When do long-run identifying restrictions give reliable results?

Many recent articles have identified behavioral disturbances in vector autoregressions by imposing restrictions on the long-run effects of shocks. This article demonstrates that this approach will be unreliable unless the underlying economy satisfies three types of strong restrictions. Although many aspects of these issues have been raised before. this article draws out and illustrates the implications for inferences under the long-run scheme. Furthermore, it provides strategies for dealing with the problems.