Statistical Procedure. 3.1.1. For the CO2 emissions and electric energy consumption from the 4 phases of a WLTP test: For the total number of N tests and the measurement results of the tested vehicles, x1, x2 «N, th[e average Xtests and the standard deviation s shall be determined: ܺ ൌ ሺ୶భା୶మା୶యାڮା୶ొሻ ௧௦௧௦ and ሺݔଵെܺ ሻଶሺݔ െܺ ሻଶ Ǥሺݔ െܺ ሻଶ ݏൌඨ ௧௦௧௦ ଶ ௧௦௧௦ ܰെͳ ே ௧௦௧௦
Statistical Procedure. We estimate the effect of spousal retirement on individual health using a standard fixed effects regression, with Yit = β Xit + vt + ui + εit , where Yit is a dependent variable of interest, measured with Mobility index, ADL index, CESD score, and self-accessed health outcomes, representing the health condition of respondents in wave i at time t; Xit represents the independent variables; vt represents time fixed effects; ui represents personal fixed effects and εit represents the residual. The previous model estimates the association between spouses being fully retired and health outcomes of respondents. However, the association could be interpreted in the opposite direction of the intended topic of the current study – namely poor health conditions of the respondents might affect spouses’ retirement decisions. Intuitively if the respondent is in poor health condition, his or her spouse may decide to work longer for the additional income or the spouse may choose to retire earlier to have more time to take care of the respondent. In fact, several studies focusing on retirees’ health conditions have found that poor health – both physical health problems and psychiatric disorders - motivates individuals to retire earlier (Xxxxx & Xxxxxxxx, 1999; XxXxxxx, 2014; Ettner et al. 1997). Moreover, Xxxxxxx and Xxxxxxxxx (2001), analyzing the patterns of joint labor supply behavior, found that both men and women were less likely to retire if their spouses appeared to have left the labor force because of health problems. To address this reverse causality issue, we drop respondents who report being diagnosed with one of the following illnesses prior to their spouses’ retirement: high blood pressure, diabetes, cancer, chronic lung disease, heart disease, stroke, arthritis, and psychiatric problems before their spouses retired are excluded from the sample. This reduces the sample to 67,741 person-wave observations after excluding these respondents. The means of both personal and panel variables in this subsample are presented in Table 4 and Table 5. The change in the financial structure after retirement obviously would have different impact on families with different levels of income. Hence in the previous models, total household income is used as a control variable, which means that it is kept constant when analyzing the effect of spouses’ retirement on respondents’ health outcomes. Nevertheless, as mentioned in the introduction section above, changes in household income ...
Statistical Procedure
