This thesis addresses the issue of estimating persistence of economic shocks using time series models. First, it is shown that the log likelihood function for ARIMA models is not strictly quadratic with respect to the persistence estimate. This result explains why the persistence literature has...