是什麼決定了旅客的錢包深度？ 影響來臺旅客每日消費額因素--以48年歷史數據為例

Abstract

摘要 我們這次的報告是要來探討1977年至2023年間決定國際旅客來臺每日平均消費額(y)的核心因素，主張分析消費產值比盲目追求人次更具實質意義。研究整合長達48年的數據，納入停留夜數(X1)、觀光目的占比(X3)、平均匯率(X4)、經濟成長率(X5)、CPI物價指數(X6)、全球GDP(X7)及來臺人次(X2)等變數進行綜合分析。在簡單迴歸分析的篩選後，我們發現代表市場規模的「來臺人次」因殘差呈現明顯U型，再做了取LOG和次方項後仍是如此，所以將其剔除以避免掩蓋其他經濟變數。同時，透過Cook's Distance與RStudent診斷圖識別並移除2021與2022年受疫情干擾的極端離群值，使模型解釋力從69%顯著提升至73.6%。為了建構最穩健 的預測模型，我們再運用前進選擇、向後消去、逐步選擇及全子集法篩選變數，並針對高消費客群波動引起的不等變異性進行統計修正，成功使原本不顯著的「全球GDP」轉為具統計意義的關鍵因子。總結來說當台幣貶值或國內物價上漲時，旅客的人均消費額會受到明顯壓抑；相對地，全球經濟環境好轉則能有效提升旅客的消費意願。據此，我們建議觀光機關應將政策核心從追求「量」轉向追求「質」，透過監控國際經濟脈動，精準吸引具備高含金量的旅客。 Abstract This study aims to investigate the key determinants of international tourists’ average daily expenditure per capita in Taiwan (y) over the period from 1977 to 2023. It argues that analyzing expenditure value provides more substantive insight than merely pursuing higher visitor numbers. Using a comprehensive dataset spanning 48 years, the analysis incorporates variables including length of stay (X1), proportion of tourism purposes (X3), average exchange rate (X4), economic growth rate (X5), Consumer Price Index (CPI) (X6), global GDP (X7), and number of inbound visitors to Taiwan (X2). Following initial screening through simple regression analysis, the variable representing market size—namely, the number of inbound visitors—was found to exhibit a pronounced U-shaped residual pattern. Even after logarithmic transformation and inclusion of polynomial terms, this issue persisted. Consequently, the variable was excluded to prevent it from obscuring the effects of other economic variables. In addition, extreme outliers corresponding to the years 2021 and 2022, which were heavily affected by the COVID-19 pandemic, were identified and removed using Cook’s Distance and RStudent diagnostic plots. This adjustment improved the explanatory power of the model from 69% to 73.6%. To construct the most robust predictive model, variable selection was further conducted using forward selection, backward elimination, stepwise regression, and all-subsets regression. Moreover, statistical corrections were applied to address heteroscedasticity arising from fluctuations in high-spending tourist segments. As a result, global GDP, which was initially statistically insignificant, emerged as a significant explanatory variable. In conclusion, depreciation of the New Taiwan Dollar or increases in domestic prices tend to suppress tourists’ per capita expenditure, whereas improvements in the global economic environment significantly enhance tourists’ willingness to spend. Accordingly, this study suggests that tourism authorities should shift their policy focus from pursuing “quantity” to enhancing “quality,” by closely monitoring global economic trends and strategically attracting high-value visitors.