Formulating wald tests of the restrictions implied by the rational expectations hypothesis
研究理性预期假说中非线性约束的Wald检验在小样本下的表现，发现检验结果对参数化方式非常敏感，乘法形式能给出最准确的检验水平，最小二乘自由度调整也能改善检验表现。
Abstract In this paper, we examine the small sample properties of alternative formulations of Wald tests of non‐linear restrictions implied by the rational expectations hypothesis. A Monte Carlo analysis is presented as well as an example using Canadian aggregate time series data. The evidence indicates that Wald test results are extremely sensitive in small samples to the way in which the non‐linear restrictions in such models are parameterized, with a multiplicative form yielding tests of most accurate size. Least squares degrees of freedom adjustments also improve the sample performance of the tests.