Cross-section bootstrap for CCE regressions

Ignace De Vos; Ovidijus Stauskas

doi:10.1016/j.jeconom.2023.105648

Cross-section bootstrap for CCE regressions

Ignace De Vos^*, Ovidijus Stauskas

^*Corresponding author for this work

Econometrics and Data Science

Research output: Contribution to Journal › Article › Academic › peer-review

Abstract

The Common Correlated Effects (CCE) methodology is now well established for the analysis of factor-augmented panel data models. Yet, it is often neglected that the pooled variant is biased unless the cross-section dimension (N) of the dataset dominates the time series length (T). This is problematic for inference with typical macroeconomic datasets, where T is often equal or larger than N. In response, we establish in this paper the theoretical foundation of the cross-section (CS) bootstrap for inference with CCE estimators in large N and T panels with TN⁻¹→τ<∞. This resampling scheme is often used to estimate standard errors, yet without theoretical justification, and with unused potential, as we show it also provides a solution to the bias problem. We derive conditions under which the scheme replicates the distribution of the CCE estimators, such that bias can be eliminated and asymptotically valid inference can ensue. In so doing, we also spend attention to the case where factors need not be common across the dependent and explanatory variables, or when slopes are heterogeneous. Since we find that the CS-bootstrap applies in each case, researchers can stay agnostic on these issues. Simulation experiments show that the asymptotic properties also translate well to finite samples.

Original language	English
Article number	105648
Number of pages	20
Journal	Journal of Econometrics
Volume	240
Issue number	1
DOIs	https://doi.org/10.1016/j.jeconom.2023.105648
Publication status	Published - Mar 2024

Bibliographical note

Funding Information:
The authors thank the Editor (Serena Ng), an Associate Editor, and 3 anonymous referees for their constructive comments and recommendations that substantially improved this paper. We also thank Joakim Westerlund, Giuseppe Cavaliere, Luca Fanelli, Francesco Ravazzolo, Vasilis Sarafidis, Simon Reese, Artūras Juodis, Yiannis Karavias, Steffen Grønneberg, Hande Karabiyik, Jan Ditzen, Peter Jochumzen and David Edgerton for valuable comments. This paper also benefited from seminar presentations at Lund University, Bank of Lithuania, Vrije Universiteit Amsterdam, University of Bologna, Free University of Bozen-Bolzano, and the University of Birmingham. Ignace De Vos is thankful to the Lund University Department of Economics, where we initiated this research. Ovidijus Stauskas is thankful to the Jan Wallander and Tom Hedelius Foundation for financial research support, and to Michael Jansson for organizing a research visit at the University of California, Berkeley, where initial work on this study was conducted. Finally, we thank SURF (www.surf.nl) for support with the Lisa and Snellius Compute Clusters that were used for the simulations in this paper.

Funding Information:
The authors thank the Editor (Serena Ng), an Associate Editor, and 3 anonymous referees for their constructive comments and recommendations that substantially improved this paper. We also thank Joakim Westerlund, Giuseppe Cavaliere, Luca Fanelli, Francesco Ravazzolo, Vasilis Sarafidis, Simon Reese, Artūras Juodis, Yiannis Karavias, Steffen Grønneberg, Hande Karabiyik, Jan Ditzen, Peter Jochumzen and David Edgerton for valuable comments. This paper also benefited from seminar presentations at Lund University, Bank of Lithuania, Vrije Universiteit Amsterdam, University of Bologna, Free University of Bozen-Bolzano, and the University of Birmingham. Ignace De Vos is thankful to the Lund University Department of Economics, where we initiated this research. Ovidijus Stauskas is thankful to the Jan Wallander and Tom Hedelius Foundation for financial research support, and to Michael Jansson for organizing a research visit at the University of California, Berkeley , where initial work on this study was conducted. Finally, we thank SURF ( www.surf.nl ) for support with the Lisa and Snellius Compute Clusters that were used for the simulations in this paper.

Publisher Copyright:
© 2023 The Author(s)

Funding

The authors thank the Editor (Serena Ng), an Associate Editor, and 3 anonymous referees for their constructive comments and recommendations that substantially improved this paper. We also thank Joakim Westerlund, Giuseppe Cavaliere, Luca Fanelli, Francesco Ravazzolo, Vasilis Sarafidis, Simon Reese, Artūras Juodis, Yiannis Karavias, Steffen Grønneberg, Hande Karabiyik, Jan Ditzen, Peter Jochumzen and David Edgerton for valuable comments. This paper also benefited from seminar presentations at Lund University, Bank of Lithuania, Vrije Universiteit Amsterdam, University of Bologna, Free University of Bozen-Bolzano, and the University of Birmingham. Ignace De Vos is thankful to the Lund University Department of Economics, where we initiated this research. Ovidijus Stauskas is thankful to the Jan Wallander and Tom Hedelius Foundation for financial research support, and to Michael Jansson for organizing a research visit at the University of California, Berkeley, where initial work on this study was conducted. Finally, we thank SURF (www.surf.nl) for support with the Lisa and Snellius Compute Clusters that were used for the simulations in this paper. The authors thank the Editor (Serena Ng), an Associate Editor, and 3 anonymous referees for their constructive comments and recommendations that substantially improved this paper. We also thank Joakim Westerlund, Giuseppe Cavaliere, Luca Fanelli, Francesco Ravazzolo, Vasilis Sarafidis, Simon Reese, Artūras Juodis, Yiannis Karavias, Steffen Grønneberg, Hande Karabiyik, Jan Ditzen, Peter Jochumzen and David Edgerton for valuable comments. This paper also benefited from seminar presentations at Lund University, Bank of Lithuania, Vrije Universiteit Amsterdam, University of Bologna, Free University of Bozen-Bolzano, and the University of Birmingham. Ignace De Vos is thankful to the Lund University Department of Economics, where we initiated this research. Ovidijus Stauskas is thankful to the Jan Wallander and Tom Hedelius Foundation for financial research support, and to Michael Jansson for organizing a research visit at the University of California, Berkeley , where initial work on this study was conducted. Finally, we thank SURF ( www.surf.nl ) for support with the Lisa and Snellius Compute Clusters that were used for the simulations in this paper.

Funders	Funder number
SURF
Jan Wallanders och Tom Hedelius Stiftelse samt Tore Browaldhs Stiftelse
University of Birmingham
Lunds Universitet
Università di Bologna
Libera Università di Bolzano

Keywords

Bias-correction
Bootstrap
Common correlated effects
Factor-augmented regression

Access to Document

10.1016/j.jeconom.2023.105648

Persistent URL (handle)

Persistent link

Cite this

@article{e20ccca782ae472bab213b68088e3980,

title = "Cross-section bootstrap for CCE regressions",

abstract = "The Common Correlated Effects (CCE) methodology is now well established for the analysis of factor-augmented panel data models. Yet, it is often neglected that the pooled variant is biased unless the cross-section dimension (N) of the dataset dominates the time series length (T). This is problematic for inference with typical macroeconomic datasets, where T is often equal or larger than N. In response, we establish in this paper the theoretical foundation of the cross-section (CS) bootstrap for inference with CCE estimators in large N and T panels with TN−1→τ<∞. This resampling scheme is often used to estimate standard errors, yet without theoretical justification, and with unused potential, as we show it also provides a solution to the bias problem. We derive conditions under which the scheme replicates the distribution of the CCE estimators, such that bias can be eliminated and asymptotically valid inference can ensue. In so doing, we also spend attention to the case where factors need not be common across the dependent and explanatory variables, or when slopes are heterogeneous. Since we find that the CS-bootstrap applies in each case, researchers can stay agnostic on these issues. Simulation experiments show that the asymptotic properties also translate well to finite samples.",

keywords = "Bias-correction, Bootstrap, Common correlated effects, Factor-augmented regression",

author = "{De Vos}, Ignace and Ovidijus Stauskas",

note = "Funding Information: The authors thank the Editor (Serena Ng), an Associate Editor, and 3 anonymous referees for their constructive comments and recommendations that substantially improved this paper. We also thank Joakim Westerlund, Giuseppe Cavaliere, Luca Fanelli, Francesco Ravazzolo, Vasilis Sarafidis, Simon Reese, Artūras Juodis, Yiannis Karavias, Steffen Gr{\o}nneberg, Hande Karabiyik, Jan Ditzen, Peter Jochumzen and David Edgerton for valuable comments. This paper also benefited from seminar presentations at Lund University, Bank of Lithuania, Vrije Universiteit Amsterdam, University of Bologna, Free University of Bozen-Bolzano, and the University of Birmingham. Ignace De Vos is thankful to the Lund University Department of Economics, where we initiated this research. Ovidijus Stauskas is thankful to the Jan Wallander and Tom Hedelius Foundation for financial research support, and to Michael Jansson for organizing a research visit at the University of California, Berkeley, where initial work on this study was conducted. Finally, we thank SURF (www.surf.nl) for support with the Lisa and Snellius Compute Clusters that were used for the simulations in this paper. Funding Information: The authors thank the Editor (Serena Ng), an Associate Editor, and 3 anonymous referees for their constructive comments and recommendations that substantially improved this paper. We also thank Joakim Westerlund, Giuseppe Cavaliere, Luca Fanelli, Francesco Ravazzolo, Vasilis Sarafidis, Simon Reese, Artūras Juodis, Yiannis Karavias, Steffen Gr{\o}nneberg, Hande Karabiyik, Jan Ditzen, Peter Jochumzen and David Edgerton for valuable comments. This paper also benefited from seminar presentations at Lund University, Bank of Lithuania, Vrije Universiteit Amsterdam, University of Bologna, Free University of Bozen-Bolzano, and the University of Birmingham. Ignace De Vos is thankful to the Lund University Department of Economics, where we initiated this research. Ovidijus Stauskas is thankful to the Jan Wallander and Tom Hedelius Foundation for financial research support, and to Michael Jansson for organizing a research visit at the University of California, Berkeley , where initial work on this study was conducted. Finally, we thank SURF ( www.surf.nl ) for support with the Lisa and Snellius Compute Clusters that were used for the simulations in this paper. Publisher Copyright: {\textcopyright} 2023 The Author(s)",

year = "2024",

month = mar,

doi = "10.1016/j.jeconom.2023.105648",

language = "English",

volume = "240",

journal = "Journal of Econometrics",

issn = "0304-4076",

publisher = "Elsevier BV",

number = "1",

}

TY - JOUR

T1 - Cross-section bootstrap for CCE regressions

AU - De Vos, Ignace

AU - Stauskas, Ovidijus

N1 - Funding Information: The authors thank the Editor (Serena Ng), an Associate Editor, and 3 anonymous referees for their constructive comments and recommendations that substantially improved this paper. We also thank Joakim Westerlund, Giuseppe Cavaliere, Luca Fanelli, Francesco Ravazzolo, Vasilis Sarafidis, Simon Reese, Artūras Juodis, Yiannis Karavias, Steffen Grønneberg, Hande Karabiyik, Jan Ditzen, Peter Jochumzen and David Edgerton for valuable comments. This paper also benefited from seminar presentations at Lund University, Bank of Lithuania, Vrije Universiteit Amsterdam, University of Bologna, Free University of Bozen-Bolzano, and the University of Birmingham. Ignace De Vos is thankful to the Lund University Department of Economics, where we initiated this research. Ovidijus Stauskas is thankful to the Jan Wallander and Tom Hedelius Foundation for financial research support, and to Michael Jansson for organizing a research visit at the University of California, Berkeley, where initial work on this study was conducted. Finally, we thank SURF (www.surf.nl) for support with the Lisa and Snellius Compute Clusters that were used for the simulations in this paper. Funding Information: The authors thank the Editor (Serena Ng), an Associate Editor, and 3 anonymous referees for their constructive comments and recommendations that substantially improved this paper. We also thank Joakim Westerlund, Giuseppe Cavaliere, Luca Fanelli, Francesco Ravazzolo, Vasilis Sarafidis, Simon Reese, Artūras Juodis, Yiannis Karavias, Steffen Grønneberg, Hande Karabiyik, Jan Ditzen, Peter Jochumzen and David Edgerton for valuable comments. This paper also benefited from seminar presentations at Lund University, Bank of Lithuania, Vrije Universiteit Amsterdam, University of Bologna, Free University of Bozen-Bolzano, and the University of Birmingham. Ignace De Vos is thankful to the Lund University Department of Economics, where we initiated this research. Ovidijus Stauskas is thankful to the Jan Wallander and Tom Hedelius Foundation for financial research support, and to Michael Jansson for organizing a research visit at the University of California, Berkeley , where initial work on this study was conducted. Finally, we thank SURF ( www.surf.nl ) for support with the Lisa and Snellius Compute Clusters that were used for the simulations in this paper. Publisher Copyright: © 2023 The Author(s)

PY - 2024/3

Y1 - 2024/3

N2 - The Common Correlated Effects (CCE) methodology is now well established for the analysis of factor-augmented panel data models. Yet, it is often neglected that the pooled variant is biased unless the cross-section dimension (N) of the dataset dominates the time series length (T). This is problematic for inference with typical macroeconomic datasets, where T is often equal or larger than N. In response, we establish in this paper the theoretical foundation of the cross-section (CS) bootstrap for inference with CCE estimators in large N and T panels with TN−1→τ<∞. This resampling scheme is often used to estimate standard errors, yet without theoretical justification, and with unused potential, as we show it also provides a solution to the bias problem. We derive conditions under which the scheme replicates the distribution of the CCE estimators, such that bias can be eliminated and asymptotically valid inference can ensue. In so doing, we also spend attention to the case where factors need not be common across the dependent and explanatory variables, or when slopes are heterogeneous. Since we find that the CS-bootstrap applies in each case, researchers can stay agnostic on these issues. Simulation experiments show that the asymptotic properties also translate well to finite samples.

AB - The Common Correlated Effects (CCE) methodology is now well established for the analysis of factor-augmented panel data models. Yet, it is often neglected that the pooled variant is biased unless the cross-section dimension (N) of the dataset dominates the time series length (T). This is problematic for inference with typical macroeconomic datasets, where T is often equal or larger than N. In response, we establish in this paper the theoretical foundation of the cross-section (CS) bootstrap for inference with CCE estimators in large N and T panels with TN−1→τ<∞. This resampling scheme is often used to estimate standard errors, yet without theoretical justification, and with unused potential, as we show it also provides a solution to the bias problem. We derive conditions under which the scheme replicates the distribution of the CCE estimators, such that bias can be eliminated and asymptotically valid inference can ensue. In so doing, we also spend attention to the case where factors need not be common across the dependent and explanatory variables, or when slopes are heterogeneous. Since we find that the CS-bootstrap applies in each case, researchers can stay agnostic on these issues. Simulation experiments show that the asymptotic properties also translate well to finite samples.

KW - Bias-correction

KW - Bootstrap

KW - Common correlated effects

KW - Factor-augmented regression

UR - http://www.scopus.com/inward/record.url?scp=85182733785&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85182733785&partnerID=8YFLogxK

U2 - 10.1016/j.jeconom.2023.105648

DO - 10.1016/j.jeconom.2023.105648

M3 - Article

SN - 0304-4076

VL - 240

JO - Journal of Econometrics

JF - Journal of Econometrics

IS - 1

M1 - 105648

ER -

Cross-section bootstrap for CCE regressions

Abstract

Bibliographical note

Funding

Keywords

Access to Document

Persistent URL (handle)

Other files and links

Fingerprint

Cite this