BOOTSTRAPPING IN VECTOR AUTOREGRESSIONS: AN APPLICATION TO THE PORK SECTOR

Standard bootstrap method is used to generate confidence intervals (CIs) of impulse response functions of VAR and SVAR models in the pork sector. In the VAR model, the bootstrap method does not produce significant different results from Monte Carlo simulations. In the SVAR analysis, on the other hand, the bootstrap CIs are significantly different from Monte Carlo CIs after a six period forecast intervals. This suggests that the choice of method used to measure reliability of IRFs is not trivial. Furthermore, bootstrap CIs in SVAR model seem to be more stable than MC CIs, which tend to be wider in the longer horizons.