18(27)
Table 3. Estimation results, from discrete time proportional hazards
model (exponentiated coefficients), for the risk of
overcrowding. A complementary log-log model, without frailty.
Clustered sandwich estimator of standard errors within
parentheses (no. of households 58,168, failures 9693).
Model 1
Model 2
Model 3
Model 4
Model 5
Reform variable:
T*[d97](t)
1.067
***
(0.011)
1.063
***
(0.011)
1.061
***
(0.010)
1.062
***
(0.010)
1.062
***
(0.010)
Decrease in
housing allowance
0.939 ***
(0.006)
0.952 ***
(0.006)
0.945 ***
(0.006)
0.949 ***
(0.006)
0.944 ***
(0.006)
Year dummy,
1997–1999 [d97]
(t)
1.173 ***
(0.038)
1.067 *
(0.035)
1.047
(0.034)
1.096 ***
(0.035)
1.051
(0.034)
Household characteristics
Age (t)
-
0.956 ***
(0.001)
0.958 ***
(0.001)
0.958 ***
(0.001)
0.957 ***
(0.001)
2 children (t)
-
2.164 ***
(0.048)
2.174 ***
(0.048)
2.182 ***
(0.048)
2.182 ***
(0.049)
More than 2
children (t)
-
4.132 ***
(0.167)
4.165 ***
(0.168)
4.157 ***
(0.168)
4.157 ***
(0.168)
Housing expenses
/income, 1996
-
1.238 ***
(0.024)
1.283 ***
(0.026)
1.243 ***
(0.025)
1.287 ***
(0.026)
Housing market characteristics
Tobin’s q (t)
-
-
0.602 ***
(0.024)
-
-
Unemployment
rate (t)
-
-
-
1.083 ***
(0.004)
-
Municipal dummies
(t)
-
-
-
-
Yes
Duration dependence
Time (t)
1.548 ***
(0.082)
1.676 ***
(0.089)
1.671 ***
(0.089)
1.353 ***
(0.073)
1.686 ***
(0.089)
Time^2 (t)
0.925 ***
(0.007)
0.927 ***
(0.007)
0.928 ***
(0.007)
0.951 ***
(0.007)
0.927 ***
(0.007)
Log likelihood
–38,192.7
–36,784.1
–36,700.4
–36,568.8
–36,665.4
Notes
: * Significant at the 10% level. *** Significant at the 1% level. (t) indicates time-varying
variables
.
