41
Table 5: Identification, estimates and residual diagnostics of the VECM and elasticity of substitution by sub-samples
|
VARIABLES |
VAR SPECIFICATION |
THEORETICAL RESTRICTIONS |
VECM DIAGNOSTIC |
ELASTICITY OF SUBSTITUTION | ||||||
|
Lag |
Rank of |
Symmetry |
Homogeneity and |
Residuals |
η21 |
η12 |
Net effect | |||
|
Gender |
Male |
3 |
2 |
3.484 (1) [.062] |
10.84 (3) [.012] |
39.66[.309] |
0.533 (.063) |
0.354 |
(.044) |
0.178 |
|
Female |
3 |
2 |
10.723 (1) [.001] |
3.517 (3) [.318] |
39.280[.325] |
0.222 (.021) |
0.140 |
(.013) |
0.081 | |
|
Education |
High |
2 |
3 |
not significant | ||||||
|
Low |
3 |
2 |
0.237(1) [.625] |
9.693 (3) [.021] |
38.828[.343] |
0.689 (.201) |
0.456 |
(.127) |
0.233 | |
|
Relative poverty |
Above |
3 |
2 |
not significant | ||||||
|
threshold |
Below |
4 |
2 |
0.170 (1) [.679] |
11.91 (3) [.007] |
47.711[.092] |
0.591(.189) |
0.394 |
(.124) |
0.196 |
|
Age |
< 35 |
2 |
2 |
not significant | ||||||
|
35 - 49 |
4 |
2 |
0.509 (1) [.475] |
12.438 (3) [.006] |
76.077[.348] |
0.393 (.192) |
0.267 |
(.130) |
0.126 | |
|
50 - 64 |
3 |
2 |
1.878 (1) [.170] |
2.947 (3) [.407] |
38.043[.376] |
0.543 (.064) |
0.359 |
(.043) |
0.184 | |
|
> 65 |
3 |
2 |
0.133 (1) [.714] |
6.583 (3) [.086] |
52.453[.037] |
0.321 (.039) |
0.201 |
(.024) |
0.119 | |
Note: High education stands for people that achieved a degree, master or PhD. Low education for the others. The criterium used to select the optimal lag order is the sequential modified LR test statistic.
The test of the rank of cointegration of the unrestricted VAR use the Johansen’s procedure. Cointegration test inferences of the AI model applied to each subsample are reported in Appendix. Degree of
freedom and p-values of the LR tests for the theoretical restrictions are reported in round and square brackets, respectively. Diagnostic autocorrelation test and estimations of substitution elasticities are
obtained by imposing symmetry and homogeneity in the VECM. Q-statistics distributed as a χ2 is reported with the p-values in square brackets. The degree of freedom are 36 for VAR lag order of 3 and 72
for lag order of 4. In round brackets are reported bootstrap standard errors of the estiamated elasticities.