Until now, all regressions contained income as control variable. The main rationale for doing so,is that the impact of consumption, consumption types, and consumption categories might vary for different income groups. Table 13 shows that excluding income from the regressions does not change the main results. Furthermore, some people might argue that subjective explanatory variables should be excluded, because their assessment is simultaneously determined with the subjective dependent variable, and hence, they might affect each other. Therefore, I re-run the regression without physical health, worn-out, and sleep issues — all three are measured non-objectively. Table 14 indicates that the relevant consumption categories are essentially unaffected. For some consumption categories, significance drops by one star (*). On the other hand, the coefficient on health care is now also highly significant for depression. This is notunexpected, since health care spending captures direct (physiological suffering) and indirect (additional required costs) consequences of bad health, now that health controls are excluded