earnings (rather than having a joint filing option for married couples) and a parental leave system that grants long paid leaves to mothers and fathers of infants. In addition to the leaves, government policies allow parents of young children to reduce their work hours for several years while keeping their old jobs (Kamerman and Kahn, 1991). The government also pushes for wage equality between men and women through its involvement in collective bargaining agreements and the fact that the state employs a very large number of women workers itself. Therefore, earnings have aided women in Sweden, but the government has played a large role in increasing women's earnings.

The Netherlands also has a very low gender poverty ratio, but its marriage, parenting, and employment rates are not unusual. It seems that "the low poverty rates and the low gender-poverty ratio in the Netherlands is largely due to the generous welfare system" (Casper et al, 1994, p. 602). In the Netherlands, single parenting is common and single mothers are not likely to be employed full-time. However, the Dutch have avoided sex differences in poverty rates because the government transfers to single mothers are generous enough to lift those women and their families out of poverty.

The cross-national comparisons confirm the importance of family status, earnings, and transfer payments in explaining poverty rates and the feminization of poverty. This research also suggests that there may be multiple ways to reduce the disparities between men's and women's poverty rates. Americans interested in reducing the feminization of poverty could design public policies to encourage any or all three of these conditions, although these policies might increase gender inequalities in other areas.


The changes in poverty rates and sex differences in poverty have occurred in two major periods, from 1950 to 1970 and from 1970 to 1996. In the decades after World War II, poverty rates declined dramatically. This was a period of strong economic growth, including rising wages, and expansion of government benefits, such as Social Security and welfare programs. Both of these forces pushed poverty rates down. As poverty rates declined, the sex-poverty ratio increased for all groups. The growing inequality in men’s and women’s poverty rates was caused by several factors. First, the wage growth that pushed poverty rates down benefited men more than women because they were more closely attached to the labor market. Second, changes in family structure hurt women more than men, mainly because women bore more responsibility for children in the growing number of unmarried households.

After 1970, poverty rates stopped declining. During the 1970s, the economy stagnated and when it recovered in the 1980s, the main beneficiaries were college graduates rather than less-educated workers living near the poverty line. The sex-poverty ratio continued to increase among the elderly, but stopped increasing among working age adults. The sex-poverty ratio may be reversing, as the gap between women’s and men’s poverty rates closes a bit. The sex-poverty ratio is increasing among the elderly partly because older women are living longer than men, and these older women often rely on a single Social Security benefit for many years. The sex-poverty ratio has stopped climbing for working age adults, even though families have continued to change in ways that leave women with a big economic burden. It seems that the stagnant economy has hurt men more than it has hurt women. Women’s wages have slowly gained on men’s over the last fifteen years, stopping the increased feminization of poverty. It is important to note, however, that women are still much more likely – about 50% more likely overall – than men to live in poverty.

Cross-national comparisons show us that the feminization of poverty is avoidable. The Netherlands keeps the sex-poverty ratio low by establishing a high income floor; the welfare state does not allow anyone to be poor, regardless of family status or employment situation. Sweden keeps the sex-poverty ratio low by subsidizing employment and by keeping wages high through union agreements. Since women’s labor force participation is very high, this insures that few women are poor. Unlike the Netherlands or Sweden, Italy has fairly high levels of poverty but the high marriage rates keep sex differences in poverty very low.


1. Other researchers also examined the feminization of poverty. See also Fuchs (1986) and Folbre (1987). back to text

2. For single people, only their own income is counted when determining poverty. If unrelated single adults share a household, each person counts as a separate "family" for determining poverty. back to text

3. Government cash transfers, such as AFDC, TANF, and Social Security payments, are counted as income when determining poverty. back to text

4. We focus on whites and blacks in this chapter. Our primary purpose is chronicling and explaining the trends in poverty rates and sex differences in poverty. Because we are considering changes across time, we need to look at stable populations (or groups within those populations). Hispanics are now a large minority group within the U.S., but there has been so much immigration during this period that it is difficult to compare the Hispanic poverty rates over time. Also, there are no data from these sources on Hispanics until 1980. For research on poverty among various Hispanic populations, see Aponte, 1993; Tienda, 1995. For information on immigrants and economic outcomes, see Portes 1996; Portes and Rumbault 1997. back to text

back to top