Predicting tax avoidance by means of social network analytics

Lismont Jasmien, Eddy Cardinaels, L.M.L. Bruynseels, Sander De Groote, B. Baesens, W. Lemahieu, J. Vanthienen

Research output: Contribution to journalArticleScientificpeer-review

Abstract

This study predicts tax avoidance by means of social network analytics. We extend previous literature by being the first to build a predictive model including a larger variation of network features. We construct a network of firms connected through shared board membership. Then, we apply three analytical techniques, logistic regression, decision trees, and random forests; to create five models using either firm characteristics, network characteristics or different combinations of both. A random forest including firm characteristics, network characteristics of firms and network characteristics of board members provides the best performance with a minimal increase of 7 pp in AUC. Hence, including network effects significantly improves the predictive ability of tax avoidance models, implying that board members exhibit specific knowledge which can carry over across firms. We find that having board members with no connections to low-tax companies lowers the likelihood of being a low-tax firm. Similarly, the higher the average tax rate of the companies a board member is connected to, the lower the chance of being low-tax. On the other hand, being connected to more low-tax firms increases the probability of being low-tax. Consistent with prior literature on firm-specific variables, PP&E has a positive influence on the probability of being low-tax, while EBITDA has a negative effect. Our results are informative for companies as to the director expertise they want to attract in their boards. Additionally, financial analysts and regulatory agencies can use our insights to predict which firms are likely to be low-tax and potentially at risk.
Original languageEnglish
Pages (from-to)13-24
JournalDecision Support Systems
Volume108
DOIs
Publication statusPublished - Apr 2018

Fingerprint

Taxation
Decision Trees
Area Under Curve
Logistic Models
Tax avoidance
Tax
Social networks
Social Networks
Avoidance
Industry
Decision trees
Logistics

Keywords

  • board interlocks
  • predictive analytics
  • social network analytics
  • social ties
  • tax avoidance
  • tax planning

Cite this

Jasmien, L., Cardinaels, E., Bruynseels, L. M. L., De Groote, S., Baesens, B., Lemahieu, W., & Vanthienen, J. (2018). Predicting tax avoidance by means of social network analytics. Decision Support Systems, 108, 13-24. https://doi.org/10.1016/j.dss.2018.02.001
Jasmien, Lismont ; Cardinaels, Eddy ; Bruynseels, L.M.L. ; De Groote, Sander ; Baesens, B. ; Lemahieu, W. ; Vanthienen, J. / Predicting tax avoidance by means of social network analytics. In: Decision Support Systems. 2018 ; Vol. 108. pp. 13-24.
@article{8eb577ff407b4fb0b679437c813ee1fa,
title = "Predicting tax avoidance by means of social network analytics",
abstract = "This study predicts tax avoidance by means of social network analytics. We extend previous literature by being the first to build a predictive model including a larger variation of network features. We construct a network of firms connected through shared board membership. Then, we apply three analytical techniques, logistic regression, decision trees, and random forests; to create five models using either firm characteristics, network characteristics or different combinations of both. A random forest including firm characteristics, network characteristics of firms and network characteristics of board members provides the best performance with a minimal increase of 7 pp in AUC. Hence, including network effects significantly improves the predictive ability of tax avoidance models, implying that board members exhibit specific knowledge which can carry over across firms. We find that having board members with no connections to low-tax companies lowers the likelihood of being a low-tax firm. Similarly, the higher the average tax rate of the companies a board member is connected to, the lower the chance of being low-tax. On the other hand, being connected to more low-tax firms increases the probability of being low-tax. Consistent with prior literature on firm-specific variables, PP&E has a positive influence on the probability of being low-tax, while EBITDA has a negative effect. Our results are informative for companies as to the director expertise they want to attract in their boards. Additionally, financial analysts and regulatory agencies can use our insights to predict which firms are likely to be low-tax and potentially at risk.",
keywords = "board interlocks, predictive analytics, social network analytics, social ties, tax avoidance, tax planning",
author = "Lismont Jasmien and Eddy Cardinaels and L.M.L. Bruynseels and {De Groote}, Sander and B. Baesens and W. Lemahieu and J. Vanthienen",
year = "2018",
month = "4",
doi = "10.1016/j.dss.2018.02.001",
language = "English",
volume = "108",
pages = "13--24",
journal = "Decision Support Systems",
issn = "0167-9236",
publisher = "Elsevier",

}

Jasmien, L, Cardinaels, E, Bruynseels, LML, De Groote, S, Baesens, B, Lemahieu, W & Vanthienen, J 2018, 'Predicting tax avoidance by means of social network analytics' Decision Support Systems, vol. 108, pp. 13-24. https://doi.org/10.1016/j.dss.2018.02.001

Predicting tax avoidance by means of social network analytics. / Jasmien, Lismont; Cardinaels, Eddy; Bruynseels, L.M.L.; De Groote, Sander; Baesens, B.; Lemahieu, W.; Vanthienen, J.

In: Decision Support Systems, Vol. 108, 04.2018, p. 13-24.

Research output: Contribution to journalArticleScientificpeer-review

TY - JOUR

T1 - Predicting tax avoidance by means of social network analytics

AU - Jasmien, Lismont

AU - Cardinaels, Eddy

AU - Bruynseels, L.M.L.

AU - De Groote, Sander

AU - Baesens, B.

AU - Lemahieu, W.

AU - Vanthienen, J.

PY - 2018/4

Y1 - 2018/4

N2 - This study predicts tax avoidance by means of social network analytics. We extend previous literature by being the first to build a predictive model including a larger variation of network features. We construct a network of firms connected through shared board membership. Then, we apply three analytical techniques, logistic regression, decision trees, and random forests; to create five models using either firm characteristics, network characteristics or different combinations of both. A random forest including firm characteristics, network characteristics of firms and network characteristics of board members provides the best performance with a minimal increase of 7 pp in AUC. Hence, including network effects significantly improves the predictive ability of tax avoidance models, implying that board members exhibit specific knowledge which can carry over across firms. We find that having board members with no connections to low-tax companies lowers the likelihood of being a low-tax firm. Similarly, the higher the average tax rate of the companies a board member is connected to, the lower the chance of being low-tax. On the other hand, being connected to more low-tax firms increases the probability of being low-tax. Consistent with prior literature on firm-specific variables, PP&E has a positive influence on the probability of being low-tax, while EBITDA has a negative effect. Our results are informative for companies as to the director expertise they want to attract in their boards. Additionally, financial analysts and regulatory agencies can use our insights to predict which firms are likely to be low-tax and potentially at risk.

AB - This study predicts tax avoidance by means of social network analytics. We extend previous literature by being the first to build a predictive model including a larger variation of network features. We construct a network of firms connected through shared board membership. Then, we apply three analytical techniques, logistic regression, decision trees, and random forests; to create five models using either firm characteristics, network characteristics or different combinations of both. A random forest including firm characteristics, network characteristics of firms and network characteristics of board members provides the best performance with a minimal increase of 7 pp in AUC. Hence, including network effects significantly improves the predictive ability of tax avoidance models, implying that board members exhibit specific knowledge which can carry over across firms. We find that having board members with no connections to low-tax companies lowers the likelihood of being a low-tax firm. Similarly, the higher the average tax rate of the companies a board member is connected to, the lower the chance of being low-tax. On the other hand, being connected to more low-tax firms increases the probability of being low-tax. Consistent with prior literature on firm-specific variables, PP&E has a positive influence on the probability of being low-tax, while EBITDA has a negative effect. Our results are informative for companies as to the director expertise they want to attract in their boards. Additionally, financial analysts and regulatory agencies can use our insights to predict which firms are likely to be low-tax and potentially at risk.

KW - board interlocks

KW - predictive analytics

KW - social network analytics

KW - social ties

KW - tax avoidance

KW - tax planning

U2 - 10.1016/j.dss.2018.02.001

DO - 10.1016/j.dss.2018.02.001

M3 - Article

VL - 108

SP - 13

EP - 24

JO - Decision Support Systems

JF - Decision Support Systems

SN - 0167-9236

ER -