Destination choice models for rock climbing in the Northeast Alps : a latent-class approach based on intensity of participation

Practitioners of outdoor sports, such as rock-climbers, are likely to exhibit preference heterogeneity that depends on the keenness' with which such sports are practiced. Such an intuition is born out in at least one study using latent class discrete choice modelling (Provencher et al. 2002). Preference heterogeneity has a reflection on the population's structure of recreational values assigned to rock-climbing destinations, to their attributes and ultimately to land management policies addressing such attributes. In this study such hypothesis is tested on a panel of destination choices by a sample of rock-climbers members of the Veneto Chapter of the Italian Alpine Club. Preliminary estimates of latent-class (finite-mixing) specifications provided evidence that intensity of participation explained heterogeneity in taste. This motivated our splitting of the sample in a high' and a low' intensity of participation sub-samples, each of which is in turn analysed for the presence of endogenous preference classes using latent-class random utility based approaches. We find evidence in support of the hypothesis that there are at least four statistically well-defined classes in each sub-sample, thereby revealing a considerable richness in the structure of preference, which would otherwise be unobservable in more conventional approaches. From the model estimates, we first focus on the derivation of posterior individual specific welfare measures for some key destination attributes, and then for a welfare neutral land management policy. One emerging feature is the strong evidence of multi-modal distribution of values, a feature that is more difficult to capture when preference heterogeneity is modelled by other means. The results also show how the proposed policy is progressive in terms of benefit distribution in the sample, and that the distribution of individual welfare changes shows markedly different patterns between high and low demand sub-samples.