Complete subset regressions

Elliott, Graham; Gargano, Antonio; Timmermann, Allan

doi:10.1016/j.jeconom.2013.04.017

Cited by 198 publications

(140 citation statements)

References 58 publications

Supporting

Mentioning

135

Contrasting

Order By: Relevance

“…Our results support the findings of Rapach, Strauss, and Zhou (2010) and Elliott, Gargano, and Timmermann (2013) that forecast combinations consistently achieve significant gains on out-of-sample predictions.…”

supporting

confidence: 87%

Model Averaging in Predictive Regressions

Liu

Kuo

2014

SSRN Journal

View full text Add to dashboard Cite

This paper considers forecast combination in a predictive regression. We construct the point forecast by combining predictions from all possible linear regression models given a set of potentially relevant predictors. We propose a frequentist model averaging criterion, an asymptotically unbiased estimator of the mean squared forecast error (MSFE), to select forecast weights. In contrast to the existing literature, we derive the MSFE in a local asymptotic framework without the i.i.d. normal assumption. This result allows us to decompose the MSFE into the bias and variance components and also to account for the correlations between candidate models. Monte Carlo simulations show that our averaging estimator has much lower MSFE than alternative methods such as weighted AIC, weighted BIC, Mallows model averaging, and jackknife model averaging. We apply the proposed method to stock return predictions.JEL Classification: C52, C53

show abstract

supporting

confidence: 87%

Model Averaging in Predictive Regressions

Liu

Kuo

2014

SSRN Journal

View full text Add to dashboard Cite

show abstract

“…Elliott et al (2013) show that in such settings it can be favorable to combine over forecasts generated by models that include only a small, k, fixed set of predictors such as five or ten variables. In a setting with K possible predictors, there are…”

mentioning

confidence: 99%

“…In empirical work for stock returns (Elliott et al 2013) and inflation rate, GDP growth, and unemployment rate forecasts (Elliott et al 2015), Elliott et al find that the CSR forecasts perform very well compared to univariate ARIMA (autoregressive integrated moving average) or common factor models. Equal-weighted combinations appear to work as well as alternative combination schemes that estimate the weights of the individual forecasting models and do not sample the models at random if the total number of models is too large to allow all possible models with k predictors to be included in the combination.…”

mentioning

confidence: 99%

Forecasting in Economics and Finance

Elliott

Timmermann

2016

Annu. Rev. Econ.

Self Cite

View full text Add to dashboard Cite

Practices used to address economic forecasting problems have undergone substantial changes over recent years. We review how such changes have influenced the ways in which a range of forecasting questions are being addressed. We also discuss the promises and challenges arising from access to big data. Finally, we review empirical evidence and experience accumulated from the use of forecasting methods to a range of economic and financial variables.

show abstract

“…But when both methods are applied on growth theory the authors do not find evidence that WALS outperforms BMA. Regarding the analysis of U.S. stock return, Elliott et al (2013) access that Complete Subset Regressions (CSR) perform better than non-equal-weighted methods like BMA. The authors use CSR which is a shrinkage technique for linear regression to run predictive regression for all models with the same number of predictors.…”

Section: Model Averagingmentioning

confidence: 99%

“…Our methodological approach to analyze the CDS-determinants is inspired by the growing literature focusing on the model space (or a subset of it) rather than single models (see, e.g., Sala-I- Martin et al, 2004;Hansen, 2007;Hansen et al, 2011;Elliott et al, 2013). The growing popularity of these approaches can mainly be attributed to the availability of computational capacities that allow analyses of large subsets of models and the recognition that in many empirical settings there exists no single model that dominates competing setups in a statistically significant way.…”

Section: Introductionmentioning

confidence: 99%

The Determinants of CDS Spreads: Evidence from the Model Space

Pelster

Vilsmeier

2016

SSRN Journal

View full text Add to dashboard Cite

The CDS market plays an important role in understanding the dynamics of financial markets and financial stability and consequently has received growing attention in the recent literature. Despite the rich literature there is still no consensus about which factors mainly drive CDS spreads. Instead we observe an increasingly large number of potential determinants being suggested. We aim to provide a transparent and robust analysis for the determinants of CDS spreads by focusing on the information provided by a large number of models (the "model space") instead of focusing on one or few "hand-picked" specifications. Our model space covers most of the specifications used elsewhere in the literature and includes several variables measuring the sensitivity of firms to extreme market movements (tail dependence). ContributionWe show that the model space includes models which allow backing many different theories by significant coefficients and good fit to the data. Thus, the information content from single model specifications is rather restricted. In contrast to previous studies, we use information from the entire model space instead of restricting ourselves to some "hand-picked" models. Doing so, we provide a transparent and robust analysis for the determinants of CDS spreads. Besides covering the informational content of most previous studies in our analysis we also give clear evidence on the impact of using different estimation procedures to measure tail dependence in CDS markets. ResultsUsing a large data-set, we find that CDS price dynamics can be mainly explained by factors describing firms' sensitivity to extreme market movements. More precisely, our results suggest that variables measuring tail dependence -based on so-called dynamic copula models -incorporate almost all essential pricing information making other potential determinants such as default risk (Merton-type) factors or variables capturing the systematic market evolutions negligible. The determinants of CDS spreads: evidence from the model space * Matthias Pelster Leuphana University of Lueneburg Johannes Vilsmeier Deutsche Bundesbank AbstractWe apply Bayesian Model Averaging and a frequentistic model space analysis to assess the pricing-determinants of credit default swaps (CDS). Our study focuses on the complete model space of plausible models covering most of the variables and specifications used elsewhere in the literature, including different copula models. The approach followed supports ultimate transparency and robustness for the empirical study at hand. Using a large data-set of CDS contracts we find that CDS price dynamics can be mainly explained by factors describing firms' sensitivity to extreme market movements. More precisely, our results suggest that dynamic copula based measures of tail dependence incorporate almost all essential pricing information making other potential determinants such as Merton-type factors or variables measuring the systematic market evolution -based on simple means or principal component analysis -negligible.

show abstract

Complete subset regressions

Cited by 198 publications

References 58 publications

Model Averaging in Predictive Regressions

Model Averaging in Predictive Regressions

Forecasting in Economics and Finance

The Determinants of CDS Spreads: Evidence from the Model Space

Contact Info

Product

Resources

About