Joblessness and Unemployment: A Review of the Literature
by Patrick Mason
This literature review attempts to gather evidence on the relationship between African American male economic potential in the formal sector of the economy and transitions in African American family structure and marital stability. By economic potential in the formal sector, we mean the capacity to obtain labor and property income from legal economic opportunities. Marital status and family transitions are addressed with reference to two particular sets of activities: a) marriage and divorce; and, b) childbirth and abortion.
Major transformations in household structure and marital status have occurred during the past 35 years. These changes have become a primary focus of much scholarly and popular debate. Similarly, the bulk of American households have experienced stagnant or declining incomes since the early 1970s. African American households have experienced a disproportionate amount of both the economic decline and the transformations in household structure and marital status.
We contend that the changes in economic wellbeing and household structure and marital status are causally related, with the primary direction of causation moving from economic outcomes to marital status and household structure. Moreover, since socioeconomic changes among African Americans often foreshadow changes among American society as a whole, the results of this study will provide insight into future changes among all American racial and ethnic groups. Finally, we pay particular attention to the role of African American men as husbands, fathers, and potential mates because: 1) they are extremely under-studied in the professional literature, 2) mate availability, i.e., a shortage of marriageable men, is often cited as a crucial issue in the sociology and economics of the family, and 3) market discrimination against women (especially African American women) and the relatively greater household responsibilities of women imply that male earning potential may be a key variable in marriage, divorce/separation, and fertility decision.Both theoretical and empirical studies are utilized to answer four interrelated questions: (1) How does male formal economic potential influence family/household structure? (2) How does male formal economic potential influence involvement in the informal and illegal sector of the economy? (3) What is the nature of the interaction between informal/illegal sector involvement and family/household structure? (4) Which economic policies, group actions, and individual strategies can increase male economic potential in the formal sector, increase African American conjugal family stability, and decrease African American male involvement in the nonformal economic sector, viz., drug dealing and violent crime?
The cumulative statistical evidence suggests four rather strong conclusions. One, mate availability is a substantive economic variable. The best evidence suggests that at least 20-40% of the difference in marriage rates between African American and White women can be explained by differences in the relative supplies of marriageable men. Two, despite a now hardened popular wisdom to the contrary, AFDC encourages neither divorce nor pre-marital childbirth. More broadly, there is no evidence of public welfare policy encouraging "underclass behavior." For the most part, it appears the primary effects of AFDC are to increase the economic wellbeing of low-income families, i.e., the purpose for the which the program was designed. Three, increases in female independence -- as expressed by increased female-male wage rates, earnings, education, and employment -- tend to increase the probability of marriage and decrease the probability of divorce. Four, increased levels of education increase the probability of marriage, decrease the probability of divorce, and reduce the probability of premarital births among teens.
This review also provides insight into the crime, unemployment, and race debate. We find that African Americans have no particular cultural (or genetic) predilection toward crime. The real culprits are low waged and inconsistent employment opportunities, which create an environment that encourages income oriented crimes, e.g., drug dealing. Violent crimes of course are often complementary to income oriented crimes. Both criminal activity and imprisonment, the government's preferred policy option for dealing with the economically marginalized, reduce the supply of marriageable males and therefore tends to further disrupt African American families, i.e., to increase the fraction of female-headed households among African Americans.
Section I begins with a brief descriptive overview of recent changes in marital status, marriage opportunity, fertility rates, and the economic well-being of families. Section II presents alternative economic theories regarding transitions in family structure. Section III provides empirical evidence on competing theoretical hypotheses. Finally, section IV discusses the results and limitations of this study within the historical context of research on the African American family.
THE CHANGING AFRICAN AMERICAN FAMILY
In the past two decades African American families have experienced dramatic changes in marital status, marriage opportunities, fertility rates and economic well-being. Between 1970 and 1994, the number of African American married couples declined by nearly 22 percentage points. By 1994, married couples constituted only 46.5% of African American families (Table 1). The number of African American female households with no spouse present rose 20 percentage points, from 28% of families in 1970 to nearly 48% of families in 1994. By contrast, White families headed by married couples declined by six percentage points, from 89% in 1970 to 83% in 1994. The number of White families headed by a woman with no spouse present increased from nine percent in 1970 to 13% in 1994. So, both African Americans and Whites have experienced a movement away from married couple headed families toward single parent families but the change has been more dramatic for African American families. Finally, Table 1 shows that the magnitude of these changes were considerably larger during 1970-1980 than during 1980-1994.
The fraction of all African American families with own children present has remained somewhat constant at 60%, even as the headship composition of these families has changed. Consider African American male householders with no spouse, which were 3.7% of families in 1970 and 5.6% of families in 1994. The presence of children in these families has expanded from 40 percent in 1970 to 53% in 1994 (Table 1). Similarly, the fraction of childless African American married couples has increased by 8 percentage points. So, among African Americans, the presence of children has increased among male and female householders with no spouse present -- especially males -- but has declined considerably in married couple households.
Table 1.
Selected Characteristics of Families, by Race: March 1994, 1990, 1980, 1970
| Types | 1994 | 1994 | 1990 | 1990 | 1980 | 1980 | 1970 | 1970 |
| Family1 Type | Afr.Am. | White | Afr.Am. | White | Afr.Am. | White | Afr.Am. | White |
| Married Couple | 46.5 | 83.4 | 50.2 | 83.0 | 55.5 | 85.7 | 68.0 | 88.7 |
| Female H'holder, No Spouse | 47.9 | 13.0 | 43.8 | 12.9 | 40.3 | 11.6 | 28.3 | 9.0 |
| Male H'holder, No Spouse | 5.6 | 3.6 | 6.0 | 4.1 | 4.2 | 2.7 | 3.7 | 2.3 |
| Family Type and Presence of Own Children |
||||||||
| All Families | ||||||||
| With No Own Child < 18 | 40.0 | 53.8 | 41.4 | 52.8 | 38.3 | 49.3 | 38.9 | 44.8 |
| With Own Children < 18 | 60.0 | 46.2 | 58.6 | 47.2 | 61.7 | 50.7 | 61.1 | 55.2 |
| Married Couple | ||||||||
| With No Own Child < 18 | 48.2 | 55.4 | 47.4 | 54.1 | 43.9 | 49.9 | 39.8 | 43.3 |
| With Own Children < 18 | 51.8 | 44.6 | 52.6 | 45.9 | 56.1 | 50.1 | 60.2 | 56.7 |
| Female H'holder, No Spouse | ||||||||
| With No Own Child < 18 | 31.2 | 43.1 | 31.8 | 42.5 | 28.1 | 41.2 | 34.0 | 52.1 |
| With Own Children < 18 | 68.8 | 56.9 | 68.2 | 57.5 | 71.9 | 58.8 | 66.0 | 47.9 |
| Male H'holder, No Spouse | ||||||||
| With No Own Child < 18 | 47.0 | 56.1 | 68.2 | 59.2 | 61.3 | 65.2 | 59.9 | 74.1 |
| With Own Children < 18 | 53.0 | 43.9 | 31.8 | 40.8 | 38.7 | 34.8 | 40.1 | 25.9 |
| Children Under 18 Years by Presence of Parents |
||||||||
| Both Parents | 33.8 | 79.1 | 37.7 | 79.0 | 42.2 | 82.7 | 58.5 | 89.5 |
| Mother Only | 54.2 | 16.1 | 51.2 | 16.2 | 43.9 | 13.5 | 29.5 | 7.8 |
| Father Only | 3.8 | 3.0 | 3.5 | 3.0 | 1.9 | 1.6 | 2.3 | 0.9 |
| Neither Parent | 8.2 | 1.8 | 7.6 | 1.8 | 12.0 | 2.2 | 9.7 | 1.8 |
Source: U.S. Department of Commerce, Current Population Reports, P20-480, page 11. Data for 1994 is for "White, not Hispanic," while prior years represents "White" without excluding Hispanics.
1. The data for families do not include families in group quarters. For March 1994, in some CPS publications the data for families include families in group quarters.
2. Excludes persons under 18 years old who were maintaining households of family groups and spouses. Data for 1994 are for children in families only.
Indeed, 1970 was the last census year in which the majority (59%) of African American children lived in families with both parents present. By 1994, the majority of African American children (54%) lived in families where only their mother was present while the majority of White children (79%) lived in two parent families. From 1970 to 1994, the fraction of White children living in mother-only families increased from 8% to 16 percent.
Table 2.
Presence and Marital Status of Parent for Children Under 18 Years Living With Only One Parent,by Age, Sex, and Race: March 1993
| TYPES | Afr.Am | White, Non-Hispanic |
| LIVING WITH MOTHER ONLY | 94.7 | 82.1 |
| Never Married | 57.6 | 16.7 |
| Husband Absent | 22.1 | 24.0 |
| Widowed | 2.4 | 5.3 |
| Divorced | 17.9 | 54.0 |
| MOTHER HOUSEHOLDER | 78.5 | 67.0 |
| Never Married | 53.4 | 13.1 |
| Husband Absent | 24.0 | 24.5 |
| Widowed | 2.8 | 5.8 |
| Divorced | 19.7 | 56.7 |
| LIVING WITH FATHER ONLY | 5.3 | 17.9 |
| Never Married | 53.7 | 25.9 |
| Wife Absent | 19.3 | 19.9 |
| Widowed | 4.3 | 4.8 |
| Divorced | 22.7 | 49.4 |
| FATHER HOUSEHOLDER | 4.2 | 15.6 |
| Never Married | 47.3 | 25.9 |
| Wife Absent | 21.1 | 19.8 |
| Widowed | 5.5 | 5.3 |
| Divorced | 25.8 | 49.0 |
Source: U.S. Department of Commerce, Current Population Reports, P20-480, page 15.
Among African American children living in mother-only families, 58% of the mothers have never married and nearly one-fifth (18%) of the mothers are divorced (Table 2). Among White children living with mothers only, 17% of the mothers have never married and over one-half (54%) of the mothers are divorced. Similarly, the majority (54%) of African American children living in father-only families do so because the father has never married. Again, among Whites, divorce is the most common (49%) path to children living in father-only families.
Information on the gender gap in earnings and differences in family income by type of family will aid in our assessment of changes in family structure and the economic circumstances of children. It is no surprise that women earn less than men. However, the gap between male and female earnings has been closing as the earnings of men have stagnated over the last two decades. Whether we examine median income, median earnings, or the median earnings of year-round, full-time workers, both African American and White men were worse off in 1993 than in 1979.
In 1993, African American women employed as year-round, full-time workers had average earnings of $19,816 per year. This is 86% of the average annual earnings of African American men with year-round, full-time jobs, $23,019. In 1969, African American women employed year-round, full-time earned just 68% of their male counterparts earned -- $14,703 versus $21,565.
The rising earnings capacity of African American wives has kept African American married couple income at 78% of White married couple income. On the other hand, African American female householders with no spouse present have lost ground relative to their White counterparts and relative to their own position in two decades ago. In 1979, their level of income was $13,405 and the interracial earning ratio was .60; by 1993, African American female householders with no spouse present were earning $11,905 and the interracial earnings ratio was .55. African American male householders with no spouse present brought home $4,700 less in 1993 compared to 1979, and their interracial income ratio declined from .70 to .65.
The earnings differences between men and women, combined with the marital status of parents, has a noticeable impact on the economic circumstances of children (Table 4). African American married couples with pre-school children have over three times the annual income ($30,850) of African American female householders with pre-school children but no spouse ($8,690). This difference is even larger for families with school age children ($43,090 versus $12,460). Incredibly, the annual income of African American female householders with children but no spouse present declined from $12,820 in 1979 to $10,380 in 1993.
African American female householders with no spouse present and no children earn $13,000 per year more than their female counterparts with children in 1993, that is, $23,290 versus $10,380. Given the average African American male earnings of $23,019 for year-round, full-time employees, African American female householders with no children and no spouse present have little economic incentive to marry -- even when these women are obviously economically attractive partners to the average male. On the other hand, the small and deteriorating income of African American female householders with children but no spouse, combined with the additional responsibility of raising children, implies that currently unmarried African American women with children may face considerable difficulty in finding a husband. Finally, the declining income of African American males suggests that there is a declining pool of men capable of sustaining a family, especially a family with children. Clearly, mate availability is a substantive issue in explaining intertemporal and interpersonal transitions in African American family structure.
Regardless of the presence of children, African American married couples still tend to earn about 80% of White married couples (Table 4). We turn now to take a look at more detailed descriptive statistics regarding recent changes in marital status (Table 5), fertility rates (Tables 6,7,8,9, and 10), and mate availability (Table 11).
Table 4.
Median Income of Families, By Family Type, Presence and Age of Children (In 1993 Dollars)
African American White Ratio
1979 1993 1979 1993 1979 1993
FAMILIES WITHOUT CHILDREN
Married Couple 27,540 32,810 39,450 40,690 0.70 0.81
Female Householder, No Spouse 17,160 23,290 28,810 28,120 0.60 0.83
FAMILIES WITH CHILDREN
Married Couple 33,920 36,660 44,350 46,380 0.76 0.79
Female Householder, No Spouse 12,820 10,380 17,690 16,020 0.72 0.65
FAMILIES BY AGE OF CHILDREN1
All Families
all under 6 years 14,290 39,130 0.37
all 6 to 17 year 22,240 44,490 0.50
Married Couple
all under 6 years 30,850 45,290 0.68
all 6 to 17 year 43,090 51,710 0.83
Female Householder, No Spouse Present
all under 6 years 8,690 11,980 0.73
all 6 to 17 year 12,460 21,320 0.58
Source: U.S. Department of Commerce, Current Population Reports, P20-480, page 27.
1. Data for Whites is "White, not Hispanic."
Table 5 shows recent trends in marital status and marriage rates for African Americans (and Whites) by gender and by age group. Relatively few African Americans are widowed. Also, the presence of widows as a fraction of the population has been declining since 1940. Except among persons 30 to 39 years of age, divorce and separation rates have tended to hold steady or decline since 1960. Among men and women in their thirties, divorce and separation rates peaked in 1980.
Table 5.
Marital Status and Marriage Rates, by Race, Sex, and Age, Selected Years, 1940-80
Percent Black Marital Status White Marital Status Age and Year Marr. Wid. DS. NM. MR. Marr. Wid. DS. NM. MR. MEN Age 16-19 1940 3.1 0.1 1.2 95.6 2.1 1.5 0.1 0.6 97.8 1.1 1950 3.1 0.1 1.7 95.1 2.1 2.3 0.1 1.1 96.5 1.7 1960 3.1 0.0 1.6 95.3 3.3 3.7 0.0 1.1 95.2 3.2 1970 3.3 0.1 1.6 95.0 2.6 3.9 0.1 1.0 95.0 3.1 1980 0.9 0.0 1.3 97.8 1.4 2.5 0.0 0.9 96.6 2.3 1985 -1987 0.5 0.0 0.3 99.2 0.1 1.5 0.0 0.3 98.2 0.6 Age 20-23 1940 31.5 0.5 5.4 62.6 12.1 20.5 0.2 2.0 77.3 8.8 1950 31.3 0.4 7.4 60.9 7.6 28.7 0.1 3.4 67.8 10.4 1960 31.2 0.2 9.2 59.4 15.6 38.0 0.1 4.6 57.3 18.0 1970 28.6 0.2 9.3 61.9 15.7 35.3 0.2 5.0 59.5 18.0 1980 13.0 0.1 5.1 81.8 6.5 24.3 0.0 4.6 71.1 11.0 1985 -1987 9.7 0.1 1.6 88.6 2.1 19.1 0.1 2.6 78.2 6.1 Age 24-29 1940 59.1 1.2 9.2 30.5 10.7 56.2 0.3 3.6 39.9 13.1 1950 61.2 0.7 11.9 26.2 10.8 68.3 0.2 4.3 27.2 14.2 1960 58.5 0.5 13.8 27.2 16.8 72.3 0.2 5.2 22.3 20.5 1970 58.3 0.7 13.2 27.8 18.6 72.3 0.2 6.4 21.1 23.1 1980 38.9 0.2 15.3 45.6 11.3 56.3 0.1 9.9 33.7 15.1 1985 -1987 35.4 0.1 8.5 56.0 4.7 51.9 0.1 8.0 40.0 12.1 Age 30-39 1940 69.4 2.5 10.9 17.2 5.3 76.5 0.9 4.6 18.0 7.2 1950 77.0 1.3 11.8 9.9 4.8 86.1 0.4 3.8 9.7 8.3 1960 68.8 1.2 16.0 14.0 11.7 84.4 0.3 5.3 10.0 10.3 1970 68.8 1.2 15.5 14.5 11.1 84.4 0.4 6.3 8.9 10.9 1980 56.9 0.7 22.6 19.8 8.8 76.7 0.2 11.8 11.3 10.4 1985 -1987 57.3 0.4 18.0 24.3 6.6 74.4 0.2 11.4 14.0 12.0 WOMEN Age 16-19 1940 19.4 0.5 3.8 76.3 12.6 12.1 0.2 1.3 86.4 8.3 1950 16.2 0.4 6.5 76.9 7.4 14.0 0.2 2.7 83.1 7.6 1960 14.7 0.2 5.0 80.1 9.6 16.9 0.1 3.1 79.9 10.4 1970 9.7 0.3 4.5 85.5 7.1 11.9 0.2 2.8 85.1 8.3 1980 3.8 0.1 2.0 94.1 3.1 9.3 0.1 2.0 88.6 6.1 1985 -1987 1.8 0.0 0.4 97.8 0.5 6.4 0.0 1.3 92.3 2.5 Age 20-23 1940 51.0 2.0 10.2 36.8 13.6 44.5 0.4 3.4 51.7 14.2 1950 45.7 1.1 16.2 37.0 14.3 55.2 0.4 5.9 38.5 15.7 1960 45.6 0.6 16.0 37.8 20.1 62.3 0.2 6.6 30.9 26.0 1970 37.1 1.1 15.5 46.3 16.8 53.2 0.6 7.7 38.5 24.8 1980 20.4 0.3 9.2 70.1 8.1 39.8 0.2 7.5 52.5 15.4 1985 -1987 13.8 0.1 5.5 80.6 2.6 33.8 0.1 5.8 60.3 11.0 Age 24-29 1940 61.8 4.0 14.3 19.9 6.8 69.8 1.0 4.6 24.6 12.0 1950 63.5 2.1 18.8 15.6 7.7 80.0 0.7 5.7 13.6 13.1 1960 60.3 1.5 21.3 16.9 16.8 82.9 0.5 6.0 10.6 19.9 1970 53.9 1.9 21.7 22.5 12.1 78.5 0.9 8.2 12.4 20.0 1980 37.8 0.9 21.6 39.7 8.5 65.7 0.4 12.5 21.4 15.6 1985 -1987 31.8 0.6 14.3 53.3 4.6 61.5 0.4 11.8 26.3 14.6 Age 30-39 1940 63.5 11.2 15.2 10.1 4.5 78.4 2.7 5.8 13.1 4.3 1950 70.5 5.6 17.1 6.8 7.9 86.6 1.6 5.1 6.7 6.9 1960 63.5 4.2 23.7 8.6 9.1 86.3 1.4 6.3 6.0 7.7 1970 57.9 4.3 26.2 11.6 7.3 83.6 1.5 8.9 6.0 8.2 1980 47.2 3.2 30.6 19.0 5.8 76.8 1.0 14.5 7.7 8.3 1985 -1987 43.1 2.3 29.2 25.4 1.2 73.6 0.9 16.0 9.5 8.0
The seemingly good news on marital disruption must be balanced against the fact that marriage is a seriously troubled institution. For all age groups there has been a large and continuing rise in the incidence of never-married men and women from 1940-1987. It is the rise in the presence of never-married persons which primarily explains the decline in the marriage rate -- especially for African Americans since 1970.
Consider individuals 24 to 29 years old-- individuals who have probably finished their formal schooling, established their own households, and (hopefully) have settled into stable work situations. Among African American males in 1985-87, 35.4% were married versus 61.2% in 1950. The fraction of new marriages among those previously never married, i.e., the marriage rate, was 4.7% in 1985-87 versus 18.6% in 1970. A clear majority of these men (56%) had never married in 1985-87, while just over one-fourth (26.3%) had never married in 1950. Among 24 to 29-year-old African American women in 1985-87, 31.8% were married versus nearly 64% in 1950. The marriage rate among African American women in their late 20s for 1985-87 was 4.6%, down from its 1960 high of 16.8 percent. In 1950, only 15.7% of these women had never married, but by 1985-87 the majority (53.2%) had never married.
By 1980, marriage was no longer the statistical norm among African American women between ages 24 to 39 or for African American men ages 24 to 29. Among African American men in their thirties, the fraction currently married stabilized at about 57% in the 1980s. By age 28, only 50% of African American women have ever married versus 80% of White women (Lichter, et al., 1992:789).
The dramatic decline in marriage among African Americans (and also Whites) means that the fraction of births to unmarried women would have increased from (say) 1960 onward, even if the average number of births per woman aged 15 to 44 had remained the same. Of course, the decline in marriage also would have led to an increase in the fraction of children living in families with only a mother present.
Consider Table 6, which records the fraction of births to unmarried women. It is abundantly clear that the majority of African American children are now born to unmarried women. In 1969, about two-thirds of African American births were to married women. In 1990, about two-thirds of African American births were to unmarried women. However, Table 6 does not tell us whether this increase was due to rising fertility rates among unmarried women or increases in the fraction of unmarried women.
Table 6.
Fraction of Live Births to Unmarried Women, by Race: 1969-90
African American White Reported/Inferred1 1990 0.65 0.20 1989 0.64 0.19 1988 0.63 0.18 1987 0.62 0.17 1986 0.61 0.16 1985 0.60 0.14 1984(2) 0.59 0.13 1983(2) 0.58 0.13 1982(2) 0.57 0.12 1981(2) 0.56 0.12 1980(2) 0.55 0.11 Estimated3 1980(2) 0.55 0.10 1979(2) 0.55 0.09 1978(2) 0.53 0.09 1977(2) 0.52 0.08 1976(2) 0.50 0.08 1975(2) 0.49 0.07 1974(2) 0.47 0.07 1973(2) 0.46 0.06 1972(2) 0.44 0.06 1971(4) 0.41 0.06 1970(4) 0.38 0.06 1969(4) 0.35 0.05
Source: U.S. Department of Health and Human Services:Vital Statistics of the United States, 1990: Volume I - Natality, page 198.
1. Data for states in which marital status was not reported have been inferred other items on the birth certificate and included with data from the reporting states.
2. Based on 100 percent of births in selected states and on a 50-percent sample of births in all other states.
3. Births to unmarried women are estimated for the United States from data for registration areas in which marital status of mother was reported.
4. Based on a 50-percent sample of births.
Table 7.
Birth Rates for Unmarried Women by Age of Mother: African Americans, 1969-1990 (Rates are live births to unmarried women per 1,000 unmarried women in specified group.
Beginning 1970 excludes births to nonresidents of the United States)
15-441 15-19 20-24 25-29 30-34 35-39 40-442 Reported/Inferred3 1990 93.9 110.1 150.6 109.0 64.0 26.5 5.3 1989(4) 93.8 107.9 147.4 106.4 62.8 26.0 5.2 1988(4) 89.3 99.1 137.8 100.5 59.6 25.2 5.2 1987(4) 85.1 93.5 129.8 94.6 55.1 23.4 5.0 1986(4) 81.2 90.6 121.2 87.4 51.8 21.5 4.6 1985(4) 79.0 89.3 116.0 82.0 49.3 21.2 4.4 1984(4,5) 77.0 87.5 110.6 80.4 45.4 20.3 4.5 1983(4,5) 78.0 86.8 109.9 82.4 45.4 20.2 5.0 1982(4,5) 79.8 86.5 112.0 85.5 45.9 20.3 5.4 1981(4,5) 81.3 86.2 113.5 85.8 47.2 20.4 5.8 1980(5) 82.9 89.2 115.1 83.9 48.2 19.6 5.6 Estimated6 1980(5) 83.2 90.3 116.0 82.9 47.0 18.5 5.5 1979(5) 83.0 91.0 114.1 80.0 44.8 19.3 5.9 1978(5) 81.1 87.9 111.4 79.6 43.9 18.5 6.2 1977(5) 82.6 90.9 110.1 78.6 45.7 19.0 6.6 1976(5) 81.6 89.7 107.2 78.0 45.0 19.2 7.0 1975(5) 84.2 93.5 108.0 75.7 50.0 20.5 7.2 1974(5) 85.5 93.8 109.8 80.3 51.8 24.3 6.7 1973(5) 88.6 94.9 116.0 84.5 57.8 27.6 7.7 1972(5) 91.6 98.2 121.2 88.3 57.4 30.4 8.5 1971(7) 96.1 98.6 130.6 99.6 68.6 32.7 10.1 1970(7) 95.5 96.9 131.5 100.9 71.8 32.9 10.4 1969(7) 90.6 90.3 125.3 99.5 70.1 34.3 10.1
Source: U. S. Department of Health and Human Services, National Center for Health Statistics, Vital Statistics of the United States, Volume I, Natality, 1990, pages 198-204.
1. Rates computed by relating total births to unmarried women, regardless of age of mother, to unmarried women aged 15-44 years.
2. Rates computed by relating births to unmarried women aged 40 years and over to unmarried women aged 40-44 years.
3. Data for states in which marital status was not reported have been inferred from other items on the birth certificate and included with data from the reporting states.
4. Rates are revised and therefore may differ from those published in Vital Statistics of the United States, Volume I, for 1989 and previous years.
5. Based on 100 percent of births in selected states and on a 50-percent sample of births in all other states.
6. Births to unmarried women are estimated for the United States from data for registration areas in which marital status of mother was reported.
7. Based on a 50% sample of births.
From at least the late 1950s to the mid-1980s, fertility rates for both African American and White women were on a downward trend1. Therefore, absolute increases in fertility rates cannot explain the increase in the incidence of births to unmarried African American women. Rather, the increase in the fraction of births to unmarried African American women is due to an increase in their relative fertility, i.e., their fertility has fallen less than the rate among married African American women.
Table 8.
Birth Rates for Unmarried Women by Age of Mother: Whites, 1969-1990 (Rates are live births to unmarried women per 1,000 unmarried women in specified group.
Beginning 1970 excludes births to nonresidents of the United States)
15-44 15-19 20-24 25-29 30-34 35-39 40-44 Reported/Inferred 1990 31.8 29.5 46.5 41.7 28.9 14.0 3.1 1989 29.2 27.2 42.4 37.9 25.9 12.7 2.8 1988 26.5 24.5 38.0 34.2 23.4 11.7 2.6 1987 24.5 22.5 35.6 31.0 21.5 10.4 2.3 1986 23.2 21.3 33.3 29.4 19.5 9.4 2.1 1985 21.8 20.3 30.8 27.5 17.7 8.7 1.9 1984 20.1 18.9 27.7 24.6 16.2 8.1 1.9 1983 19.2 18.4 26.3 22.9 15.4 7.5 1.9 1982 18.7 17.6 25.7 22.2 14.7 7.2 2.0 1981 18.1 16.9 25.0 21.5 13.6 7.0 1.8 1980 17.6 16.2 24.4 20.7 13.6 6.8 1.8 Estimated 1980 16.2 15.9 22.4 17.3 10.5 5.3 1.4 1979 14.9 14.6 20.3 15.9 10.0 5.1 1.4 1978 13.7 13.6 18.1 14.8 9.4 4.8 1.3 1977 13.5 13.4 17.4 14.4 9.3 4.9 1.4 1976 12.6 12.3 15.8 14.0 10.1 5.5 1.4 1975 12.4 12.0 15.5 14.8 9.8 5.4 1.5 1974 11.7 11.0 15.0 14.7 9.5 5.5 1.5 1973 11.8 10.6 15.5 15.9 10.6 5.9 1.7 1972 11.9 10.4 16.6 16.5 12.1 6.5 1.6 1971 12.5 10.3 18.7 18.5 13.2 7.2 1.9 1970 13.9 10.9 22.5 21.1 14.2 7.6 2.0 1969 13.4 9.9 23.0 22.5 15.1 7.6 2.0
Source: U. S. Department of Health and Human Services, National Center for Health Statistics, Vital Statistics of the United States, Volume I, Natality, 1990, pages 198-204. See notes accompanying Table 7 for additional details.
For example, in 1970 there were 96 births per 1000 unmarried African American women ages 15 to 44 and 130 births per 1000 married African American women ages 15 to 44 (see Tables 7 and 9, respectively). By 1987, the year when the fertility rate of married African American women ended its three decade decline, relative fertility rates had declined to 85 and 83, respectively. In other words, the fertility rates of both married and unmarried African American women had fallen; however, the fertility rate of married women declined at a faster pace. Indeed, for each year from 1987 to 1990 the fertility rate of unmarried African American women exceeded the rate for married African American women. This is a historic reversal of relative intraracial trends.
Interracial comparisons of fertility produce another puzzle. The fertility rate of unmarried White women increased from 14 births per 1000 unmarried women ages 15 to 44 in 1970 to 25 births per 1000 unmarried women ages 15 to 44 in 1987. During 1970-1987, the fertility rate of married White women ages 15 to 44 declined from 120 to 89.
In other words, as the fertility rate of unmarried African American women declined by 9 points the fertility rate of unmarried White women increased by 11 points. And, the considerably more rapid decline in the fertility rate of married African American women, in comparison to married White women, saw the fertility rate of married African American women fall below the fertility rate of married White women from 1979 onward. This, too, is a historic reversal of relative interracial trends.
Why did the fertility rate of unmarried African American women fall less than the fertility rate for married African American women? Why did the fertility rate of unmarried White women rise when the fertility rate of unmarried African American women was falling? Why has the fertility rate of married African American women fallen below the fertility rate of married White women? We are unaware of any study which has attempted to answer these questions.
Summing up, the stunning increase in the proportion of births to unmarried African American women has two sources. The first is the dramatic decline in marriage. The second is the fertility rates of married women declined faster than fertility rates of unmarried women. In the conclusion, we use the insights obtained from these sections to speculate on the relative inter- and intraracial changes in fertility rates.
Table 9.
Birth Rates for Married Women by Age of Mother: African Americans, 1969-1990 (Rates are live births to married women per 1,000 married women in specified group.
Beginning 1970 excludes births to nonresidents of the United States)
15-44 15-19 20-24 25-29 30-34 35-39 40-44 Reported/Inferred 1990 87.7 569.7 251.1 141.8 82.1 32.9 6.6 1989 85.8 573.4 229.7 140.7 77.7 30.6 6.7 1988 83.1 561.2 227.0 132.2 73.7 28.9 6.1 1987 82.5 481.0 216.2 129.6 72.3 28.2 5.6 1986 84.9 541.0 223.4 132.0 72.5 28.1 5.7 1985 88.0 604.0 228.3 137.8 71.3 28.5 5.5 1984 89.2 505.2 235.0 132.9 72.2 27.8 5.7 1983 88.3 476.4 226.7 126.7 69.9 27.3 5.9 1982 91.2 408.2 225.6 127.4 71.1 27.1 5.9 1981 91.1 380.7 217.3 127.2 68.6 26.3 6.0 1980 94.4 360.2 218.9 129.3 70.6 26.9 6.3 Estimated 1980 94.0 332.3 216.9 130.1 71.4 27.5 6.3 1979 95.6 315.3 222.8 131.9 69.9 27.6 6.8 1978 94.0 353.4 212.3 124.4 65.9 27.5 6.5 1977 94.7 337.3 207.7 125.1 63.1 28.8 7.3 1976 90.4 313.0 191.5 115.5 56.9 27.5 7.2 1975 91.8 318.8 188.7 117.3 54.3 27.6 8.3 1974 93.8 363.0 189.5 113.0 54.9 27.8 8.4 1973 98.2 453.7 191.8 113.0 58.3 29.8 9.9 1972 107.6 490.7 209.5 122.9 66.2 34.3 11.2 1971 121.5 511.7 238.9 139.2 76.5 40.9 13.2 1970 130.3 533.3 263.2 148.3 81.0 44.6 14.2 1969 129.1 514.6 260.2 143.5 81.3 46.4 14.6
Source: U. S. Department of Health and Human Services, National Center for Health Statistics, Vital Statistics of the United States, Volume I, Natality, 1990, pages 198-204. See notes accompanying Table 7 for additional details.
(Section II of this study presents alternative theoretical explanations for changes in marital status and fertility. Section III canvases the empirical literature on these subjects).
Before proceeding to the next sections of this report we should take note that the "baby bust" ushered in during the very late 1950s is over. It would appear that the 1980s may constitute the start of another "baby boom." For unmarried African American women, the fertility rate bottomed out in 1984 at 77 births per 1000 women ages 15 to 44. Thereafter, there has been an upward trend, reaching 94 births per 1000 women by 1990. This upward trend has occurred for all unmarried African American women of childbearing age. Teen births increased by 22 per 1000, just barely ahead of the 19 per 1000 increase among African American women in their early thirties. Unmarried African American women in their early twenties experienced a fertility rate rise from 111 per 1000 in 1984 to 151 per 1000 in 1990, the largest increase of any group. Similarly, African American women ages 25 to 29 experienced a 29 point increase in their fertility rate.
Table 10.
Birth Rates for Married Women by Age of Mother: Whites, 1969-1990 (Rates are live births to married women per 1,000 married women in specified group.
Beginning 1970 excludes births to nonresidents of the United States)
15-44 15-19 20-24 25-29 30-34 35-39 40-44 Reported/Inferred 1990 92.6 404.6 212.1 162.0 97.5 35.6 5.9 1989 91.4 377.8 209.1 157.8 93.9 33.8 5.6 1988 90.4 356.4 205.5 154.3 91.1 31.4 5.1 1987 89.8 348.6 198.9 151.9 88.1 29.6 4.7 1986 90.4 339.4 200.0 148.0 85.9 27.2 4.4 1985 92.9 343.6 202.0 149.4 85.1 26.7 4.3 1984 92.5 344.4 200.8 146.0 82.0 25.3 4.1 1983 93.1 338.8 198.7 145.8 78.7 24.3 4.1 1982 95.6 348.0 202.5 148.2 76.5 23.1 4.0 1981 95.5 327.0 198.5 147.0 72.7 21.3 3.9 1980 96.4 348.6 199.4 145.7 71.2 20.7 4.0 Estimated 1980 97.4 352.0 201.5 146.9 72.0 21.0 4.1 1979 95.8 331.8 193.3 143.5 70.2 20.8 4.2 1978 92.9 318.4 184.3 137.3 66.5 20.0 4.2 1977 94.3 305.2 185.6 138.8 64.0 20.1 4.5 1976 91.1 303.9 176.5 129.8 60.0 19.7 4.6 1975 91.5 309.4 177.0 130.1 58.2 20.1 4.9 1974 93.6 318.5 181.2 132.5 59.5 20.8 5.1 1973 93.8 328.4 179.0 131.7 60.6 22.6 5.8 1972 99.4 364.5 192.3 136.2 64.6 25.4 6.6 1971 111.7 402.5 221.3 154.6 72.1 29.3 7.6 1970 119.6 431.8 244.0 164.9 78.2 32.7 8.8 1969 117.1 423.1 242.5 162.3 79.8 34.2 9.6
Source: U. S. Department of Health and Human Services, National Center for Health Statistics, Vital Statistics of the United States, Volume I, Natality, 1990, pages 198-204. See notes accompanying Table 7 for additional details.
Among unmarried White women (Table 8), the new baby boom began a full decade earlier than among their African American counterparts. After bottoming out at about 12 births per 1000 in 1974, the fertility rate increased to about 32 per 1000 by 1990. Fertility rates among unmarried White women increased for every age group between 15 and 44 years of age. However, women in their twenties -- especially women ages 20 to 24 -- exhibited the largest increases in fertility.
Regardless of race, the baby bust among married women lasted longer and the new baby boom has been slower to materialize among them than among their unmarried counterparts. For both groups of married women 1987 appears to have been the turning point. From 1987 to 1990, the fertility rate among married African American women increased from 83 to 88, while the fertility rate among married White women increased from 90 to 93.
Across racial groups, married women in their early twenties have had the largest impact on the rise in fertility rates from 1987 to 1992. Also, measured from 1987, the rate of increase in fertility among married African American women in their twenties exceeded the rate of fertility increase of unmarried women ages 20-29, while the fertility changes among married and unmarried White women ages 20 to 29 were virtually equal.
But, the emerging baby boom among African Americans has not been matched by an increase in the frequency of marriage. Further, unless there is a turnaround in the fraction of men available for marriage, the proportion of births to unmarried women is unlikely to head downward. Mate availability for African American women is particularly discouraging. (See pages 19-25, 31-44, and 59-62 below for more extensive discussion of mate availability).
The sex ratio in Table 11 indicates the number of unmarried men per unmarried women by age and by race. For example, among unmarried 25-year-olds, there are only 681 unmarried African American males per 1000 unmarried African American women but 1,120 unmarried White males per 1000 unmarried White females.
If we restrict our attention to the ratio of unmarried and employed men per unmarried women (column 2), then there are only 447 men per 1000 women among 25-year-old African Americans. Among Whites, this ratio is near parity, 992 unmarried employed 25-year-old men per 1000 unmarried women. There are even fewer unmarried men with full-time jobs per unmarried woman (column 3). Again, for 25-year-old African Americans, there are just 276 unmarried men with full-time jobs per 1000 women.
Table 11.
Mate Availability Measures, by Race and Age: U.S. Females Ages 18-28, 1979-1986 Mate Availability Measures
Race and Age Sex Ratio Pool of Employed Men Pool of Full-Time Employed Men Pool of Men with Adequate Earnings African American 18 0.858 0.411 0.191 0.174 19 0.838 0.442 0.218 0.203 20 0.829 0.473 0.252 0.242 21 0.797 0.483 0.263 0.260 22 0.760 0.475 0.274 0.281 23 0.745 0.476 0.283 0.299 24 0.700 0.455 0.282 0.302 25 0.681 0.447 0.276 0.304 26 0.637 0.422 0.268 0.302 27 0.644 0.425 0.275 0.316 28 0.647 0.447 0.294 0.374 White 18 1.153 0.753 0.337 0.372 19 1.170 0.818 0.396 0.441 20 1.202 0.885 0.457 0.522 21 1.189 0.906 0.506 0.576 22 1.190 0.934 0.554 0.643 23 1.155 0.919 0.578 0.673 24 1.153 0.934 0.598 0.709 25 1.120 0.922 0.610 0.720 26 1.108 0.928 0.628 0.746 27 1.075 0.906 0.630 0.749 28 1.085 0.893 0.597 0.762
Source: Lichter, et al, 1992, page 791, Table 3.
Finally, column 5 records the ratio when the relevant pool of men is unmarried men whose earnings exceed the poverty threshold for a family of four. There are 304 unmarried 25-year-old African American males with adequate earnings per 1000 unmarried 25-year-old African American women. Yet, there are 720 unmarried 25-year-old White men with adequate earnings per 1000 unmarried 25-year-old White women.
By any measure of mate availability, there is a shortage of African American men. On the other hand, the sex ratio shows a surplus of unmarried White men per White woman. For all measures of mate availability, White women have a larger selection of males to choose from and, unlike African American women, their choices do not decline as they age from 18 to 28.
Economists, sociologists, public policymakers, and others have utilized alternative theoretical perspectives to explain the recent changes in family structure within and between racial groups. This review will pay particular attention to the theoretical and empirical work among economists, although we will freely incorporate work from other disciplines. Among the competing theoretical perspectives, the "African American structural model" and the "new household economics" have generated the most widely accepted explanations of transitions in African American family structure. We discuss each of these theoretical perspectives in the next section. Section III then evaluates the major empirical hypotheses affiliated with these perspectives.
ECONOMIC THEORIES ON TRANSITIONS IN FAMILY STRUCTURE
The African American Structural Model3
Darity and Myers (1995a), like others, wish to explain the rise in female-headed households4. They suggest an economic model of household formations which conforms with Bowman's (1995, 1989, 1988) social-psychological model of intra-family relations. And, in accordance with Wilson (1987) who attributes the rise to the unavailability of marriageable mates, and Bowman who attributes the rise to provider role strain, Darity and Myers suggest that the rise in female-headed households and the rise in the fraction of never-married African Americans, is strongly determined by the economic marginalization -- social unwantedness and economic redundancy -- of African American males.
Marginalized males are clearly unavailable as mates if they have suffered a premature death (violent crimes and accidents) or if they are involuntarily institutionalized, e.g., jail/prison or mental hospital. If not involuntarily institutionalized, marginalized males may be economically unsuitable as mates due to insufficient or unstable earnings. Or, marginalized males may be socially unsuitable as mates because of current participation in illegal activities. Marginalization also may result from insufficient access to health care and information, as well as over-representation in hazardous work, e.g., combat assignment in the military or hazardous industrial employment. By reducing the supply of marriageable males, marginalization is a contributory factor to the rise of female-headed households. Darity and Myers outline a two stage dynamic model of African American male marginalization5. In short, an increase in the probability of marginalization this year lowers the ratio of men in the marriage market, which in turn tends to raise the probability of marginalization next year.
Indeed, as we discuss below, there is solid empirical evidence that for African American men, limited legal earnings opportunities are the primary determinant of drug dealing (Myers, 1992), and increases in joblessness tend to be associated with increases in the incarceration rate of African American men (Myers and Sabol, 1987). So, in the Darity-Myers model of transitions in the African American family structure, secular trends and cyclical shifts in the macroeconomic structure strongly influence microeconomic decision-making in the marriage market through endogenous changes in the ratio of marriage eligible males to females.
Bowman (1995, 1989,1988) also presents a structural analysis of the relationship between economic transformations and family role strains. Provider role strain is likely to be most intense for two groups of men: prime age working class men who have been laid off or who face an insecure employment future; and, young working class men facing a future of limited employment opportunities. Bowman argues that de-industrialization and economic isolation create role strains for African American men, women, and children. However, the provider role strain suffered by African American men has "ripple effects" within the family on women and children. Deindustrialization and economic isolation creates economic insecurity for African American males. This economic insecurity may take the form of an inability to find work, or an inability to find work at a family supporting wage, or an inability to find stable work at a family supporting wage. Consequently, African American males subject to the forces of deindustrialization and economic isolation suffer provider role strain.
Mothers, who bear the primary responsibility for childcare, then are forced to take on additional responsibilities. Further, teenage children of fathers suffering from provider role strain also feel the pressure to replace the income which the father might normally provide. Male provider role strain then increases the role strain on mothers and teenage children and affects the quality of family life.
Short and long run policy responses can have an impact on the quality of family life through two means. They can ameliorate the speed and impact of deindustrialization and social isolation and hence, reduce or increase provider role strain among fathers. Two, policy responses may aid (or worsen) the internal coping strategies of African Americans. Conversely, the coping strategies of African Americans may have an impact on the nature and formation of social policies designed to eliminate deindustrialization and social isolation.
Marginalization has severely reduced both the male-female sex ratio and the fraction of marriage eligible men. It has been reported that a 25% increase in a husband's income reduces the probability of divorce by 8 percent. Among women ages 23-28, African American women are 51% as likely to enter first marriage as White women if one does not statistically control for male marriageability. However, when one controls for male marriageability young African American women are 71% as likely to enter first marriage as young White women. Similarly, if the sample is restricted to women who expect to marry within 5 years the relative probability of marriage increases from 46% to 59% when the pool of fulltime employed men is included as an explanatory factor. Had the Darity-Myers index (DMI) of mate availability remained unchanged from 1976 to 1985, the expected proportion of female-headed households among African Americans would have been roughly .37 in 1985 instead of .45.
The African American structural perspective also argues that strong neighborhoods (Jewell, 1988; Milburn and Bowen, 1991; Hill, 1993) help produce strong families. A family's ability to make positive contributions to society and successfully socialize its offspring depends on the social barriers encountered by the family, the family's economic resources, the institutional support received by the family, and the strength of the family's informal networks. In addition, the strength of the family's informal network is dependent upon the composition of the neighborhood, the nature of commodities exchanged between households, the conditions for exchanging goods and services, and the availability of alternative sources of support.
Jewell argues that after 1960, some social barriers were lowered and some institutional support increased [and per capital family resources increased].That is to say, the 1960s saw the elimination of Jim Crow in the South, higher welfare payments, and higher family income for African Americans. Therefore, Jewell argues, if interfamily networks had remained at the same level of strength, African American families would be much better off, i.e., the average African American family's ability to make positive contributions to society and successively socialize its offspring would have increased.
Together with the Darity-Myers focus on the marginalization of African American males and Bowen's analysis of male role strain and its ripple effects, Jewell's argument then suggests that declining informal ties within African American neighborhoods may be an important causal factor in the decline of marriage.
The New Household Economics on Marriage and Divorce
Among economists, Gary Becker's A Treatise on the Family (1991) summarizes the most widely accepted analysis of marriage, divorce, and fertility. Often referred to as "the new household economics," Becker's choice theoretic approach suggests that individuals marry or divorce if the marginal utility of marriage exceeds or falls below, respectively, the marginal utility of being single, i.e., the individual's share of the marriage output6 must exceed (be less than) his/her household output when he/she remains single.
Individuals combine market goods and time to produce household commodities, e.g., "the quality of meals, the quality and quantity of children, prestige, recreation, companionship, love, and health status" (Becker, 1976, p. 207). Market wages are taken as given. Each individual's ability to combine market goods and time to produce household commodities determines the individual's value of time in household production. Given the wage rate, the price of market goods, and household productivity, each individual divides her/his time between market work, housework, and leisure so as to maximize total household output. Neither involuntary unemployment or underemployment -- economic marginalization -- exists in the new household economics. At a given wage, each person can engage in market work as much or as little as she/he so chooses.
The new household economics suggests strong economic incentives for most adults to marry. First, specialization and trade within the household tends to raise family income above the combined total of two individual incomes. If one person specializes in housework while another specializes in market work, then the combined household output will exceed the output of two otherwise identical single persons. With equitable trading (sharing), both partners can raise their standard of living through marriage. Second, assuming that the demand for one's own children is a major reason for marriage, it is cheaper and more productive to raise one's children in a married household than in separate households of single persons.
So, the supply of wives (husbands) is an increasing function of gains from marriage. The demand for wives (husbands) is a decreasing function of the cost of marriage. An increase in the number of men (women) will also increase the gains to marriage for women (men) and increase the number of marriages. If men are relatively more scarce than women, for example, then they will have more bargaining power regarding the distribution of household output, i.e., the competition for wives among men will be less intense than the competition for husbands among women. Men will gain more from marriage than otherwise identical women. Hence, when the sex ratio is less than one, men should have higher marriage rates than women.
An efficient marriage market will positively sort mates, e.g. highly efficient men tend to marry highly efficient women, if the efficiency of men and women in household production are mutually reinforcing (complementary). Efficiency here is related to the traits of men and women, e.g. intelligence, education, etc., not to market outcomes such as wages. Of course, some traits (such as education) increase market income and some traits (such as love of children) do not increase market income, although the latter traits clearly affect one's relative position in the marriage market. Therefore, the least productive men and women are the most likely to be unmarried. For example, if education increases both household and market productivity then the least educated men and women are the most likely to be unmarried.
New Household Economics on Childbirth and Abortion
The demand for children declines with an increases in the relative price of conceiving and rearing a child. Factors which affect the relative price of children include food and housing costs (inputs in rearing children), child productive contributions (market work by children), government subsidies (AFDC), and the value of parent's time (especially mothers).
The new household economics argues that AFDC encourages fertility because it reduces the relative cost of children and because its reduces female labor force participation. According to Becker, two-thirds of the total cost of raising and producing children is mother's time. Therefore, a decrease in the wage rate reduces the opportunity cost of children, which then encourages an increase in the demand for children7. Because fathers spend relatively little time with children or in housework, Becker reports that changes in the husband's wage rate tend to encourage the demand for children in some studies and reduce the demand for children in others.
Becker surmises that husband's wage has more of a pure income effect on fertility8. Simple theory says that demand for children is positively related to the income of parents. Empirically, the number of children and father's income are negatively related in industrial societies. A more sophisticated construction of the theory says that demand for children and quality of children (measured by such things as the quantity and quality of a child's education) are related to level of parents' income. An increase in income increases parents' demand for improvements in child quality per child (which simultaneously raises the cost of each child) and the number of children. An interaction takes place. Increases in the number of desired children raises the cost of child quality. The rise in the cost of children then reduces the demand for the quantity of children. The quality-quantity interaction implies demand for children is highly responsive to price and income, even though there are no close substitutes for children9. So, at higher levels of income parents will tend to have fewer but higher quality children.
FOUR HYPOTHESES ON TRANSFORMATIONS IN FAMILY STRUCTURE
The empirical literature has focused on four hypotheses regarding transformations in family structure: marginalization of African American males; increases in female independence; the effects of AFDC and other public policies; and more extended schooling of men and women. Women, it is argued, are especially less likely to combine marriage and schooling. The marginalization of African American males hypothesis can be approached in two ways: According to Darity and Myers and the African American structural model, as the economic marginalization of African American males increases, the pool of marriage-eligible males will decrease relative to the number of marriage eligible women and the proportions of female-headed households and never married women should rise also. A straightforward implication of the African American structural model is a higher probability of divorce among African Americans because of relatively greater role strain.
On the other hand, mate availability is purely a demographic variable in Becker's "new household economics" analysis. Assuming economically rational persons, all men and all women are in the market at all times. Some are simply more competitive at finding or retaining mates than others. Given the mate availability ratios, Becker's model clearly predicts that African American males will have a higher marriage rate than African American females and that African American males will have a higher marriage rate than White males -- all other factors being equal.
The female independence hypothesis suggests that increases in female labor force participation or higher wages relative to males will be associated with lower fertility and a lower inclination on the part of women to marry. And, among married women, as economic independence increases there will be a reduced inclination to remain in relationships that are unfulfilling; hence, the divorce rate should rise. Also, since increased schooling may raise a woman's future earnings and therefore make her more attractive in the marriage market and recent school leavers are less likely to marry than those who have been out of school for some time, then increased schooling will tend to delay the age of first marriage and delay the age of first childbirth.
Transfers payments (especially Aid to Families with Dependent Children ) generally have been considered to provide a negative effect on all aspects on family stability, among the proponents of new household economics. A woman is eligible for AFDC if she has a child and (for the most part) is unmarried. Thereby, it is argued that AFDC encourages divorce, non-marriage among pregnant women, and reduced attention to the possibility of pregnancy among sexually active unmarried women (and men). Indeed, once pregnant, it is argued that AFDC pushes the odds of pregnancy resolution away from abortion and toward live birth.
In the African American structural model, on the other hand, AFDC and other transfer payments simply keep families out of poverty. There is no significant hypothesized impact on family transitions. Other factors, such as the availability of marriageable males, dominate the effects of transfer payments.
We now examine the empirical literature with respect to each of these hypotheses and their effects on marriage, divorce, live births, and abortion.
Marriage and Divorce
Male Marginalization
Research into the value African American men place on work and the provider role indicates that African American men identify responsible parenthood as a primary issue in achieving manhood (Connor, 1988). Moreover, to African American men, being a good father means being able to provide economically for one's family (Connor, 1988; Furstenberg, Hershberg & Model, 1975). These findings have been mirrored in samples of White men.
A few studies indicate that African American men regardless of their marital status (or relationship with the mother), employment status, or age, take great pride in being fathers and place a high value on being able to provide for their families. Cazenave's (1979) study of 54 nonprofessional middle-income African American men suggests that men with sufficient financial means to support their families perceive themselves as actively involved fathers and are more likely to engage in other paternal roles. Similar results have been reported by McAdoo (1990). Data from a subsample of 771 adult men taken from the 1979 National Survey of Black Americans indicates that a majority of African American men perceive themselves as good providers.
Several studies report that employed fathers, who are better able to provide for their children, are better able to cope with parental strain than unemployed fathers. Employed fathers of both races appear to be less irritable and more actively involved than unemployed fathers. Danziger and Radin (1990) obtained data from 289 interviews with teenage mothers (of both races and receiving public assistance) on father's willingness to assume paternal responsibilities. The most important predictor of an absent father's involvement with his children was his employment status over the last year. No significant racial differences were found. Grossman, Pollack, and Golding (1988) using a small White sample of intact families found that among employed fathers, greater job satisfaction tends to improve slightly the quality of father-child relationships.
Research on the importance of fulfilling the provider role in men's lives may shed some light on the rise in fatherless households, particularly among African American families. Scholars attribute the rise in female-headed households to the high rate of unemployment among African American men--which is twice that of White men across every demographic variable (Connor, 1988; Furstenberg et al., 1975; Joe, 1987; Johnson & Sum, 1987; McAdoo, 1985-1986; McAdoo, 1988; McLoyd, 1990; McLoyd, Jayaratne, Ceballo & Borquez, 1994). Furthermore, chronic male unemployment is also related to the decline in marriages among African Americans (Connor, 1988; Johnson & Sum, 1987; Testa, Astone, Krogh & Neckerman, 1989).For example, demographic data from the birth certificates of 11,143 babies born to either African American or White teenage mothers indicated that failure to marry among teenage parents is largely due to chronic male unemployment (Hardy & Duggan, 1988). Connor (1988) also reports evidence from several studies that greater economic wellbeing is associated with a higher percentage of marriages and fewer consensual unions. Through interviews with 20 unwed African American teenage fathers, Hendricks (1980) reported that fathers were unhappy with their inability to provide adequate financial support for their children and that such failure was often the source of strained relationships with the maternal grandparents. It is plausible that when faced with such a high risk of failing to fulfill their provider role that many African American men may simply avoid marriage and fail to act as fathers.
Christmon (1990) concluded from questionnaires administered to 43 unwed African American teenage fathers that teenagers with positive self-images were more likely to assume a role in their children's lives. Coupled with the consistent finding that unemployment and underemployment diminish self-esteem and general wellbeing, one implication of this study could be that the large-scale resistance of African American young men to commit to paternal responsibilities may be due, at least in part, to their high levels of under- and unemployment which undermines their self-esteem.
Unemployed young adults appear to be hardest hit psychologically by joblessness. Likewise, self-evaluations of one's performance as a father and provider are more favorable among middle-aged men than among younger men. A substantial implication of this research is that improving the job opportunities of male youth in poor communities may foster their assumption of parental obligations, and in so doing, not only decrease welfare dependency of single mothers but also positively contribute to the well-being of poor children in single-parent homes (Allen, 1985; Danziger & Radin, 1990; McAdoo, 1988; McAdoo, 1990).
Wood (1995) uses 1970 and 1980 census data for 76 Standard Metropolitan Statistical Areas (SMSAs) to examine the relationship between the change in the presence of ever-married African American women ages 20 to 34 (êEVM) and the change in the ratio of marriageable African American males ages 20 to 34 per African American women ages 20 to 34 (êMMPI). Wood also controlled for changes in the fraction of non-marriageable males (ê(nonMarrM/F)).
Following Wilson's (1987) male marriageable pool index, men employed full-time or in the military were considered marriageable; otherwise, the man was considered nonmarriageable. Wood also experimented with income based definitions of marriageability, where the minimum level of income required for achieving marriageability was $6,000, $8,000, and $10,000, respectively, in 1979 dollars10. Men earning above the minimum were considered marriageable11.
Changes in the total ratio of unmarried men to unmarried women consists of changes in the ratio of marriageable men to unmarried women (êMMPI) plus changes in the ratio of nonmarriageable men to unmarried women (ê(nonMarr M/F)). When full-time employment is used as the marriageability criterion, Wood finds that the marriage rate is affected equally by changes in the marriageable male-female ratio and changes in the nonmarriageable male-female ratio. Both variables have a positive correlation with the marriage rate. Therefore, Wood concludes that when full-time employment is the marriageability criterion, the marriage rate is affected by the changes in the sex ratio of unmarried persons, rather than by changes in the marriageable male to female ratio.
However, when alternative levels of income are used as the criterion, an increase in the fraction of unmarried but nonmarriageable males per unmarried female has a negative but statistically insignificant effect on the marriage rate12. Moreover, the impact of a change in the ratio of unmarried but marriageable males per unmarried female is about twice as large when income is the criterion (either $8,000 or $10,000) than when full employment is the marriageability criterion.
In Wood's sample, 55.6% of African American women were ever married in 1980, down from 72.3% in 1970, i.e., a decline of -17 percentage points. Using the $8,000 annual income definition of marriageability, 39.5% of African American males were marriage eligible in 1980, a drop of 4.7 percentage points from 1970 when the rate was 44.2 percent. At the $8,000 criterion, Wood's statistical model suggests that the decline in the marriageability index can account for about 9%13 of the decline in young ever-married African American females14.
Mare and Winship (1991) attempt to capture the determinants of the timing of first marriage. They use data from the 1940 to 1980 Public Use Microdata Sample (PUMS) and from the Current Population Survey (CPS). The sample is limited to persons between the ages 16 to 39 who had never been married or who entered their first marriage during the year prior to the census15. Among African American males a 0.10 increase in the expected probability of employment after marriage increased the odds of marriage by 25% for African American men 20 to 23 years olds and by 15% for African American men ages 30 to 3916. Mare and Winship also found that decreases in expected employment have tended to account for roughly 20% of the decline in the probability of marriage among African American males age 16 to 29 and 14% of the decline for African American males age 30 to 39.
Hoffman and Duncan (1995) examine the economic determinants of divorce for women entering into their first marriage between 1967 and 1983. And, if divorced, the relative probability of entering into divorce with and without receiving AFDC within two years of the postmarriage period17.
Hoffman and Duncan do not include a male marriageability index; however, they do include a variable for race and for husband's earnings18. Hoffman and Duncan report (but do not show) that when race is interacted with all four income variables (wife's wage, husband's income, child support, AFDC guarantee) the interaction terms are insignificant. Accordingly, in the Hoffman-Duncan statistical analysis, African American and White women do not have racially distinct marital stability and welfare receipt probabilities for changes in income-related variables. Although the exact theoretical interpretation of the race variable is ambiguous, Hoffman and Duncan present strong statistical evidence that African American women are more likely to experience divorce than White women -- net of all other variables included in the model19. African American women are both more likely to be self-supporting (i.e. not utilizing AFDC) and divorced and receiving AFDC than White women. However, when age at first marriage is included in the statistical model, the latter is no longer true.
Husband's earnings have a positive impact on remaining married. Column 3 (Spouse Earnings) of Table 12 shows the simulated effect on the baseline divorce and welfare receipt rates of a 25 percent increase ($125.26 per week in 1985 dollars) in husband's earnings. The divorce rate would decline for all women, roughly 8%, regardless of length of marriage, with the steepest percentage point declines for marriages of 1 to 3 years duration. The year 5 cumulative divorce rate would decline by over 7% at the higher male earnings level.
Table 12.
Simulated Annual and Cumulative Divorce and Conditional Welfare Receipt Rates
25 Percent Increase in Baseline Rate AFDC Benefits Spouse Earnings Wife's Wage Rate A. Annual Divorce Rate and Conditional Welfare Receipt Duration 1 - 3 years Divorce rate (%) 5.40 5.64 4.97 4.70 Welfare receipt (%) 28.33 34.96 28.33 19.77 Duration 4 - 5 years Divorce rate (%) 4.34 4.45 3.99 3.87 Welfare receipt (%) 16.06 20.66 16.06 10.66 Duration > 5 years Divorce rate (%) 2.44 2.48 2.24 2.20 Welfare receipt (%) 9.37 12.32 9.37 6.05 B. Cumulative Divorce Rate (%) Year 5 22.50 23.30 20.90 20.00 Year 10 31.50 32.40 29.40 28.40 Year 15 39.50 40.30 36.90 35.90 Hoffman and Duncan (1995), Table 7, page 37
Lichter, et al. (1992) present compelling evidence that previous social science literature has not properly operationalized the demographic and geographic components of the "marriage market."20 First, previous research has simply assumed that all single women are in the marriage market. Lichter, et al. argue that a woman should be considered in the marriage market only if she is either searching for a spouse or expects to marry soon. Accordingly, Lichter, et al., considered a women in the marriage market if she was between 23 and 28 years of age or responded affirmatively to the query of whether marriage was anticipated in the next 5 years. Second, Lichter, et al. define the geographic boundaries of the marriage market on the basis of "residential propinquity." They argue:
Potential mates meet within geographically delimited areas that circumscribe their daily activities at work, in neighborhoods, and in various organizations. In this sense, the local marriage market differs from person-to-person. We use data from the PUMS-D file of the 1980 U.S. Census ... to measure mate availability in local marriage markets. The PUMS-D file includes a labor market area (LMA) geocode for each individual in the sample. These LMAs, which total 382, are identified on the basis of journey-to-work patterns (Lichter, et al.; 1992:786).
The marriage market indicators change with age and individual residential change. Given the marriage market, an unmarried male - unmarried female sex ratio is constructed for each women in each age-race-LMA group.
Alternative measures of mate availability are also constructed, where the numerator is, alternatively, all employed unmarried men, all fulltime employed unmarried men, and all unmarried men with above poverty earnings21.
When the pool of full-time employed men is the MMPI, 5.42% of African American women are predicted to enter first marriage during a given year versus 12.45% of White women22. If African American women faced the same MMPI as White women, 6.94% would enter first marriage during a given year. If White women faced the same MMPI as African American women, their marriage rate would decline to 10.58 percent. So, the marginalization of African American males explains about 1.1 to 1.9 percentage points (or 18 to 32%) of the 6.0 percentage point gap between young African American and White women entering their first marriage.
When Lichter, et al. estimate the probability of entry into first marriage for young women actually in the marriage market, i.e., women ages 23-28 or women expecting to marry within five years, the pool of full-time employed men has a larger impact on interracial differences in marriage23. Among women ages 23 28, African American women are 51% as likely to enter first marriage as White women if one does not statistically control for male marriageability. However, when one controls for male marriageability, young African American women are 71% as likely to enter first marriage as young White women24. Similarly, if the sample is restricted to women who expect to marry within 5 years the relative probability of marriage increases from 46% to 59% when the pool of fulltime employed men is included as an explanatory factor. These estimates show that 24-41 percent of the interracial difference in entry into first marriage is due to African American male marginality.
From the perspective of the African American structural model of household formations, mate availability is not a strictly demographic variable. Rather, mate availability is at least partially determined by macroeconomic cycles and trends, macrosocial policies, and microeconomic decisionmaking which is strongly regulated by these larger social and economic processes. African American male marginalization then is, at most, only tangentially related to dysfunctional individual decisionmaking in a supposedly meritocratic society which has sufficient opportunities for all.
Darity and Myers (1995a, 1990) provide the most sophisticated statistical model of the relationship between marginalization and the rise of female-headed households among African Americans. Their model has three elements. First, mate availability is determined by age, education, residential location, region, unemployment rate, male institutionalization, the mortality ratio, and whether the individual has any children. Second, both the homicide rate (which influences the mortality ratio) and the incarceration rate (which is a component of male institutionalization) are determined by age, education, region, the unemployment rate, percent of population which is African American, percent of population receiving welfare, and residential location. Finally, the proportion of female-headed households is determined by the generosity of welfare, mate availability, number and age of children, age, and education level.
In the Darity-Myers model, an increase in homicides or the incarceration rate will decrease mate availability, as measured by either the Wilson MMPI or the DMI, which in turn increases the proportion of female-headed households25. An increase in the local unemployment rate also directly decreases mate availability and indirectly decreases mate availability through a rise in homicides and incarcerations. According to Darity and Myers a 10 percent reduction in the homicide rate among African Americans would ultimately reduce the number of female-headed households by about 1 percent. A 10 percent reduction in the incarceration rate ultimately will reduce the proportion of female-headed households by 3/10 of 1 percent.
A 10 percent cut in expected welfare would reduce the number of female headed households by 3/4 of 1 percent26. For example, Darity and Myers (1990:28) found that for 1985 a $1,000 drop in welfare benefits would lower the proportion of black female-headed families from 44 to 38 percent, which still is nearly 2 percentage points higher than the 1976 rate of black female-headed families. This 6 percentage point drop in [female-headed families], of course, would require not only that all welfare benefits be eliminated but that former welfare recipients repay the Treasury for the $439.61 difference between their average expected AFDC income of $560.30 (in 1984) and the $1,000 welfare cut.
Darity and Myers (1995a) report four policy simulations and their effects on the proportion of female-headed households: expected welfare payments unchanged between 1976 and 1985; DMI unchanged between 1976 and 1985; African Americans with White DMI ratios; and African Americans with White expected welfare.
Had the Darity-Myers Index (DMI) of mate availability remained unchanged from 1976 to 1985, the expected proportion of female-headed households among African Americans would have been roughly 37 percent in 1985 instead of 45 percent. If African Americans had the White DMI, the 1985 expected proportion of female-headed households would have been 39 percent.
Expected welfare declined for both African Americans and Whites between 1976 and 1985, even as the fraction of female-headed households increased. Specifically African American expected welfare declined from $760 to $567 (after adjusting for inflation), while White expected welfare declined from $127 to $105. Among African Americans, if expected welfare had remained at the 1976 level, the 1985 expected proportion of female-headed households would have been just 46% -- rather than 45 percent.
Gyimah-Brempong (1986), Myers and Sabol (1987), Myers (1992), and Darity and Myers (1990) all look at the process of marginalization, race, and crime. After adjusting for racial differences in poverty, age, and unemployment, Gyimah-Brempong found that race has no statistically significant impact on a series of crimes: robbery, burglary, motor vehicle theft, larceny, murder, rape. However, increases in the nonWhite population of an area were associated with reductions in the probability of aggravated assault. Indeed, contrary to the conventional wisdom, Darity and Myers (1990) found that after controlling for age, education, region, state unemployment rate, percentage of population on welfare, and residential location -- as the African American share of the population increases homicide and incarceration rates decline.
Similarly, Myers (1992) found that if African American males earned the same legal and illegal wages as White males, drug dealing among African American males would drop 90 percent. The clear suggestion from the Myers study is that drug dealing is an all too rational response to low wages. Finally, Myers and Sabol (1987) provide evidence that imprisonment is not simply a process for protecting society from harmful persons -- it is also a process for warehousing the economically marginal. Utilizing census data for 1850 to 1980, Myers and Sabol present strong statistical evidence that "black imprisonment (and Northern imprisonment in general) [is inversely linked to] manufacturing output and [directly linked to] black unemployment" (page 189).
Thus it is safe to conclude that African Americans have no particular cultural affinity for crime. Rather, to some degree, low-waged and inconsistent employment opportunities create an environment which encourages income-oriented crimes, e.g., drug dealing, and thereby also the violent crimes that are often complementary to income oriented crimes. And, it is clear that imprisonment is one of the government's preferred policy options for dealing with the economically marginalized. However, both criminal activity and the state's policy response of greater incarceration tends to increase the fraction of female-headed households among African Americans.
Female Independence
Utilizing time series data from the Current Population Survey for the years 1954-1984, McCrate (1987) constructed an index of women's economic independence as the ratio of their nonmarital standard of living to their marital standard of living27. McCrate shows that the index of women's economic independence rose continuously from 1960-1984 (there was no clear trend from 1954 to 1960) and that African American women's independence index was greater in each year than White women's independence index. She also shows that the percentage of currently married (not separated or divorced) women declines more or less steadily from 1954 to 1984 and that a smaller fraction of African American women are married in any given year than White women. This is especially true after 1966 when the marriage rate for African American women began to decline much faster than the rate for White women and much faster than its own previous trend.
McCrate establishes the minimum set of correlations required for the increased independence hypothesis; however, the correlations do not seem to hold in studies using more detailed regression analyses. It may be the case, especially among African Americans, that McCrate's independence index grows only by comparison to the growing marginalization of African American males.
Accounting for male marriageability also undermines McCrate's correlations. For example, Wood (1995) found that changes in the wage rate of young African American women over the interval 1970-1980 had no statistical impact on changes in the fraction of these women ever married. In fact, Wood found that though statistically insignificant, increases in the wage rate tend to be associated with increases in the fraction of African American women ever married28.
Similarly, Mare and Winship find only weak evidence to affirm the female independence hypothesis for African American women. For African American women under 30 years of age, expected employment had no statistically discernible impact on entry into first marriage. For African American women age 30 to 39, a ten percentage point increase in the probability of employment reduces the odds of marriage by about 4 percent. Contrary to the female independence hypothesis, among African American women ages 16 to 39, an increase in weekly earnings tends to increase the probability of marriage (although weekly earnings is statistically insignificant for women 30-39), but the impact of weekly earnings of entry into marriage is much smaller among African American women than among their male counterparts29. Perhaps, male provider role expectations, i.e., being the "breadwinner," implies a less elastic response in the marriage rate for a given increase in weekly earnings.
Hoffman and Duncan provide evidence that increases in a wife's wage tend to reduce the probability of divorce. Additionally, a wife's wage has a large negative (and statistically significant) effect on the probability of being divorced and receiving welfare as a result of the divorce and not receiving welfare30. In short, as the wife's wage increases, if a divorce occurs, she is more likely to go into a divorce-no receipt of welfare situation than a divorce-receipt of welfare situation. Consider Column 4 (Wife's Wage Rate) of Table 12. Regardless of the duration of marriage, a 25 percent increase in the wife's wage will yield a larger decrease in the probability of divorce than an equal percentage increase in her husband's earnings. Similarly, a 25 percent decrease in the wife's wage will increase the probability of welfare dependence more than an equiproportionate increase in AFDC payments.
Lichter, et al. (1992) also found evidence which contradicts the female independence hypothesis. Among women ages 18 to 28, they found that the probability of entry into first marriage increases with current employment, annual earnings, and the level of education. It may be the case that currently employed, highly educated, and high earning women do have more economic independence -- relative to their unemployed, less educated, and lower earning sisters. However economic independence appears to either elevate a woman in the marriage queue or make her better able to afford a husband and children or both.
AFDC and Public Policy
Wood (1995) found that changes in the generosity of Aid to Families with Dependent Children (AFDC) had a positive but statistically insignificant effect on changes in the fraction of evermarried African American women31. Lichter, et. al found that AFDC had a positive but insignificant effect on entry into first marriage among young women. Among White women, the effect of more generous welfare payments was negatively related to marriage rates (as predicted by the new household economics model) but it was statistically insignificant. In short, Wood and Lichter, et al, as well as Darity and Myers (1995a, 1990), find minimal to no correlation between AFDC and the fraction of women ever married.
Hoffman and Duncan (1995) also found that AFDC has little or no impact on the probability of divorce. For example, in the hazard model specification, an increase in AFDC payment guarantee is marginally associated with an increase in the probability of divorce, but only if the number and age of children is omitted32. This statistical specification then cannot distinguish the effect on AFDC on divorce from the effect of the presence of young children on divorce. That is to say, AFDC marginally increases the probability of divorce if we are willing to assume that the presence of children is not a relevant decision variable in the decision to divorce. In the competing risks specification, Hoffman and Duncan found that AFDC payments have a positive and significant impact on those divorced with welfare receipt, assuming a divorce has occurred. The AFDC effect falls more heavily on welfare receipt than on the probability of divorce. And AFDC-UP33 has no statistically significant impact on remaining married.
Column 2 (AFDC Benefits) of Table 12 quantifies the effect of AFDC on the probability of divorce and divorce with welfare receipt. The impact of AFDC guarantee is less than half the absolute value of the effect of husband's income (Column 3) on the probability of divorce. For example, a 25 percent increase in AFDC will raise the probability of divorce for women married 4 to 5 years from 4.34 to 4.45, while a similar percentage increase in spousal income will decrease the probability of divorce from 4.34 to 3.99. On the other hand, if a divorce has occurred, an increase in the AFDC guarantee does have a large positive impact on the probability that a women will receive AFDC. For women married 4 to 5 years, a 25 percent increase in AFDC will raise the probability of receiving AFDC when divorced by about 25 percent, i.e., from 16.1 to 20.7.
Schooling
Mare and Winship (1991) show that current school enrollment reduces the probability of marriage, especially for women, but the impact of school enrollment on marriage declines with age. Among 20- to 23- year-old African Americans: male students were less than one-half as likely to enter marriage as nonstudents, while female students were less than one-third as likely to enter marriage as nonstudents. For African American women in their thirties, students were only half as likely to enter marriage as nonstudents. Thus, African American men are more likely than African American women to combine marriage and schooling34.
Similarly, among Whites, men also are more likely to combine school and marriage than women. But African American men are even more likely to combine school and marriage than White men and African American women are more likely to combine school and marriage than White women. For example, among White women, students are only 5% (ages 16 to 19), 18% (ages 20 to 23), 40% (ages 24 to 29), and 26% (ages 30 to 39) to marry as nonstudents. Among African American women, students are 12 percent (ages 16 to 19), 31 percent (ages 20 to 23), 58 percent (ages 24 to 29), and 52 percent (ages 30 to 39) to marry as nonstudents.
Among African American women, except teenagers, the probability of marriage rises with educational attainment. For teenagers, high school dropouts are more likely to marry than women with 12 years of education but about equally as likely to marry as women with some postsecondary education. Among African American men ages 16-23, high school graduates (with no postsecondary education) were less likely to marry than dropouts; otherwise, the level of educational attainment was not a significant determinant of entry into marriage. By comparison, among White teens, high school dropouts are more likely to marry than their more educated counterparts. Among 20- to 23- year-olds, White male dropouts continue to be more likely to marry than others but more educated White females were more likely to marry than White female dropouts. Among 24- to- 39-year old White males, dropouts maintained a higher probability of marrying than graduates (although the substantive effect of educational attainment was small). But the probability of entering into marriage increased with educational attainment for White women ages 24 to 39 (with the absolute value of the female coefficient at least three times as large as the male coefficients).
These results lead Mare and Winship (page 192) to conclude,
These patterns suggest that long-term increases in educational attainment may have contributed to declines in the marriage rate among White teenagers since 1960. Among blacks, however, the within-year effect of educational attainment was weak for men at all ages and for teenage women. For women 20 to 39 years old, moreover, education improved marriage prospects, suggesting that educational attainment trends have tended to raise teenage marriage rates.
Current enrollment explained 12% of the decline in marriage for African American male teenagers, 6% for men ages 20 to 23, and virtually none of the decline for African American men 24 to 39 years of age.
Other demographic factors also appear to affect marriage rates. Among young African American women, Wood (1995) found that residing in the South was a positive and statistically significant predictor of changes in the fraction of ever married women. He estimates that "declines in black marriage rates were 3 to 6 percentage points smaller in the South than in the rest of the country."35 Similarly, Lichter, et al. (1992) found that among both African American and White young women, Southern women were 27% more likely to enter first marriage during ages 18 - 28. On the other hand, total size of own-racial population does not have an impact on marriage for either African Americans or Whites (Wood, 1995). They also found that entry into first marriage is uncorrelated with urban residence36.
Mare and Winship (1991) also examine the intertemporal trend in the probability of entry into first marriage. As we have discussed, during a given year, the labor market and schooling variables do have strong effects on the probability of entry into marriage. Yet, the net year effects reported by Mare and Winship suggest that trends in labor market and school variables do not account for trends in the entry into marriage37.
For all Whites, except 24 to 29 year-old males in 1980 and 30 to 39 year-old males in 1950, the net year effects are positive and generally peak around 1960 or 1970. For African Americans, the probability of marriage peeked in 1960. Indeed, 20 to 23 year-old males (females) had a 40 percent (35%) lower probability of marriage in 1980 than 1960. For African American women ages 16 to 19, the net year effect was negative for every census year (but insignificant for 1960). By 1980 a 16- to 19- year-old African American woman had a 35% lower probability of marriage than in 1940 -- exclusive of labor market and schooling effects.
Hoffman and Duncan (1995) present evidence that short duration marriages (1-3 years) are more likely to end in divorce and are more likely to receive welfare (once divorced) rather than remain independent. The exact opposite holds for longer term marriages (greater than 5 years). Further, marriage at less than 20 years of age increases the probability of divorce but has no statistically significant impact on the probability of welfare receipt. Hoffman and Duncan also confirm that the presence of young children (less than age 6) tends to reduce the probability of divorce. Interestingly, however, expected child support has a small negative and insignificant impact on the probability of divorce-no receipt versus divorce-receipt of AFDC.
Despite the statistical strength of their model, Hoffman and Duncan (1995:37-38) offer a rather negative evaluation of the substantive impact of their results:
Factoring in the average changes in income from 1963 to 1971, we find that the divorce rate would have fallen by .6 percentage points if all else had been constant, whereas, in fact, the divorce rate increased to 4.0 percent. Thus, we find no support for assigning blame for the rise in divorce rates to changes in incomes. Throughout the 1970s, our model predicts a very small increase in the divorce rate, and then a very small decrease thereafter. Overall, our evidence suggests that male incomes, wages, and AFDC benefits did not play a large role in the change in the divorce rate over the past few decades and that the trend reflects primarily either changes in behavior or changes in noneconomic factors.
Lichter, et al. found that women with more "traditional" values regarding gender roles were about 2% more likely to enter first marriage than their more non-traditional counter-parts.
Summary
The collective results on marriage and divorce are more consistent with the African American structural model than the new household economics. First, the sex ratio is not simply a demographic variable. It is also an economic variable. By the $8,000 annual income marriageability criterion, Wood found that mate availability can account for 9% of the decline in the fraction of young evermarried African American women. The African American male marriage rate increases with increases in employment and earnings. Indeed, the probability of a wife experiencing divorce declines by 8% with a 25 percent increase in husband's earnings. Further, roughly 20-40% of the interracial difference regarding entry into first marriage can be explained by male marriageability. We also have evidence that when African American male opportunities are block within the formal sector illegal activities increase, which further reduce African American mate availability.
Two, the female independence hypothesis is not consistent with the facts we have uncovered. Regardless of race, increases in a woman's income tend to increase the probability of marriage, decrease the probability of divorce, and decrease the probability of receiving welfare if a divorce does occur. In fact, the negative impact of wife's wages on divorce exceeds the negative impact of husband's wages on divorce. For most women, increases in education, employment, and earnings increase the probability of marriage.
Three, the much abused AFDC program does not encourage divorce and it has no effect on entry into marriage. Fourth, schooling has separate but related effects on marriage. Entry into marriage and current school enrollment are negatively correlated. The impact is larger for women than for men and for Whites than for African Americans. But, the impact of current school enrollment on the marriage rate declines with age. However, for African American women (except teenagers) increases in the level of education are associated with an increase in the probability of marriage. For all African American teenagers and African American men ages 16 to 23, dropouts are more likely to marry than high school graduates with no post-secondary education. Current enrollment delays the age of first marriage, but the probability of marriage rises with years of schooling among persons not currently enrolled in school.
Births and Abortion
Introduction
In the National Longitudinal Survey of Youth (NLSY) data, the African American abortion rate is less than a third of the White abortion rate (Table 13)38. However, Lundberg and Plotnick (1995) provide evidence that both African Americans and Whites underreport the actual number of abortions. Drawing on the work of Henshaw et al. (1989) and the National Center for Health Statistics (1986), Lundberg and Plotnick estimate that the actual ratio of abortions to premarital births is 0.856 for African Americans and 2.060 for Whites (versus 0.145 and 1.313 in the NLSY sample, respectively). Actual African American abortion odds are only 1/2 the White abortion odds. And, if an abortion has occurred, African Americans are dramatically less likely to report it. These differences are consistent with the "child centeredness" of African American culture (McDaniel, 1994, 1990).
Duncan and Hoffman (1991) evaluate the hypothesis that public policy has contributed to "underclass" behavior by teens. Specifically, they evaluate the claim that teen pregnancies and high school dropouts have increased because public policy created an environment that encourages irresponsible behavior.
Political conservatives have argued that as the cost for irresponsible behavior (teen pregnancy and high school dropout) decreases, more individuals become irresponsible. However, the fraction of African American women breaking both rules of "good" behavior, i.e., neither dropping out of high school nor having an out-of-wedlock birth, actually has decreased substantially, while the fraction of White women doing so has slightly increased (Table 14). Finally, regardless of race, there has been a decrease in the fraction of women breaking at least one rule.
Table 13.
Pregnancy, Abortion, and Marital Outcomes of Samples - Weighted Observations, in Thousands
Whites African Number of Persons 4543.4 805.7 Premarital Pregnancies 1103.0 388.2 24.3% 48.2% Less: stillbirths and miscarriages 122.1 48.2 Equal: pregnancies to be resolved by choice 980.9 340.0 Abortions 363.7 40.5 37.1% 11.9% Live Births: 617.2 299.5 Mother married before birth 321.0 19.2 Mother did not marry before birth 296.2 280.3 48.0% 93.6% Percentage of all cases with out-of-wedlock birth 6.6% 34.8% Source. Tabulations of National Longitudinal Survey of Youth. Lundberg and Plotnick (1995), Table 1, page 182.
Even as the fraction of women engaged in "good" behavior (high school graduates and no baby before marriage) has increased, the reward for "good" behavior has decreased. Among African American women, the reward for "good" behavior has decreased relatively and absolutely. And, penalties for "bad" behavior (baby before marriage and/or high school dropout) have increased substantially. For African American teenagers with "good" behavior, 15%, 12%, and 13% lived in impoverished households (upon reaching age 25) during 1967-72, 1973-79, and 1980-85. For African American women with "bad" behavior, the poverty rate at age 25 went from 25% (1967-72) to 24% (1973-79) to 48% (1980-1985). The inflation-adjusted average annual family income (at age 25) for the former group declined from $26,100 to $17,700 (1985 dollars), while the average family income for the latter group declined even more precipitously from $14,300 to $8,30039. Changes in the probability of a young woman dropping out of high school and/or giving birth before marriage do not appear to be caused by public policies which encourage irresponsible attitudes.
Table 14.
High School Graduation and Teenage Out-of-Wedlock Births for Women, by Race, Selected Periods, 1967-85
White African High school graduate, no out-of-wedlock birth 1967-72 82 49 1973-79 84 57 1980-85 86 51 Not high school graduate, and/or had out-of-wedlock birth 1967-72 18 51 1973-79 16 43 1980-85 14 48 Not high school graduate, and had out-of-wedlock birth 1967-72 1 19 1973-79 2 16 1980-85 3 15 Source: Panel Study on Income Dynamics. a. Time period is the calendar year in which a woman was age 25. Duncan and Hoffman (1991: Table 2, page 160)
Childhood Family Structure and Teen Births40
To date, there is little information on the earnings effect of pre-marital parenthood for either teenage or mature fathers. It is a well established fact in the economics literature that married men earn more than single men; however, there is no existing information on whether unmarried fathers earn more or less than either married fathers or men with neither children nor a wife. Hence, we will confine our discussion of teenage pregnancy to the economic causes and consequences for women.
Duncan and Hoffman (1991) estimate the probability of a young African American woman giving birth prior to marriage41. Although they did not include a variable for the presence of a father in the home, Duncan and Hoffman do include measures of family income. In particular, as the level of parents' income increases the probability of out-of-wedlock teen births declines. And, since two parent homes have substantially higher incomes than single parent homes, we may reasonably infer that when a father is present in the home the probability of out-of-wedlock teen births decline.
Olsen and Farkas (1990) provide direct evidence regarding the importance of African American fathers in the home of economically disadvantaged youth42. They measure the number of times the father was present in the home when the girl lived at ages 1, 5, and 10. Olsen and Farkas then find that increasing the presence of the father in the home by one additional time raises the probability of marriage by age 19 years 10 months by approximately 17 percent43. Similarly, Lichter, et al. (1992) found that regardless of race or socioeconomic status, young women who reside with both parents at age 14 are over 16% more likely to enter first marriage during a given year than similar young women not residing with both parents at age 14.
Lundberg and Plotnick (1995) use a nested logit model to examine the sequential outcomes of pregnancy, abortion, and marriage among girls aged 14 16 in 1979 but who were observed from 1979 to 1986. Their analysis is limited to the timing and resolution of first pregnancy. All data is from the NLSY. Unfortunately, the pregnancy and marriage equations of the model performed quite poorly for African Americans. Nevertheless, Lundberg and Plotnick were able to determine that African American young women living in mother only families are more likely to have an unmarried birth, given that the young woman has become pregnant and decided not to abort. Young women from mother only families were more likely to become pregnant and less likely to abort, but the coefficients were not statistically significantly differ from zero44.
Economic Marginalization and Teen Births
Duncan and Hoffman (1991) estimated that a 25 percent increase in a young woman's family income at age 25 would decrease the proportion of African American women having out-of-wedlock births by 2 percentage points. And Olsen and Farkas found that employment opportunity45 has a much larger impact on the probability of out-of-wedlock births among disadvantaged African Americans.
Increasing employment opportunity by 10%, from its mean value of 0.27 to 0.30, increases the probability the young woman has married by about age 20 years, 10 months by 0.008 percentage points (a 16 percent increase). Simultaneously, increasing employment opportunity by 10% reduces the probability the young woman has given birth by age 17.5 years by about 0.062 percentage points (16.67 %). Economic opportunity then, increases the probability of marriage and reduces the probability of childbirth prior to marriage.
Lundberg and Plotnick (1995) found that both African American and White daughters of working mothers and mothers with higher levels of education are more likely to abort46. Among Whites, daughters of working mothers also are more likely to have a marital than nonmarital birth.
AFDC, Abortion Policy, and Teen Births
According to Duncan and Hoffman (1991), a 25 percent increase in AFDC benefits, which had a statistically insignificant effect in their econometric model, would increase the proportion of out-of-wedlock births by only 1 percentage point. This constitutes exactly one-half of the expected impact of a 25 percent decrease in the young woman's family income. Lundberg and Plotnick (1995) find that AFDC increases the probability that a young White woman will remain unmarried prior to birth, but it has no statistically significant impact on the probability of young African American women marrying prior to childbirth47. They find that for young White women, abortion funding by states (whether for all abortions, abortions required for medical and emotional reasons, or limited to cases involving rape, incest, or life endangerment) tends to increase the probability of abortion. Among young African American women, abortion funding has a negative impact on the probability of abortion, but it is statistically significant only when the state funds all abortions. However, the coefficient is quite large, and a negative sign contradicts economic theory.
Among Whites, the availability of abortion increased the probability of abortion48. Among African Africans, the availability of abortion had a negative but insignificant association with the probability of abortion. Lundberg and Plotnick also found that the restrictiveness of abortion laws was negative and statistically insignificant for both African Americans and Whites49. Finally, Lundberg and Plotnick found that restricting the availability, advertisement, or sales of contraceptives increases the probability of young White women becoming pregnant but has a negative and insignificant impact on young African American women.
Additional Factors: Neighborhoods, Schools, and Family Environment
Empirical studies have tried to account for the fact that both cultural attitudes toward and the economic consequences of fertility may vary by region and family environment. The results, however, have been rather mixed. For example, Olsen and Farkas (1990) found in their study of economically disadvantaged African American girls that an increase in the girl's mother's age at first birth by 5 years increases the probability the probability of marriage or cohabitation by about age 20 by approximately 0.002 (a 4 percent increase in marriage and cohabitation). Four additional years of education by the girl's mother also increases the probability the girl has married or cohabited by about age 20 by about 0.002. Duncan and Hoffman (1991) found that neither Southern residence nor urban Northern residence were related to teen birth rates among young African American women during the 1970s.
Utilizing the Public Use Microdata Sample (PUMS) for 1970, Crane (1991) found that "neighborhood quality" has a discontinuous effect on the probability of a young woman, age 16 to 19, having a child50. Neighborhood quality was proxied by the percentage of workers in the neighborhood who held professional and managerial jobs. Teen births might bear a negative correlation with this index because of role model effects, community resources effects, or a biased selection process. As the fraction of professional and managerial workers in the neighborhood increases, the demonstrable payoffs associated with achieving economic success through the "traditional" paths of hard work, deferred gratification, and education tends to raise young people's expectations of the future. Presumably, this would alter youths' behavior away from teenage pregnancy and dropping out of high school.
Or, a rise in the index might be associated with middle-income people using their affluence to bring greater political and economic resources into community, thereby increasing the quality of institutions and services which have an impact on the community. However, Crane argues that the best explanation for the explanatory power of the index is that affluent people have a large set of options regarding neighborhood location and, not surprisingly, they choose to live in "good" neighborhoods. If few professional and managerial workers live in a particular neighborhood, it is most likely the case that the neighborhood is unattractive.
For both African Americans and Whites, an increase in the fraction of professional and managerial workers in the neighborhood tended to decrease the probability of a teenager giving birth51. For a given fraction of high status workers in the neighborhood, African Americans had a higher probability of teenage births than Whites. Except at the lowest level of the index, African American teenagers in the largest cities (population of at least 1 million) have a lower probability of giving birth than African Americans as a whole. Among Whites, except at the lowest level of the index, there are no discernible differences in the probability of giving birth by size of town.
For all African American females, when the index is at 31.2% (the maximum) the probability of having a child is 8.2 percent. As the index declines to 5.6%, the probability of giving birth increases to 12.4 percent. Roughly speaking, a 10 percentage point decrease in the index raises the probability of having a child by 1.64 percentage points. However, a major discontinuity occurs as the index falls below 5.6 percent. When the index is at its lowest value (3.5%), the probability of a teenage African American giving birth rises to 16.1 percent.
For African American females living in the largest cities, the probability of giving birth is 7.4% when 31.2% of the neighborhood workers have professional or managerial occupations. The probability of giving birth increases to 12% as the index of high status workers declines to 6 percent. For big city neighborhoods with the lowest percentage of professional and managerial workers, the probability of an African American teenager giving birth rises to 19.8%; for every 1000 teens, nearly 200 will have an out-of-wedlock birth.
For White females residing in large cities, when the index is at 46.3% (the maximum) the probability of having a child is 0.10 percent. As the index declines to 7.5%, the probability of giving birth increases to 1 percent. Roughly speaking, a 10 percentage point decrease in the index raises the probability of having a child by 0.23 percentage points. So, above 7.5%, extremely large changes in the fraction of White neighborhood workers with professional or managerial occupations have very little impact on the probability of a White teenage female giving birth.
However, a major discontinuity occurs as the index falls below 7.5 percent. When the index is at its lowest value (3.5%), the probability of a teenage White giving birth rises to 10.2 percent52. White teenage females living in the least affluent neighborhoods of large cities appear to have more in common with similarly situated African American teenagers than with other Whites.
If the preferences for abortion vary by neighborhood (as they certainly do by race), then we are unable to tell whether inter-neighborhood differences in the probability of teenage birth are due to inter-neighborhood differences in the probability of getting pregnant versus differences in the probability of aborting a pregnancy. For teenage women living in the least affluent neighborhoods, the timing of births may have only a marginal impact on expected lifetime earnings. Thereby, given that a pregnancy has occurred, there is less of an incentive to abort (relative to teenagers living in more affluent neighborhoods) and hence the probability of giving birth will rise for such teenagers.
Before we can derive useful policy suggestions from the research on teenage childbearing, we must have a clearer idea of how relevant explanatory variables affect the probability of abortion versus the probability of getting pregnant. For example, low income African Americans have very strong religious attachments which tend to reduce the probability of abortion. Hence, what is often invidiously referred to as "underclass behavior" -- a high birth rate among teenage women within low income African American communities -- may simply reflect a lower probability of abortion due to the child-centeredness of African American culture (Hill, 1993; Billingsley, 1992) and strong religious attachments of low income African Americans.
Utilizing data from the 1980 High School and Beyond Survey, Mayer (1991) found that the probability of a girl having a baby between tenth and twelfth grade declines with increases in the socioeconomic status of the girl's parents, as well as the school she attended. For African American girls of average socioeconomic status53, the probabilities of giving birth between the 10th and 12th grade are 8.6%, 7.9%, and 7.6% if she attends a low, average, or high socioeconomic school, respectively. For African American girls of high socioeconomic status, the probability of giving birth is roughly 5% to 5.5%, with the socioeconomic status of the school being relatively unimportant. Similarly, for low socioeconomic status African American girls the socioeconomic status of the school shows little substantive impact; the probabilities are 13% (low SES school), 12.3% (average SES school), and 11.6% (high SES school).
Among African American girls, parent's SES is a considerably more important predictor of teen parentage than the SES of the school. African American girls from low SES homes are over two and one-third times more likely to become teen mothers than African American girls from high SES homes. By contrast, African American girls attending low SES schools are about 12-13% more likely to become teen mothers than girls attending high SES schools. On the other hand, for a given parent-school SES, White girls have a lower probability of giving birth between 10th and 12th grade than African American girls. School SES also has a larger impact among Whites. And, given school SES, the differential effects of parent's SES is much larger among Whites than African Americans. For White girls of average socioeconomic status, the probabilities of giving birth between the 10th and 12th grade are 6.4%, 4.2%, and 2.8% if she attends a low, average, or high socioeconomic school, respectively. For White girls of high socioeconomic status, the probabilities of giving birth are 3.7%, 2.4%, and 1.6 percent. Similarly, the probabilities are 10.9% (low SES school), 7.4% (average SES school), and 5.0% (high SES school). Low SES Whites attending low SES schools tend to have probabilities resembling low SES African Americans, rather than any other group of Whites. White girls from low SES homes are about 3 times more likely to become teen mothers than White girls from high SES homes. By contrast, White girls attending low SES schools are well over 2 times more likely to become teen mothers than girls attending high SES schools.
The intergroup differences between African American and White teen parenting probabilities due to intragroup changes in either school or parent SES suggest that cultural attitudes regarding pregnancy resolution (giving birth versus abortion) and the utilization of pregnancy prevention measures exercise a differential response between African Americans and Whites. However, since Mayer does not control for region, family structure, or mate availability we must be somewhat cautious in accepting this interpretation.
The Hispanic 54 probabilities aid our understanding of relative differences in culture explanation. Like African Americans, according to Mayer, the SES of the school attended by Hispanic girls has almost no impact on the probability of becoming a parent, ranging from 10% to 12.5 percent. On the other hand, low SES Hispanic girls are over three times as likely as to become mothers as high SES Hispanic girls. For example, a low SES Hispanic girl attending an average SES school has a .104 probability of becoming a parent, while a high SES girl would have a .034 probability of becoming a parent if she attends an average SES school.
Summary
This section provides additional evidence consistent with the African American structural model. It also provides additional evidence of different cultural preferences regarding African American and White patterns in family structure and marital stability. For example, African American women ages 14-23 have a lower abortion to live births ratio than similar age White women and, if an abortion has occurred, African American women are much less likely to report it. Indeed, among Whites, interstate differences in abortion funding and the availability of abortion increase the probability of abortion; however, these factors have no impact on interstate differences in African American abortion rates.
Further, we find no evidence that public policy has created an environment which encourages irresponsible behavior. Quite the contrary, public policy may be working to discourage "good" behavior. For example, simple descriptive statistics show that among African American women reaching 25 years of age between 1980 and 1985, high school graduation increased and the likelihood of premarital birth decreased from 1967 to 1985, even as the economic payoff to this "good" behavior was decreasing. More complex econometric analysis showed that a 25 percent increase in AFDC had no effect on out-of-wedlock births among young women. However, AFDC does increase the probability a young White woman will remain unmarried prior to birth, even as it has no impact on the behavior of young African American women.
Teenage out-of-wedlock births have a negative correlation with parent's income and the presence of a father in the home. For economically disadvantaged young African American women, increases in the probability of marriage by (about) age 20 and decreases in the probability of childbirth by (about) age 18 are quite elastic with respect to increases in the employment opportunities available to economically disadvantaged young African Americans (both male and female).
Finally, there is some evidence that the characteristics of neighborhoods are related to family structure. In particular, an increase in the fraction of professional and managerial workers in a neighborhood has a negative correlation with the incidence of teen births. White teens living in the less affluent neighborhoods of large cities have teen birth rates quite comparable to African American teens. For a given fraction of professional and managerial workers in a neighborhood, the African American teen birth rate exceeds the White teen birth rate. However, the socioeconomic status of schools is relatively unimportant for African American teen births, even though it has a larger impact among Whites. For African American teens, parent's socioeconomic status is more important than the socioeconomic status of a teen's school.
THE HISTORICAL CONTEXT OF OUR RESULTS: WHAT HAVE WE LEARNED?
Summary of Results
The cumulative statistical evidence suggests the following conclusions.
- One, mate availability is a substantive economic variable. The best evidence suggests that at least 20-40% of the difference in marriage rates between African American and White women can be explained by differences in the relative supplies of marriageable men.
- Two, AFDC encourages neither divorce nor premarital childbirth. More broadly, there is no evidence of public welfare policy encouraging "underclass behavior." For the most part, it appears the primary effects of AFDC are to increase the economic well-being of low income families, i.e., the purpose for the which the program was designed.
- Three, increases in female independence -- as expressed by increased female-male wage rates, earnings, education, and employment -- increases the probability of marriage and decreases in the probability of divorce.
- Four, increased levels of education increases the probability of marriage, decreases the probability of divorce, and reduces the probability of premarital births among teens.
Collectively, the results of this study suggest that the African American structural model provides the most appropriate framework for understanding current changes and future challenges confronting African American families. This model indicates that increases in economic potential among males also increases father involvement, fathering activities, and the African American marriage rate, while decreasing the probability of divorce. Simultaneously, increases in legitimate economic potential among males reduces the probability of involvement in illegal economic activities. Reductions in illegal activities increases the ratio of marriageable males thereby raising the probability of marriage.
How does male formal economic potential influence family/household structure? Children raised in married couple families will have a higher standard of living -- and all of its attendant benefits -- than children raised in single parent families. Children raised in male householder families with no spouse present will have a higher standard of living than children raised in female householder families with no spouse present. The level of income of African American married couple families has exhibited considerably greater stability from 1979 to the present than either male or female householder families.
The economic marginalization of African American males and its accompanying role strain reduces the probability of marriage and increases the probability of divorce. African American men identify responsible parenthood as a primary issue in achieving manhood. Being a good father means being able to provide economically for one's family. Regardless of their marital status (or relationship with the mother), employment status, or age, African American men take great pride in being fathers and place a high value on being able to provide for their families. Men with sufficient financial means to support their families perceive themselves as actively involved fathers and are more likely to engage in other paternal roles. A majority of African American men perceive themselves as good providers.
Before we can derive useful policy suggestions from the research on teenage childbearing, we must have a clearer idea of how relevant explanatory variables affect the probability of abortion versus the probability of getting pregnant. For example, low income African Americans have very strong religious attachments which tend to reduce the probability of abortion. Hence, what is often invidiously referred to as "underclass behavior" -- a high birth rate among teenage women within low income African American communities -- may simply reflect a lower probability of abortion due to the child-centeredness of African American culture (Hill, 1993; Billingsley, 1992) and strong religious attachments of low income African Americans.
Among African American girls, parent's SES is a considerably more important predictor of teen parentage than the SES of the school. African American girls from low SES homes are over 2 1/3 times more likely to become teen mothers than African American girls from high SES homes. By contrast, African American girls attending low SES schools are about 12-13% more likely to become teen mothers than girls attending high SES schools.
How does male formal economic potential influence involvement in the informal and illegal sector of the economy? What is the nature of the interaction between informal/illegal sector involvement and family/household structure? The best estimate is that a 10 percent reduction in the homicide rate among African Americans would ultimately reduce the number of female-headed households by about 1 percent. By contrast, a 10 percent cut in expected welfare would reduce the number of female headed households by 3/4 of 1 percent. A 10 percent reduction in the incarceration rates will ultimately reduce the proportion of female-headed households by 3/10 of 1 percent.
Gyimah-Brempong found that race has no statistically significant impact on a series of crimes: robbery, burglary, motor vehicle theft, larceny, murder, rape; however, increases in the nonWhite population of an area were associated with reductions in the probability of aggravated assault. Indeed, contrary to the conventional wisdom, Darity and Myers (1990) found that after controlling for age, education, region, state unemployment rate, percentage of population on welfare, and residential location, as the African American share of the population increases the homicide and incarceration rates decline.
Further, Myers (1992) found that if African American males earned the same legal and illegal wages as White males, drug dealing among African American males would drop 90 percent. The clear suggestion from the Myers study is that drug dealing is an all too rational response to low wages. Finally, Myers and Sabol (1987) provide evidence that imprisonment is not simply a process for protecting society from harmful persons -- it is also a process for warehousing the economically marginal. African American imprisonment responds pro-cyclically to macroeconomic fluctuations and fluctuations in the African American rate of unemployment.
African Americans then have no particular cultural affinity for crime. Rather, to some degree, low waged and inconsistent employment opportunities create an environment which encourages income oriented crimes, e.g., drug dealing, and thereby also the violent crimes that are often complimentary to income oriented crimes. And, it is clear that imprisonment is one of the government's preferred policy options for dealing with the economically marginalized. However, both criminal activity and the state's policy response of greater incarceration tends to increase the fraction of female-headed households among African Americans.
Which economic policies, group actions, and individual strategies can increase influence male economic potential in the formal sector, increase African American conjugal family stability, and decrease African American male involvement in the nonformal economic sector, viz., drug dealing and violent crime? The primary policy/action objective should be to increase the rate of marriage (reduce the fraction of never married African American men and women). And, among the married, public policy and the group action of African Americans should aim to reduce the probability of divorce. These are long range objectives which require persistent and broad-based action.
More employment, better jobs, and greater compensation, especially among African American men with a high school diploma or less, must be at the core of any strategy to increase the likelihood and the stability of marriage among African Americans. Concomitantly, both the scope and the economic attractiveness of illegal activities must be reduced. Crime reduction undoubtedly will strengthen neighborhoods and thereby help to promote stronger interfamily networks.
Despite the positive effects of schooling on marriage and the probability of pre-marital childbirth, it is doubtful that more education per se will be a significant contributor to either earnings or the marriage rate. Rapid educational attainment has been one of the most outstanding accomplishments of African American families. African American men and women have made enormous strides toward equality in educational attainment. To date, there is less than one-half of a school year difference in the level of educational attainment of young African Americans and Whites.
Rather, educational improvements designed to strengthen families should be more specifically targeted. The quality of primary and secondary education available to African American children living in neighborhoods with less than the median level of family income should be improved. It is a reasonable conjecture that such improvements will aid neighborhood retention of more upwardly mobile African Americans. Also, postsecondary educational assistance should be increased so as to return African American college matriculation rates to the heights experienced during the early 1970s.
There is no monocausal explanation of the decline in marriage among African Americans. We offer the following scenario as broadly consistent with the reported results. During 1954-1974 African Americans made tremendous strides in educational attainment (United States, 1979). In 1940, 25- to 34- year-old African Americans had a median of 6.9 years of completed schooling, 3.5 years less than the White median. By 1960, 25- to 34- year-old African Americans had a median of 10.3 years of education, just 2 years less than the White median. In 1970 and 1975, young African Americans averaged 12 and 12.4 years of completed education, respectively, or 0.6 and 0.4 years less than the White median. The largest gains were among Southern African Americans during the 1960s. By 1975, Southern and nonSouthern African Americans had virtually identical median years of completed schooling. As high school graduation and postsecondary education became the norm between (say) 1960 and 1975, and therefore current enrollment increased, African American men and women increasingly delayed entry into marriage.
However about the time many of these new high school and college graduates would normally enter into marriage, the long decline in the average real wage rate of men and the increasing inequality in male earnings from 1973 to the present had become a fact of economic life. The declining wage trend was super-imposed on the decrease in African American labor force participation which started in 1940 and bottomed out in the mid-1980s. So, during the post-1973 era male marginalization and earnings instability may have become the lead explanatory factors for explaining the decline in marriage.
The combination of increased schooling and long term marginalization of men then are probably the dominant causal factors behind the structural shift toward "nonmarriage" as the norm among 16- to 29- year olds. As the marriage rate fell among the youngest adults the marriage stock (fraction of currently married persons) began to decline. Further, as the marriage rate fell among the youngest and most fertile women, the relative fertility of unmarried to married women increased.
Despite the accomplishments of this report there are clear limitations. For example, we have left the interactions among race, family formation, and social class underexamined. Yet, American society is rift with class divisions and these divisions have an impact on family formation and marital stability, as well as public policy orientation and implementation. Also, our nearly exclusive focus on economics may lead one to conclude (incorrectly) that we view the relationship between culture and the family as a secondary issue. We cannot adequately address these limitations here, but we can outline how our research might be extended to confront these issues. Partially addressing these limitations presents us with the opportunity to locate our report within the context of the historically developmental approach (Dubois, 1908) versus the culturally emulative (Frazier, 1939) perspective on the African American family.
Class
The African American structural model is part of a larger perspective which emphasizes the role of class divisions in American society. The construction and implementation of public policy, especially social welfare policies, job creation programs, health insurance, legislation dealing with crime and punishment, and other policies which directly affect the social and economic health of families, may be better understood if we examine the nature of class differences.
American class divisions are derived from two sources, the ownership of and control over capital and the control over the analysis, construction, and administration of social policy (Darity, 1991). Nearly four-fifths of the nation's productive property is controlled by just 10% of the nation's families. The economic elite has a near monopoly of the means of production, while the primary income earning asset of most families is the family's supply of labor power (Kennickell and Shack-Marquez, 1992; Smith, 1986). The post-1973 average wage rate has been declining and the level of inequality among wage earning families, as well as the level of inequality between wage income and property income families, has been increasing as family structure and marital stability have been undergoing a dramatic transformation (Darity and Myers, 1995b). This rising inequality between and within social classes has lowered the economic wellbeing of the majority of African American and other families, especially families which depend on the support of younger men, less educated men, and men working in so-called blue collar occupations. We also suspect that the longterm deteriorating economic circumstances of wage earners is a causal factor in the declining fraction of currently married men and women.
Further, the American political and economic elite is racially homogenous (White) while the American nation is decisively and increasingly multiracial (Fullerton, 1989). And, as we have mentioned in the African American structural model, involuntary joblessness is a continuously reproduced feature of the economy. Involuntary joblessness helps to regulate the average wage rate, the intensity of labor, and the quality of working conditions. Finally, the competitive process can sustain persistent racial discrimination in the earnings and employment process (Mason, 1995; Williams and Kenison, forthcoming), in addition to sustaining persistent discrimination in housing and credit markets (Dymski, forthcoming; Fix and Struyk, 1993).
The class divisions between workers and the economic elite and the racial struggles between Whites and African Americans is mediated by armies of private and social managers. Managerial control over ideas, information, and knowledge has become a major source of power. Social managers are the manufacturers of policy. They are concerned with social structure stability rather than social structure transformation.
Their vehicle for autonomous action -- independent of the direct control of capital -- has been the central government.... From the perspectives of the managers, both capital and the working class must be contained for the social good. Of course, the managers themselves necessarily define the social good. Idealization of "pluralist democracy" represents the essence of managerial thinking on conflict containment (Darity, 1983:62-65).
Black social managers are the intermediaries between the African American working class and the White managerial and capitalist classes. By contrast, White social managers -- the executive partners of the managerial class -- are the intermediaries between the racially privileged social classes, as well as the intermediaries between racially privileged and African American social classes.
Future sociological, economic, and political science research should demonstrate the precise manner in which public policy and group action are governed by competing racial and class interests. Evaluations of this process should aid in the development of more effective strategies for combating marital instability, the rise in the fraction of nevermarried adults, and the declining standard of living for African American and other families.
Culture
Dubois (1908) presented the first scientific analysis of the African American family. Among his lasting contributions, Dubois "established the agenda for studying the African family as a unique cultural population" (McDaniel, 1994:59). That is to say, slavery did not destroy the African family but rather created political, social and economic conditions which lead to the transformation of the African family. The cultural prerogatives which Africans brought to America -- child centered families, consanguine rather than conjugal kinship systems, extended versus nuclear families, marriage as a merger of families rather than simply the joining of individuals -- conditioned the manner in which slaves responded to the exegeses of slavery and the abusive power of the slave master (McDaniel, 1990). The unique cultural heritage of African Americans suggests that African American family structure and marital stability would differ from Whites, even if there were no interracial differences in social and economic circumstances. But, African American families have been confronted with social and economic circumstances of a qualitatively and quantitatively different character than those confronting White families.
Of course, this more extended perspective on the African American structural model, which combines an emphasis on mate availability with an understanding of social classes, persistent discrimination, and a unique cultural heritage, differs sharply from the new household economics and the sociological tradition which asserts that African American family life is "pathological" (Moynihan, 1965).
Moynihan's infamous report on the African American family was constructed on the basis of four unquestioned axioms, some of which are also easily derived from the new household economics and all of which are at variance with the African American structural model. First, both Moynihan and the new household economics assume that family structure and marital stability determine economic well-being. For example, he argued that, "At the heart of the deterioration of the fabric of Negro society is the deterioration of the Negro family. It is the fundamental source of the weakness of the Negro community at the present time" (Moynihan, 1965:5).
On the other hand, Moynihan strongly suggests that the impact of economic circumstances on family structure and marital stability are minimal. For Moynihan and many of today's social managers, family structure and marital stability are, for the most part, determined by the cultural practices of social groups. Loury (1989) refines this line of argument by focusing on the importance of household values and the social capital of neighborhoods, where social capital refers to market functional values, attitudes, and behaviors. Murray (1984) and Mead (1986) extend the logic of this axiom one step further when they argue that public assistance to the needy and targeted employment programs ultimately reduce social capital. Therefore, they assert that these programs create more unstable families and marriages which thereby worsen problems of poverty, joblessness, crime, and welfare dependency. Murray and Herrnstein (1994) pushed the issue back to the late 19th century, arguing that genetics is a more powerful determinant of family structure and stability than culture.
Third, Moynihan assumed that African Americans do not possess a unique or distinctive culture of their own. African American culture is assimilative. After arriving in American, Africans shed all of their cultural past and begin to emulate the values, customs, traditions, and (to the extent possible) the family structure of Whites. Thereby, differences between African American and White families indicates pathology in African American families.
In agreement with the new household economics of today, Moynihan accepted the axiom that American society is meritocratic. Equal opportunity for upward mobility and full employment are the norms of the economy. The market is a great equalizer. Left to its own devices, a competitive economy will eliminate racial discrimination in market outcomes (Becker, 1957). Just as White ethnic groups were able to overcome discrimination through stable families, deferred gratification, hard work, the pursuit of education, and the avoidance of crime -- or so the argument goes -- African Americans can overcome social and economic barriers confronting them once they have learned to sufficiently emulate White or "mainstream" cultural values. For Moynihan and the new household economics, interracial differences in the economic wellbeing of families is ultimately related to interracial differences in family values; differences in social class and the pervasive impact of racism are irrelevant.
However, as we have suggested, the bulk of empirical evidence tends to support the African American structural model. Certainly, there are remaining theoretical and empirical issues which this framework must address. But, the extended African American structural model's emphasis on mate availability, social classes, the unique cultural heritage of African Americans, and the continuing impact of racial discrimination in market processes presents the most advantageous framework from which to derive productive solutions to problems confronting African American families.
REFERENCES
Allen, W. (1985). Race, income and family dynamics: A study of adolescent male socialization processes and outcomes. In M. B. Spencer, G. K. Brookins, & W. R. Allen (Eds.). Beginnings: The social and affective development of Black children. Hillsdale, NJ: Erlbaum.
Becker, G. (1957). The economics of discrimination. Chicago: University of Chicago Press.
Becker, G. (1976). The economic approach to human behavior. Chicago: University of Chicago Press.
Becker, G. (1991). A treatise on the family (Enlarged ed.). Cambridge, MA: Harvard University Press.
Bianchi, S. & Spain, D. (1986). American women in transition. New York: Russell Sage Foundation.
Billingsley, A. (1992). Climbing Jacob's Ladder: The enduring legacy of African-American families. New York: Touchstone.
Bowman, P. (1988). Post-industrial displacement and family role strains: Challenges to the Black family. In P. Voydanoff & L. Majka (Eds.), Families & economic distress (pp. 75-101). Newbury Park, CA:.
Bowman, P. (1989). Research perspectives on Black men: Role strain and adaptation across the adult life cycle. In R. L. Jones (Ed.), Black adult development and aging (pp. 117-150). Berkeley, CA: Cobbs & Henry.
Bowman, P. (1995). Commentary. In M. Belinda Tucker & Claudia Mitchell-Kernan (Eds.), The decline in marriage among African Americans: Causes, consequences, and policy implications (pp. 309-322). New York: Russell Sage Foundation.
Butz, W. & Ward, M. (1977). The emergence of countercyclical U.S. fertility. (RAND Report R-1605-NIH). Santa Monica, CA: RAND.
Cazenave, N. (1979). Middle income Black fathers: An analysis of the provider role. Family Coordinator, 28, 583-593.
Christmon, K. (1990). Parental responsibility and self-image of African American fathers. Journal of Contemporary Human Services, 71, 563-567.
Connor, M. (1988). Teenage fatherhood: Issues confronting young Black males, in J. T. Gibbs (Ed.). Young,, Black and male in America: An endangered species (pp. 188-218). Dover, MA: Auburn House.
Cox, O. (1940). Sex ratio and marital status among Negroes. American Sociological Review, 5, 937-947.
Crane, J. (1991). Effects of neighborhoods on dropping out of school and teenage childbearing. In C Jencks & P. E. Peterson (Eds.). The Urban Underclass (pp. 299-320). Washington, D.C.: The Brookings Institution.
Danziger, S. & Radin, N. (1990). Child support in paternity cases. Social Service Review, 64(3), 458-474.
Darity, Jr., W. (1983). Reaganomics and the Black community. In S. Weintraub & M. Goodstein (Eds.). Reaganomics in the stagflation economy (pp. 59-77).
Darity, Jr., W. (1984). Does welfare dependency cause female headship. Journal of Marriage and the Family, 46, 765-779.
Darity, Jr. W. (1990). Impacts of violent crime on Black family structure. Contemporary Policy Issues, 8(4), 15-29.
Darity, Jr., W. (1991). Underclass and overclass: Race, class, and economic inequality in the managerial age. In E. Hoffman (Ed). Essays on the economics of discrimination (pp.67-84). Kalamazoo, MI.: W. E. Upjohn Institute for Employment Research.
Darity, Jr., W. & Myers, Jr., S. (1995a). Family structure and the marginalization of Black men: Policy implications. In M. B. Tucker & C. Mitchell-Kernan (Eds). The decline in marriage among African Americans: Causes, consequences, and policy implications (pp.171-203), New York: Russell Sage Foundation.
Darity, Jr. W. & Myers, Jr., S. (1995b). The widening gap: A summary and synthesis of the debate on increasing inequality. Unpublished manuscript, University of Minnesota.
Dubois, W. (1908). The Negro American family. Atlanta, GA: Atlanta University Press.
Duncan, G. & Hoffman, S. (1991). Teenage underclass behavior and subsequent poverty: Have the rules changed? In C. Jencks & P. E. Peterson (Eds.). The urban underclass (pp. 155-174). Washington, D.C.: The Brookings Institution.
Dymski, G. (in press). Why does race matter in housing and credit markets: Current research and future directions. In P. Mason & R. Williams (Eds.). Race, markets, and social outcomes. Norwell, MA: Kluwer Academic Publishers.
Ellwood, D. & Crane, J. (1990). Family change among Black Americans: What do we know? Journal of Economic Perspectives, 4(4), 65-84.
Fix, M. & Struyk, R. (1993). Clear and convincing evidence: Measurement of racial discrimination in America. Washington, DC: Urban Institute Press.
Frazier, E. (1939). The Negro family in the United States. Chicago: University of Chicago Press.
Fullerton, H. (1989 November). New labor force projections, spanning 1988 - 2000. Monthly Labor Review, 6-74.
Furstenberg, F., Hershberg, T., & Model, J. (1975). The origins of the female-headed Black family: The impact of the urban experience. Journal of Interdisciplinary History, 6(2), 211-233.
Grossman, F., Pollack, W., & Golding, E. (1988). Fathers and children: Predicting the quality and quantity of fathering. Developmental Psychology, 24(1), 82-91.
Gyimah-Brempong, K. (1986). Empirical models of criminal behavior: How significant a factor is race? Review of Black Political Economy, 15(1), 27-44.
Hardy, J. & Duggan, A. (1988). Teenage fathers and the fathers of infants of urban, teenage mothers. American Journal of Public Health, 78(8), 919-922.
Hendricks, L. (1980). Unwed adolescent fathers: Problems they face and their sources of social support. Adolescence, 15(60), 861-869.
Henshaw, S., Kenney, A., Somberg, D., & Van Vort, J. (1989). Teenage pregnancy in the United States: The scope of the problem and state responses. New York: Alan Guttmacher Institute.
Herrnstein, R., & Murray, C. (1994). The bell curve: Intelligence and class structure in American life, New York: Free Press.
Hill, R. (1993). Research on the African American family: A holistic perspective. Westport, CT: Auburn House.
Hoffman, S. & Duncan, G. (1995). The effect of incomes, wages, and AFDC benefits on marital disruption. Journal of Human Resources, 30(1), 19-41.
Jewell, K. (1988). Survival of the Black family: The institutional impact of U.S. social policy. New York: Praeger.
Joe, T. (1987). The other side of Black female-headed families: The status of adult Black men. Family Planning Perspectives, 19(2), 74-76.
Johnson, C. & Sum, A. (1987). Declining earnings of young men: Their relation to poverty, teen pregnancy, and family formation, Washington, DC: Children's Defense Fund.
Kennickell, A. & Shack-Marquez, J. (1992). Changes in family finances from 1983 to 1989: Evidence from the Survey of Consumer Finances, Federal Reserve Bulletin,78,1-18.
Lundberg, S. & Plotnick, R. (1995). Adolescent premarital childbearing: Do economic incentives matter? Journal of Labor Economics, 13(2), 177-200.
Lichter, D., McLaughlin, D., Kephart, G., & Landry, D. (1992). Race and the retreat from marriage: A shortage of marriageable men? American Sociological Review, 57(December), 781-799.
Loury, G. (1989). Why do we care about group inequality? In S. Shulman and W. Darity, eds., The question of discrimination: Racial inequality in the U.S. labor market, Middletown, CT: Wesleyan University Press, pages 268 - 292.
Macunovich, D. (1995). The Butz-Ward fertility model in the light of more recent data, Journal of Human Resources, 30(2), 229-255.
Mare, R. & Winship, C. (1991). Socioeconomic change and the decline of marriage for Blacks and Whites, in Christopher Jencks and Paul E. Peterson, editors, The urban underclass, Washington, D.C.: The Brookings Institution, pages 175-202.
Mayer, S. (1991). How much does a high school's racial and socioeconomic mix affect graduation and teenage fertility rates? in Christopher Jencks and Paul E.Peterson, editors, The urban underclass, Washington, D.C.: The Brookings Institution, pages 321-341.
McAdoo, J. (1985-1986). A Black perspective on father's role in child development, Marriage and Family Review, (3-4),117-133.
McAdoo, J. (1988). Changing perspectives on the role of the Black father, in P. Bronstein and C. P. Cowan, editors, Fatherhood today: Men's changing role in the family, New York: John Wiley, pages 79-93.
McAdoo, J. (1990). Understanding African American teen fathers, in P. E. Leone,editor, Understanding troubled and troubling youths, New York: Sage, pages 229-245.
McCrate, E. (1987). Trade, merger, and employment: Economic theory on marriage, Review of Radical Political Economics, 19(1), 73-89.
McDaniel, A. (1990). The power of culture: A review of the idea of Africa's influence of family structure in antebellum America, Journal of Family History, 15(2), 225-238.
McDaniel, A. (1994). Historical racial differences in living arrangements of children, Journal of Family History, 19(1):57-77.
McLoyd, V. (1990). The impact of economic hardship on Black families and children: Psychological distress, parenting, and socioeconomic development, Child Development, 61, 311-346.
McLoyd V, Jayaratne, T., Ceballo, R., & Borquez, J. (1994). Unemployment and work interruption amng African American single mothers: Effects on parenting and adolescent socioemotional functioning, Child Development, 65, 562-589.
Mead, L. (1986). Beyond entitlement: The social obligations of citizenship, New York: The Free Press.
Milburn, N. & Bowen, P. (1991). Neighborhood life, in James S. Jackson, editor, Life in Black America, Newbury Park, CA: Sage Publications, Inc., pages 31-45.
Morgan, S., McDaniel, A., Miller, A., & Preston, S. (1993). Racial differences in household and family structure at the turn of the century, American Journal of Sociology, 98(4), 798-828.
Moynihan, D. (1965). The Negro family: The case for national action. Washington, D.C.: Office of Policy Planning and Research, U. S. Department of Labor.
Murray, C. (1984). Losing ground: American social policy, 1950-1980, New York: Basic Books.
Myers, S. (1992). Crime, entreprenuership, and labor force withdrawal, Contemporary Policy Issues, 10(2), 84-97.
Myers, S. & Sabol, W. (1987). Unemployment and racial differences in imprisonment. Review of Black Political Economy, 16(1-2), 189-209.
National Center for Health Statistics. (1986). Advance Report of Final Natality Statistics,1984. Monthly Vital Statistics Report, Vol. 35, No. 4 Supplement. Washington,DC: U.S. Government Printing Office.
Olsen, R. & Farkas, G. (1990). The effect of economic opportunity and family background on adolescent cohabitation and childbearing among low-income Blacks, Journal of Labor Economics, 8(3), 341-362.
Smith, J. (1986). The concentration of wealth in the United States: Trends in the distribution of wealth among American families. Washington, DC: Joint Economic Committee, United States Congress.
Testa, M., Astone, N., Krogh, M., & Neckerman, K. (1989). Employment and marriage among inner-city fathers, The Annals of the American Academy of Political and Social Science, 501, 79-91.
U.S. Department of Commerce. Bureau of the Census. (1979). The Social and EconomicStatus of the Black Population in the United States: An Historical View, 1790-1978. Washington, DC: U.S. Government Printing Office.
U.S. Department of Commerce, Bureau of the Census, Current Population Reports, P20-480. (1995). The Black Population of the United States: March 1994 and 1993.Washington, DC: U.S. Government Printing Office.
U.S. Department of Health and Human Services, National Center for Health Statistics.(1995). Vital Statistics of the United States, 1990: Volume I -Natality.Washington, DC: U.S. Government Printing Office.
Williams, R. & Kenison, R. (forthcoming). The way we were?: Discrimination, competition, and inter-industry wage differentials in 1970. Review of Radical Political Economics.
Wilson, W. (1987). The truly disadvantaged: The inner city, the underclass, and public policy. Chicago: University of Chicago Press.
Wood, R. (1995). Marriage rates and marriageable men: A test of the Wilson hypothesis, Journal of Human Resources, 30(1), 163-193.
NOTES
1 See Bianchi and Spain (1986) for data on the years prior to 1969.2 The tables clearly show larger increases in the fertility of teens than any other group. However, only 2% of teens had ever married during 1987 (see Table 5). Hence, the teenage impact on the overall increase in marital fertility is quite limited.
3 "African American structural model" refers to a line of thought which Darity and Myers (1995) report begins with Cox (1940). However, neither Darity and Myers nor other studies actually employ this phrase. It is used in this report for two purposes. One, it provides a convenient shorthand for summarizing an important body of work which greatly differs from and is not compatible with the work of Gary Becker (discussed below), the dominant voice among both sociologists and economists. Second, although African Americans are not the only contributors to the "African American structural model," the phrase does force one to recognize that the manner in which African Americans understand their own history and families is somewhat incompatible with interpretations provided by the predominantly White social science academy.
4 The works of Antonio McDaniel (1994 and 1990) and Morgan, et al. (1993) also suggest that African American cultural history conditions the manner in which African Americans structure their family in the face of racism, economic exploitation, and social isolation. That is to say, African American families and transitions in family structure are not mere imitations of Whites. Rather, African Americans possess a unique cultural heritage which offers its own solutions for adjusting to either economic hardship or economic prosperity. The relatively greater prevalence of extended families among African Americans - even after controlling for such things as income and education - offers a case in point. Unfortunately, there are no studies in the economics literature which address this point.
Lerman (1989) attempts to interpret the African American structural model, viz., the works Wilson (1987) and Darity and Myers (1995a), from the perspective of Becker's new household economics. Of course, he then finds the data inconsistent with his forced assimilation of these models.
5 For Darity and Myers, marriage-eligible males are those unmarried men who are either labor force participants or enrolled in school. The current marriage eligible ratio of men to women is a negative function of the extent of last period's marginalization (incarceration and mortality). However, it is also the case that the current period probability of marginalization is a positive function of the current period marriage eligible ratio of men to women. All unmarried women are considered marriage eligible. Darity and Myers make imaginative use of Bayes' theorem to develop their model. Let P(ULF|male) represent the Bayesian prior probability distribution of the probability that a male is an unmarried labor force participant or student. Let P(male) represent the probability of being male. Then the Bayesian posterior probability distribution of being an unmarried labor force participant or student is P(UMLF)=P(ULF|male)*P(male).
6 Let Qm represent the composite output of a married couple household, with Qmm and Qmf representing the amount of the composite married couple output accruing to men and women, respectively. Let Qsm and Qsf represent single person composite household outputs for men and women. The gain to marriage is Qmf - Qsf = Qm - Qmm - Qsf. No woman will marry if Qmm - Qsf < 0 and men gain (women lose) from marriage as Qmm rises. The supply of (demand for) wives rises (falls) as Qmf - Qsf increases.
7 In this instance, opportunity cost refers to reductions in wage income that occur when a woman reduces her time working in the market in order to increase the amount of time associated with childcare.
8 Wage changes have a "substitution" and "income effect" on fertility. Consider a wage increase. The opportunity cost of children will now be higher, thereby husbands will tend to demand fewer children, preferring instead to substitute more market hours for hours that might otherwise have been spent in childcare. On the other hand, a higher wage rate means higher income; hence, husbands will now be able to afford more children. For men, the argument is that the latter income effect is much greater than the former substitution effect since men spend little time in childcare.
9 Becker then derives a number of hypotheses. First, as quality per child increases the relative price of children increases which leads to a decline in fertility. Second, as the price elasticity of quality increases the relative price of children decreases which leads to an increase in fertility. Third, as the ratio of average fixed to variable cost of children (quality) increases the relative price of children increases (decreases) which leads to a decrease (increase) in fertility.
10 In 1979, $7,421 per year was required to keep a family of four out of poverty.
11 EVM is a stock variable, hence it is the result of present and past MMPIs. A flow variable should be used but is unavailable in dataset. Let Z represent time varying but SMSA constant factors, e.g., year, F represent variables which affect the marriage rate across SMSAs but which remain constant over time, and X represent factors which vary over time and over SMSAs. Wood's empirical model is:
(1)EMVit = a'Zt + bêMMPIit + g'êXit + f'Fi + eit.
He argues that "some elements of F are unobserved and correlated with MMPI, so OLS on (1) would yield a biased estimate of bä In order to avoid this source of biasä" the equation is differenced to yield:
(2)êEVMit = a + bMMPIi + g'êXi + ni, and we might also try
(2A)êEMVit = a* + b1êMMPIi + b2*'ê(nonMarr M)/Fi + m*i.
The equation actually estimated:
êEVMit = a* + b1êMMPIi + b2*'ê(nonMarr M)/Fi + b3South + b4êlog(black population) + b5êlog (wf) + b6êlog(Max AFDC) + m*i.
Equation was estimated with and without b2*'ê(nonMarr M)/Fi for each of the four definitions of the MMPI; hence 8 regressions were estimate for both African Americans and Whites.
12 In the Lichter, et al. (1992) model, which is discussed below, the authors argue that the local sex ratio consists of economic (MMPI) and noneconomic components (nonMarr F/F). They find that regardless of the MMPI used, the noneconomic or demographic component is statistically insignificant. However, both the coefficients and standard errors of MMPI uniformly rise quite substantially when nonMarr M/F is included in the model with only minor changes in the coefficient of the race variable. This leads Lichter, at al. to conclude, "These results support an economic interpretation that stresses, among other factors, the primacy of men's employment circumstances in the marital decisions of young women" (page 796).
13 This represents 1.6 percentage points out of a total of 17 percentage point decline.
14 For Whites, the coefficient on êMMPI was substantively small and statistically insignificant. Among young White women, 73.4% were ever married in 1980, down -7.3 percentage points from 1970. At the $8,000 criterion, 68.3% of White men age 20-34 were marriage eligible in 1980, up from 67% in 1970.
15 The model estimated by Mare and Winship had the form:
Probability of entry into marriage = f(age, race, sex, educational attainment, whether currently enrolled in school, weekly earnings squared, postmarriage expected employment status, census year).
Mare and Winship recognize that the actual postmarriage employment status may be endogenous a cause as well as a consequence of marriage. Accordingly, they obtain "an estimate of employment potential based on educational attainment and work experience during the previous year, characteristics that are not affected by marriage. We constructed this marriage by regressing the employment status for all respondents at the time of the census on their weeks of work during the previous year and their educational attainment" (page 190).
16 "A $100 increase in weekly earnings raised the odds of marriage by about 30 percent for black men age 20 to 23 and by about 20 percent for those age 30 to 39. The employment and earnings effects for white men were remarkably similar to those for black men in every age group. For both races, these effects were large" (Mare and Winship, 1991:191).
17 Hoffman and Duncan do not distinguish between marriage and cohabitation or between divorce and separation. Their data comes from 1968 1987 cross-year family-individual file of the Panel Study on Income Dynamics (PSID). From a methodological perspective, there is no reason to distinguish marriage from cohabitation, even though cohabiting couples are more likely to "divorce" (Billingsley, 1992). Indeed, Lichter, et al. (1992) found that cohabitation among 18- to 28- year-old young women increases the probability of entry into first marriage by 67% - only race had a larger statistical impact.
18 The theoretical structure of the Hoffman-Duncan model is described below.
(1)VijtM = VM(aIjtM, Wit, Zi, Xijt, mijt)
(2)VitD = VD(IitD, Wit, Zit, Xijt)
VijtM = discounted utility stream for woman i married to husband j, evaluated at marital duration t
aIjtM = discounted stream of earned income of j from duration t to the end of the planning horizon and a is the proportion of IjtM that is available for i's consumption
IitD = nonmarital exogenous income (AFDC, child support, income from a future spouse)
Wit = wage income of woman i evaluated at marital duration t
Zi = time invariant individual traits
Xijt = observable characteristics of the marriage, e.g., number and age of children
mijt = quality of marriage
Duncan and Hoffman use both a hazard rate specification and a nested logit model competing risk to estimate probability of divorce and utilization of AFDC.
19 The hazard specification included both women with and without children. It employed the following variables:
Probability of Divorce (married women w/children) = f(wife's wage rate, husband's earnings, child support, 5 year average AFDC guarantee, race, duration 1 3 years, duration > 5 years, AFDC-UP available, South, age at marriage < 20, year 1968 1974, year 1979-83, number of children, whether child < age 6).
AFDC-UP is a dummy variable for the presence of the Aid to Families with Dependent Children - Unemployed parent program.
The competing risks model was limited to married women with children. It is specified as:
Prob Divorce/Welfare Receipt Married woman w/children = f[wife's wage rate (dw), wife's wage rate (m), husband's earnings (dw), husband's earnings (m), child support (m), child support (dn), 5 year average AFDC guarantee (dw), 5 year average AFDC guarantee (m), race (dw), race (m), duration 1-3 years (m), duration 1-3 years (dw), duration > 5 years (m), duration greater than 5 years (dw), AFDC-UP available (m), age at marriage < 20 (dw), age at marriage < 20 (m), inclusive value].
The inclusive value was statistically insignificant in all specifications, even negative in the FIML model, and reduced to less than half its original value in the fully specified equation. This indicates the model is similar to a bivariate probit, where the decision is to remain married versus getting a divorce. The divorce alternatives are perfect substitutes, a change in the relative attractiveness of divorce-receive AFDC versus divorce-no AFDC does not affect the probability of remaining married, only the probability of which state of divorce to enter.
20 The Lichter, et al. model of the probability of entry into first marriage is:
Prob. of marriage = f(race, age, family background, current living arrangement, value orientation, economic independence, local marriage market).
Family background includes the socioeconomic status of parents when the woman was age 14, whether woman lived with both parents at age 14, and years of education of mother. Current living arrangements include whether woman is cohabiting, living with a parent, has own child present, or is enrolled in school. Value orientation includes whether woman lives in an urban area, the South, or has traditional values, i.e., believes "a woman's place is in the home," "traditional husband (wife) roles are best," or "women are happier in traditional roles." Economic independence includes years of education, employment status, earnings, and whether receives public assistance. Alternative measures of the local marriage market include sex ratio, pool of employed men, pool of fulltime employed men, pool of men with adequate earnings.
The data include 6,288 noninstitutionalized young women ages 14 to 22 years on January 1, 1979. The data were examined for women 18 years of age and older in the 1979 through 1986 waves of the NLSY. Only first marriages after age 18 are included in the sample. Women in the military were excluded from sample and the measures of local mate availability also excluded men in the military.
21 The poverty threshold for the sample was $7,412/year (1979 dollars) for a family of four.
22 We focus on fulltime employed men as the appropriate MMPI because among all alternative measures of the MMPI it was consistently significant regardless of whether sample was limited to women ages 23 28, women expecting to marry, or all women. And, it always had the largest coefficient and consistently had the greatest impact on the size of the race coefficient. (If the MMPI is the sole or a major explanatory factor in interracial differences in the probability of marriage, then the size and statistical significance of coefficient on race should decline as accurate measures of the MMPI are included in the model.)
23 Women ages 18 22 are much more likely to be completing their education and therefore are less likely to be in the marriage market than 23-28 year olds.
24 In fact, when the pool of fulltime employed men is included in the model the coefficient on the race variable becomes statistically insignificant.
25 DMI represents the Darity-Myers index of mate availability, which is the fraction of unmarried men either employed or enrolled in school divided by the fraction of unmarried women. The Wilson MMPI limits marriageable males to unmarried but employed men.
26 These elasticities are from Darity and Myers, 1990. Darity and Myers (1995) report an expected welfare elasticity of 0.0924, i.e., a 10 percent cut in expected welfare will reduce the proportion of female-headed households by 9/10 of 1 percent.
27 Specifically, the index of women's economic independence = yf/[a(yf + ym)], where yi includes all earned and unearned income and a = women and children's share of married-couple family income = (K + 1)/(K + 2), i.e., women, men, and children share equally in married-couple income, where K is the number of children.
28 In contrast, the impact of the economic opportunities of White women on changes in the fraction eve married was very large, negative, and statistically significant. Wood (1995:187) writes:
My estimates suggest that a 0.10 log increase in this index should result in a 4 to 5 percent decline in the marriage rates of whites [across SMSAs]. However, unlike median earnings for black women, which rose sharply over the 1970s, median earnings for white women remained unchanged over the period. Thus, although greater female economic opportunity is strongly negatively associated with marriage rates for whites, my results suggest that none of the decline in white marriage rates between 1970 and 1980 should be attributed to the increasing earnings of white women.
29 Mare and Winship find more support for the female independence hypothesis among White women. An increase in expected employment reduces the probability of marriage and the size of this coefficient increases with age. "A 0.10 increase in the expected probability of employment reduced the odds of marriage by 3 percent for women age 16 to 19ä[and] 13 percent for those 30 to 39." The latter effect was similar to the size of the corresponding positive coefficient for White men. An increase in earnings tended to increase the probability of marriage among White women over 20, but reduce the probability of marriage of White women age 16-19. In any event, the substantive impact was quite small.
30 Statistical significance means that we can rule out the possibility that the measured impact of one variable (say, wife's wage) on another variable (say, probability of divorce) is simply a chance correlation in the data. Statistical insignificance means that we cannot rule out the possibility that changes in wife's wage are not correlated with changes in the probability of divorce, any measured correlation between the two variables is coincidental and not "likely" to be observed in repeated sampling.
31 The generosity of AFDC, commonly referred to as "welfare," is captured by the maximum AFDC payment available to a family of four in the state where the women resided.
32 By marginally associated, I mean it has a 10% level of statistical significance.
33 AFDC-UP was an optional program extending AFDC benefits to married couples when a parent became unemployed. One would expect it to decrease the probability of divorce.
34 For African American males ages 24 to 39, current enrollment has a negative but statistically insignificant impact on entry into marriage.
Lichter, et al. (1992), also found that school enrollment has a negative but statistically insignificant impact on the probability of entry into first marriage among women ages 18 to 28.
35 Southern residence had no impact on White marriage rates.
36 The coefficient was negative but statistically insignificant.
37 Net year effects, captured by dummy variables for 1950, 1960, 1970, and 1980, explain changes in the probability of marriage after accounting for age, schooling, and labor market effects, and account for 75% 129% of the total proportion of changes in the probability of marriage since 1940.
38 This data is for girls 14 16 years of age in 1979, but who were observed from 1979 to 1986.
39 For White teenagers with "good" behavior, no more than 2 to 3% lived in impoverished households (upon reaching age 25) each year between 1967 1985. For White women with "bad" behavior, poverty rate at age 25 went from 6% (1967 72) to 10% (1973 79) to 22% (1980 85). Average family income (at age 25) declined from $30,600/year to $28,600/year (1985 dollars), while average family income for the latter group declined even more precipitously from $21,700 to $17,600.
40 Empirical attempts to confirm the theoretical hypothesis of the economic model of fertility have met with disappointment. Macunovich (1995) provides a definitive evaluation of the Butz-Ward (1977) aggregate specification of Becker's model. Utilizing individual level data from the 1964 1987 Current Population Survey, Macunovich shows that for women aged 20 24, 25 29, and 30 34 that neither total fertility nor the fertility of married women has the expected correlations with husband's income, wife's wage rate (market participants), or the value of wife's time (nonmarket participants).
Economic theory predicts a positive correlation between husband's income and wife's fertility. Macunovich shows a statistically significant negative relationship. Theory predicts a negative relationship between the value of a wife's time (market wage for working wives or fraction of husband's income for nonmarket participants); the data show precisely the opposite. These are "perverse" results.
41 The Duncan and Hoffman sample is drawn from the PSID. It includes African American women who were teenagers between 1968 and 1978, were 25 years of age between 1980 and 1985, and remained in the PSID throughout their teenage years. If the woman received AFDC anytime within two years of childbirth, the birth was treated as AFDC related.
The econometric results (with standard errors in parenthesis) were:
Probability of out-of-wedlock teenage birth = 3.264 - 0.524(1og earned family income at age 25)
(3.407) (0.235) + 0.132(log AFDC) - 0.197(Parental income $10,000 - $20,000) (0.311) (0.392) - 0.734 (Parental income > $20,000) + 0.596(Parental income from AFDC) (0.445) (0.330) +0.355 (Urban North) - 0.189 (South) (0.493) (0.513) X2 = 24.46 N=232
All income variables are measured in 1985 dollars.
42 The Olsen-Farkas data were derived from the Youth Incentive Entitlement Pilot Projects (YIEPP), a program aimed at creating employment for the economically disadvantaged. Individuals were 14-19 years of age in June 1978 and had not graduated from high school. The youths were surveyed in the spring of 1978, 1979, 1980, and 1981.
The mean values for age at which girl's mother gave birth = 26 years, education of girl's mother = 9.35 years, number of times biological father was present at ages 1, 5, and 10 = 1.73, number of times grandfather was present at ages 1, 5, and 10 = .05, and employment opportunity = .27.
Probability of cohabitation = f(age at which girl's mother gave birth, education of girl's mother, number of times biological father was present at ages 1, 5, and 10, number of times grandfather was present at ages 1, 5, and 10, employment opportunity), where employment opportunity = local employment rate for African American, 17-year-olds, same-local-labor-market youths on a month specific basis (male and female rates are combined to yield a single rate), and cohabitation = either marriage or living together.
Probability of nonmarital births = f(age, first interval, employment opportunity) (This is a fixed effects model so time invariants characteristics, e.g., family background, are differenced out of the model.)
The mean values for age at which girl's mother gave birth = 26 years, education of girl's mother = 9.35 years, number of times biological father was present at ages 1, 5, and 10 = 1.73, number of times grandfather was present at age I, 5, and 10 = .05, and employment opportunity = .27.
43 The presence of a grandfather in the home of economically disadvantaged African American girls is strongly negatively correlated with the probability of marriage (its coefficient is 2.24 times larger than the coefficient for the presence of girl's father), but the data does not discriminate between grandfathers in extended households versus grandfathers as primary care givers because of a missing parents.
44 Among Whites, girls living in mother-only families were less likely to have a marital birth and more likely to get pregnant and equally likely to abort.
45 Employment opportunity is measured by the local employment rate for African American 17 year olds, same-local-labor-market youths on a month specific basis with male and female rates combined to yield a single rate.
46 Women ages 18 28 with more educated mothers are also less likely to enter first marriage than otherwise identical women with less educated mothers (Lichter, et al., 1992).
47 Lundberg and Plotnick find that, for Whites, a 20 percent increase in AFDC raises the probability of pregnancy from .246 to .249, raises the conditional probability of carrying the pregnancy to term from .595 to .631, and increases the conditional probability of being unmarried from .431 to .536. Hence, a 20 percent increase in AFDC - about $116/week (1983 dollars) - raises the unconditional probability of premarital birth from .063 to .084.
48 Availability was defined as, "Percentage of population living in counties with facilities providing at least 400 abortions annually, as of 1980" (Lundberg and Plotnick, 1995:197).
49 Restrictiveness is determined by the date abortion became legal in a state, the presence of parental consent laws, and whether second trimester abortions had to be performed in a hospital.
50 The sample was limited to teenagers living with parents. Crane (page 302) argues that,
[S]ocial problems almost certainly affected both the probability that teenagers would move out of their parents' home and the quality of the neighborhoods into which they moved. In a model that assumed one-way causality and made no adjustment for this simultaneity effect, the effects of neighborhoods on teenage behaviors would probably be overestimated.
In addition, Crane made no distinction between married and unmarried teenagers. Presumably, relatively few teenagers living at home were married, and fewer still were married prior to giving birth.
51 The probability of a teenager giving birth is modeled as a function of family income, parents' education, occupation of family head, household structure, rural origin, race, population size, region, residual mobility, and neighborhood effects. Accessibility and liberality of public assistance as an explanatory variable is not considered. Only the neighborhood effects were reported by Crane.
52 For all White females the latter probability is about 3.5 percent.
53 Socioeconomic status is a composite variable constructed from father's and mother's education, whether family is a professional, teacher, manager, proprietor, or owner, whether family owns a home, whether family owns two or more cars, and whether family owns a dishwasher. The components of SES were weighted according to their relative within-school effects on teenage childbearing. Low SES students have SES scores at least one deviation below the mean. High SES students have SES scores at least one deviation above the mean. All other students were classified as medium SES. The school SES is the mean SES of students attending the school.
54 Mayer does not report the fraction of the Hispanic population which is of Mexican, Cuban, Puerto Rican, or other national origin.
Year:
1996 