Our analysis assumes no effect of marriage on the labor supply of parents. The best evidence, as reported in Ribar’s extensive literature review in 2004, indicates that marriage increases male labor supply and seems not to depress the average female labor supply in the more recent groups of women studied.47
Our analysis assumes no behavioral effect of marriage on parenting skills. If marriage reduces stress on parents, which leads to better parenting, then this approach underestimates the true taxpayer costs of family fragmentation.
Given the limits of the available empirical evidence, we implicitly assume that these three reasons exactly offset any of the effects of childhood poverty that are due to unobserved lower motivation and/or skills present among single parents. To the extent this assumption is wrong and it leads to overstating taxpayer costs because of the use of Holzer’s research, the magnitude of the overestimate would have to be viewed in light of the magnitude of our underestimates as described in the next subsection. As suggested below, these underestimates are likely quite substantial.
3. The use of Thomas and Sawhill’s research overestimates the impact of marriage on reducing poverty.
Thomas and Sawhill estimate that marriage would lift 65.4 percent of single- mother households out of poverty.48 In their microsimulation they place individu- als in the March 1999 CPS in “plausible” marriages until they obtain a marriage rate similar to 1970. Attempting to marry all single-mother households would likely fall short because of a lack of marriageable men—prisoners are disproportionately men, as are the unemployed, and men have lower life expectancies than women. The dearth of marriageable men is one reason that we use a 60 percent figure instead of the 65.4 percent estimate from Thomas and Sawhill. In addition, Thomas and Sawhill assume no behavioral effects of marriage on male labor supply, which sug- gests they underestimate the effect of marriage on poverty reduction. For these two reasons, our use of Thomas and Sawhill’s research should not lead to an overesti- mate of the taxpayer cost of family fragmentation.
4. The main assumption of this study—the percentage of the costs of government programs due to family fragmentation is proportional to the percent of poverty due to family fragmentation—overestimates the taxpayer costs.
To the contrary, the following thought experiment suggests that this assumption likely leads to an underestimate of the taxpayer cost of family fragmentation. Suppose there was an antipoverty program that cost taxpayers a total of $100 bil- lion. Also suppose this program provided $5,000 per year to 10 million married households and $5,000 per year to 10 million single-mother households. In addi- tion, suppose that 20 million married households were eligible for the program but only 10 million used it, while all 10 million single-mother households eligible for the program used it.