In this paper we use the two waves of the British Retirement Survey (1988/89 and 1994) to quantify the relationship between socio-economic status and health outcomes. We find that, even after conditioning on the initial health status, wealth rankings are important determinants of mortality and the evolution of the health indicator in the survey. For men aged 65 moving from the 40th percentile to the 60th percentile in the wealth distribution increases the probability of survival by between 2.4 and 3.4 percentage points depending on the measure of wealth used. A slightly smaller effect is found for women of between 1.5 and 1.9 percentage points. In the process of estimating these effects we control for non-random attrition from our sample. British welfare reform debate in recent years. This debate is informed by tax-benefit modelling, yet accurate modelling of Family Credit is fraught with potential problems. The main model input data are found to under-sample Family Credit recipients considerably, but those who it does sample seem representative of the Family Credit recipient population. Substantial mismatch is found between those reporting Family Credit receipt and those modelled as en Titled. We show that regression techniques can be used to adjust model results for the fact of non take-up, but that data constraints leave no obvious way to deal with the equally significant problem of famlies who receive benefit but are not modelled as en titled. The difficulties posed by the input data's under-sampling and by the significant number of claimants without modelled en Titlement lead us finally to consider the use (and the limitations) of calibrating results.
