The performance of mixed and penalized effects models in predicting the value of the ecological footprint of tourism

Abstract

In recent decades, the issue of ecological footprint (EF) in the world has become a serious anxiety among environmental stakeholders. This anxiety is more in top tourism attracting countries. The purpose of this research is the performance of mixed and penalized effects models in predicting the value of the EF of tourism in the top eight countries of tourism destinations. The World Bank and Global Footprint Network databases have been used in this study. Penalized regression and MCMC models have been used to estimate the EF over the past 19 years (2000-2018). The findings of the research showed that the amount of ecological footprint in China, France and Italy is much higher than other countries. In addition, based on the results, a slight improvement in the performance of penalized models to linear regression was observed. The comparison of the models shows that in the Ridge and Elastic Net models, more indicators were selected than Lasso, but Lasso has a better predictive performance than other models on ecological footprint. Therefore, the use of penalized models is only slightly better than linear regression, but they provide the selection of appropriate indices for model parsimoniousness. The results showed that the penalized models are powerful tools that can provide a significant performance in the accuracy and prediction of the EF variable in tourism attracting countries.