Modeling and forecasting tourism flow in Portugal: Perspectives for a strategic management

• Autor(a) principal: Ramos, F. R.
• Data de Publicação: 2022
• Outros Autores: Lacerda, L., Curto, J.
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10071/27927
• topic: Tourism; Time series; Exponential smoothing models; Forecasting; Business management; Strategic management

Purpose: The increase of Tourism in Portugal, as well as the companies related to it, it is necessary to analyze and forecast the flow of tourists so that the management of the business is endowed with a competitive strategy. Given the changes in the 'recent' dynamics of tourism data, this article discusses the contributions and limitations of using classical forecasting methodologies, when applied to this sector, namely to the number of overnight stays in tourist accommodation establishments in Portugal. Methodology: The study focuses on the modeling and forecasting of time series related to the number of monthly overnight stays, in tourist accommodation establishments in Portugal, between January 2002 and March 2022. As a result of some suggestions contained in the scientific literature, it was resorted the Exponential Smoothing (ETS) methodologies. In computational terms, we used the Jupyter computational environment, with the Python programming language (version 3.7.3). Findings: The results were presented and discussed through the analysis of two time series: (1) Total number of overnight stays in tourist accommodation establishments in Portugal – Total series; (2) Number of overnight stays spent by residents in Portugal in tourist accommodation establishments in Portugal – Residents series. Overall, from the analysis of the time series, there was a growth of Tourism in Portugal since 2002, with a visible drop in 2020, due to the pandemic situation. Regarding the ETS methodologies used in the modeling and forecasting, although they corresponded positively in the forecast of the Total series (with some error), the same did not happen in the Residents series. In this series, due to the recent dynamics that are completely atypical, it appears that the ETS methodologies, potentially more adequate, do not converge, in general. However, it is important to mention that it was the overnight stays of residents that, in the pandemic period, dictated the dynamics present in the Total series. Research limitations: The literature points to a good performance of ETS methodologies in time series with characteristics present in the series under study (with the presence of a trend cycle and clear seasonality), a fact that motivated its choice. However, the difficulty of these methodologies in dealing with abrupt breaks in the data history was evident in this study. Despite how adjusted the forecasts are, the highlight is the non-convergence of some models that could be better adjusted to the historical data. In this sense, it is necessary to search for alternative forecasting methodologies, where Machine Learning methodologies, namely Deep Learning (Deep Neural Networks) have been pointed out in the scientific literature as quite promising. This will be the next step of the investigation. Originality: Given the importance that Tourism has both in the economic and social dimension of Portugal, and being a very volatile and constantly changing sector, it is imperative to define a strategy for future action to understand how, internally, the sector can define policies to avoid situations of external dependence. In addition to a current analysis of the data history, resulting from an atypical period of pandemic, we need to critically evaluate the predictive capacity of (classical) econometric models, which can be used by the industry related to tourism. This not only contributes to a better understanding of the phenomenon under study, but also constitutes a tool to support decision-making.