Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
educe the quality of dwellings. Therefore, they are consistent with the theoretical predictions<br />
of cases I and S.<br />
Table 2 Probit coefficients and marginal effects (total effect)<br />
Variable Coefficient Standard error Marginal effect<br />
Intercept 1.41 ∗∗∗ 0.056<br />
Tenant-owned housing −0.12 ∗∗∗ 0.036 −0.02<br />
Landlord-owned housing −0.34 ∗∗∗ 0.034 −0.06<br />
Built pre–1970 −1.33 ∗∗∗ 0.039<br />
Built 1971–1980 −0.90 ∗∗∗ 0.039<br />
Built 1981–1990 −0.46 ∗∗∗ 0.041<br />
SRC 0.13 ∗∗∗ 0.046<br />
Block −0.20 ∗ 0.118<br />
Others 0.07 0.086<br />
Room 0.04 ∗∗∗ 0.005<br />
Story 0.16 ∗∗∗ 0.020<br />
Income 0.05 ∗∗∗ 0.005<br />
Number of observations 44772<br />
Log-likelihood 4012.12<br />
∗∗∗ indicates significance at 1%; ∗ indicates significance at 10%<br />
Table 3 reports probit coefficients and marginal effects for tenure security on housing<br />
quality. In line with the theoretical predictions of both cases I and S, tenure security has an<br />
ambiguous effect on tenant-owned housing quality. For example, consider housing built before<br />
1970. Housing moved into between 1991 and 1998 is more likely to be unsound than is housing<br />
moved into before 1970 (with respective probabilities of 27% and 17%). The same is true of<br />
housing built between 1971 and 1980. Consider landlord-owned housing. The more recent the<br />
move-in date, the smaller the absolute value of the coefficient (except for housing moved into<br />
between 1981 and 1990 constructed before 1970). This implies that the longer the tenancy<br />
duration, the greater the extent of deterioration. Thus, tenure security further reduces the<br />
quality of landlord-owned housing. This result confirms the theoretical predictions of both<br />
cases I and S.<br />
21