19(27)
The estimates of the coefficients of the reform variable seem to remain
very stable throughout the introduction of new covariates. These results
indicate that the allowance reduction may have negatively affected the
households’ housing consumption: after the 1997 reform, the risk of
moving into overcrowded conditions seems to increase significantly in the
treatment group by approximately 6% for every thousand SEK reduction in
the allowance.
12
The variables controlling for duration dependence indicate that the risk of
moving into overcrowded conditions increases with the duration of being a
housing allowance recipient, but at a decreasing rate.
To incorporate the heterogeneity issue, gamma and normal frailty models
were estimated, and the results are presented in Table 4. The frailty terms,
or the unobserved heterogeneity, are found to be significant; however,
incorporating normal or gamma frailty has no effect on the estimated
coefficient of the policy variable.
12
We also included variables controlling for gender (not significant), age squared, having children
older than 16, and age of the youngest child; doing so did not alter the results.