12
the weighting factors with regard to gender and age have been calculated so that they correspond to
those of the population.
One of the purposes of the research was to detect whether altruism or fairness plays a role in the
evaluation of the public pension system. We tested for the effects of these variables by estimating
different specifications of eqs. (3) where the differences are due to the inclusion or exclusion of
variables that might deal with feelings of altruism and fairness. Table 4.4 presents for 9 variants the
estimated coefficients for the younger generation by means of WLS regression.
[ INSERT TABLE 4.4 ABOUT HERE ]
Obviously, the income position of the respondents themselves (Kj) plays a key role in the evaluation
of the pension system. It is significant and positive in all variants considered. Further, it is rather robust.
Its average value amounts to 1.38.
The introduction of altruism (measured via income of others) and / or fairness (measured via the rate
of return) improves the explanatory power of the equation considerably. In particular, adding altruism
(variant 1) or altruism and the rate of return for the young themselves (variant 3) doubles or triples
the explanatory power. Adding next rates of returns for other generations leads to problems regarding
the statistical significance of the variables and/or their signs (compare e.g. variant 3 with variant 5).
The generation’s own income gives a larger contribution to utility than income of other generations.
Incorporating both the fairness terms with respect to the elderly (rʃ and r0y2) and the income of other
generations (Ym +Y0s, results in a negative sign for the latter variable. In particular, these statistical
results are due to the fact that the fairness terms for the middle-aged, the fairness terms for the elderly
and Ym are0 are strongly correlated. So, the size and / or the significance of variables measuring
altruism and fairness is affected by the specification. When all rates of return are included (variant 4)
fairness turns out to be completely insignificant which, as noted before, is probably due to
multicollinearity problems. The same problems make it empirically impossible to include all the rates
of return and the income of other generations at the same time in the regression equation (variant 8).
On the other hand, the fact that it is difficult to discriminate between altruism and fairness also indicates
that individuals do not clearly separate altruistic feelings from notions of fairness. From the estimated
coefficients we have derived the elasticities of altruism and fairness. The elasticity of altruism, which