conclude that the augmented flexibility of the random parameters logit model
does not justify its higher computational costs with these data.
We completed the analysis by using the obtained marginal willingness-to-pay
estimates and investment cost to generate insights into the extent to which free-
rider effects may undermine the social benefits of a financial support program.
We found that for some 50% of the households, the willingness-to-pay exceeds the
observed cost, a share that drops only slightly when allowing for the possibility
that households incur additional hidden costs.
Our findings are of special interest in Europe, given that a recent directive of
the European Union requires that member states introduce political measures to
decrease energy end-use by 9%. To the extent that measures such as Germany’s
grants program suffer from extensive free-riding - and our results suggest that
they do - an immediate issue arises as to whether these political targets should
recognize free-rider effects, and make corresponding adjustments. The analysis
presented in this paper provides the first step in articulating such an adjustment
by quantifying the magnitude of the problem. Having done so, two useful endeav-
ors for future research emerge. The first would involve devising methodological
approaches for quantifying the level of hidden costs associated with renovation
activities, perhaps by drawing on experimental techniques. The second extension
would estimate the determinants of free-riding, with the ultimate aim of identi-
fying options for excluding free-riders from program participation by means such
as market discrimination.
23