An individual receives utility at two points during a period. First, she receives utility associ- ated with alternatives mt, st, ct, and et. Then, later in the period, if she becomes pregnant she
receives utility based on the abortion decision made and her consumption of a composite good. A random preference shock for each combination of mt, st, ct, andet, denoted mscet , is known at
the beginning of the period, while a shock for each abortion alternative,a
t, is only realized after a
pregnancy.
The following utility functions represent the individual’s preferences within a period, where
Umsceis flow utility associated with the alternativesmt=m,st=s,ct=c, andet =e, andUais
flow utility associated with the alternativeat=a:
Umsce=XNt α0+ 2 X m=1 1[mt=m]Xmt α1+ S X s=1 1[st=s]Xstα2+ 2 X e=1 1[et=e]Xetα3+mscet (3.3) Ua= Ct1−θ 1−θ+at(α4+α5Rt+α6Zt+α7P Ct+α8CDt+µ 6) +B t(α9LBt+α10LBt2) + a t (3.4)
9In the simulations presented in 6.3 the level effects of abortion policy in the partner offer functions are fixed
across all policy regimes so that only the interactions of sexual activity with policy create differences in the partner offer distributions across the regimes.
1067 percent of the sample remain in the same state of residence for all years observed. Of the remaining 33 percent,
39 percent change states once and 37 percent change twice. Those who switch states often move to neighboring states or states in the same region of the country, which may have similar unobserved characteristics in their market for partners.
The vectorXNt contains number of children, number of children squared, and allows for pref- erences for children to vary with race. The other Xvectors contain variables that shift the utility of marriage, sex, and employment. Specifically,
Xmt = [1, Nt, (1−1[mt−1 =m]), rt, µ3m] Xst = [1, 1[mt= 0], 1[mt= 1], 1[mt = 0]Rt, C X c=1 1[ct=c], µ4s] Xet = [1, Nt, Bt−1, 1[mt= 0], 1[mt= 1], µ5e] The parameters µ3
m, µ4s, and µ5e capture permanent unobserved heterogeneity in preferences
for each of the m, s, and e alternatives.11 The parameters in the vector α
1 capture the utility (disutility) of romantic partnerships and the complementarity of partnership with the number of children in the household.12 To capture the initial fixed costs of cohabiting or becoming married
an indicator of the current partnership is interacted with an indicator of not being in the same type of partnership in the prior period, following Keane and Wolpin (2010). The parameters inα2 capture the marginal utility of sexual behavior and allow for the possibility that marginal utility of sex depends on partnership status and religiosity. The fifth element of Xs
t captures the utility (disutility) of contraceptive use.13 The parameters inα
3 capture the utility (disutility) of work and school, and the complementarity of working and being in school with children and partnership.
In equation (3.4),Ctis a composite consumption good,P Ctis an indicator of a binding parental
consent restriction, andCDtis an indicator of a binding mandatory counseling and delay law. Re-
call that the competitive effect depends on heterogeneity in preferences and costs of obtaining
11It is noted that the elements of theαvectors that multiply these permanent unobserved factors (i.e., the factor
loadings) are 1. Alternatively, the permanent unobserved factors could be fixed across alternatives and the factor loadings for each alternative could differ.
12With respect to partners, women in the model have preferences for the presence of a partner, the type of partner
(married or cohabiting), and the amount of income that a partner earns (discussed below). Preferences for having a partner and partner type vary by characteristics of the woman and the jointly chosen sexual activity, contraceptive use, and employment/schooling alternatives.
13Recall that contraceptive use is a percentage of sexual encounters for which the individual uses contraception.
abortion. The parameters inUacapture how abortion and pregnancy impact period utility directly.
The marginal utility of abortion is shifted by an individual’s religiosity, age, the requirement of parental consent if less than 18 years old, and the requirement that an individual receive state man- dated counseling and wait a state mandated amount of time before undergoing an abortion. The permanent unobserved factorµ6 allows for unobserved heterogeneity in abortion preferences. If the individual becomes pregnant and chooses to give birth (i.e.,Bt = 1), her utility from having
a child depends on the amount of years that have passed since she last gave birth,LBtandLB2t,
which allows for preferences over child spacing.14
It is noted that without observations on the monetary and time costs of specific actions, such as marrying or attending school, monetary, time, and psychic (utility) costs are not separately identified. For this reason, some of the parameters that are included in the utility function capture two or three of these types of costs. For example, the coefficient on attending school includes effort and time costs.