Forecasting air passenger numbers with a GVAR model
该研究使用全球向量自回归模型，结合经济驱动因素，预测全球最繁忙20个机场及亚太、拉美加勒比地区的航空旅客数量，发现该模型在部分机场的预测准确性优于其他基准模型。
This study employs a GVAR model on the passenger numbers of the top 20 busiest airports of the world and the Asia-Pacific and Latin America-Caribbean regions. With air passenger numbers representing a demand measure, country-level proxies for economic drivers are included as domestic and foreign variables. In terms of ex-ante forecast accuracy, the GVAR model performs best for several airports – yet not for the entirety of airports – compared to four benchmarks for horizons one and three quarters ahead. It also achieves several second and third ranks for these and two other horizons and when all horizons are evaluated jointly. Considering the connectivity of airports is worthwhile to achieve accurate and economically interpretable air passenger demand forecasts.