The Gender-specific Role of Body Weight for Health, Earnings and Life Satisfaction in Piecewise and Simultaneous Equations Models

Based on the German Socio-Economic Panel, the correlation between the body mass index, health, earnings and life satisfaction is analysed by gender. The previous literature has found no consistent results. This might have several reasons. The purpose of this paper is to analyse the gender-specific role of weight in single equation, piecewise and simultaneous equations models. We ask whether this distinction is important for the degree of association between health, earnings, satisfaction and body weight. In our context, piecewise modelling means a separate inspection of weight coefficients for under- and overweight people, allowing the detection of non-linear influences. As a benchmark, we begin our estimations under the assumption that the association between health, earnings, satisfaction, and weight is the same for under- and overweight people, and that there are no jointly dependent influences between our three outcome variables. The basic results are: health worsens, income declines and satisfaction is poorer with higher body mass index. If the association with weight is separately determined for over- and underweight people, the estimates show striking differences between overweight men and women. Underweight women earn more and overweight less than others. For normal-weight men the income is on average higher than for over- and underweight men but this difference is insignificant. When matching and instrumental variables procedures are applied, the health outcome for overweight people matches that of independent and unmatched estimates. Stronger positive effects on health are found for underweight women. No clear-cut advantages in income of overweight women can be found. Underweight women and especially underweight men tend to be less happy. For overweight men this influence is ambiguous but more speaks in favour of a lesser level of satisfaction. Overweight women seem to be happier.