Exploring scale effects of best/worst rank ordered choice data to estimate benefits of tourism in Alpine grazing commons

In many environmental valuation applications standard sample sizes for choice modelling surveys are impractical to achieve. This is the case in our research context, where we interviewed mountain visitors about the external effects of maintaining Alpine grazing commons. In these situations coupling rank-ordered data with efficient experimental design seems to be an effective way to obtain high quality information. In our case we find that the number of design replicates needed to obtain the same standard error for the rank-ordered model is only about one-third that necessary for the favorite choice multinomial logit model. One can improve data quality using more in-depth surveys administered to fewer respondents. The resulting “exploded logit” choice model, estimated on 64 responses per person, was used to study the willingness to pay for external benefits by visitors for policies which maintain the cultural heritage of alpine grazing commons. In terms of the estimated distributions of marginal WTPs at the respondent level, we found that they respond in a plausible fashion to variation of covariates, which suggests that the best-worst approach produces valid WTP estimate for policy analysis. Local politicians may therefore be advised that a visitors’ access fee to support the upkeep of Alpine grazing commons is a viable proposition, but it might not be sufficient by itself to collect the necessary revenue. Finally, our best model suggests that a sharp trade-off exists between the access fee and number of visitors, and hence revenue, probably due to the large number of perceived substitute destinations accessible for free in the Alps.