Application of copula-based Markov Switching Unrelated Regression to tourism demand modeling
本研究用两状态马尔可夫转换无关联回归模型分析旅游需求，考虑非线性、结构变化和不同客源市场间的相关性，发现R-Vine Copula能更好拟合数据。
This study examines the tourism demand using a two-regime Markov Switching Seemingly Unrelated Regression (MS-SUR) model. This approach provides relevant methodological contributions that account for three characteristics that are common in tourism data: nonlinear and structurally dynamic nature of tourism behavior, correlation between different origin markets and non-normal distribution. Based on actual monthly tourism data, the model identifies two distinct demand phases and accounts for changes in both the expected mean and variance over time. The analysis also tests the suitability of various copula models, namely Elliptical, Archimedean, and Vine structures—to describe the relationship between different origin markets. Model selection criteria such as AIC and BIC suggest that the Regular Vine (R-Vine) copula provides a better fit than alternative approaches. The results highlight the presence of changing patterns in tourism demand and show that flexible dependency structures can improve the modeling of such changes.