RESETTest

RESETTest, short for the Ramsey Regression Equation Specification Error Test, is a general econometric test used to detect misspecification in a regression model. Introduced by James B. Ramsey in 1969, it assesses whether a linear regression with a given set of regressors is correctly specified or whether nonlinear relationships or omitted variables may be present.

The standard procedure begins by estimating the baseline regression model by ordinary least squares. After obtaining the fitted values Ŷ, the test augments the regression by including powers of Ŷ (commonly Ŷ^2, Ŷ^3, and sometimes higher powers). The augmented model is re-estimated, and a joint test is conducted to determine whether the additional terms significantly improve the fit. In compact form, the null hypothesis is that the coefficients on the added powers are zero, implying no misspecification; the alternative is that at least one of those coefficients is nonzero.

Interpreting the result is straightforward but non-specific. A rejection of the null suggests that the original functional form may be incorrect, that nonlinearities exist, or that relevant variables or interactions have been omitted. However, the test does not indicate the exact source of misspecification. It is commonly used as a general diagnostic tool after an OLS fit, and it can be implemented with robust standard errors to mitigate concerns about heteroskedasticity (yielding a robust RESET).

Limitations include sensitivity to the chosen number of added powers and limited power against certain alternatives. Therefore, a rejected RESET typically motivates further specification checks and additional diagnostic tests rather than providing a definitive correction.