146 Chapter 4 Table A14. Two-stage least squares (2SLS) regression estimates for individual outcomes with alternative treatment definition: supported housing + institutional care Effect of eligibility for supported housing First-stage Observations coefficient (se) coefficient (se) Calendar year after application Supported housing admission 0.354*** (0.056) 1.062*** (0.043) 7,953 Use of home care -0.138** (0.064) 1.062*** (0.043) 7,953 Use of mental health care -0.04 (0.049) 1.060*** (0.043) 7,852† Total expenditure (€) 21,690*** (4,708) 1.062*** (0.043) 7,953 Supported housing expenditure (€) 13,649*** (2,446) 1.062*** (0.043) 7,953 Home care expenditure (€) -178 (512) 1.062*** (0.043) 7,953 Medical expenditure (€) -806 (855) 1.060*** (0.043) 7,852† Total mental health care expenditure (€) 7,244* (3,903) 1.060*** (0.043) 7,852† Outpatient mental health care expenditure (€) 980 (617) 1.060*** (0.043) 7,852† Inpatient mental health care expenditure (€) 6,263 (3,894) 1.060*** (0.043) 7,852† All-cause mortality -0.016 (0.013) 1.062*** (0.043) 7,953 Working -0.085* (0.037) 1.073*** (0.043) 7,596† Income from work (€) -2,425*** (695) 1.073*** (0.043) 7,596† Personal Income (€) -1,490** (713) 1.073*** (0.043) 7,596† Fourth calendar year after application All-cause mortality 0.034 (0.029) 1.062*** (0.043) 7,953 Working -0.087 (0.052) 1.075*** (0.040) 7,218† Income from work (€) -3,763* (1,435) 1.075*** (0.040) 7,218† Personal Income (€) -2,276*** (718) 1.075*** (0.040) 7,218† Notes: Robust standard errors in parentheses, clustered at the assessor level; *** p<0.01, ** p<0.05, * p<0.1. †The smaller number observations is due to missing data for the outcomes coming from health insurance claims and tax returns databases. All regressions include the main specification controls described in section Empirical implementation. Estimates above are results of 2SLS when the instrument is defined as eligibility to supported housing or institutional care (inpatient mental health care, nursing homes or residences for the disabled). Results are similar to main estimates, reinforcing the idea that the small share of individuals from the control group that is granted eligibility to institutional care has little influence the LATEs of our main specification.
RkJQdWJsaXNoZXIy MjY0ODMw