Time varying parameter and fixed parameter linear AIDS: An application to tourism demand forecasting

Abstract

This study develops time varying parameter (TVP) linear almost ideal demand system (LAIDS) models in both long-run (LR) static and short-run error correction (EC) forms. The superiority of TVP-LAIDS models over the original static version and the fixed-parameter EC counterparts is examined in an empirical study of modelling and forecasting the demand for tourism in Western European destinations by UK residents. Both the long-run static and the short-run EC-LAIDS models are estimated using the Kalman filter algorithm. The evolution of demand elasticities over time is illustrated using the Kalman filter estimation results. The remarkably improved forecasting performance of the TVP-LAIDS relative to the fixed-parameter LAIDS is illustrated by a one-year-to four-years-ahead forecasting performance assessment. Both the unrestricted TVP-LR-LAIDS and TVP-EC-LAIDS outperform the fixed-parameter counterparts in the overall evaluation of demand level forecasts, and the TVP-EC-LAIDS is also ranked ahead of most other competitors when demand changes are concerned.

Modelling substitution between domestic and outbound tourism in Australia: A system-of-equations approach

“…Thus the STATIC-AIDS specification, without incorporating the dynamic adjustment mechanism in the short run, often fails the tests for theoretical restrictions such as homogeneity and symmetry, and the subsequent long-run elasticity estimates may not be accurate (Edgerton, Assarsson, Hummelmose, Laurila et al, 1996). In tourism studies, Cortés-Jiménez, et al (2009), Durbarry and Sinclair (2003), Li, et al (2004Li, et al ( , 2006 and Wu, Li, and Song (2011) incorporated error correction (EC) mechanisms into AIDS modelling (from here on referred to as EC-AIDS model). Empirical evidence has shown that EC-AIDS models can improve theoretical conformability as well as the forecasting performance (see, for example, Cortés-Jiménez et al, 2009, Li et al, 2004.…”

Tourism Demand Modelling and Forecasting

“…For instance, Li et al (2004) showed EC-LAIDS model's superior forecasting performance over its static counterpart. Li et al (2006a) demonstrated that TVP-LR-AIDS and TVP-EC-LAIDS outperformed their fixed-parameter counterparts in the overall evaluation of demand level forecasts. De Mello and Nell (2005) also examined the forecasting performance of the static AIDS in comparison with three VAR models.…”

“…Specifically, this variable was measured by total tourist arrivals from an origin to a destination, which could be decomposed further into holiday tourist arrivals, business tourist arrivals, tourist arrivals for visiting friends and relatives (VFR) purposes (e.g., Turner and Witt, 2001a, and Kulendran and Wong, 2005, and tourist arrivals by air (Coshall, 2005;Rosselló-Nadal, 2001). Some studies used tourist expenditure in the destination as the demand variable (such as Li et al, 2004Li et al, , 2006aLi et al, and 2006b) and others employed tourist expenditure on particular tourism product categories, such as meal expenditure (Au and Law, 2002), sightseeing expenditure , and shopping . Other tourism demand variables used in the literature include tourism revenues (Akal, 2004), tourism employment and tourism import and export (Smeral, 2004).…”

“…For example, De Mello and Fortuna (2005), Durbarry and Sinclair (2003), Li et al (2004), and Mangion et al (2005) all combined ECM with the linear AIDS (LAIDS) model (i.e., EC-LAIDS). Li et al (2006a) further combined the TVP model along with the long-run LAIDS and EC-LAIDS to establish TVP-LR-AIDS and TVP-EC-LAIDS models, respectively. In addition to the demand elasticity analysis, forecasting performance of various AIDS models was evaluated in five out of eleven studies.…”

A combination selection algorithm on forecasting

“…Three main types of forecasting models (Li, Song & Witt, 2005; are Time series model (Cao, Ewing & Thompson, 2012;Cho, 200;Goshall & Charlesworth, 2011), Causal econometric model (Li, Song & Witt, 2006;Naude & Saayman, 2005;Page, Song & Wu, 2012;Roget & Gonzalez, 2006) and new emerging Artificial Intelligence based model, such as neural network, fuzzy time-series theory, grey theory, genetic algorithms, and expert systems (Cao, Ewing & Thompson, 2012;Carbonneau, Laframboise & Vahidov, 2008;Bodyanskiy & Popov 2006;Chen & Wang, 2007;Cho, 2003;Hadavandi, Ghanbari , Shahanaghi & Abbasian-Naghneh, 2011;Law & Au, 1999;Pai & Hong, 2005;Wong, Xia & Chu, 2010;Wu & Akbarov, 2011). From these studies, researchers often seek to identify the best individual model to generate a forecast.…”