This paper shows how the empirical entropy (also known as exponential likelihood or non-parametric tilting) method can be used to test general parametric hypothesis in time series regressions. To capture the weak dependence of the observations, the paper uses blocking techniques which are also...