Compression of morbidity and the labour supply of older people

Abstract

This article tests whether there is evidence of compression of morbidity using Health and Retirement Study (HRS) data and analyses the effects of this on the labour supply of older people. We find younger cohorts to suffer less from functional problems than older cohorts at given ages. Furthermore, we observe that instrumentalized disability has a negative effect on labour force participation. According to the cohort analysis and the multivariate analysis, it can be concluded that individuals will be able to work longer because of the delay in the onset of disability problems.

Acknowledgements

The author acknowledges the financial support provided through the European Community's Human Potential Programme under contract HPRN-CT-2002-00235, [AGE]. This article was written during my stay at the University Ca’Foscari of Venice. I thank the Economics Department and specially Agar Brugiavini for her support and her contributions. I also thank David M. Blau, Mette Christensen, Christina Wübbeke and Johann Fuchs for their comments.

Notes

¹ For women aged 15–49 (Source: OECD).

²Zweifel et al. (1999) argue that what is determining in terms of health care expenditure is the proximity of death and not the age at which death occurs, so ageing of the population will not have major effects on aggregate health care expenditures.

³ Like osteoarthritis, depression, isolation, Alzheimer.

⁴ Concretely, the first 5 waves of the RAND version of the HRS.

⁵ The delay in the onset of functional problems might also be evidence of what we have defined in the last section as status quo, where the delay in the onset of health problems is equal to the delay in death. However, since we are comparing birth cohorts which are close we do not expect large delays in longevity and, therefore, we consider significant delays in the onset of health problems as prima facie evidence of compression of morbidity.

⁶ We repeat this calculation for every wave, so that this cumulative measure changes for every age at which the individual is observable.

⁷ See also Almond et al. (2004).

⁸ These results coincide with the results obtained in the medicine literature (Crimmins and Saito, 2001; Manton and Gu, 2001; Freedman et al., 2002; Fries, 2003).

⁹ F-tests are listed on the regression tables.

¹⁰ Disability is an index which can take values between 0 (nondisability) and 2 (when the ADL index takes value 1 and the IADL index takes value 1).

¹¹ Only for persons working.

¹² There are not enough observations in order to also include CODA (born between 1924 and 1930).

¹³ Calculations have been carried out following Long and Freese (2001).

¹⁴ The change in disability is the difference in disability level between two cohorts for ages between 51 and 60 (ages at which both cohorts are comparable).