Dynamic clustering of multivariate panel data

Igor Custodio João; André Lucas; Julia Schaumburg; Bernd Schwaab

doi:10.1016/j.jeconom.2022.03.003

Dynamic clustering of multivariate panel data

Igor Custodio João, André Lucas^*, Julia Schaumburg, Bernd Schwaab

^*Corresponding author for this work

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

145 Downloads (Pure)

Abstract

We propose a dynamic clustering model for uncovering latent time-varying group structures in multivariate panel data. The model is dynamic in three ways. First, the cluster location and scale matrices are time-varying to track gradual changes in cluster characteristics over time. Second, all units can transition between clusters based on a Hidden Markov model (HMM). Finally, the HMM's transition matrix can depend on lagged time-varying cluster distances as well as economic covariates. Monte Carlo experiments suggest that the units can be classified reliably in a variety of challenging settings. Incorporating dynamics in the cluster composition proves empirically important in a study of 299 European banks between 2008Q1 and 2018Q2. We find that approximately 3% of banks transition per quarter on average. Transition probabilities are in part explained by differences in bank profitability, suggesting that factors contributing to low profitability for some banks can lead to long-lasting changes in financial industry structure.

Original language	English
Article number	105281
Pages (from-to)	1-18
Number of pages	18
Journal	Journal of Econometrics
Volume	237
Issue number	2, Part B
Early online date	6 May 2022
DOIs	https://doi.org/10.1016/j.jeconom.2022.03.003
Publication status	Published - Dec 2023

Bibliographical note

Funding Information:
This work was supported by the Dutch National Science Foundation (NWO), Netherlands [ 406.18.EB.011 to I.C.J. and A.L., VI.VIDI.191.169 to J.S]. The views expressed in this paper are those of the authors and do not necessarily reflect the views or policies of the European Central Bank.

Funding Information:
This work was supported by the Dutch National Science Foundation (NWO), Netherlands [406.18.EB.011 to I.C.J. and A.L., VI.VIDI.191.169 to J.S]. The views expressed in this paper are those of the authors and do not necessarily reflect the views or policies of the European Central Bank.

Publisher Copyright:
© 2022 Elsevier B.V.

Funding

This work was supported by the Dutch National Science Foundation (NWO), Netherlands [ 406.18.EB.011 to I.C.J. and A.L., VI.VIDI.191.169 to J.S]. The views expressed in this paper are those of the authors and do not necessarily reflect the views or policies of the European Central Bank. This work was supported by the Dutch National Science Foundation (NWO), Netherlands [406.18.EB.011 to I.C.J. and A.L., VI.VIDI.191.169 to J.S]. The views expressed in this paper are those of the authors and do not necessarily reflect the views or policies of the European Central Bank.

Keywords

Bank business models
Dynamic clustering
Hidden Markov Model
Panel data
Score-driven dynamics

Access to Document

10.1016/j.jeconom.2022.03.003

Dynamic clustering of multivariate panel dataFinal published version, 1.69 MBLicence: Article 25fa Dutch Copyright Act

Persistent URL (handle)

Persistent link

Cite this

@article{62a78c793a5147338f06d22ad13667ff,

title = "Dynamic clustering of multivariate panel data",

abstract = "We propose a dynamic clustering model for uncovering latent time-varying group structures in multivariate panel data. The model is dynamic in three ways. First, the cluster location and scale matrices are time-varying to track gradual changes in cluster characteristics over time. Second, all units can transition between clusters based on a Hidden Markov model (HMM). Finally, the HMM's transition matrix can depend on lagged time-varying cluster distances as well as economic covariates. Monte Carlo experiments suggest that the units can be classified reliably in a variety of challenging settings. Incorporating dynamics in the cluster composition proves empirically important in a study of 299 European banks between 2008Q1 and 2018Q2. We find that approximately 3% of banks transition per quarter on average. Transition probabilities are in part explained by differences in bank profitability, suggesting that factors contributing to low profitability for some banks can lead to long-lasting changes in financial industry structure.",

keywords = "Bank business models, Dynamic clustering, Hidden Markov Model, Panel data, Score-driven dynamics",

author = "{Custodio Jo{\~a}o}, Igor and Andr{\'e} Lucas and Julia Schaumburg and Bernd Schwaab",

note = "Funding Information: This work was supported by the Dutch National Science Foundation (NWO), Netherlands [ 406.18.EB.011 to I.C.J. and A.L., VI.VIDI.191.169 to J.S]. The views expressed in this paper are those of the authors and do not necessarily reflect the views or policies of the European Central Bank. Funding Information: This work was supported by the Dutch National Science Foundation (NWO), Netherlands [406.18.EB.011 to I.C.J. and A.L., VI.VIDI.191.169 to J.S]. The views expressed in this paper are those of the authors and do not necessarily reflect the views or policies of the European Central Bank. Publisher Copyright: {\textcopyright} 2022 Elsevier B.V.",

year = "2023",

month = dec,

doi = "10.1016/j.jeconom.2022.03.003",

language = "English",

volume = "237",

pages = "1--18",

journal = "Journal of Econometrics",

issn = "0304-4076",

publisher = "Elsevier B.V.",

number = "2, Part B",

}

TY - JOUR

T1 - Dynamic clustering of multivariate panel data

AU - Custodio João, Igor

AU - Lucas, André

AU - Schaumburg, Julia

AU - Schwaab, Bernd

N1 - Funding Information: This work was supported by the Dutch National Science Foundation (NWO), Netherlands [ 406.18.EB.011 to I.C.J. and A.L., VI.VIDI.191.169 to J.S]. The views expressed in this paper are those of the authors and do not necessarily reflect the views or policies of the European Central Bank. Funding Information: This work was supported by the Dutch National Science Foundation (NWO), Netherlands [406.18.EB.011 to I.C.J. and A.L., VI.VIDI.191.169 to J.S]. The views expressed in this paper are those of the authors and do not necessarily reflect the views or policies of the European Central Bank. Publisher Copyright: © 2022 Elsevier B.V.

PY - 2023/12

Y1 - 2023/12

N2 - We propose a dynamic clustering model for uncovering latent time-varying group structures in multivariate panel data. The model is dynamic in three ways. First, the cluster location and scale matrices are time-varying to track gradual changes in cluster characteristics over time. Second, all units can transition between clusters based on a Hidden Markov model (HMM). Finally, the HMM's transition matrix can depend on lagged time-varying cluster distances as well as economic covariates. Monte Carlo experiments suggest that the units can be classified reliably in a variety of challenging settings. Incorporating dynamics in the cluster composition proves empirically important in a study of 299 European banks between 2008Q1 and 2018Q2. We find that approximately 3% of banks transition per quarter on average. Transition probabilities are in part explained by differences in bank profitability, suggesting that factors contributing to low profitability for some banks can lead to long-lasting changes in financial industry structure.

AB - We propose a dynamic clustering model for uncovering latent time-varying group structures in multivariate panel data. The model is dynamic in three ways. First, the cluster location and scale matrices are time-varying to track gradual changes in cluster characteristics over time. Second, all units can transition between clusters based on a Hidden Markov model (HMM). Finally, the HMM's transition matrix can depend on lagged time-varying cluster distances as well as economic covariates. Monte Carlo experiments suggest that the units can be classified reliably in a variety of challenging settings. Incorporating dynamics in the cluster composition proves empirically important in a study of 299 European banks between 2008Q1 and 2018Q2. We find that approximately 3% of banks transition per quarter on average. Transition probabilities are in part explained by differences in bank profitability, suggesting that factors contributing to low profitability for some banks can lead to long-lasting changes in financial industry structure.

KW - Bank business models

KW - Dynamic clustering

KW - Hidden Markov Model

KW - Panel data

KW - Score-driven dynamics

UR - http://www.scopus.com/inward/record.url?scp=85129976936&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85129976936&partnerID=8YFLogxK

U2 - 10.1016/j.jeconom.2022.03.003

DO - 10.1016/j.jeconom.2022.03.003

M3 - Article

AN - SCOPUS:85129976936

SN - 0304-4076

VL - 237

SP - 1

EP - 18

JO - Journal of Econometrics

JF - Journal of Econometrics

IS - 2, Part B

M1 - 105281

ER -

Dynamic clustering of multivariate panel data

Abstract

Bibliographical note

Funding

Keywords

Access to Document

Persistent URL (handle)

Other files and links

Cite this