Application of vector autoregressive models to Hong Kong air pollution data

Abstract

In this thesis, the levels of the major air pollutants in Hong Kong are modelled by various vector autoregressive (VAR) models, including the general VAR model, structural VAR model and the Space-Time autoregressive (STAR) model. Regular and seasonal unit root tests are performed to test whether the time series being studied are stationary. The results show that all the time series of the major air pollutants in Hong Kong are stationary or trend stationary and the levels of nitrogen dioxide (NO2) and ozone (O3) increase continuously. The structural VAR model gives more accurate forecasts than other Box-Jenkins models based on the mean absolute percentage error (MAPE). The relationship among the pollutants studied by the impulse response function and the variance decomposition of the VAR model is consistent with the chemical and environmental theories. And finally, the model is further improved by using the STAR model, which considers the levels of air pollutants in other stations as well. The forecasting accuracies of all the final models are quite satisfactory.

Date of Award	4 Oct 2004
Original language	English
Awarding Institution	City University of Hong Kong
Supervisor	Miu Han Iris YEUNG (Supervisor)