254 AMERICAN SOCIOLOGICAL REVIEW Table 1.Instrumental Variables Estimates for the Economic Consequences of Separation for Married and Cohabiting Men:Panel Study of Income Dynamics,1982 to 1992 White Men Black Men Standard of Standard of Type of Real Per Capita Living Scale Real Per Capita Living Scale Separation IncomeIncome ELES Poverty Income Income ELES Poverty Gross Private Household Income Separation from -.397* .385*+ -.005 -037 -.383** .546** .069 .068 marriage (.040) (.050) (.043) (.044) (.112) (.164) (.137) (.139) Separation from ,443 .349* -.033 -.107 -.885** .129 -,429* .379 cohabitation (.098) (.094) (.092) (.096) (.141) (.243) (.182) (.189) Average percentage impact of separation if: Separated from -33 47 4 -32 73 marriage Separated from -36 42 -3 -10 -59 14 -35 -32 cohabitation Net Household Income after Taxes,Transfers,and Support Payments Separation from -.513** ,293** -,116*块 -.146** .527* .516* -.030 -.032 marriage (.035) (.040) (.035) (.036) (.084) (.114) (.094) (.099) Separation from -.553* ,254* -.130° -.219* -.860* .260 -.363* -.313 cohabitation (.063) (.068) (.059) (.066) (.103) (.175) (.135) (.136) Average percentage impact of separation if: Separated from -40 34 -11 -14 68 3 -3 marriage Separated from 42 29 12 -20 -58 30 -30 -27 cohabitation Note:Numbers in parentheses are robust standard errors.Number of person-years for white men is 16,403; the number of person-years for black men is 5,410. p<.05 "p<.01 ""p<.001 (two-tailed tests) LEVELING MECHANISMS AND THE who separate.6 Along the way,we demon- AVERAGE IMPACT OF SEPARATION strate that the economic windfall that is con- ventionally attributed to men in general ac- Our next task is to unpack the market and tually applies to only a select group of men nonmarket mechanisms shaping the finan- who go through a separation or divorce. cial aftermath of separation.Married and co- There are clear economic winners;but there habiting men vary in their exposure to mar- are also clear economic losers. ket risk from the loss of their partner's in- The regression estimates in Table 2 show come,and separation almost certainly has the impact of separation on change in nomi- different consequences for sole breadwin- nal income for men according to breadwin- ners than for men who rely heavily on their ner status in the pre-disruption household. partner's income.Welfare state,judicial,and private,voluntary responses to the separa- tion may exacerbate or narrow these differ- 6 Married men in our sample had better post- ences.In the remainder of the analyses we separation outcomes than did cohabiting men,in part because they contributed proportionally use breadwinner status rather than marital more to pre-disruption household income.In status to show how variation in men's pre- supplemental analyses restricted to the sub- disruption household situation contributes to sample of married men,we obtained results simi- a heterogeneous set of outcomes for men lar to those presented here. This content downloaded from 129.96.252.188 on Mon,15 Feb 2016 15:26:54 UTO All use subject to JSTOR Terms and Conditions254 AMERICAN SOCIOLOGICAL REVIEW Table 1. Instrumental Variables Estimates for the Economic Consequences of Separation for Married and Cohabiting Men: Panel Study of Income Dynamics, 1982 to 1992 White Men Black Men Standard of Standard of Type of Real Per Capita Living Scale Real Per Capita Living Scale Separation Income Income ELES Poverty Income Income ELES Poverty Gross Private Household Income Separation from -.397*** .385*** -.005 -.037 -.383*** .546*** .069 .068 marriage (.040) (.050) (.043) (.044) (.112) (.164) (.137) (.139) Separation from -.443** .349*** -.033 -.107 -.885*** .129 -.429* -.379* cohabitation (.098) (.094) (.092) (.096) (.141) (.243) (.182) (.189) Average percentage impact of separation if: Separated from -33 47 0 -4 -32 73 7 7 marriage Separated from -36 42 -3 -10 -59 14 -35 -32 cohabitation Net Household Income after Taxes, Transfers, and Support Payments Separation from -.513*** .293*** -.116*** -.146*** -.527*** .516*** -.030 -.032 marriage (.035) (.040) (.035) (.036) (.084) (.114) (.094) (.099) Separation from -.553*** .254*** -.130* -.219*** -.860*** .260 -.363** -.313* cohabitation (.063) (.068) (.059) (.066) (.103) (.175) (.135) (.136) Average percentage impact of separation if: Separated from -40 34 -11 -14 -41 68 -3 -3 marriage Separated from -42 29 -12 -20 -58 30 -30 -27 cohabitation Note: Numbers in parentheses are robust standard errors. Number of person-years for white men is 16,403; the number of person-years for black men is 5,410. *p <.05 ** < .01 ***p < .001 (two-tailed tests) LEVELING MECHANISMS AND THE AVERAGE IMPACT OF SEPARATION Our next task is to unpack the market and nonmarket mechanisms shaping the finan￾cial aftermath of separation. Married and co￾habiting men vary in their exposure to mar￾ket risk from the loss of their partner's in￾come, and separation almost certainly has different consequences for sole breadwin￾ners than for men who rely heavily on their partner's income. Welfare state, judicial, and private, voluntary responses to the separa￾tion may exacerbate or narrow these differ￾ences. In the remainder of the analyses we use breadwinner status rather than marital status to show how variation in men's pre￾disruption household situation contributes to a heterogeneous set of outcomes for men who separate.6 Along the way, we demon￾strate that the economic windfall that is con￾ventionally attributed to men in general ac￾tually applies to only a select group of men who go through a separation or divorce. There are clear economic winners; but there are also clear economic losers. The regression estimates in Table 2 show the impact of separation on change in nomi￾nal income for men according to breadwin￾ner status in the pre-disruption household. 6 Married men in our sample had better post￾separation outcomes than did cohabiting men, in part because they contributed proportionally more to pre-disruption household income. In supplemental analyses restricted to the sub￾sample of married men, we obtained results simi￾lar to those presented here. This content downloaded from 129.96.252.188 on Mon, 15 Feb 2016 15:26:54 UTC All use subject to JSTOR Terms and Conditions