Wild bootstrap tests for autocorrelation in vector autoregressive models (Q1685299)

The authors consider the problem of testing for the autocorrelation error in VAR models and in vector error correction models. All results are obtained under the assumptions that the errors represent a martingale difference sequence with at least 8th moments, with satisfied convergence of conditional moments and restrictions for the fourth-order cumulants. All tests are based on the Lagrange multiplier statistic. The authors propose wild bootstrap Lagrange multiplier tests for the autocorrelation error and show that these tests are asymptotically valid. Some applications of the proposed tests are discussed on credit default swap prices and Euribor interest rates.