Simple probit |
Seemingly unrelated probits^ |
Simultaneous probits^ | |||||
Public housing |
No |
Yes |
No |
Yes |
Yes |
Yes | |
in neighborhood eq. |
None |
None |
None |
None |
None |
Occupational |
Spouse’s |
in unemployment eq. |
(ɪ) |
(2) |
(3) |
(4) |
(5) |
status (6) |
nationality (7) |
Deprived neighborhood Coefficient 0.173*** |
0.631** |
0.788*** |
0.869 |
0.805 | |||
Correlation |
(0.042) |
0.0992*** |
0.0604** |
(0.270) -0.274* |
(0.116) -0.390*** |
(0.107) -0.426*** |
(0.114) -0.396*** |
Marginal effect |
0.0213 |
(0.0255) |
(0.0257) |
(0.155) |
(0.065) |
(0.060) |
(0.064) |
(0.00593) |
(0.00602) |
(0.00587) |
(0.00700) |
(0.00626) |
(0.00630) |
(0.00624) | |
[0.00591] |
[0.00596] |
[0.00563] |
[0.00705] |
[0.00611] |
[0.00627] |
[0.00612] | |
Public housing Marginal effect |
0.0315 |
0.0386 |
0.0327 | ||||
Log likelihood |
-2351.34 |
-8164.32 |
-7883.84 |
-8162.24 |
(0.00632) |
(0.00658) |
(0.00628) |
Pseudo-R2 |
5.88 |
9.55 |
12.66 |
9.57 |
12.86 |
12.11 |
12.68 |
Notes: Control variables include, unless otherwise mentioned, individual’s age, nationality, education level, occupational status, spouse’ age,
nationality, education level and number of children, (a) Neighborhood variable is not in the unemployment equation, (b) Neighborhood variable
is in the unemployment equation. Mean marginal effects are calculated as 1/N ∑(P(yn = 1∣⅜2 = I)- Р(.Уи = l∣!∕i2 = 0)) for neighborhood type
and l∕N∑(P(yil = l,τ∕i2 = ⅛3) + P(yn = 1>⅜2 = t¾3) - P(yil = 0, yi2 = ⅛3) - P(yn = 0, yi2 = 0∣⅜3)) for public housing. Predicted
probabilities are calculated for each individual vector of characteristics. Standard errors of marginal effects are calculated by the delta method.
Standard errors between square brackets are corrected for dependencies within neighborhoods.
Table 6: Effect of public housing and neighborhood on unemployment probability