Abstract
In this chapter we continue our estimation of adult mortality hazard functions. Here we use the Retirement History Survey (RHS), which covers heads of households aged 58–63 in 1969 and for whom death records are taken from Social Security records for the period 1969–1979 (though the data for 1978 and 1979 are incomplete because of late posting of the death data).31 Section 2.2 provides more detail about this data source. As discussed in Section 2.3, we use these data to estimate mortality hazard functions that can be interpreted as conditional demand functions or dynamic decision rules in which the right-side variables reflect decisions and the state of the world up to the time of the data collection.32
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
References
Data on deaths through 1981 have been collected, but the Census Bureau has refused to link the new information to the RHS because of legal constraints.
Also see Section 2.3 for discussion of possible estimation problems.
For the time period covered by the RHS Social Security death records are nearly 100 percent accurate (Duleep 1986). Examination of the more limited data on death provided primarily by widows in the RHS provides few contradictions with Social Security death records.
However, this is unlikely since for many wives life insurance is not sufficient to cover the costs of replacing their services. In 1985 the average amount of adult women life insurance was $21,000 (American Council of Life Insurance, 1986).
Defined benefit pensions are either proportional to the earnings base used or are progressive when a company integrates its pension plan with Social Security benefits (as many do).
We omit the physical activity transformation which is highly collinear with the dummies used.
Figure 5.2 displays the observed and estimated hazards for the proportional hazard model. The fit is good with an with an R2= 0.88.
For example, see Shulman (1987), Kahn and Sherer (1988) Andrisani (1977), Welch (1973), Smith (1984), Orazem (1987), Darity (1982), Ashenfelter (1977), Freeman (1973), Smith and Welch (1977), and Welch (1974).
The figure is limited to people aged 60 through 66 (even though our data include ages 58–73) in part because of the smaller sample sizes for other ages (arising from the age and panel structure of the RHS) and in part because of the incomplete information on death after 1977.
However, the white hazard would be about 11 percent higher if whites had blacks’ observed characteristics rather than the 18 percent figure when differences in pension income are based on the 1975 figures.
We have also run the equation for non-black women. The results are given in Table 5.8A. The marital status and income variables change by about 10 percent e.g., from-2.0 to-2.2, but the patterns and significance levels are similar.
Author information
Authors and Affiliations
Rights and permissions
Copyright information
© 1998 Springer Science+Business Media New York
About this chapter
Cite this chapter
Behrman, J.R., Sickles, R.C., Taubman, P. (1998). Mortality Hazard Estimates From the Retirement History Survey: Education, Pensions and Marital Status and Black-White and Gender Differences. In: Causes, Correlates and Consequences of Death Among Older Adults. Springer, Dordrecht. https://doi.org/10.1007/978-94-011-4393-6_5
Download citation
DOI: https://doi.org/10.1007/978-94-011-4393-6_5
Publisher Name: Springer, Dordrecht
Print ISBN: 978-94-010-5887-2
Online ISBN: 978-94-011-4393-6
eBook Packages: Springer Book Archive