Precise tourist spending estimates under budget constraints: A Bayesian solution

Abstract

Reliable estimates of visitor spending are essential for Tourism Satellite Accounts (TSA) and tourism policy, yet International Visitor Surveys (IVS) are costly and often infrequent in small and developing economies. This study proposes a Bayesian power-prior approach as a resource-efficient alternative to frequentist estimation of mean visitor expenditure. Using Monte Carlo simulations based on a log-Normal data-generating process, historical survey data are incorporated through a power parameter ( $a_0$ ) that regulates information borrowing, combined with a commensurability diagnostic to address prior–data conflict. Simulation results show that, when historical and current data are aligned, Bayesian estimators achieve accuracy comparable to frequentist methods while reducing interval widths by up to 80% in small samples. An empirical application further demonstrates how Bayesian updating can support cost-effective survey designs while maintaining statistical reliability under budget constraints.