This paper provides statistical methods that permit the association of socioeconomic status and health to be partially unraveled in panel data by excluding some postulated causal paths, or delimiting their range of action. These methods are applied to the Asset and Health Dynamics of the Oldest Old (AHEAD) Panel to test for the absence of causal links from socioeconomic status (SES) to health innovations and mortality, and from health conditions to innovations in wealth. We conclude that in this elderly American population, where Medicare covers most acute care and pension income is not affected by ability to work, the evidence supports the hypothesis of no direct causal link from SES to mortality and to incidence of most sudden onset health conditions (accidents and some acute conditions), once initial health conditions are controlled, but there is some association of SES with incidence of gradual onset health conditions (mental conditions, and some degenerative and chronic conditions), due either to causal links or to persistent unobserved behavioral or genetic factors that have a common influence on both SES and innovations in health. There is mixed evidence for an association of health conditions and innovations in wealth. The death of a spouse appears to have a negative effect on the wealth of the survivor; this is plausibly a direct causal effect. There is evidence for some association of health conditions with increased dissaving from liquid wealth for intact couples and singles. From these findings, we conclude that there is no evidence that SES-linked therapies for acute diseases induce mortality differentials. The question of whether SES-linked preventative care influences onset of chronic and mental diseases remains open. The appendix to this paper containing the detailed model estimates, the data, and the programs used for data preparation and estimation, can be found at http://elsa.berkeley.edu/wp/hww/hww202.html. (C) 2002 Elsevier Science B.V. All rights reserved.
