There's no place like home: the relationship of nonstandard employment and home ownership over the 1990s

American Journal of Economics and Sociology, The, Oct, 2004 by Barbara A. Wiens-Tuers

Each NLSY79 survey year from 1994 to 1998 the respondent worked at all is categorized by type of employment. Dummy variables are used to represent regular or standard employees (Reg), agency temporaries (Agftem) and in-house temporaries (Intern). Consultants and contractors (CC) and employees of contractors and others (Other) are respectively grouped together. The employment dummies are then broken down into current employment, or the type of employment reported in the 1998 survey, and past employment or type of employment, if working, in the 1994 or 1996 surveys. Four separate regressions are run. The first regression (1) includes no employment variables. The second regression (2) includes no employment arrangements, the third regression (3) includes past employment arrangements, and the final regression (4) includes both current and past employment. The variables in the regressions are weighted (inverse probability of appearing in the sample) using the weights assigned to the 1998 survey.

The coefficients of the independent variables ([beta]) are reported as odds ratios. Hosmer and Lemeshow (1989) define the odds ratio. denoted as [PSI], as the ratio of probabilities for a discrete independent variable, x = 1 to x = 0. For example, if the dependent variable HO denotes whether or not the household is purchasing its home and x = 1 denotes the respondent is male, [psi] = 2 indicates that a male is twice as likely to be a homeowner than is a female. For a continuous independent variable, the coefficient gives a change in the log odds for a "1" unit increase in x. The odds ratio for each independent variable is calculated controlling for the other independent variables (at the mean of the other variables).

Regression results (see Table 5) for the usual players in the models of home ownership show results similar to that of other researchers cited earlier in this paper. Relative to the lowest income level, there is a greater probability of owning a home as the level of income increases. Married households are over four times more likely to own a home than are unmarried households. Relative to households of six or more people, smaller households are more likely to own a home. Respondents living in the southern or north central United States are much more likely to be homeowners than are those living in the western United States.

In 1998, even after reforms in lending and other institutional changes, race is still a significant predictor of home ownership. Relative to white households, black households are a third as likely to own their homes, and Hispanic households are about half as likely. In all fairness, one would have to compare the coefficient of race over several periods to see if it has become less important over time (see Bostic and Surette 2000). The level of education is a significant predictor of home ownership. Respondents with less than a high school education are much less likely to own a home than are respondents with a high school diploma, and those with college education and above are more likely to own a home than are high school graduates. Segal and Sullivan (1998) note in their study that education has greatly increased in importance as an indicator of home ownership. Their study shows that the gap in home ownership between respondents with less than a high school degree and those with a college degree rose from less than 6 percentage points in 1977 to over 20 percentage points in 1997.