Close
Martin Weale, Silvia Sze Wai Lui and Jan Van de Ven, National Institute of Economic and Social Research
(Project no. 30125)
The project will produce estimates of healthy life expectancy based on the British Household Panel Survey (BHPS) and to find the influence of education on these, data from the ONS LS will make it possible to correct for under-recording of mortality.
The analysis proposed will involve the calculation of age-dependent transition matrices between health states identified in the BHPS. It is a development of a paper published in the Journal of the Royal Statistical Society, Series A. Vol 171. p. 203-222. A key issue raised, and addressed only in aggregate was the under-recording of mortality in the BHPS. The study did not address the question of migration from households to care institutions. This study is intended to address both the mortality and the migration questions by making use of data on mortality rates from the 1991 and 2001 Censuses living in care institutuions.
Deaths of people identified in the two censuses will make it possible to estimate death rates by age for people aged 65+ and to see how they have changed between the 1991 population and the 2001 population. Interpolating/extrapolating the prevalence of institutional living and mortality patterns for the years covered by BHPS, it will then be possible to calculate, using the official life tables to provide whole population mortality rates, mortality rates for people living in households. Migration rates to institutions can also be deduced. These will be compared with migration and mortality rates identified in BHPS. An underlying assumption is that old people move from households into institutions but not in the other direction. For this reason it is not clear that it is useful also to calculate mortality rates for people reported as living in households in the 2001 Census- although the utility of such data will also be explored.
The final aim of the study will be to estimate health state transition equations using an ordered probit model - with mortality (and probably migration) treated as "health states" ranked below very poor health in the ordered probit. Educational status and possibly income will be used as explanatory variables in addition to initial health state, age and sex. The information on aggregate mortality and migration rates will be used to identify restrictions on the coefficients of the probit model. A framework will also be developed to handle the statistical problem that under-recording of mortality and migration may be different in form from other forms of attrition from the BHPS.