03.06.2016 Views

Download PDF - Ivie

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

savings, since retirement and social security income for the elderly is the result of household<br />

savings throughout the life cycle (Lee, 2001; Lillard and Weiss, 1997), while the ratio of mortgage<br />

payments to house value reflects the couple’s capacity and willingness to save rather than consume<br />

(Wolff, 1998; Deaton and Paxson, 1994; Lin, et al., 2000; Hurst, et al., 1998).<br />

The regressors are the age and the educational level, the latter defined as the number of<br />

completed years of schooling, of each partner (spouse); the number of household members or of<br />

each partner’s own children living in the household; and a dummy variable for the length of time<br />

that the household has occupied the current residence (5 years or less), as a proxy for the duration of<br />

the relationship. Unfortunately, the Census does not record the total number of either the head or<br />

the partner's children but only the number of those who are currently members of the household.<br />

This is particularly regrettable in regard to our elderly-couples sample, in which the discrepancy<br />

between the actual total number of children and the total recorded by the Census would be<br />

particularly large. 7 . We account for economic conditions controlling for the total individual<br />

incomes of both heads and partners (spouses) in 1990, creating an average individual income in<br />

1990 by state, race, sexual orientation, cohabitation status, age (5-year brackets) and education (4<br />

groups, high-school dropouts, high school graduates, some college, college and college plus), which<br />

we merge by these characteristics to our sample in year 2000. As our analysis concerns current<br />

retirement income and the ratio of mortgage to house value by different types of households, we<br />

want to control for “lagged” income as a measure of the wealth and income that an individual in a<br />

given reference group is expected to have, on average.<br />

Although an individual’s health status is not recorded in the US Census, his or her disability<br />

status along several dimension of impairment (ambulatory, cognitive, and/or vision impairments,<br />

and independent living and work disabilities) is, and so we construct and control for the<br />

corresponding dummy variables in our estimation of the elderly sample. State dummy variables are<br />

7 A record of the total number of children borne by a woman is available in the Census, but only for women and only<br />

until 1990.<br />

16

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!