【Abstract】Many economic analyses revolve around the impact of shocks. However, we are not always able to observe the variables causing and reacting to these shocks at a sufficiently high frequency. As a result, if the response of one variable to a shock to another takes place ‘in the nick of time,’ this response remains intractable. We introduce a structural vector-auto regression model with Markov-switching heteroskedasticity in the data generating process that allows us to study instantaneous impulse-response relationships with the proper selection of a supporting ‘catalyst’, which can be easier to find than an instrumental variable. Our method is especially useful for - but is in no way limited to - the study of financial markets, where information usually diffuses very quickly. In our application to the crude oil futures market, we find evidence that the inflow of index traders as contributed significantly to excess returns.
【Keywords】SVAR；Identification；Markov-switching；Commodity prices；Index Trading