Einde inhoudsopgave
Directors' liability (IVOR nr. 101) 2017/3.4.3
3.4.3 Which behavioural and contextual legal case factors have a significant effect on directors’ liability in cases not involving directors’ ‘subjective bad faith’?
mr. drs. N.T. Pham, datum 09-01-2017
- Datum
09-01-2017
- Auteur
mr. drs. N.T. Pham
- JCDI
JCDI:ADS393753:1
- Vakgebied(en)
Ondernemingsrecht / Rechtspersonenrecht
Voetnoten
Voetnoten
The regression analysis based on the total sample of 158 cases did not reveal other significant legal case factors (see also Table 7 asterisk 2). This was presumably because one third of the sample (59 cases) was explainable as ‘subjective bad faith’.
I applied logistic regression analysis hierarchically to test several models. The Nagelkerke R square increased significantly when I tested the model involving the five predictor variables as presented in Table 7. The Nagelkerke R square ranges from 0 – 1 and indicateshow well a model fits the data. I obtained a value of 0.8.
These results must be viewed with appropriate consideration of the following. I have used legal case factors: i.e. legally interpreted case factors occurring in a court decision. Consequently, these legal case factors by their character, are very close to the outcome of a court case. This may explain in part the high predictive value of the model obtained in Table 7.
logit = −3.106 + 4.649 ∗ 0 + 4.096 ∗ 1 − 0.550 ∗ 0 + 4.123 ∗ 0 + 3.695 ∗ 1 = 4.685P = 1 / (1+e– 4.685) ~ 0.99.
In paragraph 3.2.1, I discussed that, under established case law, courts are bound to apply an objective test when assessing if director conduct is subject to ‘serious reproach’. By no means do Dutch courts require a claimant to prove a director’s ‘subjective bad faith’. Moreover, courts do not need to infer a director’s ‘subjective bad faith’ to assume ‘serious reproach’ and find a director personally liable. It goes without saying that, when pleaded well, a director’s ‘subjective bad faith’ action may influence a court’s judgement. Consider for instance De Rouw v. Dingemans where a director acted fraudulently. The ‘subjective bad faith’ action on its own caused the court to assume that the director was personally liable, without regard to other factors. In paragraph 3.4.2, I demonstrated that, in 59 of the 158 sample cases, a director’s ‘subjective bad faith’ action led a court to rule that a director bore personal liability. Moreover, I observed no variation in case outcome among the 59 ‘subjective bad faith’ cases. This empirical finding bears an important consequence. Within the group of cases not involving director’s ‘subjective bad faith’ (99 cases), I did observe variation in case outcome. To predict the outcome of these 99 cases not involving directors’ ‘subjective bad faith’ requires a separate analysis of this group of cases.1 To pursue this line of research, I conducted a bivariate analysis on the 99 cases not involving issues of ‘subjective bad faith’, the results of which are presented in Table 6.
Legal case factors
% Liability within categories1
Exact Sig. (2-sided)
Cramér’s V
Behavioural legal case factors:
1
0
Unreasonably informed
46%
44%
1.000
.019
Norm violation
78%
26%
.000
.511
Foreseeability of damage
87%
33%
.000
.459
Unreasonable risk takingUnreasonable risk taking
36%
51%
.209
.142
Incompetence2
–
–
–
–
Dereliction of duty
95%
34%
.000
.482
Conflict of interest
69%
42%
.079
.186
Limitation of liabLimitation of liabilityility
63%
23%
.000
.391
Contextual legal case factors:
1
0
Misrepresentation of information
90%
35%
.000
.431
Fraud
82%
38%
.001
.337
Policy failure
80%
39%
.005
.293
Mismanagement
92%
30%
.000
.543
Bankruptcy
48%
27%
.336
.129
These involve the percentages of cases in which the court judged a director personally liable under the condition that the legal case factor (predictor variable) occurred (=1) and did not occur (=0) in the cases under study.
This legal case factor did not meet the requirement of the chi-square test (all expected counts are greater than 1 and no more than 20% of the expected counts is less than 5).
The research results indicate a strong association with directors’ liability for 2 legal case factors (V ≥ 0.5): ‘mismanagement’ (V = 0.54) and ‘norm violation’ (V = 0.51); and a medium strong association (0.3 ≤ V ≤ 0.5) for 5 legal case factors: ‘dereliction of duty’ (V = 0.48), ‘foreseeability of damage’ (V = 0.46), ‘misrepresentation of information’ (V = 0.43), ‘limitation of liability’ (V = 0.39) and ‘fraud’ (V = 0.34).
As I have explained in paragraph 3.3.4, bivariate analysis only provides insight into the independent effect of each of the predictor variables on directors’ liability. It does not provide insight into how certain predictor variables relate to the case outcome as a group. The central focus of this research is to analyse the effect of multiple legal case factors on the case outcome (whether the director was deemed liable or not). Moreover, I wish to investigate whether the factors of ‘norm violation’ and ‘foreseeability of damage’, which I identified as important factors in the court’s simple legal decision model (paragraph 3.2.5), may find empirical support for their importance. The results are shown in Table 7.2
Legal case factors1
B
Wald2
Sig.
Exp(B)
Behavioural legal case factors:
Foreseeability of damage
4.649
13.617
.000
104.480
Norm violation
4.096
12.808
.000
60.129
Dereliction of duty
-.550
.674
.412
.577
Contextual legal case factors:
Misrepresentation of information
4.123
8.576
.003
61.770
Mismanagement
3.695
8.226
.004
40.263
Constant
-3.106
4.263
.039
.045
Nagelkerke R square
0.80
Overall accuracy of the model
88.9%
All legal case factors have been tested for multicollinearity. The results yield no concerns. The legal case factors have been analysed for potential interaction effects. The results do not give rise to different conclusions. At least, on the basis of this data set, potential interaction effects cannot be excluded.
The model was tested for robustness using the bootstrap method. The results do not give rise to different conclusions. The model was also applied to the total sample of 158 cases. The results do not give rise to different conclusions.
All predictor variables are highly significant (sig. ≤ 0.05) except for ‘dereliction of duty’. Analysing the predictor variables as a group shows that the legal case factors of ‘foreseeability of damage’ (Wald = 13.617) and ‘norm violation’ (Wald = 12.808) are the strongest predictors in the model in comparison to the other factors. In terms of effect size of the model, Nagelkerke R square of 0.80 was found and an overall prediction rate of 88.9%.3 These research results indicate that the Supreme Court’s simple legal decision model functions effectively. Courts apply the factors ‘norm violation’ and ‘foreseeability of damage’ fairly consistently. Arguably, there is empirical support for the conclusion that these two predictor variables are the most influential legal case factors adjudicating directors’ liability.
Based on the B-values in Table 7, it is now possible to predict the outcome of a hypothetical case. Let us imagine that in a given case there was ‘mismanagement’ – for instance, the company’s administrative system was failing – and that the director acted in conflict with a norm (‘norm violation’) and other factors did not occur. Applying Formula 1, there is a 99% chance that the case would end with the director bearing personal liability.4
Table 8 presents several more hypothetical cases. The chance of finding a director liable increases when more legal case factors are considered.
Legal case factors
Case 1
Case 2
Case 3
Case 4
Case 5
Foreseeability of damage
–
X
X
–
–
Norm violation
X
X
–
X
–
Dereliction of duty
–
–
–
–
–
Misrepresentation of information
–
–
–
–
X
Mismanagement
X
–
–
–
–
Chance of directors’ liability
99%
99%
82%
73%
73%