Encompassing and Automatic Model Selection*

Doornik, Jurgen A.

doi:10.1111/j.1468-0084.2008.00536.x

Cited by 50 publications

(40 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For high non-centralities, the default-mode gauge is increased by about 1-2 percentage points (see §5.2). Doornik (2008) shows that encompassing checks against the GUM help stabilize performance. …”

Section: Simulation Results For N = 10mentioning

confidence: 99%

“…Autometrics improves further by a treesearch to detect and eliminate statistically-insignificant variables, and handling N > T . At any stage, a variable is removed only if the new model is a valid reduction of the GUM (i.e., the new model must encompass the GUM at the chosen significance level: see Doornik, 2008). A path terminates when no variable meets the reduction criterion.…”

Section: Comparisons Of 1-cut Selection and Automated Getsmentioning

confidence: 99%

“…When there are more variables than observations, expanding and contracting searches are used, as in Doornik (2009), so specific-to-general search also plays a role. Encompassing tests ensure a parsimonious congruent model of the GUM is selected, see Doornik (2008).…”

Section: Comparisons Of 1-cut Selection and Automated Getsmentioning

confidence: 99%

See 2 more Smart Citations

Evaluating Automatic Model Selection

Castle¹,

Doornik²,

Hendry³

2011

Journal of Time Series Econometrics

View full text Add to dashboard Cite

We outline a range of criteria for evaluating model selection approaches that have been used in the literature. Focusing on three key criteria, we evaluate automatically selecting the relevant variables in an econometric model from a large candidate set. General-tospecific selection is outlined for a regression model in orthogonal variables, where only one decision is required to select, irrespective of the number of regressors. Comparisons with an automated model selection algorithm, Autometrics (Doornik, 2009), show similar properties, but not restricted to orthogonal cases. Monte Carlo experiments examine the roles of post-selection bias corrections and diagnostic testing, and evaluate selection in dynamic models by costs of search versus costs of inference.JEL classifications: C51, C22.

show abstract

Section: Simulation Results For N = 10mentioning

confidence: 99%

Section: Comparisons Of 1-cut Selection and Automated Getsmentioning

confidence: 99%

See 1 more Smart Citation

Evaluating Automatic Model Selection

Castle¹,

Doornik²,

Hendry³

2011

Journal of Time Series Econometrics

View full text Add to dashboard Cite

show abstract

“…The general model often has more candidate variables N than observations T , so block contracting and expanding searches are required, as implemented in Autometrics: see Doornik (2009), andDoornik (2007). Once a feasibly estimable model is obtained with n << T , it is evaluated for congruence by a range of misspecification tests, and if not rejected, reduction continues till no insignificant variables remain, then that terminal model is checked by encompassing: see Doornik (2008).…”

Section: Introductionmentioning

confidence: 99%

Model Discovery and Trygve Haavelmo’s Legacy

Hendry

Johansen

2014

Econom. Theory

View full text Add to dashboard Cite

Trygve Haavelmo's Probability Approach aimed to implement economic theories, but he later recognized their incompleteness. Although he did not explicitly consider model selection, we apply it when theory-relevant variables, {x t }, are retained without selection while selecting other candidate variables, {w t }. Under the null that the {w t } are irrelevant, by orthogonalizing with respect to the {x t }, the estimator distributions of the x t 's parameters are unaffected by selection even for more variables than observations and for endogenous variables. Under the alternative, when the joint model nests the generating process, an improved outcome results from selection. This implements Haavelmo's program relatively costlessly.JEL classifications: C51, C22.

show abstract

“…For model selection, we follow an automatic selection method along the lines of Doornik (2008), Hyndman and Athanasopoulos (2013), and Hendry and Doornik (2014). More precisely, we perform a backwards stepwise regression starting with a model containing all potential predictors and removing one predictor at a time.…”

Section: Forecasting Models' Comparisonmentioning

confidence: 99%

Data-Driven Decision-Making and Its Application to the Corporate Cash Management Problem

Molina¹

View full text Add to dashboard Cite

Encompassing and Automatic Model Selection*

Cited by 50 publications

References 18 publications

Evaluating Automatic Model Selection

Evaluating Automatic Model Selection

Model Discovery and Trygve Haavelmo’s Legacy

Data-Driven Decision-Making and Its Application to the Corporate Cash Management Problem

Contact Info

Product

Resources

About