Model selection when there are multiple breaks

Castle, Jennifer L.; Doornik, Jurgen A.; Hendry, David F.

doi:10.1016/j.jeconom.2012.01.026

Cited by 124 publications

(81 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The power loss from tighter significance levels is usually not substantial relative to, say, a t-distribution with few degrees of freedom. However, Castle, Doornik and Hendry (2010) show that impulse-indicator saturation (see Hendry, Johansen andSantos, 2008, andNielsen, 2009) is a successful antidote for fat-tailed error processes.…”

Section: Selection Effects and Bias Correctionsmentioning

confidence: 99%

Evaluating Automatic Model Selection

Castle¹,

Doornik²,

Hendry³

2011

Journal of Time Series Econometrics

Self Cite

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: Selection Effects and Bias Correctionsmentioning

confidence: 99%

Evaluating Automatic Model Selection

Castle¹,

Doornik²,

Hendry³

2011

Journal of Time Series Econometrics

Self Cite

View full text Add to dashboard Cite

show abstract

“…When applied to models with fat-tailed error distributions, much of the non-normality can be picked up by the impulse indicators, so that using critical values based on normality for inference is reasonable (Castle, Doornik, and Hendry 2012;Doornik and Hendry 2014, Ch. 15.6).…”

mentioning

confidence: 99%

Does institutional quality resolve the Lucas Paradox?

2017

View full text Add to dashboard Cite

The Lucas Paradox observes that capital flows predominantly to relatively rich countries, contradicting the neoclassical prediction that it should flow to poorer capital-scarce countries. Alfaro, Kalemli-Ozcan, and Volosovych (2008) (AKV) argue that cross-country variation in institutional quality can fully explain the Paradox, contending that if institutional quality is included in regression models explaining international capital inflows, a country's level of economic development is no longer statistically significant. We replicate AKV's results using their cross-sectional IFS capital flow data. Motivated by the importance of conducting inference in statistically adequate models, we focus on misspecification testing of alternative functional forms of their empirical model of capital flows. We show that their resolution of the Paradox relies on inference in a misspecified model. In models that do not fail basic misspecification tests, even though institutional quality is a significant determinant of capital inflows, a country's level of economic development also remains a significant predictor. The same conclusions are reached using an extended dataset covering more recent IFS international capital flow data, first-differenced capital stock data and additional controls.

show abstract

“…Multi-path search algorithms with tight critical values can handle more candidate variables, N , than T . As implemented in automatic model selection algorithms like Autometrics (see Doornik, 2009), IIS enables jointly locating breaks with selection over variables, functional forms and lags: see Castle, Doornik and Hendry (2012). Some well-known procedures are variants of IIS, such as the Chow (1960) test (sub-sample IIS over T − k + 1 to T without selection), and recursive estimation, which is equivalent to IIS over future samples, reducing indicators one at a time.…”

Section: Unpredictability and Model Selectionmentioning

confidence: 99%

Unpredictability in economic analysis, econometric modeling and forecasting

Hendry

Mizon

2014

Journal of Econometrics

Self Cite

View full text Add to dashboard Cite

Unpredictability arises from intrinsic stochastic variation, unexpected instances of outliers, and unanticipated extrinsic shifts of distributions. We analyze their properties, relationships, and different effects on the three arenas in the title, which suggests considering three associated information sets. The implications of unanticipated shifts for forecasting, economic analyses of efficient markets, conditional expectations, and inter-temporal derivations are described. The potential success of general-to-specific model selection in tackling location shifts by impulse-indicator saturation is contrasted with the major difficulties confronting forecasting.JEL classifications: C51, C22.Keywords: Unpredictability; 'Black Swans'; Distributional shifts; Forecast failure; Model selection; Conditional expectations. IntroductionUnpredictability has been formalized as intrinsic stochastic variation in a known distribution, where conditioning on available information does not alter the outcome from the unconditional distribution, as in the well-known prediction decomposition, or sequential factorization, of a density (see Doob, 1953).Such variation can be attributed (inter alia) to chance distribution sampling, 'random errors', incomplete information, or in economics, many small changes in the choices by individual agents. A variable that is intrinsically unpredictable cannot be modeled or forecast better than its unconditional distribution.However, the converse does not hold: a variable that is not intrinsically unpredictable may still be essentially unpredictable because of two additional aspects of unpredictability. The first concerns independent draws from fat-tailed or heavy-tailed distributions, which leads to a notion we call 'instance * This research was supported in part by grants from the Open Society Foundations and the Oxford Martin School. We are indebted to Gunnar Bärdsen, Jennifer L. Castle, Neil R. Ericsson, Søren Johansen, Bent Nielsen, Ragnar Nymoen, Felix Pretis, Norman Swanson and two anonymous referees for helpful comments on earlier versions. Forthcoming, Journal of Econometrics. Contact details: david.hendry@nuffield.ox.ac.uk and grayham.mizon@soton.ac.uk. 1 unpredictability'. Here the distribution of a variable that is not intrinsically unpredictable is known, as are all conditional and unconditional probabilities, but there is a non-negligible probability of a very discrepant outcome. While that probability is known, it is not known on which draw the discrepant outcome will occur, nor its magnitude, leading to a 'Black Swan' (as in Taleb, 2007), with potentially large costs when that occurs (see Barro, 2009). The third aspect we call 'extrinsic unpredictability', which derives from unanticipated shifts of the distribution itself at unanticipated times, of which location shifts (changes in the means of distributions) are usually the most pernicious. Intrinsic and instance unpredictability are close to 'known unknowns' in that the probabilities of various outcomes can be correctly pre-ca...

show abstract

Model selection when there are multiple breaks

Cited by 124 publications

References 14 publications

Evaluating Automatic Model Selection

Evaluating Automatic Model Selection

Does institutional quality resolve the Lucas Paradox?

Unpredictability in economic analysis, econometric modeling and forecasting

Contact Info

Product

Resources

About