Demographic Factors Affecting the Level of Poverty
One of the fundamental indicators of inequality in a population is its level of poverty, which is typically measured as the proportion of households whose income falls below some specified poverty threshold (i.e., the poverty rate).30 Because poverty thresholds vary by household size (and, to a lesser extent, composition), Bumpass and Sweet (1981) note that processes of family formation and change can have direct effects on the incidence of poverty. These demographic factors may also have indirect effects via consequent reductions in labor supply or earnings capacities. Households may thus leave or enter poverty as the result of such demographic changes as marital dissolution, aging, retirement, death, childbirth, remarriage, the departure of children from the household, and the formation of new households.
The particular demographic issue that has been prominent in the poverty literature is the extent to which changes in family structure over the past few decades have contributed to the rate of poverty in the U.S. Earlier qualitative and ethnographic
29 The one major exception here is Treiman (1977), who also uses data for developing nations. The Treiman constant is thus the one generalization that may be said to pertain to both developed and developing nations.
30 Despite Sen's (1976) seminal contribution to the methodology of poverty indices, most empirical research seems to be based on the simple proportion of the poor in the total population (perhaps because Sen's  index may be more unreliable than a proportion when the incomes of the poor are reported with significant error).
research by sociologists often interpreted family structure as being adaptive to poverty (e.g., Rainwater 1968; Schulz 1969; Stack 1974). According to this "situational view'' of poverty, nonmarital fertility and increased proportions of female-headed households are seen as being in part a consequence of low wages, unemployment, and economic inequality rather than the primary causes of them (Valentine 1968). By contrast, the ''culture of poverty'' view maintains that poverty is substantially promoted and increased by the rejection of middle-class social values and normative behaviors (Banfield 1968). The resulting high rates of female-headed households and high school dropouts are predicted to be fundamental and important sources of increased poverty rates (Lewis 1966).
Bane's demographic study (1986) is one of the first to systematically investigate the effect of family structure on poverty rates. Her analysis is based on the U.S. Census Bureau definition of poverty, which is an absolute measure (i.e., it is invariant with respect to changes in average standards of living or household income in society). Using standardization methods and demographic data from 1960 to 1983, she finds that "the poverty rate in 1979 would have been about 16 percent lower than it was had family composition remained as it was in 1959'' (Bane 1986: 214). For the black population separately, the corresponding figure is 22% (Bane 1986: 215). Bane interprets her results as lending some support to the "situational view'' because ''most poverty, even that of female-headed families, occurs because of income or job changes'' (Bane 1986: 231). For African Americans, Bane (1986: 231) concludes that ''although there has indeed been a dramatic and shocking increase in female-headed households among blacks and an equally dramatic feminization of black poverty, one cannot conclude that much of the poverty could have been avoided had families stayed together.''
Another important study is Danziger and Gottschalk (1995), which also uses the U.S. Census Bureau definition of poverty. Unlike Bane (1986), however, Danziger and Gottschalk seek to evaluate simultaneously the effects of economic changes as well as demographic changes. To do this, they develop a decomposition that incorporates changes in mean household income, household income inequality, and race-specific family structure. Consistent with Bane's (1986) general conclusion, they find that changes in family structure (in particular, the increase in female-headed households) had a moderate effect on increasing poverty from 1973 to 1991. Danziger and Gottschalk's (1995: 102) results for this period also indicate, however, that economic factors had larger effects than did family structure, including the increase in mean household income (which reduced poverty) and the increase in household income inequality (which increased poverty). Economic factors, particularly the increase in mean household income, had the overwhelmingly largest impact on reducing poverty during the earlier period from 1949 to 1969 (during which time family structure did not change significantly).
An update and extension of Danziger and Gottschalk's (1995) analysis is provided by Iceland (2003), who uses both the U.S. Census Bureau definition as well as a relative measure of poverty for the period from 1949 to 1999. With regard to his results for the U.S. Census Bureau definition, Iceland also finds (as did Danziger and Gottschalk) that economic factors—particularly increases in mean household income—overwhelmingly had the dominant impact on reducing poverty from 1949 to 1969. However, in the subsequent period, from 1969 to 1990, Iceland's (2003: 512-513) results indicate that changes in family structure played a significant role, especially for African Americans. Even so, the total impact of economic factors was still larger than was the effect of family structure for this period. During the last period in their analysis, from 1990 to 1999, the poverty rate did not change much, and economic factors clearly had the greatest impact—while the distribution of family structure remained relatively constant.
An important contribution of Iceland (2003) is to extend the analysis to a relative definition, according to which the poverty threshold increases as the average standard of living increases in society.31 Although absolute poverty (i.e., the U.S. Census Bureau definition) declined dramatically from 1949 to 1969, Iceland (2003: 507) finds that relative poverty was reduced only slightly. During the subsequent two periods, from 1969 to 1990 and 1990 to 1999, relative poverty actually increased—in marked contrast to the declines in absolute poverty. Whereas according to the official (i.e., absolute) definition, the poverty rate declined from 13.9% in 1969 to 11.8% in 1999, by Iceland's relative measure, the poverty rate substantially increased from 20.6% in 1969 to 26.1% in 1999.
Iceland's (2003: 509-512) decomposition results generally show that increases in mean household income have much less impact on reducing relative poverty than on reducing absolute poverty and that the former is more highly affected by changes in household income inequality. This result is not surprising given that the thresholds for relative poverty are defined as a percentage of the median household income. In addition, the decomposition results indicate that the impact of family structure for the period from 1969 to 1990 was slightly larger for relative poverty than for absolute poverty, especially in the case of African Americans.32 From 1990 to 1999, however, family structure again had a negligible impact in the case of relative poverty.
In sum, the studies reviewed in this section used decomposition methods to analyze the impact of changes in family structure on the trend in the poverty rate in the U.S. All of these studies indicate that changes in family structure—and in particular, the increase in female-headed households—have increased poverty at least to some degree before 1990. This conclusion is probably somewhat more applicable to African Americans and to results that are based on a relative measure of poverty. Since 1990, however, changes in family structure have been minor and have not had a significant impact on the poverty rate.
At the same time, these studies seem to agree that although changes in family structure have sometimes played a significant role, economic factors have been the primary factors associated with the trend in poverty in the U.S. In the case of an absolute measure, the trend in poverty is more affected by changes in mean household income. In the case of a relative measure, the trend in poverty is more affected by changes in household income inequality. In reaching these general conclusions, however, we point out that these studies have not investigated whether changes in family structure may have had "indirect effects'' on the poverty rate by affecting trends in mean household income and household income inequality. This may be a fruitful issue to explore in future research.
31 In particular, Iceland (2003: 503) defines the poverty threshold as being one-half of the median household income after accounting for differences in household size.
32 In a related study, Smith (1988) argues that the effect of family structure on poverty among blacks was slightly more substantial during this time period than had been concluded by Bane (1986). These somewhat different conclusions probably in part reflect their different measures of poverty because Smith (1988) uses a type of relative measure.
Before concluding this section, we point out that the social and economic disadvantages of female-headed families are also evident in other industrialized nations (Garfinkel and McLanahan 1994; Lichter 1997). Data for the early 1990s indicate significantly higher relative poverty rates for female-headed households in Canada, France, Germany, and the United Kingdom (Garfinkel and McLanahan 1994: 209). As discussed by McLanahan and Sandefur (1994), female-headed households tend to be inevitably disadvantaged (as compared to two-parent families) in terms of parenting, economic resources, and social ties to the community. Family structure therefore probably plays a significant role in influencing the level of poverty in other nations as well.
Nonetheless, comparative studies also indicate clearly that the impact of female-headed households on the level of poverty is highly variable and conditional on the role of government programs (Smeeding, Torrey, and Rein 1988). For example, the relative poverty rates of female-headed families in Sweden and the Netherlands are quite low, and these countries are notable for the effectiveness of their welfare programs (Garfin-kel and McLanahan 1994). In other European countries as well, government benefits are far more effective (than is the case in the U.S.) in mitigating the adverse economic disadvantages faced by female-headed households (Garfinkel and McLanahan 1994; Rainwater 1995). To be sure, such programs are more costly but they do indicate that poverty among female-headed families can be reduced substantially without increasing appreciably the incidence of female-headed households (Bergmann 1996; Garfinkel and McLanahan 1994).
In addition to its relationship with the poverty rate, household income inequality is an important issue in itself. With regard to couple-headed families, Treas (1987: 265) notes that income inequality can be viewed as a function of income inequality among husbands plus income inequality among wives plus the covariance between the incomes of husbands and wives.33 Thus, income inequality among couple-headed families would increase if wealthier men tended to be married to wealthier women but would decrease if wealthier men tended to be married to poorer women. The pattern of assortative mating can thus have a direct impact on the level of income inequality among couple-headed families.
Treas's (1987) review of the literature from the 1960s and 1970s finds that most studies report that the earnings of wives tend to reduce family income inequality. These results seem to derive primarily from (at least for whites) the higher labor force participation among the wives of husbands who have lower incomes. That is, the wives of wealthier white husbands were less likely to have an income through work in the paid labor force. In short, due to the reduced labor force participation and hence lower earnings of the wives of wealthier husbands, the correlation between the incomes of white spouses during this time period appears to be negative, which reduces the level of family income inequality (Cancian, Danziger, and Gottschalk 1993: 210).34
33 The same basic function could also be applied to gay and lesbian couples.
34 The negative correlation is evident only when nonworking wives' earnings are set equal to zero; the correlation between spouses' earnings among couples where both spouses work was actually slightly positive even during this time period (Cancian, Danziger, and Gottschalk 1993: 210).
During the late 1970s and the 1980s, the labor force participation of women continued to increase. Partly as a result of this increase, the correlation between the earnings of spouses became positive (Cancian, Danziger, and Gottschalk 1993: 210). Although this positive correlation should increase the level of family inequality, the effect was counterbalanced by a reduction in the inequality of earnings among women. That is, the variation in the earnings among women declined significantly as more women worked more hours in the paid labor force. The reduction in inequality in the earnings of women is greater than the effect of the increase in the correlation between the earnings of spouses. The overall result is that during the 1970s, 1980s, and early 1990s, women's earnings reduced family income inequality (Cancian and Reed 1999).
Perhaps the most thorough and informative analysis of this issue is provided by Cancian and Reed (1999). This study develops a decomposition approach that factors in the proportion of households consisting of single persons or single parents. In addition, Cancian and Reed clarify the importance of stipulating the "counterfactual" when assessing the impact of women's earnings on the level of family income inequality. For example, comparison of the observed level of family income inequality and the level that would be obtained if all wives had zero earnings yields a different result than comparison of the observed level of family income inequality and that which would be obtained if the earnings of wives were reduced only slightly. For 1994, the former comparison increases family income inequality while in the latter case inequality is reduced (Cancian and Reed 1999: 180). However, when comparing the level of family income inequality in 1967 with that in 1994, Cancian and Reed conclude that the increase in the latter period would have been substantially greater if wives' earnings in 1994 were the same as they had been in 1967 (i.e., if the mean and variance of the distribution of wives' earnings did not change over this time period). Thus, ''the growth in family income inequality cannot be attributed primarily to changes in wives' earnings. Changes in husbands' earnings are substantially more important in explaining recent trends'' (Cancian and Reed 1999: 184).
Although this conclusion is important, so is the caveat mentioned by Treas (1987: 275) that ''working women trade off time for money. Women perform many valued services around the home (e.g., housekeeping, meal preparation, child care, entertaining, emotional support). When the wife goes to work, the family must either forego some of these services, find another family member to provide them, or spend money to purchase them in the marketplace.'' This loss in the value of home production is not reflected in the statistics on income distribution analyzed in the studies discussed above. On an hourly basis, however, working-class wives earn appreciably less than do upper middle-class wives, so the former group must work longer hours in order to obtain earnings that are equivalent to those of middle-class wives; working-class families may thus be forgoing more home production. For this reason, we agree with with Treas's (1987: 275) speculation that the equalizing effect of women's earnings on the distribution of full income among households may be overstated. Future research on this topic should address this issue.
An additional issue worth pursuing is the effect of women's earnings on men's earnings. This topic was mentioned by Dooley and Gottschalk (1985: 31) in their analysis of the increasing earnings inequality among men. In particular, the increased labor force participation of women may increase the level of market competition experienced by men and perhaps drive down the wages of less educated men through the sort of occupational queuing process described by Reder (1955).35 The consequence is that, as more highly educated women work more and increase their labor market opportunities, the inequality in the earnings of men may increase. As we have seen, this inequality is an important source of the level of poverty and the recent increase in family income inequality. Thus, to the extent that such an effect is evident, the equalizing effect of women's earnings on the distribution of family income may be overstated.
Regarding international studies, we are aware of only two publications that have focused on the effects of wives' earnings on inequality in the distribution of family income. The United Kingdom was investigated by Layard and Zabalza (1979), and Gronau (1982) considered the case of Israel. Using data for the 1970s, when the labor force participation and earnings of married women were relatively low in these countries, both of these studies found that the earnings of wives significantly reduced inequality in the distribution of family income.36 These results are consistent with the basic conclusion for the U.S. during this period (Treas 1987).
In addition to the issue of assortative mating, we also note some recent research on the effect of population growth on household income inequality. While the debate on the relationship between population growth and economic development is well known (e.g., Ahlburg 1998), the relationship between population growth and economic inequality has received little consideration.37 It is worth noting, however, that Nielsen
(1994) and Nielsen and Alderson (1995) consistently find large positive effects of the natural rate of population increase on household income inequality using data for 88 nations from 1952 to 1988. This result is evident in regression models that control for a variety of independent variables and use different specifications. It is thus a robust finding that merits further consideration.
Was this article helpful?
If Pregnancy Is Something That Frightens You, It's Time To Convert Your Fear Into Joy. Ready To Give Birth To A Child? Is The New Status Hitting Your State Of Mind? Are You Still Scared To Undergo All The Pain That Your Best Friend Underwent Just A Few Days Back? Not Convinced With The Answers Given By The Experts?