Robust Nonnested Testing and the Demand for Money
提出针对动态非嵌套模型的稳健检验方法，能处理未知序列相关和条件异方差，通过蒙特卡洛模拟比较了自助法和固定b渐近法的效果，并应用于美国货币需求函数中消费与收入作为规模变量的选择。
Nonnested hypothesis testing procedures recently have been extended to dynamic nonnested models. We propose robust tests that generalize the J test and the F test for nonnested dynamic models with unknown serial correlation and conditional heteroscedasticity in errors and regressors. We investigate the finite-sample properties of our test statistics and propose to use bootstrap methods or the fixed-b asymptotics developed by Kiefer and Vogelsang to improve the asymptotic approximation to the sampling distribution of the test statistics. We compare the semiparametric bootstrap and the fixed-b approaches with the standard normal or chi-squared approximations using Monte Carlo simulations and find that they give markedly superior approximations. We also present an application to U.S. money demand models, where consumption seems to be a better variable than income as a “scale” variable in the money demand function.