Automatic selection of indicators in a fully saturated regression

By Carlos Santos, David F. Hendry, SÃ¸ren Johansen. Computational statistics (Zeitschrift), 2008

297 citations recorded by the source corpus.

Subfields: Applied Mathematics, Economics and Econometrics

View Paper: https://www.semanticscholar.org/paper/871495f1befb35c80ff5fd7b699951f1ebcdc78e

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Applied Mathematics

In Economics and Econometrics

Citations in the corpus

Information and Efficiency: Goal Arrival in Soccer Betting (2014, methodology): “Inserting a dummy variable for (almost) every observation and testing their statistical significance would be consistent with an alternative strategy of ‘dummy saturation’ advanced by Hendry et al. (2008).”
An Automatic Test of Super Exogeneity (2010, background): “model should not have to be over-identified under the alternative of a failure of super exogeneity, as needed for tests in the class proposed by (say) Revankar and Hartley (1973). Fifthly, the test must have power against a large class of potential failures of super exogeneity in the conditional mod”
Climate change awareness: Empirical evidence for the European Union (2020, methodology): “Fixed effects can however be accounted for, also when time-invariant regressors are included in the model, by following a general to specific procedure, implemented through an autometrics/saturation algorithm (Hendry et al., 2008; Johansen and Nielsen, 2009; Doornik, 2009).”
Model selection when there are multiple breaks (2012, methodology): “Impulse saturation (see Hendry et al., 2008, and Johansen and Nielsen, 2009) includes an indicator for every observation, entered (in the simplest case) in blocks ofT/2, with the significant outcomes retained.”
MODEL DISCOVERY AND TRYGVE HAAVELMO’S LEGACY (2014, background): “Johansen and Nielsen (2009) show that the splithalf approach in Hendry et al. (2008) can be generalized to dynamic regression models, and despite including T additional candidate indicators, the limiting distributions of the retained parameters of interest have the same characteristics as if IIS…”
Detecting Location Shifts during Model Selection by Step-Indicator Saturation (2015, methodology): “The analytic derivations below use the split-half one-cut approach in [3] for establishing the distribution under the null of no shifts, checked by simulations that also compare split-half sequential selection outcomes and a multi-path search algorithm, extended to non-null alternatives.”