121 The effects of supported housing for individuals with mental disorders the nature of the intervention, the institutional setting, and the selection of the study population. As compared to the full study population, the relative likelihood that the compliers have certain characteristics is presented in column 3 of Table 2. Compliers are more likely than the average individual to be young adults, have worked in the prior year, to have a personality disorder or no treatment/diagnosis of mental health disorder in the previous 365 days; and less likely to be late-age adults, to live with a partner at the time of the application, to have prior home care use or high prior health care expenditure, have a psychotic disorder or to be in the category other diagnosis. Parental outcomes Table 5 displays the effects on parental mental health care use and labour outcomes for the subpopulation for whom information about parents is available. Information about at least one parental outcome is available for 72% of the individuals in the full study population, ranging from 48% to 66% depending on the outcome13. This group differs from the overall population by being on average younger, more likely to live with the parents, work ahead of the application, and to have a disorder diagnosed during childhood (Table A7 of Appendix). Overall, effects of supported housing eligibility on individual outcomes among those with information on at least one parental outcome are qualitatively and statistically consistent with those in the study population, although the magnitude of the effects is smaller for supported housing admission and more precisely estimated for a decrease in the individual labour participation in the short-run (Table A9 of the appendix). Over time, child eligibility for supported housing progressively increases the likelihood of parents being in a paid job, and leads to 15.1 pp (se: 7.2 mothers) and 17.8 pp (se: 7.6 fathers) increases on the forth calendar year after the application. The relative size of these effects is larger for the mothers, for which the average population employment is lower than for fathers (36% vs. 51%). No precise estimates are found for the other parental outcomes. 13 There are several reasons for parental outcomes not being available. On the one hand, the identity of the father and the mother might not be known (29% and 17% of all missing observations for parents, respectively): parents might have never lived in the Netherlands and child-parent linkages are only available for parents alive from 1995. On the other hand, outcome data might be missing for parents observed in the data: parents have moved abroad or have died, either before the application was placed or between the application and the measurement of outcomes. Importantly, there is no missing data for parental outcomes in the subgroup of individuals living with their parents at the time for the application, for which we conduct a robustness analysis. 4
RkJQdWJsaXNoZXIy MjY0ODMw