Maximum Likelihood Estimation of a Unit Root Bilinear Model with an Application to Prices

We estimate a unit root bilinear process using the Maximum Likelihood method with log-likelihood function constructed by means of the Kalman filter, and evaluate the finite sample properties of this estimator. One hundred and six world-wide price series are tested for unit root bilinearity applying the test suggested by Charemza et al. (2002b). Applying the Maximum Likelihood estimator based on the Kalman filter, the null hypothesis of no bilinearity is rejected for 40 out of 106 series at the 5% level of significance. Most of the significant unit root bilinear coefficient estimates are explosive