EMPIRICAL ANALYSIS OF TOURISM FLOW FORECASTING IN CENTRAL ASIA BASED REGRESSION MODELS

Abstract

Over the past ten years, tourism in Central Asia has grown significantly, especially in Uzbekistan, Kazakhstan,
Kyrgyzstan, and Tajikistan. However, due to external shocks and macroeconomic volatility, forecasting tourist arrivals
remains challenging. This study forecasts tourist arrivals for the period 2010–2024 and examines the key determinants
of tourism demand using a multiple regression framework. Annual data were collected from regional and international
statistical sources, including indicators such as GDP, inflation, exchange rates, tourist arrivals, and trade openness. All
variables were transformed into logarithmic form to estimate elasticities, and stationarity was tested using the Augmented
Dickey–Fuller (ADF) test. A log-linear regression model was estimated using the Ordinary Least Squares (OLS) method
in R. Model robustness was assessed through diagnostic tests, including Durbin–Watson, Variance Inflation Factor
(VIF), Breusch–Pagan, and normality tests. Forecasting accuracy was evaluated using RMSE and MAPE. The results
indicate that GDP and exchange rates are the most influential determinants of tourism demand, while trade openness
and inflation exhibit statistically significant but relatively smaller effects. The model demonstrates strong short-term
forecasting performance. These findings highlight the importance of effective macroeconomic management in sustaining
tourism growth in Central Asia and provide a solid methodological and policy-relevant foundation for future econometric
research in the region.