23
Appendix 2:
Relation between national election vote and individual wealth considerations.
To support the idea that national election vote is related to individual wealth, we have regressed variable
CON on the per capita income. As exact data on local income per capita is not available, we have used the
estimated per capita income index (
PCII) that is elaborated by “La Caixa” financial institution, referring to
year 2000. This index lies in a range from 0 thru 10, where 10 is the maximum level of (estimated) local
income per capita. Results are shown in table A-1.
Table A-1: Relation between left-wing constituency and income per capita.
Explained variable: CON
PCII
-0.0335927
(-5.86)***
Constant
0.7056736
(15.72)***
F
34.36+++
R-squared
0.1657
Adjusted R-squared
0.1609
N
175
Notes:
*** significant at 1% level, ** at 5%, * at 10%.
+++ null hypothesis rejected at 1% level, ++ at 5%, + at 10%.
In brackets, t-statistic values (null hypothesis: value equal to zero).
There is a clear negative relation between national election vote and individual income, as it is seen in
the results of the estimation. Besides, we are to empirically prove that this relation is stronger at a national
election level than at a local election one, which is a critical issue of our paper. To do so, we estimate probit
models relating governing party
’s ideological identity and social majorities with the income per capita index.
Results are shown in table 5.
Table A-2: A comparison of the influence of per capita income on national and local election vote.
Explained dummy:
Left-wing mayor
Explained dummy:
Left-wing majority (>50% votes) in
national election.
PCII
-0.2025349
(-2.32)**
-0.6692689
(-4.92)***
Constant
1.666457
(2.44)**
4.259356
(4.29)***
F
5.45++
32.48+++
Log likelihood
-118.09257
-79.028506
Pseudo R-squared
0.0226
0.1705
N
175
175
Notes:
*** significant at 1% level, ** at 5%, * at 10%.
+++ null hypothesis rejected at 1% level, ++ at 5%, + at 10%.
In brackets, z-statistic values (null hypothesis: value equal to zero).
It is readily seen that the relation between income per capita and mayor
’s ideological identity is much
weaker than the one between income and social majority (national election vote). For instance, the log
likelihood and the Pseudo R-squared are much higher in the latter model.
Convention