Momentvalg

Momentvalg is a term used in statistics, particularly in Danish-language econometrics, referring to the selection of moment conditions used in estimation by the Generalized Method of Moments (GMM). Moments are expectations of functions of the data and parameters; common examples include the mean and variance, but moment conditions can be more complex.

Because many moment conditions may be valid but redundant or noisy, momentvalg seeks a subset that yields reliable and efficient parameter estimates while avoiding ill-conditioning and overfitting.

Techniques include penalized GMM, where a penalty discourages including many moments; information-criterion-based selection adapted to moment sets; sequential testing using overidentifying restrictions and Hansen’s J-test to prune moments; and cross-validation or out-of-sample predictive checks.

Used in econometrics, finance, and social sciences, momentvalg helps improve estimator efficiency and numerical stability, especially in models with many instruments or moment conditions relative to sample size.

Discarding valid moments can bias estimates; the results can be sensitive to the chosen selection method; theoretical guarantees typically rely on large samples and correct model specification. See also generalized method of moments, moment conditions, and model selection.