Similarly, to evaluate the performance of the test for threshold constancy, F2 (the constantthreshold against the time-varying threshold model), we set γ0 = 0, β0 = 1, β1 = 1, and β2 = {1.5, 2}. Then the rejection frequencies under the DGP given by (16) with {γ1 = γ2 = 0} and {γ1 = γ2 = 1} are the size and power of the test statistic F2, respectively.Table 2 reports size and power properties of the test statistics F1 and F2. The results show that the empirical size is close to the 5% critical value, implying the test statistics F1 and F2 are correctly sized. Meanwhile, the probability of rejection increases as the sample size gets larger, and the power performance is generally satisfactory especially when the sample size T is not less than 100. Overall, the proposed test statistics work well in finite samples.