What Gradient? A BMA Approach to Untangling Income, Inequality, and Mortality

In this update to an earlier analysis using BMA, I consider both a wider variety and some recoded plausible controls that might affect the observed relationship between income and mortality, and inequality and mortality. I also suggest that the much of the empirical investigation of inequality and health outcome (including mortality) rests on the incorrect proposition that income has a direct causal link to health, whereas its influence is likely to only ever be indirect. Consistent with this speculation, I find that when the right controls are selected from a sufficient number and variety of candidates, the relationship between income and mortality disappears, as does the relationship between inequality and mortality. This is a novel finding with cautionary implications for public health policy in the US, where a known bias is to assume that income growth will solve local health problems