First of all, causality requires identification. Vector autoregressions (VARs) do not provide any automatic or free identification. To do policy analysis with a VAR (as opposed to agnostic forecasting) one has to make the same type of untestable identifying assumptions here as one does in the older, explicitly simultaneous equation, Cowles commission approach.<\/p>\n
The most common way of identifying a VAR (ordering the variables and performing a\u00a0Cholesky decomposition) is EXACTLY the same as using exclusion restrictions to identify a system of equations. Other structural VARS do NOT remove the need for identifying assumptions. VARS are not a free lunch.<\/p>\n
That is Kevin Grier, in a\u00a0post<\/a>\u00a0over at Kids Prefer Cheese.<\/p>\n I was looking forward to someone finally saying it, as many of the media discussions of the 2011 Nobel prize for economics somehow made it sound as though the work of Sargent and Sims allowed us to estimate causal relationships.<\/p>\n Not so. The estimation of causal relationships is difficult in the social sciences for the simple reason that we almost never observe the counterfactual. This is especially true in macroeconomics, where the opportunities to run an experiment are few and far between. And even in more micro contexts, we’re not completely sure about much<\/a>.<\/p>\n","protected":false},"excerpt":{"rendered":" First of all, causality requires identification. Vector autoregressions (VARs) do not provide any automatic or free identification. To do policy analysis with a VAR (as…<\/p>\n