I have difficulty understanding the Structural Vector Autoregression (SVAR). I have some books about it, and have read them, but still cannot grasp the idea behind that concept. Can someone explain me the distinctiveness of SVAR compared to a VAR in simple words? In which situations is a SVAR more appropriate than a simple VAR?
According to Lütkepohl's Applied Time Series Econometrics, the difference between simple VAR and structural VAR is that instead of identifying the coefficients, identification focuses on the errors of the system, which are interpreted as (linear combinations of) exogenous shocks.
Restrictions based on the theory is imposed on the relations between the variables and the rest of variable dynamics are considered exogenous shocks. In a simple VAR no restrictions are imposed in advance and the coefficients of the lagged values of the variables included are identified.
So in one sentence, in a SVAR restrictions are imposed on the variable dynamics beforehand and the rest is considered exogenous shocks, while in the VAR the coeffients of the lags are identified.
I strongly recommend this book: H. Lutkepohl & M. Kratzig. Applied Time Series Econometrics. Cambridge University Press, The Edinburgh Building, Cambridge CB2 2RU, UK, 2004.