Modelling and Forecasting the European Tourist Arrivals in Nepal: A Case of the SARIMA Model Approach

Abstract

Background: Tourism is a major foreign-exchange earner and employment generator for Nepal, and Europe is a stable, high-value long-haul source market. Existing Nepalese forecasting research has concentrated on aggregate arrivals, leaving source-market dynamics largely unexamined. Methods: Monthly European tourist arrivals in Nepal (January 2020–February 2026, 74 observations) were modelled implementing the Seasonal Autoregressive Integrated Moving Average (SARIMA) approach, with January 2020–December 2024 (60 observations) as the training sample and January 2025–February 2026 (14 observations) withheld for out-of-sample testing. Stationarity was assessed via the Augmented Dickey-Fuller and Kwiatkowski-Phillips-Schmidt-Shin tests, and candidate seasonal specifications were compared using information criteria and the Ljung-Box test. Results: After first-order regular and seasonal differencing (d = 1, D = 1), the series was stationary. The multiplicative "airline model," SARIMA(0,1,1)(0,1,1)12, was selected, with significant non-seasonal (θ1 = -0.487, p = 0.014) and seasonal (Θ1 = -0.439, p = 0.004) moving-average terms. Out-of-sample validation yielded a mean absolute error of 6,401 arrivals, an RMSE of 7,466, and a MAPE of 44 percent, reflecting the difficulty of forecasting the unusually sharp October 2025 peak. Forecasts for March 2026–December 2028 indicate continuation of a bimodal seasonal pattern, peaking in spring and autumn and troughing in the summer monsoon. Conclusion: SARIMA adequately captures the trend and seasonal structure of European arrivals to Nepal, with implications for destination marketing, capacity planning and seasonal-workforce scheduling. Novelty: This is the first study to build a source-market-specific SARIMA model for European tourist arrivals in Nepal, rather than aggregate arrivals, spanning the COVID-19 collapse and recovery.