Conference Agenda

Substitution is central to welfare measurement in travel cost recreation demand, yet single-site count models often treat substitutes either implicitly (by omission) or by adding many substitute travel-cost variables—a “wide-net” strategy that can induce multicollinearity and unstable consumer-surplus estimates. Using a national online survey covering recreation across all 23 Polish national parks (n = 5,855), we benchmark negative binomial single-site models against the utility-consistent linked site-choice/count framework of Hausman et al. (1995). We then evaluate two practical substitute controls designed for data-limited settings: (i) restricting the substitute set using correlation-based screening, and (ii) constructing low-dimensional substitute-price indices via principal components. Both approaches mitigate multicollinearity while preserving the welfare impact of substitution, yielding consumer-surplus estimates close to the linked-model benchmark. In an application to Tatra National Park, models that ignore substitutes overstate welfare, while models that include an unrestricted set of substitutes can substantially understate it. The results provide implementable guidance for improving substitute treatment in single-site travel cost studies when rich multi-site visitation data are unavailable.