Allowing the Data to Speak Freely: The Macroeconometrics of the Cointegrated Vector Autoregression
12 Pages Posted: 21 Nov 2007
Date Written: November 16, 2007
An explication of the key ideas behind the Cointegrated Vector Autoregression Approach. The CVAR approach is related to Haavelmo's famous "Probability Approach in Econometrics" (1944). It insists on careful stochastic specification as a necessary groundwork for econometric inference and the testing of economic theories. In time-series data, the probability approach requires careful specification of the integration and cointegration properties of variables in systems of equations. The relationship between the CVAR approach and wider methodological issues and between it and related approaches (e.g., the LSE approach) are explored. The specific-to-general strategy of widening the scope of econometric models to identify stochastic trends and cointegrating relations and to nest theoretical economic models is illustrated with the example of purchasing-power parity.
Keywords: cointegrated vector autoregression, VAR, CVAR, cointegration, Johansen, Juselius, Haavelmo, time-series models, Walrasian methodology, Marshallian methodology
JEL Classification: B41, C30, C32, C50, C51, C52
Suggested Citation: Suggested Citation