Divorce Rates, Marriage Rates, and the Problematic Persistence of Traditional Marital Roles

Family Law Quarterly

Volumn 34, Issue 1, 2000, Pages 1-42

  Ellman, Ira Mark ^a  

^a NONE

Author keywords

[No Author keywords available]

Indexed keywords

EID: 0034563015     PISSN: 0014729X     EISSN: None     Source Type: Journal    
DOI: None     Document Type: Article

References (137)

1. The two states are Louisiana and Arizona. LA. REV. CODE § 9-224 et seq.; ARIZ. REV. STAT. § 25-901 et seq.
2. The two states are Louisiana and Arizona. LA. REV. CODE § 9-224 et seq.; ARIZ. REV. STAT. § 25-901 et seq.
3. Ira Ellman, The Misguided Movement to Revive Fault Divorce, 11 INT'L J. L., POLICY & FAM. 216 (1997); Ira Ellman & Sharon Lohr, Dissolving the Relationship Between Divorce Law and Divorce Rates, 18 INT'L REV. L. & ECON. 341 (1998).
4. Ira Ellman, The Misguided Movement to Revive Fault Divorce, 11 INT'L J. L., POLICY & FAM. 216 (1997); Ira Ellman & Sharon Lohr, Dissolving the Relationship Between Divorce Law and Divorce Rates, 18 INT'L REV. L. & ECON. 341 (1998).
5. Marital Status and Living Arrangements: March 1994, BUREAU OF THE CENSUS, CURRENT POPULATION REPORTS, SERIES P-20-484, (1996), Table A-9.
6. See the sources described in IRA MARK ELLMAN, PAUL KURTZ & ELIZABETH SCOTT, FAMILY LAW: CASES, TEXT, PROBLEMS 930-31 (3d ed. 1998).
7. See the discussion of the Swedish experience in Hoem & Hoem, infra note 94.
8. Ellman & Lohr, supra note 2. Similar chart on all the states is provided in Ellman & Lohr, which also corrects the data to exclude regional effects on divorce rates. Such regional influences are very great, as explained in more detail later in this article. Because there are also regional patterns in divorce laws, a careful examination of the impact on rates of a state's adoption of no-fault requires some kind of control for these regional effects. One approach is to look at a state's residual changes in divorce rates over time, after removing the average changes of the other states in its region, over the same time period. Examination of the data after such an adjustment confirms the point in the text. See Ellman & Lohr, supra note 2.
9. An article appearing after Ellman & Lohr claims to find a relationship between no-fault divorce laws and divorce rates. Leora Friedberg, Did Unilateral Divorce Raise Divorce Rates?, 88 AM. ECON. REV. 608 (1998). Friedberg relies on Brinig's compilation of the year in which the states adopted no-fault divorce. In several cases we used different years, believing Brinig to be incorrect, but it is not obvious that this difference alone would explain the different results. A more likely explanation is a combined effect of two other differences between her methodology and ours. If one posits that no-fault divorce was the result, not the cause, of rising divorce rates, one would expect a pattern in which each state's shift to no-fault occurs some years after divorce rates in that state begin to rise steeply. That is in fact the pattern that Lohr and I found. But we looked at the divorce rates between 1960 and 1992, while Friedberg examines the period 1968 to 1988. Her truncated time series excludes for many states a period during the 1960s that preceded any legal change but which included steeply rising divorce rates. She might therefore overestimate the impact of no-fault in the early-adopting states. This possible problem could be exacerbated by her method for correcting for the nonlegal factors that affect divorce rates over time. She finds a significant association of divorce rates with no-fault divorce only if she includes in her regression a term that corrects for state-specific trends in divorce laws. But if for each state the rates tend to rise steeply for some years before enactment of no-fault, and level out or decline within a few years after enactment (a pattern we found common), then correcting for each state with state-specific trends effectively excludes this real phenomenon from the analysis, thus biasing it against the competing claim that no-fault was the result rather than the cause of rising divorce rates. Nonetheless, some correction for trends over time is appropriate. We made the correction in a different way. Our assumption was that nonlegal factors affecting divorce rate trends, such as social or demographic changes, would be relatively homogenous within regions, so that for each state the average of the other states within its region could provide a basis for the correction that was not affected by the date at which that particular state changed its law. For this purpose we divided the country into ten regions. Our method may also have its drawbacks, and we do not claim certainty on the best approach to this problem. Friedberg's truncated time series may alone explain her results, however. Professor Lohr ran Friedberg's model (including her correction for state-specific effects), extending the time series back to 1960 and forward to 1992. She also used our dates, rather than Brinig's, for the year in which each state adopted no-fault. The resulting coefficient for no-fault was not significant.
10. For a fuller account of the material that follows, see IRA ELLMAN ET AL., supra note 4, at 191-98, on which this summary relies.
11. ANDREW CHERLIN, MARRIAGE, DIVORCE, REMARRIAGE 46 (rev. ed. 1992).
12. Id. at 126.
13. As cited in notes 14, 15, and 16 of Stephanie Ventura et al., Declines in Teenage Birth Rates, 1991-97: National and State Patterns, NATIONAL CENTER FOR HEALTH STATISTICS, 47 NAT'L VITAL STAT. REP. No. 12 (Dec. 17, 1998).
14. See Ventura et al., supra note 11, notes 14 through 17.
15. Id.
16. The rate declined from 82.1 birth per 1,000 unmarried women in 1994 to 75.9 in 1995 and 74.4 in 1996. CENTER FOR DISEASE CONTROL, NATIONAL CENTER FOR HEALTH STATISTICS, 46 MONTHLY VITAL STAT. REP. No. 11, Supplement (June 30, 1998), at Table 18.
17. The birth rate for unmarried black women peaked at 90.7 per 1,000 in 1989; by 1996 it had declined to 74.4 per 1,000. The rate for whites increased from 30.2 in 1989 to 38.3 in 1994, before declining to the next two years, to 37.6. Id. 16. The property crime rate is less than half the 1973 rate. Robbery rates fell 32% just between 1991 and 1997, while homicide rates fell 31%. The national data is taken from Decline of Violent Crimes Is Linked to Crack Market, N.Y. TIMES, Dec. 28, 1998, at A16. Homicide rates in New York City have now fallen below their levels in 1964, when the Beatles first appeared on the Ed Sullivan Show. N.Y. TIMES, Dec. 24, 1998, at A1.
18. The birth rate for unmarried black women peaked at 90.7 per 1,000 in 1989; by 1996 it had declined to 74.4 per 1,000. The rate for whites increased from 30.2 in 1989 to 38.3 in 1994, before declining to the next two years, to 37.6. Id. 16. The property crime rate is less than half the 1973 rate. Robbery rates fell 32% just between 1991 and 1997, while homicide rates fell 31%. The national data is taken from Decline of Violent Crimes Is Linked to Crack Market, N.Y. TIMES, Dec. 28, 1998, at A16. Homicide rates in New York City have now fallen below their levels in 1964, when the Beatles first appeared on the Ed Sullivan Show. N.Y. TIMES, Dec. 24, 1998, at A1.
19. Rates from 1870 through 1960 taken from Table 9 in 700 Years of Marriage and Divorce Statistics, United States, 1867-1967, NATIONAL CENTER FOR HEALTH STATISTICS, VITAL AND HEALTH STAT. SERIES 21, No. 24, DEP'T OF HEALTH, ED. & WELFARE PUB. NO. (HRA) 74-1902 (1973). Rates from later years taken from the Marriage and Divorce, NATIONAL CENTER FOR HEALTH STATISTICS, III VITAL STATISTICS OF THE UNITED STATES, 1987, DHS Pub. No. (PHS) 91:1103, and the equivalent reports issued annually in the preceding years and in 1988. Subsequent to 1988 the NCHS stopped reporting these statistics on a 4-region basis. It stopped compiling them altogether in 1990. From that time forward only provisional divorce statistics for each year are available on a state by state basis. Divorce statistics are apparently difficult to compile and the NCHS chose to save the costs of improving their collection. See the notice at 60 Fed. Reg. 64,437-64,438 (1995), also available at 〈http:// www.cdc.gov/nchswww/datah/datasite/frnotice.htm〉.
20. Rates from 1870 through 1960 taken from Table 9 in 700 Years of Marriage and Divorce Statistics, United States, 1867-1967, NATIONAL CENTER FOR HEALTH STATISTICS, VITAL AND HEALTH STAT. SERIES 21, No. 24, DEP'T OF HEALTH, ED. & WELFARE PUB. NO. (HRA) 74-1902 (1973). Rates from later years taken from the Marriage and Divorce, NATIONAL CENTER FOR HEALTH STATISTICS, III VITAL STATISTICS OF THE UNITED STATES, 1987, DHS Pub. No. (PHS) 91:1103, and the equivalent reports issued annually in the preceding years and in 1988. Subsequent to 1988 the NCHS stopped reporting these statistics on a 4-region basis. It stopped compiling them altogether in 1990. From that time forward only provisional divorce statistics for each year are available on a state by state basis. Divorce statistics are apparently difficult to compile and the NCHS chose to save the costs of improving their collection. See the notice at 60 Fed. Reg. 64,437-64,438 (1995), also available at 〈http:// www.cdc.gov/nchswww/datah/datasite/frnotice.htm〉.
21. Rates from 1870 through 1960 taken from Table 9 in 700 Years of Marriage and Divorce Statistics, United States, 1867-1967, NATIONAL CENTER FOR HEALTH STATISTICS, VITAL AND HEALTH STAT. SERIES 21, No. 24, DEP'T OF HEALTH, ED. & WELFARE PUB. NO. (HRA) 74-1902 (1973). Rates from later years taken from the Marriage and Divorce, NATIONAL CENTER FOR HEALTH STATISTICS, III VITAL STATISTICS OF THE UNITED STATES, 1987, DHS Pub. No. (PHS) 91:1103, and the equivalent reports issued annually in the preceding years and in 1988. Subsequent to 1988 the NCHS stopped reporting these statistics on a 4-region basis. It stopped compiling them altogether in 1990. From that time forward only provisional divorce statistics for each year are available on a state by state basis. Divorce statistics are apparently difficult to compile and the NCHS chose to save the costs of improving their collection. See the notice at 60 Fed. Reg. 64,437-64,438 (1995), also available at 〈http:// www.cdc.gov/nchswww/datah/datasite/frnotice.htm〉.
22. What could be true is that the West's high divorce rates made it more receptive to no-fault reforms, or that cultural or demographic factors distinctive to the West that contribute to its high divorce rates also contribute to a political climate hospitable to no-fault divorce.
23. Although it appears that for Americans born after 1930, a Catholic upbringing has no association with the likelihood of divorce, even though it did for Americans born before then. William Sander, Catholicism and Marriage in the United States, 30 DEMOGRAPHY 373 (1993). Interfaith marriages remain associated with a higher divorce rate, Evelyn Lehrer & Carmel Chiswick, Religion as a Determinant of Marital Stability, 30 DEMOGRAPHY 385 (1993), and perhaps they are more common in the West than in the Northeast - although one would not expect them to be more common in the South than in the Northeast.
24. Although it appears that for Americans born after 1930, a Catholic upbringing has no association with the likelihood of divorce, even though it did for Americans born before then. William Sander, Catholicism and Marriage in the United States, 30 DEMOGRAPHY 373 (1993). Interfaith marriages remain associated with a higher divorce rate, Evelyn Lehrer & Carmel Chiswick, Religion as a Determinant of Marital Stability, 30 DEMOGRAPHY 385 (1993), and perhaps they are more common in the West than in the Northeast - although one would not expect them to be more common in the South than in the Northeast.
25. A third possibility, that those who move are disproportionately in an age range during which persons are also more likely to divorce, is probably wrong. In 1998, over 30% of adults in their twenties moved, while 22% of those between 30 and 34 did so. The percentages drop considerably with increasing age, until one gets to persons older than 74. Geographical Mobility March 1997 to March 1998, BUREAU OF THE CENSUS, CURRENT POPULATION REPORTS P20-520 (Jan. 2000), at Table B. For those born between 1948 and 1950, the average age at divorce was 33.8, a decline of several years from the average age at divorce for those born between 1908 and 1912. Because the age at first marriage has steadily increased for birth cohorts after 1950, the age at divorce probably has as well. CHERLIN, supra note 9, at 68 (including note 5). It thus seems that those who move tend to be younger than those who divorce.
26. A third possibility, that those who move are disproportionately in an age range during which persons are also more likely to divorce, is probably wrong. In 1998, over 30% of adults in their twenties moved, while 22% of those between 30 and 34 did so. The percentages drop considerably with increasing age, until one gets to persons older than 74. Geographical Mobility March 1997 to March 1998, BUREAU OF THE CENSUS, CURRENT POPULATION REPORTS P20-520 (Jan. 2000), at Table B. For those born between 1948 and 1950, the average age at divorce was 33.8, a decline of several years from the average age at divorce for those born between 1908 and 1912. Because the age at first marriage has steadily increased for birth cohorts after 1950, the age at divorce probably has as well. CHERLIN, supra note 9, at 68 (including note 5). It thus seems that those who move tend to be younger than those who divorce.
27. We originally reported this statistic in note 25 of Ellman & Lohr, supra note 2. Note that for this calculation, as for most calculations reported in that article, we omitted Louisiana and Nevada - the first because its data was incomplete, the second because it is an outlier in divorce statistics, its rates being affected by a large number of divorces granted to de facto nonresidents.
28. By "in-migrant" I mean anyone who moved into the region, most of whom move in from another region, rather than from another country.
29. For performing these calculations particularly, I wish to thank Lynn Tobin.
30. This is known as the problem of "ecological correlations." See DAVID FREEDMAN ET AL., STATISTICS 140-41 & A-7 (2d ed. 1991).
31. Five year migration totals, by region, were obtained from Table A, Interregional Migration: 1965-1970, 1970-75, and 1975-1980, at 1, in Geographical Mobility: 1975 to 1980, BUREAU OF THE CENSUS, CURRENT POPULATION REPORTS, SERIES P-20, NO. 368 (1981). The population by region for the relevant years was also taken from Census Bureau reports. Using these figures, I calculated the ratio of in-migrants during each five-year period, over the population in the earliest year of that period, as follows: In-Migrants (for the 5-year period)/Population (first year of 5-year period) Years NE Midwest South West 1965-70 0.0268 0.0373 0.0527 0.0717 1970-75 0.0216 0.0306 0.0650 0.0674 1975-80 0.0225 0.0344 0.0604 0.0735 The average divorce rates for the relevant periods can be calculated from the data presented in Table 1 in the text, and are as follows: Average Divorce Rate (per 1,000 population) Years NE Midwest South West 1971-75 2.64 4.04 4.88 6.02 1976-80 3.38 4.86 5.8 6.48 1981-85 3.66 4.6 5.7 5.98
32. I compare in-migrant rates with subsequent rather than contemporaneous divorce rates because my intuition was not that movers concurrently divorce, but that willingness to migrate may identify persons in the population at a higher risk of divorce over time - because, e.g., of their temperament, or employment instability, or level of discontent generally. A colleague pointed out that one does not get similar correlations between divorce rate and in-migrant rate if one examines changes over time, within regions, rather than comparing regions at different times, as I did. However, that alternative analysis is affected by the generally increasing divorce rate everywhere during these particular time periods, a powerful general trend that probably swamps the mobility factor. (I owe this point to Sharon Lohr.)
33. This data is taken from Table 17, at 43, of 100 Years of Marriage and Divorce Statistics, United States, 1867-1967, NATIONAL CENTER FOR HEALTH STATISTICS, VITAL AND HEALTH STATISTICS SERIES 21, No. 24, DEP'T OF HEALTH, ED. & WELFARE PUB. NO. (HRA) 74-1902 (1973). In fairness one must also observe that by 1960 the percentage of divorces issued to couples married in that state had converged, and the South's rate was only very slightly below that of the Midwest. But the shift in the two regions' relative percentages over this time period is still dramatic.
34. The Census provides this data on the web. The 1990 data I used here is at 〈http://www.census.gov/population/socdemo/migration/pob-rank.txt〉.
35. As previously noted, good data on divorce rates began disappearing in the 1990s. For this calculation, I relied upon the Advance Report of Final Divorce Statistics, 1989 and 1990, in 43 MONTHLY VITAL STAT. REP. No.9, Supp. (Mar. 22, 1995), published by the National Center for Health Statistics. There is no final report, and this "advance report" has only incomplete data for Indiana, Louisiana, and New Mexico. I therefore omitted these three states, as well as Nevada (as always) from the calculations I report here.
36. I thank Sharon Lohr for suggesting this calculation to me. A relationship between mobility and divorce analysis may also shed light on the relatively high divorce rate for Americans as compared to Western Europeans, in that Americans are more mobile than are Europeans. I leave it to others to find the data to test this hypothesis.
37. CHERLIN, supra note 9, at 53.
38. Sanford L. Braver, M. Whitley & C. Ng, Who Divorced Whom? Methodological and Theoretical Issues, 20 J. DIV. & REMARRIAGE 1 (1993).
39. See Sam Cameron, A Review of Economic Research into Determinants of Divorce, 17 BRIT. REV. ECON. ISSUES 1 (1995), for a general review of the literature. He concludes that there is wide consensus on the positive association of divorce with female wage rates and its negative association with male wage rates. A more recent attempt to overcome the methodological challenge, which also contains a more recent if less comprehensive review of the literature, is Steven Ruggles, The Rise in Divorce and Separation in the United States, 1880-1990, 34 DEMOGRAPHY 455 (1997). Ruggles looks at census data on marriage and divorce and on male and female employment, by local area, and attempts to prove the theory by showing an association between local areas with a higher percentage of employed women and higher divorce rates. He finds this association for each of the decades he examines between 1880 and 1990. He finds an even stronger relationship over this time period between local divorce rates and male employment, but with a negative sign, also consistent with prevailing theory. But male employment patterns cannot explain the persistent rise in divorce rates over the last 100 years because there has been no corresponding long-term decline in male employment - while of course there has been a corresponding long-term increase in female employment. Male unemployment might thus explain certain short-term changes in divorce rates over particular periods or in particular locales, but not the long-term general trend. For criticisms of Ruggles, see Anne Preston, Comment on Steven Ruggles's "The Rise of Divorce and Separation in the United States, 1880-1990," 34 DEMOGRAPHY 473 (1997), and Valerie Oppenheimer, Comment on Steven Ruggles's "The Rise of Divorce and Separation in the United States, 1880-1990," 34 DEMOGRAPHY 467 (1997). Preston focuses on the problem of separating the cultural and employment explanations, while Oppenheimer argues from the perspective of one of the few social scientists who does not believe the prevailing theory. Ruggles' response is at 34 DEMOGRAPHY 473 (1997). See also Ian Smith, Explaining the Growth of Divorce in Great Britain, 44 SCOT. J. POL. ECON. 519 (1997). He compares trends in Scotland with those in England, given that divorce law changed at different times, and concludes the law had little impact on the divorce rate, while the rising real earnings of women did - but not women's rising relative earnings (to men).
40. See Sam Cameron, A Review of Economic Research into Determinants of Divorce, 17 BRIT. REV. ECON. ISSUES 1 (1995), for a general review of the literature. He concludes that there is wide consensus on the positive association of divorce with female wage rates and its negative association with male wage rates. A more recent attempt to overcome the methodological challenge, which also contains a more recent if less comprehensive review of the literature, is Steven Ruggles, The Rise in Divorce and Separation in the United States, 1880-1990, 34 DEMOGRAPHY 455 (1997). Ruggles looks at census data on marriage and divorce and on male and female employment, by local area, and attempts to prove the theory by showing an association between local areas with a higher percentage of employed women and higher divorce rates. He finds this association for each of the decades he examines between 1880 and 1990. He finds an even stronger relationship over this time period between local divorce rates and male employment, but with a negative sign, also consistent with prevailing theory. But male employment patterns cannot explain the persistent rise in divorce rates over the last 100 years because there has been no corresponding long-term decline in male employment - while of course there has been a corresponding long-term increase in female employment. Male unemployment might thus explain certain short-term changes in divorce rates over particular periods or in particular locales, but not the long-term general trend. For criticisms of Ruggles, see Anne Preston, Comment on Steven Ruggles's "The Rise of Divorce and Separation in the United States, 1880-1990," 34 DEMOGRAPHY 473 (1997), and Valerie Oppenheimer, Comment on Steven Ruggles's "The Rise of Divorce and Separation in the United States, 1880-1990," 34 DEMOGRAPHY 467 (1997). Preston focuses on the problem of separating the cultural and employment explanations, while Oppenheimer argues from the perspective of one of the few social scientists who does not believe the prevailing theory. Ruggles' response is at 34 DEMOGRAPHY 473 (1997). See also Ian Smith, Explaining the Growth of Divorce in Great Britain, 44 SCOT. J. POL. ECON. 519 (1997). He compares trends in Scotland with those in England, given that divorce law changed at different times, and concludes the law had little impact on the divorce rate, while the rising real earnings of women did - but not women's rising relative earnings (to men).
41. See Sam Cameron, A Review of Economic Research into Determinants of Divorce, 17 BRIT. REV. ECON. ISSUES 1 (1995), for a general review of the literature. He concludes that there is wide consensus on the positive association of divorce with female wage rates and its negative association with male wage rates. A more recent attempt to overcome the methodological challenge, which also contains a more recent if less comprehensive review of the literature, is Steven Ruggles, The Rise in Divorce and Separation in the United States, 1880-1990, 34 DEMOGRAPHY 455 (1997). Ruggles looks at census data on marriage and divorce and on male and female employment, by local area, and attempts to prove the theory by showing an association between local areas with a higher percentage of employed women and higher divorce rates. He finds this association for each of the decades he examines between 1880 and 1990. He finds an even stronger relationship over this time period between local divorce rates and male employment, but with a negative sign, also consistent with prevailing theory. But male employment patterns cannot explain the persistent rise in divorce rates over the last 100 years because there has been no corresponding long-term decline in male employment - while of course there has been a corresponding long-term increase in female employment. Male unemployment might thus explain certain short-term changes in divorce rates over particular periods or in particular locales, but not the long-term general trend. For criticisms of Ruggles, see Anne Preston, Comment on Steven Ruggles's "The Rise of Divorce and Separation in the United States, 1880-1990," 34 DEMOGRAPHY 473 (1997), and Valerie Oppenheimer, Comment on Steven Ruggles's "The Rise of Divorce and Separation in the United States, 1880-1990," 34 DEMOGRAPHY 467 (1997). Preston focuses on the problem of separating the cultural and employment explanations, while Oppenheimer argues from the perspective of one of the few social scientists who does not believe the prevailing theory. Ruggles' response is at 34 DEMOGRAPHY 473 (1997). See also Ian Smith, Explaining the Growth of Divorce in Great Britain, 44 SCOT. J. POL. ECON. 519 (1997). He compares trends in Scotland with those in England, given that divorce law changed at different times, and concludes the law had little impact on the divorce rate, while the rising real earnings of women did - but not women's rising relative earnings (to men).
42. See Sam Cameron, A Review of Economic Research into Determinants of Divorce, 17 BRIT. REV. ECON. ISSUES 1 (1995), for a general review of the literature. He concludes that there is wide consensus on the positive association of divorce with female wage rates and its negative association with male wage rates. A more recent attempt to overcome the methodological challenge, which also contains a more recent if less comprehensive review of the literature, is Steven Ruggles, The Rise in Divorce and Separation in the United States, 1880-1990, 34 DEMOGRAPHY 455 (1997). Ruggles looks at census data on marriage and divorce and on male and female employment, by local area, and attempts to prove the theory by showing an association between local areas with a higher percentage of employed women and higher divorce rates. He finds this association for each of the decades he examines between 1880 and 1990. He finds an even stronger relationship over this time period between local divorce rates and male employment, but with a negative sign, also consistent with prevailing theory. But male employment patterns cannot explain the persistent rise in divorce rates over the last 100 years because there has been no corresponding long-term decline in male employment - while of course there has been a corresponding long-term increase in female employment. Male unemployment might thus explain certain short-term changes in divorce rates over particular periods or in particular locales, but not the long-term general trend. For criticisms of Ruggles, see Anne Preston, Comment on Steven Ruggles's "The Rise of Divorce and Separation in the United States, 1880-1990," 34 DEMOGRAPHY 473 (1997), and Valerie Oppenheimer, Comment on Steven Ruggles's "The Rise of Divorce and Separation in the United States, 1880-1990," 34 DEMOGRAPHY 467 (1997). Preston focuses on the problem of separating the cultural and employment explanations, while Oppenheimer argues from the perspective of one of the few social scientists who does not believe the prevailing theory. Ruggles' response is at 34 DEMOGRAPHY 473 (1997). See also Ian Smith, Explaining the Growth of Divorce in Great Britain, 44 SCOT. J. POL. ECON. 519 (1997). He compares trends in Scotland with those in England, given that divorce law changed at different times, and concludes the law had little impact on the divorce rate, while the rising real earnings of women did - but not women's rising relative earnings (to men).
43. See Sam Cameron, A Review of Economic Research into Determinants of Divorce, 17 BRIT. REV. ECON. ISSUES 1 (1995), for a general review of the literature. He concludes that there is wide consensus on the positive association of divorce with female wage rates and its negative association with male wage rates. A more recent attempt to overcome the methodological challenge, which also contains a more recent if less comprehensive review of the literature, is Steven Ruggles, The Rise in Divorce and Separation in the United States, 1880-1990, 34 DEMOGRAPHY 455 (1997). Ruggles looks at census data on marriage and divorce and on male and female employment, by local area, and attempts to prove the theory by showing an association between local areas with a higher percentage of employed women and higher divorce rates. He finds this association for each of the decades he examines between 1880 and 1990. He finds an even stronger relationship over this time period between local divorce rates and male employment, but with a negative sign, also consistent with prevailing theory. But male employment patterns cannot explain the persistent rise in divorce rates over the last 100 years because there has been no corresponding long-term decline in male employment - while of course there has been a corresponding long-term increase in female employment. Male unemployment might thus explain certain short-term changes in divorce rates over particular periods or in particular locales, but not the long-term general trend. For criticisms of Ruggles, see Anne Preston, Comment on Steven Ruggles's "The Rise of Divorce and Separation in the United States, 1880-1990," 34 DEMOGRAPHY 473 (1997), and Valerie Oppenheimer, Comment on Steven Ruggles's "The Rise of Divorce and Separation in the United States, 1880-1990," 34 DEMOGRAPHY 467 (1997). Preston focuses on the problem of separating the cultural and employment explanations, while Oppenheimer argues from the perspective of one of the few social scientists who does not believe the prevailing theory. Ruggles' response is at 34 DEMOGRAPHY 473 (1997). See also Ian Smith, Explaining the Growth of Divorce in Great Britain, 44 SCOT. J. POL. ECON. 519 (1997). He compares trends in Scotland with those in England, given that divorce law changed at different times, and concludes the law had little impact on the divorce rate, while the rising real earnings of women did - but not women's rising relative earnings (to men).
44. See Sam Cameron, A Review of Economic Research into Determinants of Divorce, 17 BRIT. REV. ECON. ISSUES 1 (1995), for a general review of the literature. He concludes that there is wide consensus on the positive association of divorce with female wage rates and its negative association with male wage rates. A more recent attempt to overcome the methodological challenge, which also contains a more recent if less comprehensive review of the literature, is Steven Ruggles, The Rise in Divorce and Separation in the United States, 1880-1990, 34 DEMOGRAPHY 455 (1997). Ruggles looks at census data on marriage and divorce and on male and female employment, by local area, and attempts to prove the theory by showing an association between local areas with a higher percentage of employed women and higher divorce rates. He finds this association for each of the decades he examines between 1880 and 1990. He finds an even stronger relationship over this time period between local divorce rates and male employment, but with a negative sign, also consistent with prevailing theory. But male employment patterns cannot explain the persistent rise in divorce rates over the last 100 years because there has been no corresponding long-term decline in male employment - while of course there has been a corresponding long-term increase in female employment. Male unemployment might thus explain certain short-term changes in divorce rates over particular periods or in particular locales, but not the long-term general trend. For criticisms of Ruggles, see Anne Preston, Comment on Steven Ruggles's "The Rise of Divorce and Separation in the United States, 1880-1990," 34 DEMOGRAPHY 473 (1997), and Valerie Oppenheimer, Comment on Steven Ruggles's "The Rise of Divorce and Separation in the United States, 1880-1990," 34 DEMOGRAPHY 467 (1997). Preston focuses on the problem of separating the cultural and employment explanations, while Oppenheimer argues from the perspective of one of the few social scientists who does not believe the prevailing theory. Ruggles' response is at 34 DEMOGRAPHY 473 (1997). See also Ian Smith, Explaining the Growth of Divorce in Great Britain, 44 SCOT. J. POL. ECON. 519 (1997). He compares trends in Scotland with those in England, given that divorce law changed at different times, and concludes the law had little impact on the divorce rate, while the rising real earnings of women did - but not women's rising relative earnings (to men).
45. This point is noted in Preston, supra note 33. See also William Johnson & Jonathan Skinner, Labor Supply and Marital Separation, 76 AM. ECON. REV. 455 (1986), who found that women typically increase their labor supply in the three years preceding a marital separation. That behavior could reflect their anticipation of the separation, or it could be a factor contributing to the separation, or both.
46. This point is noted in Preston, supra note 33. See also William Johnson & Jonathan Skinner, Labor Supply and Marital Separation, 76 AM. ECON. REV. 455 (1986), who found that women typically increase their labor supply in the three years preceding a marital separation. That behavior could reflect their anticipation of the separation, or it could be a factor contributing to the separation, or both.
47. Geographic Mobility, supra note 20, at 1.
48. The Maricopa County (Phoenix) clerk issued 6,224 marriage licenses between August 1, 1998, and January 31, 1999. The new covenant marriage license law took effect on August 21, 1998, and from that day through January 31, 1999, 31 covenant marriage licenses were issued. These figures actually overstate the proportion of all Maricopa County marriage licenses that are covenant marriages, because justice courts and city clerks in Maricopa County also issue marriage licenses, but not covenant marriage licenses. E-mail from Maureen Ramroth, Maricopa County Clerk's Office, February 22, 1999. Professor Steven Nock of the University of Virginia found that in the year following adoption of Louisiana's covenant marriage law, only 1.6% of all new marriages in that state were covenant marriages. The percentage remained essentially unchanged during the first six months of the succeeding year. I thank Brian Bix for sharing with me the e-mail he received from Steven Nock containing this data.
49. One recent rejection occurred in Colorado. ARIZ. CAPITOL TIMES, Mar. 5, 1999, at 13. The defeated Colorado measure was HB 1194. And while Arizona did follow Louisiana in adopting covenant marriage, it passed a much-weakened version. Under the Arizona law, a spouse may petition for divorce with the claim that he or she expects that the parties will live apart for the required two years. The actual divorce decree is then deferred until the two-year period has run, but the court may in the interim issue temporary orders of support. See ARIZ. REV. STAT. § 25-903. These concessions from the language of the original Louisiana statute were required for passage, and even then the law passed by only one vote.
50. One recent rejection occurred in Colorado. ARIZ. CAPITOL TIMES, Mar. 5, 1999, at 13. The defeated Colorado measure was HB 1194. And while Arizona did follow Louisiana in adopting covenant marriage, it passed a much-weakened version. Under the Arizona law, a spouse may petition for divorce with the claim that he or she expects that the parties will live apart for the required two years. The actual divorce decree is then deferred until the two-year period has run, but the court may in the interim issue temporary orders of support. See ARIZ. REV. STAT. § 25-903. These concessions from the language of the original Louisiana statute were required for passage, and even then the law passed by only one vote.
51. See The Big Picture, AM. DEMOGRAPHICS (Aug. 1997), at 35. Marriage rates and divorce rates (as conventionally measured) affect one another. Divorce rates, conventionally measured as divorces per 1,000 people, necessarily decline with marriage rates (all else being equal), since the number of people eligible for divorce declines with the marriage rate. So declining marriage rates accounts for some of the divorce rate decline (but only some). Divorces per 1,000 married persons is a better measure of divorce rates because it does not suffer from this difficulty. It also shows that American divorce rates peaked at about 1980 and began declining after that. See CHERLIN, supra note 9, at 21.
52. See The Big Picture, AM. DEMOGRAPHICS (Aug. 1997), at 35. Marriage rates and divorce rates (as conventionally measured) affect one another. Divorce rates, conventionally measured as divorces per 1,000 people, necessarily decline with marriage rates (all else being equal), since the number of people eligible for divorce declines with the marriage rate. So declining marriage rates accounts for some of the divorce rate decline (but only some). Divorces per 1,000 married persons is a better measure of divorce rates because it does not suffer from this difficulty. It also shows that American divorce rates peaked at about 1980 and began declining after that. See CHERLIN, supra note 9, at 21.
53. NATIONAL CENTER FOR HEALTH STATISTICS, 46 MONTHLY VITAL STAT. REP. No. 12 (July 28, 1998), at 3.
54. The age at first marriage was relatively stable from 1950 to 1976, increasing over that period by only one year. But it then took only five more years for it to increase again by one year, and then again in four years. The increase in age at first marriage now seems to be leveling off. See the following chart, derived from Bureau of the Census, Table MS-2, Estimated Median Age at First Marriage, by Sex, 1890 to Present (Jan. 7, 1999) 〈http://www.census.gov/population/socdemo/ms-la/tabms-2.txt〉. Increase in Median Age at First Marriage, in Years, over Selected Time Periods Period Length of Period Women Men 1950-1976 26 years 1 1 1976-1981 5 1 1 1981-1985 4 1 0.7 1986-1992 7 1.1 1 1992-1998 6 0.6 0.2 Median Age at First Marriage in 1950 20.3 22.8 Median Age at First Marriage in 1998 25 26.7
55. One estimate, made in 1991, was that the percentage who never marry in their lifetime will increase from 5 to "not more than ten" percent, with others simply marrying later. FRANCINE BLAU, MARIANNE FERBER & ANNE WINKLER, THE ECONOMICS OF WOMEN, MEN AND WORK 274 (3d ed. 1998). At least one careful researcher does seem to think the story is marriage delay more than marriage avoidance. See Valerie Oppenheimer, Women's Rising Employment and the Future of the Family in Industrial Societies, 20 POP. & DEV. REV. 293 (1994). Oppenheimer looks at several cohorts, beginning with white women aged 20-24 in 1965 and continuing through those reaching that age in 1990, and shows that the percent still single tended to converge once they were in their 30s. The problem, of course, is that the most recent cohorts were not yet in their 30s when this data was collected. Id. at 306. Another problem is that her data on black women suggests the contrary, that the younger cohorts are in fact less likely to ever marry.
56. One estimate, made in 1991, was that the percentage who never marry in their lifetime will increase from 5 to "not more than ten" percent, with others simply marrying later. FRANCINE BLAU, MARIANNE FERBER & ANNE WINKLER, THE ECONOMICS OF WOMEN, MEN AND WORK 274 (3d ed. 1998). At least one careful researcher does seem to think the story is marriage delay more than marriage avoidance. See Valerie Oppenheimer, Women's Rising Employment and the Future of the Family in Industrial Societies, 20 POP. & DEV. REV. 293 (1994). Oppenheimer looks at several cohorts, beginning with white women aged 20-24 in 1965 and continuing through those reaching that age in 1990, and shows that the percent still single tended to converge once they were in their 30s. The problem, of course, is that the most recent cohorts were not yet in their 30s when this data was collected. Id. at 306. Another problem is that her data on black women suggests the contrary, that the younger cohorts are in fact less likely to ever marry.
57. One estimate, made in 1991, was that the percentage who never marry in their lifetime will increase from 5 to "not more than ten" percent, with others simply marrying later. FRANCINE BLAU, MARIANNE FERBER & ANNE WINKLER, THE ECONOMICS OF WOMEN, MEN AND WORK 274 (3d ed. 1998). At least one careful researcher does seem to think the story is marriage delay more than marriage avoidance. See Valerie Oppenheimer, Women's Rising Employment and the Future of the Family in Industrial Societies, 20 POP. & DEV. REV. 293 (1994). Oppenheimer looks at several cohorts, beginning with white women aged 20-24 in 1965 and continuing through those reaching that age in 1990, and shows that the percent still single tended to converge once they were in their 30s. The problem, of course, is that the most recent cohorts were not yet in their 30s when this data was collected. Id. at 306. Another problem is that her data on black women suggests the contrary, that the younger cohorts are in fact less likely to ever marry.
58. The Big Picture, supra note 38.
59. See the studies cited by CHERLIN, supra note 9, at 51-52.
60. BLAU ET AL., supra note 41, at 271-73.
61. Oppenheimer, supra note 41, at 316.
62. It may be that where the parties are equal earners, they will still benefit from specializing, because neither may maximize earning potential if they do not specialize. Or, they may find that domestic tasks they both deem important, such as child care, cannot be performed at a level either or both of them believe necessary, unless one of them specializes in it. So the parties may conclude that they maximize their utility, in income and parenting combined, if one works full-time while the other works parttime. See also Amy Wax, Bargaining in the Shadow of the Market: Is There a Future for Egalitarian Marriage?, 84 VA. L. REV. 509 (1998). Wax offers a feminist perspective as well as an economic analysis. Agreeing that an egalitarian marriage is not necessarily one in which the parties have identical roles, Wax employs bargaining theory to reach a pessimistic conclusion about the prospects for egalitarian marriage in a broader sense - marriage in which the utility gains are more nearly equal - and worries that the declining attractiveness of marriage to women may indeed be the most important threat to the institution.
63. At least one feminist writer has urged that strategy on women. RHONA MAHONY, KIDDING OURSELVES: BREADWINNING, BABIES, AND BARGAINING POWER 215-38 (1995). Amy Wax concludes that Mahony's suggestion "is unlikely to work very well," and it seems certain she is correct. See Wax, supra note 46, at 644-45.
64. At least one feminist writer has urged that strategy on women. RHONA MAHONY, KIDDING OURSELVES: BREADWINNING, BABIES, AND BARGAINING POWER 215-38 (1995). Amy Wax concludes that Mahony's suggestion "is unlikely to work very well," and it seems certain she is correct. See Wax, supra note 46, at 644-45.
65. In what must be the most extreme cultural manesfestation of the view that women should marry "up," the most elite clans of certain provinces in northern India routinely committed infanticide on infant daughters who had no prospect of marriage because there was no place "up" to go. That the phenomena was more complex than a simple cultural bias against daughters is evidenced by the fact that the practice was in proportion to the social status of the family: lesser elites killed only later born daughters, while the lowest ranking clans kept some or all their daughters. SARAH BLAFFER HRDY, MOTHER NATURE 326-27, 338-40 (1999). Amy Wax observes that women's "emotional attraction to men of higher status . . . may be just as deeply ingrained and difficult to alter as men's penchant for younger women." Wax, supra note 46, at 645.
66. In what must be the most extreme cultural manesfestation of the view that women should marry "up," the most elite clans of certain provinces in northern India routinely committed infanticide on infant daughters who had no prospect of marriage because there was no place "up" to go. That the phenomena was more complex than a simple cultural bias against daughters is evidenced by the fact that the practice was in proportion to the social status of the family: lesser elites killed only later born daughters, while the lowest ranking clans kept some or all their daughters. SARAH BLAFFER HRDY, MOTHER NATURE 326-27, 338-40 (1999). Amy Wax observes that women's "emotional attraction to men of higher status . . . may be just as deeply ingrained and difficult to alter as men's penchant for younger women." Wax, supra note 46, at 645.
67. Indeed, because the likelihood an individual marrying is affected by so many variables, simple connections between any one important variable, and the marriage rate, may easily be concealed. Oppenheimer, who has been skeptical of the hypothesis that women's rising earnings have contributed to the decline in marriage rates, concludes from data on white women that the likelihood of a woman marrying does not appear to be associated with her level of education. Yet, as Oppenheimer herself carefully points out, this does not necessarily mean that a woman's chance of marrying is not reduced by the rising earning potential associated with greater education. Rather, the lack of association may reflect the conflicting impact of several variables. More highly educated women probably have better access to marriage markets, not only from higher education itself, and the social networks that arise from it, but also in the job environments that higher education allows them to enter. Oppenheimer, supra note 41, at 315.
68. The data in Figure 3 is derived from Table F-13, Work Experience of Husband and Wife, in the Historical Income Table - Families, posted by the Census Bureau at 〈http://www.census.gov/income/histinc/〉.
69. Oddly, the media attention given "stay-at-home" dads, while often portraying them as a new trend, usually provides further evidence of their relative scarcity. The stories typically have a "man bites dog" quality. See, e.g., At-Home Fathers Step Out to Find They Are Not Alone, N. Y. TIMES, Jan. 2, 2000, § 1, at 1. The article reports on a convention of an organization called "At-Home Dads." A national organization, the convention drew 85 male attendees from 20 states - up from 35 attendees at its inaugural meeting in 1996, but down from 100 at the 1998 convention. Strategies for dealing with the isolation experienced by these house-husbands was a principal topic addressed at the convention.
70. Note that the average age differential between husbands and wives remains fairly stable, even as the average age at first marriage rises for both. See the data at supra note 40.
71. Indeed, there may be reason to think they are fewer even than they once were. The labor force participation rates of men declined over the final two decades of the last century, probably as a result of declining labor force participation rates of older men. It thus may be that more of this 5% consists of retired men and their working wives today than was the case in 1980.
72. The first line in this table is derived from the historical income tables posted on the web by the Census Bureau: Table F-13, at 〈http://www.census.gov/income/ f13.txt〉; Table F-14, at 〈http://www.census.gov/hhes/income/ histinc/f14.html〉, and Table F-19, at 〈http://www.census.gov/hhes/income/histinc/ f19.html〉. From these tables it is possible to get the number of marriages in which both husband and wife had earnings, with the wife earning more than the husband. I then divided this figure, for each year, into the total number of marriages in which the husband worked, also available from these tables. The three sets of figures in the second line are taken Bureau of Labor Statistics, Highlights of Women's Earnings in 1998, Report 928 (Apr. 1999), at Table 13. This data is based upon median weekly earnings of "full-time wage and salary workers." The BLS defines full-time as 35 hours or more per week. One often sees data on "dual-earner" marriages that report higher figures than those provided in Line 1 of Table 3. See, e.g., Anne Winkler, Earnings of Husbands and Wives in Dual-earner Families, MONTHLY LAB. REV. (Apr. 1998), at 42 (observing that the proportion of dual-earner marriages in which the wife earned more than the husband increased from 16% in 1981 to 23% in 1996). The higher figures in these accounts refer, however, to the proportion of dual earner marriages in which the wife earns more than the husband. They overstate the relative income position of wives, to husbands, because they ignore one-earner marriages, in which husband-breadwinners are far more common than wife-breadwinners, as Figure 3 shows. A better figure might be the proportion of all marriages in which the wife earns more than the husband. These figures, for 1981, 1992, and 1997, respectively, are 13.2%, 18.5%, and 19.1%, as derived from the historical income tables noted above. These numbers have a different problem, however: they include the marriages in which the wife earns more than the husband because the husband does not work, and some proportion of these involve younger wives with older, retired, husbands. The calculation provided in Line One of Table 3 avoids this problem by looking only at marriages in which the husband works (full- or part-time), and giving the percentage of these marriages in which the wife earns more. One must note, however, that these figures are not perfect either, because they effectively exclude the small proportion of true, complete, role-reversal marriages from the calculation.
73. The first line in this table is derived from the historical income tables posted on the web by the Census Bureau: Table F-13, at 〈http://www.census.gov/income/ f13.txt〉; Table F-14, at 〈http://www.census.gov/hhes/income/ histinc/f14.html〉, and Table F-19, at 〈http://www.census.gov/hhes/income/histinc/ f19.html〉. From these tables it is possible to get the number of marriages in which both husband and wife had earnings, with the wife earning more than the husband. I then divided this figure, for each year, into the total number of marriages in which the husband worked, also available from these tables. The three sets of figures in the second line are taken Bureau of Labor Statistics, Highlights of Women's Earnings in 1998, Report 928 (Apr. 1999), at Table 13. This data is based upon median weekly earnings of "full-time wage and salary workers." The BLS defines full-time as 35 hours or more per week. One often sees data on "dual-earner" marriages that report higher figures than those provided in Line 1 of Table 3. See, e.g., Anne Winkler, Earnings of Husbands and Wives in Dual-earner Families, MONTHLY LAB. REV. (Apr. 1998), at 42 (observing that the proportion of dual-earner marriages in which the wife earned more than the husband increased from 16% in 1981 to 23% in 1996). The higher figures in these accounts refer, however, to the proportion of dual earner marriages in which the wife earns more than the husband. They overstate the relative income position of wives, to husbands, because they ignore one-earner marriages, in which husband-breadwinners are far more common than wife-breadwinners, as Figure 3 shows. A better figure might be the proportion of all marriages in which the wife earns more than the husband. These figures, for 1981, 1992, and 1997, respectively, are 13.2%, 18.5%, and 19.1%, as derived from the historical income tables noted above. These numbers have a different problem, however: they include the marriages in which the wife earns more than the husband because the husband does not work, and some proportion of these involve younger wives with older, retired, husbands. The calculation provided in Line One of Table 3 avoids this problem by looking only at marriages in which the husband works (full- or part-time), and giving the percentage of these marriages in which the wife earns more. One must note, however, that these figures are not perfect either, because they effectively exclude the small proportion of true, complete, role-reversal marriages from the calculation.
74. The first line in this table is derived from the historical income tables posted on the web by the Census Bureau: Table F-13, at 〈http://www.census.gov/income/ f13.txt〉; Table F-14, at 〈http://www.census.gov/hhes/income/ histinc/f14.html〉, and Table F-19, at 〈http://www.census.gov/hhes/income/histinc/ f19.html〉. From these tables it is possible to get the number of marriages in which both husband and wife had earnings, with the wife earning more than the husband. I then divided this figure, for each year, into the total number of marriages in which the husband worked, also available from these tables. The three sets of figures in the second line are taken Bureau of Labor Statistics, Highlights of Women's Earnings in 1998, Report 928 (Apr. 1999), at Table 13. This data is based upon median weekly earnings of "full-time wage and salary workers." The BLS defines full-time as 35 hours or more per week. One often sees data on "dual-earner" marriages that report higher figures than those provided in Line 1 of Table 3. See, e.g., Anne Winkler, Earnings of Husbands and Wives in Dual-earner Families, MONTHLY LAB. REV. (Apr. 1998), at 42 (observing that the proportion of dual-earner marriages in which the wife earned more than the husband increased from 16% in 1981 to 23% in 1996). The higher figures in these accounts refer, however, to the proportion of dual earner marriages in which the wife earns more than the husband. They overstate the relative income position of wives, to husbands, because they ignore one-earner marriages, in which husband-breadwinners are far more common than wife-breadwinners, as Figure 3 shows. A better figure might be the proportion of all marriages in which the wife earns more than the husband. These figures, for 1981, 1992, and 1997, respectively, are 13.2%, 18.5%, and 19.1%, as derived from the historical income tables noted above. These numbers have a different problem, however: they include the marriages in which the wife earns more than the husband because the husband does not work, and some proportion of these involve younger wives with older, retired, husbands. The calculation provided in Line One of Table 3 avoids this problem by looking only at marriages in which the husband works (full- or part-time), and giving the percentage of these marriages in which the wife earns more. One must note, however, that these figures are not perfect either, because they effectively exclude the small proportion of true, complete, role-reversal marriages from the calculation.
75. See Joyce Jacobsen & Laurence Levin, Effects of Intermittent Labor Force Attachment on Women's Earnings, MONTHLY LAB. REV. (Sept. 1995), at 14.
76. U.S. DEPARTMENT OF LABOR, REPORT ON THE AMERICAN WORKFORCE (1999), Table 3-5, at 113.
77. Thirty-five hours a week seems a low figure to use in defining full-time work. In fact, 95% of working American men put in more than 35 hours a week, as do 82% of working American women, according to an analysis of 1986 data from the Luxembourg Income Survey. Rachel Rosenfeld & Gunn Elizabeth Birkelund, Woman's Part-Time Work: A Cross National Comparison, 11 EUROPEAN SOC. REV. 111 (1995), at Table 3.
78. June O'Neil & Solomon Polacheck, Why the Gender Gap in Wages Narrowed in the 1980's, 11 J. LAB. ECON. 205, 208 n. 1. One can also note that the gap between men's and women's hourly earnings is smaller than the gap between the earnings of wives and husbands working full-time, which also suggests that a difference in hours worked is essential to explaining the total gap. In 1998, wives working full-time earned only 71% of what husbands working full-time earned - yet, the median hourly earnings of women paid hourly rates was 82% of the male hourly rate. Highlights of Women's Earnings in 1998, U.S. DEPARTMENT OF LABOR, BUREAU OF LABOR STATISTICS, REP. 928 (Apr. 1999), at Table 17, at 29. Of course, this hourly figure is not applicable to those not paid in hourly rates, and can thus tell only part of the story. What of people not paid by the hour? Victor Fuchs has reported that of married women with eighteen years or more of schooling and at least one child under twelve, only one in ten worked more than 2,250 hours per year, equivalent to 45 hours per week, 50 weeks a year. Yet half the husbands of these women put in 2,250 hours, and one-third worked more than 2,500 hours. VICTOR FUCHS, WOMEN'S QUEST FOR ECONOMIC EQUALITY 47-48 (1988).
79. June O'Neil & Solomon Polacheck, Why the Gender Gap in Wages Narrowed in the 1980's, 11 J. LAB. ECON. 205, 208 n. 1. One can also note that the gap between men's and women's hourly earnings is smaller than the gap between the earnings of wives and husbands working full-time, which also suggests that a difference in hours worked is essential to explaining the total gap. In 1998, wives working full-time earned only 71% of what husbands working full-time earned - yet, the median hourly earnings of women paid hourly rates was 82% of the male hourly rate. Highlights of Women's Earnings in 1998, U.S. DEPARTMENT OF LABOR, BUREAU OF LABOR STATISTICS, REP. 928 (Apr. 1999), at Table 17, at 29. Of course, this hourly figure is not applicable to those not paid in hourly rates, and can thus tell only part of the story. What of people not paid by the hour? Victor Fuchs has reported that of married women with eighteen years or more of schooling and at least one child under twelve, only one in ten worked more than 2,250 hours per year, equivalent to 45 hours per week, 50 weeks a year. Yet half the husbands of these women put in 2,250 hours, and one-third worked more than 2,500 hours. VICTOR FUCHS, WOMEN'S QUEST FOR ECONOMIC EQUALITY 47-48 (1988).
80. June O'Neil & Solomon Polacheck, Why the Gender Gap in Wages Narrowed in the 1980's, 11 J. LAB. ECON. 205, 208 n. 1. One can also note that the gap between men's and women's hourly earnings is smaller than the gap between the earnings of wives and husbands working full-time, which also suggests that a difference in hours worked is essential to explaining the total gap. In 1998, wives working full-time earned only 71% of what husbands working full-time earned - yet, the median hourly earnings of women paid hourly rates was 82% of the male hourly rate. Highlights of Women's Earnings in 1998, U.S. DEPARTMENT OF LABOR, BUREAU OF LABOR STATISTICS, REP. 928 (Apr. 1999), at Table 17, at 29. Of course, this hourly figure is not applicable to those not paid in hourly rates, and can thus tell only part of the story. What of people not paid by the hour? Victor Fuchs has reported that of married women with eighteen years or more of schooling and at least one child under twelve, only one in ten worked more than 2,250 hours per year, equivalent to 45 hours per week, 50 weeks a year. Yet half the husbands of these women put in 2,250 hours, and one-third worked more than 2,500 hours. VICTOR FUCHS, WOMEN'S QUEST FOR ECONOMIC EQUALITY 47-48 (1988).
81. Thomas W. Harrell, The Association of Marriage and MBA Earnings, 72 PSYCH. REP. 955, 959-60 (1993) (men who had earned M.B.A.'s from Stanford between 1973-85 averaged work weeks of 55.7 hours; women averaged 47.1 hours).
82. Carol S. Weisman et al., Sex Differences in the Practice Patterns of Recently Trained Obstetricians-Gynecologists, 67 OBSTETRICS & GYNECOLOGY 776, 778 (1986). Women ob-gyn's were much more likely than men to work in settings, such as health maintenance organizations as compared to private practice, in which their hours could be limited.
83. See id. at 779 (presence of children at home increases the number of work hours for men and reduces the number of work hours for women). See also the data reported by FUCHS, supra note 58.
84. See id. at 779 (presence of children at home increases the number of work hours for men and reduces the number of work hours for women). See also the data reported by FUCHS, supra note 58.
85. The title of Arlie Hochschild's book popularized the term "second shift" for this phenomenon.
86. Child care or maternity leaves certainly have a lasting impact on earnings history, as a variety of studies have shown. For one example, see Joyce Jacobsen & Laurence Levin, Effects of Intermittent Labor Force Attachment on Women's Earnings, MONTHLY LABOR REV. (Sept. 1995), at 14.
87. FUCHS, supra note 58, at 64-70.
88. Id. at 62.
89. REPORT ON THE AMERICAN WORKFORCE, supra note 56, at 100.
90. CHERLIN, supra note 9, Figure 2-5, at 58.
91. This point is also made by FUCHS, supra note 58, at 30. The theory of cognitive dissonance was first proposed by Leon Festinger in A THEORY OF COGNITIVE DISSONANCE (1957).
92. FUCHS, supra note 58, at 45. Marriage increases the proportion of white women working part-time by 13 points, but for African-American women, it is only 2 points.
93. CHERLIN, supra note 9, at 61, makes the same point.
94. This chart is taken from REPORT ON THE AMERICAN WORKFORCE, supra note 56, at 102.
95. Figure 5 is taken from id. at 97.
96. See id., Chart 3-21, at 97. That chart is not reprinted here.
97. Oppenheimer has argued that rising women's wages cannot explain declining marriage rates because their wages did not in fact increase. Her data, however, (based entirely on whites) does not dispute the steady increase in the ratio of women's earnings to men's between the mid-1970s and 1989. Indeed, she bolsters this point, because she shows that the ratio changed in this way when one looks at men and women out of school 1 to 3 years - a key population in the marriage market. Oppenheimer, supra note 41, at 327. Her point is that, at least until the mid-1980s, men's declining real wages explained this trend, rather than women's rising real wages. That point is not inconsistent with the argument in the text, however, which is based on the ratio of women's to men's income, not their absolute amounts.
98. The Bureau is part of the Division of Labor Force Statistics. I wish to thank Howard Hayghe of the Bureau for his patient help in locating and deciphering this data for me.
99. See Historical Income Tables, Table P-11, at 〈http://www.census.gov/hhes/ income/histinc/p11.html〉. The precise 1997 median income figure reported in that table for "males, married, spouse present," is $31,983.
100. It is the case that wives with a high school education begin working less at a higher income point than wives with a college education, with the result that the proportion working full-time is slightly higher than college educated wives in the $35,000 to $50,000 income range. While no obvious explanation for this departure from the otherwise prevailing patterns presents itself, it does not alter the general point that for each level of educational attainment, there is a husband's income point above which wives' labor force participation declines.
101. One can again only speculate as to why the wives of lower earning husbands are less likely themselves to work full-time. It may be that older, semi-retired couples are disproportionately represented among the lower earners, a phenomenon one might think more likely in the group with no minor children than in the group that has them.
102. FUCHS, supra note 58, at 47.
103. See, e.g., a survey commissioned in 1990-91 by the U.S. Information Agency asked men and women whether they agreed with these two statements: "A job is all right, but what most women really want is a home and children" and "Being a housewife is just as fulfilling as working for pay." Fifty-two percent of the American women agreed with the first statement, and 73% agreed with the second. (42% and 25% disagreed, the balance apparently having no clear opinion). Men were slightly less likely to agree (50% and 66%, with 37% and 23% disagreeing). OFFICE OF RESEARCH AND MEDIA REACTION, USIA, A WORLD VIEW OF WOMEN: SOCIAL , POLITICAL AND ECONOMIC ATTITUDES 29 (undated softcover). See also David L. Chambers, Accommodation and Satisfaction: Women and Men Lawyers and the Balance of Work and Family, 14 L. & SOC. INQ. 251 (1989) (reporting on survey conducting between 1981 and 1986, of 1976-79 University of Michigan Law School graduates; three-fourths of women respondents described themselves as spending more time on their personal life, and less on their professional life, than men in similar positions; most reported themselves "quite satisfied" with this balance; 80 of the married women lawyers had spouses with comparable or greater incomes than they had; only one had a house-husband. Only 2 of 160 married men reported that their wives earned more than they did; 23% reported that their wives were homemakers). It is sometimes observed that dual earner parents may stagger their work hours so that one of them is available to care for their children during most of the day. Yet such couples, when interviewed, show very different motivations of the husbands and wives. When asked why they worked weekends or evenings, only 8% of fathers mention child care responsibilities; 75% say simply that is a requirement of their job. On the other hand, most mothers answer the same question with the explanation that care of their family members, mainly their job, is the reason for their schedule. Harriet Presser, Can We Make Time for Children? The Economy, Work Schedules, and Child Care, 26 DEMOGRAPHY 523, 530-31 (1989).
104. See, e.g., a survey commissioned in 1990-91 by the U.S. Information Agency asked men and women whether they agreed with these two statements: "A job is all right, but what most women really want is a home and children" and "Being a housewife is just as fulfilling as working for pay." Fifty-two percent of the American women agreed with the first statement, and 73% agreed with the second. (42% and 25% disagreed, the balance apparently having no clear opinion). Men were slightly less likely to agree (50% and 66%, with 37% and 23% disagreeing). OFFICE OF RESEARCH AND MEDIA REACTION, USIA, A WORLD VIEW OF WOMEN: SOCIAL , POLITICAL AND ECONOMIC ATTITUDES 29 (undated softcover). See also David L. Chambers, Accommodation and Satisfaction: Women and Men Lawyers and the Balance of Work and Family, 14 L. & SOC. INQ. 251 (1989) (reporting on survey conducting between 1981 and 1986, of 1976-79 University of Michigan Law School graduates; three-fourths of women respondents described themselves as spending more time on their personal life, and less on their professional life, than men in similar positions; most reported themselves "quite satisfied" with this balance; 80 of the married women lawyers had spouses with comparable or greater incomes than they had; only one had a house-husband. Only 2 of 160 married men reported that their wives earned more than they did; 23% reported that their wives were homemakers). It is sometimes observed that dual earner parents may stagger their work hours so that one of them is available to care for their children during most of the day. Yet such couples, when interviewed, show very different motivations of the husbands and wives. When asked why they worked weekends or evenings, only 8% of fathers mention child care responsibilities; 75% say simply that is a requirement of their job. On the other hand, most mothers answer the same question with the explanation that care of their family members, mainly their job, is the reason for their schedule. Harriet Presser, Can We Make Time for Children? The Economy, Work Schedules, and Child Care, 26 DEMOGRAPHY 523, 530-31 (1989).
105. See, e.g., a survey commissioned in 1990-91 by the U.S. Information Agency asked men and women whether they agreed with these two statements: "A job is all right, but what most women really want is a home and children" and "Being a housewife is just as fulfilling as working for pay." Fifty-two percent of the American women agreed with the first statement, and 73% agreed with the second. (42% and 25% disagreed, the balance apparently having no clear opinion). Men were slightly less likely to agree (50% and 66%, with 37% and 23% disagreeing). OFFICE OF RESEARCH AND MEDIA REACTION, USIA, A WORLD VIEW OF WOMEN: SOCIAL , POLITICAL AND ECONOMIC ATTITUDES 29 (undated softcover). See also David L. Chambers, Accommodation and Satisfaction: Women and Men Lawyers and the Balance of Work and Family, 14 L. & SOC. INQ. 251 (1989) (reporting on survey conducting between 1981 and 1986, of 1976-79 University of Michigan Law School graduates; three-fourths of women respondents described themselves as spending more time on their personal life, and less on their professional life, than men in similar positions; most reported themselves "quite satisfied" with this balance; 80 of the married women lawyers had spouses with comparable or greater incomes than they had; only one had a house-husband. Only 2 of 160 married men reported that their wives earned more than they did; 23% reported that their wives were homemakers). It is sometimes observed that dual earner parents may stagger their work hours so that one of them is available to care for their children during most of the day. Yet such couples, when interviewed, show very different motivations of the husbands and wives. When asked why they worked weekends or evenings, only 8% of fathers mention child care responsibilities; 75% say simply that is a requirement of their job. On the other hand, most mothers answer the same question with the explanation that care of their family members, mainly their job, is the reason for their schedule. Harriet Presser, Can We Make Time for Children? The Economy, Work Schedules, and Child Care, 26 DEMOGRAPHY 523, 530-31 (1989).
106. Hungary was the only country survey in which the proportion was higher than in the United States. THE GALLUP ORGANIZATION, GENDER AND SOCIETY, STATUS AND STEREOTYPES (Mar. 1996) (multinational polling).
107. Sixty-five percent of American women 16-64 are employed, compared with 54% of German women, and 60% of married American women with preschool children are employed, compared with 46% of comparable German women. Moreover, while only a quarter of employed American women are working part-time, a third of the employed German women are. Sonja Drobnic, Hans-Peter Blossfeld & Götz Rohwer, Dynamics of Women's Employment Patterns over the Family Life Course: A Comparison of the United States and Germany, 61 J. MARRIAGE & FAM. 133, 134 (1999).
108. I say "appear" because the data collection methods are not entirely comparable. But with that caveat, it seems that the percentage of British women who work part-time rather than full-time is much higher than in the United States, despite the apparently greater rejection of the sexual division of labor by British men and women as compared with Americans. See Rosenfeld & Birkelund, supra note 57, at 111.
109. Bretta Hoem, The Way of the Gender-Segregated Swedish Labour Market, in GENDER AND FAMILY CHANGE IN INDUSTRIALIZED COUNTRIES 279 (Karen Mason & An-Magritt Jensen eds., 1995).
110. Id.
111. This characterization of the Swedish work environment is made by CHERLIN, supra note 9, at 47. It is confirmed by the data in Janet Gornick & Jerry Jacobs, Gender, the Welfare State, and Public Employment: A Comparative Study of Seven Industrialized Countries, 63 AM. SOC. REV. 688 (1988). From Tables 1 and 6 in Gornick & Jacobs one can derive a female/male earnings ratio of .8 in Sweden, as compared to .69 in the United States, using comparable 1991-92 measures from both countries.
112. This characterization of the Swedish work environment is made by CHERLIN, supra note 9, at 47. It is confirmed by the data in Janet Gornick & Jerry Jacobs, Gender, the Welfare State, and Public Employment: A Comparative Study of Seven Industrialized Countries, 63 AM. SOC. REV. 688 (1988). From Tables 1 and 6 in Gornick & Jacobs one can derive a female/male earnings ratio of .8 in Sweden, as compared to .69 in the United States, using comparable 1991-92 measures from both countries.
113. Cherlin cited a 1985 source that reported that the female/male ratio of part-timers is almost 6 to 1 in Sweden, much higher than in America. CHERLIN, supra note 9, at 47. See also Table 3 of Rosenfeld & Birkelund, supra note 57, which uses 1986 data from the Luxembourg Income Survey to find that about 95% of working men work more than 35 hours a week in both Sweden and the United States - but while 82% of American working women also do so, only 53% of Swedish women do.
114. Cherlin cited a 1985 source that reported that the female/male ratio of part- timers is almost 6 to 1 in Sweden, much higher than in America. CHERLIN, supra note 9, at 47. See also Table 3 of Rosenfeld & Birkelund, supra note 57, which uses 1986 data from the Luxembourg Income Survey to find that about 95% of working men work more than 35 hours a week in both Sweden and the United States - but while 82% of American working women also do so, only 53% of Swedish women do.
115. Hoem, supra note 84, at 281, 283.
116. Id. at 284-85.
117. Id. at 286-87.
118. Id. at 293.
119. Id. at 295.
120. The marriage rates during that same time period was 4.3 marriages per 1,000 inhabitants in Sweden, as compared to 9.4 in the United States. U.N. DEPARTMENT OF ECONOMIC AND SOCIAL AFFAIRS STATISTICS DIVISION, 52 MONTHLY BULL. STAT. No. 12 (Dec. 1998).
121. For more on the Swedish cohabitation tradition, see Bretta Hoem & J.M. Hoem, The Swedish Family: Aspects of Contemporary Development, 9 J. FAM. ISSUES 397 (1988).
122. Robert Schoeni, Marital Status and Earnings in Developed Countries, 8 J. POP. ECON. 351 (1995). This article looks at 14 developed countries, relying on the Luxembourg Income Study for data. It shows that married men earn more than single men (but not widowed men) nearly everywhere; it does not attempt to determine why.
123. The association between male earnings and marital status could have three different sources: men with good earnings prospects have more success as suitors because their earning potential itself increases their appeal to women; men with good earnings prospects have more success as suitors because there is an overlap in the attributes, which may remain unidentified, that lead to good earnings and those that appeal to women; or, marriage itself enhances men's earnings prospects. Teasing these apart is methodologically difficult, but it seems certain that the first contributes to the association, even if it does not account for all of it. See, e.g., Christopher Cornwell & Peter Rupert, Unobservable Individual Effects, Marriage and the Earnings of Young Men, 35 ECON. INQUIRY 285 (1997) (finds that most of the male earnings premium attributed to marriage "is associated with unobservable individual effects that are correlated with marital status and wages"); Yinon Cohen & Yitchak Haberfeld, Why Do Married Men Earn More Than Unmarried Men?, 20 SOC. SCI. RES. 29 (1991) (finds no evidence that marriage generally confers a career advantage on men).
124. The association between male earnings and marital status could have three different sources: men with good earnings prospects have more success as suitors because their earning potential itself increases their appeal to women; men with good earnings prospects have more success as suitors because there is an overlap in the attributes, which may remain unidentified, that lead to good earnings and those that appeal to women; or, marriage itself enhances men's earnings prospects. Teasing these apart is methodologically difficult, but it seems certain that the first contributes to the association, even if it does not account for all of it. See, e.g., Christopher Cornwell & Peter Rupert, Unobservable Individual Effects, Marriage and the Earnings of Young Men, 35 ECON. INQUIRY 285 (1997) (finds that most of the male earnings premium attributed to marriage "is associated with unobservable individual effects that are correlated with marital status and wages"); Yinon Cohen & Yitchak Haberfeld, Why Do Married Men Earn More Than Unmarried Men?, 20 SOC. SCI. RES. 29 (1991) (finds no evidence that marriage generally confers a career advantage on men).
125. Pamela Smock & Wendy Manning, Cohabiting Partners' Economic Circumstances and Marriage, 34 DEMOGRAPHY 331 (1997).
126. See, e.g., in 1993 43% of black women aged 30 to 34 had never married, as compared with only 15.5% of white women in that age range. Arlene Saluter, Marital Status and Living Arrangements, March 1993, U.S. BUREAU OF THE CENSUS, CURRENT POPULATION REPORTS, SERIES P20-478 (1994). In 1998, 5 million of the 11 million black women 15 and older (45%) had never married, while 25 million of the 85 million white women of that age (29%) had never married. Table MS-1, Marital Status of the Population 15 Years Old and Over, by Sex and Race: 1950 to Present, Census Bureau, Jan. 7, 1999, 〈http://www.census.gov/population/socdemo/ms-la/tabms-1.txt〉. For some of the work on the connection between the low black marriage rates and the economic circumstances of black men, see the literature review in Michael J. Brien, Racial Differences in Marriage and the Role of Marriage Markets, 32 J. HUM. RES. 741 (1998).
127. See, e.g., in 1993 43% of black women aged 30 to 34 had never married, as compared with only 15.5% of white women in that age range. Arlene Saluter, Marital Status and Living Arrangements, March 1993, U.S. BUREAU OF THE CENSUS, CURRENT POPULATION REPORTS, SERIES P20-478 (1994). In 1998, 5 million of the 11 million black women 15 and older (45%) had never married, while 25 million of the 85 million white women of that age (29%) had never married. Table MS-1, Marital Status of the Population 15 Years Old and Over, by Sex and Race: 1950 to Present, Census Bureau, Jan. 7, 1999, 〈http://www.census.gov/population/socdemo/ms-la/tabms-1.txt〉. For some of the work on the connection between the low black marriage rates and the economic circumstances of black men, see the literature review in Michael J. Brien, Racial Differences in Marriage and the Role of Marriage Markets, 32 J. HUM. RES. 741 (1998).
128. See, e.g., in 1993 43% of black women aged 30 to 34 had never married, as compared with only 15.5% of white women in that age range. Arlene Saluter, Marital Status and Living Arrangements, March 1993, U.S. BUREAU OF THE CENSUS, CURRENT POPULATION REPORTS, SERIES P20-478 (1994). In 1998, 5 million of the 11 million black women 15 and older (45%) had never married, while 25 million of the 85 million white women of that age (29%) had never married. Table MS-1, Marital Status of the Population 15 Years Old and Over, by Sex and Race: 1950 to Present, Census Bureau, Jan. 7, 1999, 〈http://www.census.gov/population/socdemo/ms-la/tabms-1.txt〉. For some of the work on the connection between the low black marriage rates and the economic circumstances of black men, see the literature review in Michael J. Brien, Racial Differences in Marriage and the Role of Marriage Markets, 32 J. HUM. RES. 741 (1998).
129. HRDY, supra note 48, at 378.
130. Id. at 409.
131. Id. at 501.
132. Hrdy describes the modern young couple who start out with the intention of caring equally for their newborn, but are soon affected by both the baby's preference for the mother, and by the mother's inclination to respond more rapidly to infant cries of discomfort or need than do fathers, a difference Hrdy thinks is probably innate. (There is no difference in the response of mothers and fathers when the baby gives cries of real distress.) See HRDY, supra note 48, at 211-13.
133. For supporting authorities for all these propositions, see Linda Waite, Does Marriage Matter?, 32 DEMOGRAPHY 483 (1995).
134. In 1994, 37 percent of children in single-parent households were living with a divorced parent, but 36 percent were living with a never-married parent. The remainder were living with a separated parent. This is a significant change since 1983, when he proportion living with a divorced parent was nearly double (42%) those living with a never-married parent (24%). Marital Status and Living Arrangements: March 1994, BUREAU OF THE CENSUS, CURRENT POPULATION REPORTS. SERIES P20-484 (1996), at Table E.
135. Compare, e.g., Neal v. Neal, 570 P.2d 758 (Ariz. 1977) (no reason for divorce court to retain jurisdiction to provide for alimony in an era of "feminine equality"), with Mori v. Mori, 603 P.2d 85 (Ariz. 1979) (court ordered to retain jurisdiction so it may extend alimony indefinitely at termination of 23 year marriage with grown children).
136. Compare, e.g., Neal v. Neal, 570 P.2d 758 (Ariz. 1977) (no reason for divorce court to retain jurisdiction to provide for alimony in an era of "feminine equality"), with Mori v. Mori, 603 P.2d 85 (Ariz. 1979) (court ordered to retain jurisdiction so it may extend alimony indefinitely at termination of 23 year marriage with grown children).
137. See Domestic Partners, in AMERICAN LAW INSTITUTE, PRINCIPLES OF THE LAW OF FAMILY DISSOLUTION ch. 6 (Tentative Draft No. 4, 2000), which provides for divorce remedies at the dissolution of most cohabiting relationships with children.