The sequential estimation of subset VAR with forgetting factor and intercept variable

In this paper we propose a forward time update algorithm to recursively estimate subset vector autoregressive models (including an intercept term) with a forgetting factor, using the exact window case. The proposed recursions cover, for the first time, subset vector autoregressive models (VAR) with a forgetting factor and an intercept variable. We then present two applications. In the first application we apply the proposed estimation algorithm to the quarterly aluminium prices on the London Metal Exchange. The findings show that the proposed algorithm can improve the forecasting performance. In the second application a bivariate system investigates the relationship between the Australian's All Ordinaries Share Price Index (SPI) futures and BHP share price (BHP). The proposed algorithm also introduces the Monte Carlo Integration approach into the proposed algorithm to generate error bands for the impulse responses. These results confirm that the SPI Granger causes BHP, but not vice versa.