Chytanya S. Agarwal
Quasi-criminal cases occupy a middle ground between civil and criminal law. They often involve civil disputes with elements of criminality, such as fraud, corruption, or contempt, which carry heavier consequences for the parties involved. These include civil proceedings where a penalty may be akin to a criminal case. Because quasi-criminal cases blur the distinction between civil disputes and criminal liability, they raise questions regarding the appropriate standard of proof.
In this essay, I put forth a two-pronged argument. First, in civil suits involving quasi-criminal subject matter, Indian case law indicates a tacit adoption of the ‘clear and convincing’ standard of proof. I base this claim not on an empirical analysis of cases but by using cases as symptomatic of a given subject matter. Second, using a Law and Economics approach, I rationalise the adoption of this standard. In particular, using game theory and behavioural economics, I explain why higher sanctions and loss aversion lead judges to adopt a higher standard than the ‘preponderance of probabilities’. Using the same reasoning, I show why a standard below ‘beyond reasonable doubt’ is applied by judges in quasi-criminal offences.
Please note that this essay confines itself to an economic description of the manner in which judges approach quasi-criminal matters as it is. Any normative implications of this analysis would be largely inferential and not intended by the scope of this article.
Tracing the Standards: A Doctrinal Analysis of Case Laws
Per Wigmore, ‘standard of proof’ denotes the level of conviction or the ‘decisional threshold’ that enables the fact-finder to decide in favour of the party carrying the burden of proof. In the Indian context, Section 3 of the Evidence Act defines ‘proved’ but does not distinguish between its differing standards. In this regard, Ram Razik v. Jaswant Singh Chouhan (¶15) observed that the standard of proof must be “[j]udged by the standard of [a] prudent man,” making courts adopt two distinct standards of proof – the ‘preponderance of probabilities’ in civil cases and ‘proof beyond reasonable doubt’ in criminal convictions.
This bifurcation of standards of proof is a hallmark of adversarial justice systems and is non-existent in the inquisitorial systems of continental Europe. However, in some jurisdictions, particularly the United States, courts employ a third and intermediate standard of proof, called the ‘clear and convincing evidence’ standard, in civil cases involving moral turpitude. Being ‘intermediate’, it rests midway between ‘preponderance of probabilities’ and ‘proof beyond reasonable doubt’ standards. Expressed mathematically, the ‘degree of confidence’ required by the ‘preponderance of probabilities’ standard is above 50% probability; ‘proof beyond reasonable doubt’ requires 90-95% probability, and the ‘clear and convincing’ standard requires around 70-75% probability (see United States v. Fatico). Here, the ‘degree of confidence’ is a real number which lies in the range of 0 and 1. It signifies the probability of the truth of all the material facts to be established by the plaintiff to discharge their burden of persuasion (p.33). (Note: This probabilistic view of proof has been both defended and criticised; this article does not digress to this debate).
Indian Evidence law has been heavily inspired by the English common law which was adversarial and not inquisitorial. This influenced the Indian jurisprudence on standards of proof. While common law does not have the ‘clear and convincing’ standard of proof, per Bater v. Bater (pp.459-460) (‘Bater’), the ‘preponderance of probabilities’ standard is neither fixed nor absolute; it is flexible enough to accommodate varying ‘degrees of probability’ depending upon the suit’s subject-matter. This introduced an element of flexibility in affixing the standard of proof by courts in civil cases. Bater is veritably the current position of law due to its verbatim reproduction in many Indian cases including M. Siddiq (Ram Janmabhumi Temple-5 J.) v. Suresh Das (¶721-725); Lakshmi Rani Dhar v. Falakata Industries Ltd. (¶59); Shakuntala Rani v. Om Prakash Kohli (¶11); and Divisional Controller v. Ravindra Adhar (¶19).
Akin to Bater, the Australian High Court opined in Briginshaw v. Briginshaw that the ‘strength of evidence’ requisite to satisfy the ‘preponderance of probabilities’ depends on the “nature of what is sought to be proven.” This context-specific iteration of the standard of proof in civil cases, called the ‘Briginshaw standard’, is identical to Bater’s holding. However, Briginshaw expressly rejected the existence of a “third standard of persuasion” in common law. Nonetheless, scholars such as Williams and Fridman, the Briginshaw standard raises the ‘quantum of evidence’ required to tilt the ‘preponderance of probabilities’, blurring its difference with the clear and convincing standard of proof. Per Williams, Briginshaw’s flexibility has allowed courts to implement a flexible and intermediate standard of proof when the civil action involves criminal elements, grave sanctions, and the “intrinsic unlikelihood” of the event sought to be proved. This holds for Bater and subsequent common law as well. In the post-Bater common law context, Frank Bates discovered a heightened standard of proof adopted by English courts in civil cases based on the gravity of issues. This, he pointed out, was similar to the clear and convincing standard.
This trend of raising the evidentiary bar above that of the ‘preponderance of probabilities’ is apparent in Indian case law as well. In cases concerning paternity and illegitimacy, the Supreme Court and High Courts have consistently held that ‘preponderance of probabilities’, due to its low threshold, could potentially do grave injustice to the child if it was deemed illegitimate (see Kamti Devi, Banarsi Dass, Ashok Kumar, and W v. H). These cases held that deciding a child’s illegitimacy solely on the tilting of probabilities was odious. I acknowledge that most of these cases do not speak in the language of the ‘clear and convincing’ standard. Still, they invariably require a standard higher than the ‘preponderance of probabilities’. Notably, in W v. H (¶55), a standard between the ‘preponderance of probabilities’ and ‘proof beyond reasonable doubt’ was applied.
Similarly, a threshold higher than ‘preponderance of probabilities’ has been applied in civil suits involving a quasi-criminal subject matter (e.g., allegations of corruption, civil contempt, election fraud, forgery, violation of administrative law, etc.). In some of these cases, the courts have simply assumed that the standards of ‘preponderance of probabilities’ and ‘proof beyond reasonable doubt’ are mutually exclusive and exhaustive binaries – the inapplicability of the former implies the applicability of the latter. This stems from the binary treatment of civil and criminal cases – a dichotomy which, as argued by Cornwell, is no longer relevant in light of quasi-criminal cases. In only one of these cases, i.e., Heinz India (P) Ltd. v. State of U.P (‘Heinz India’), did the Supreme Court explicitly adopt the ‘clear and convincing’ standard. Similarly, in M. Krishnan v. State (‘M Krishnan’), the court held that ‘proof beyond reasonable’ doubt is merely a ‘subjective’ term used to “guide the subjective mind of the Judge.” In M Krishnan too, the court noted that standards of proof are judge-made; the definitions of ‘proved’ and ‘not proved’ in Section 3 of the Evidence Act give no guidance on the standard of proof. After referring to American jurisprudence on the ‘clear and convincing’ standard, M. Krishnan acknowledged the existence of an “intermediate standard of proof between the civil standard and the criminal standard commensurate with the occasion applies to civil cases.” Regarding Section 3 of the Act, the Law Commission of India too has recognised that there could be “different degrees of proof within the standard, both in civil and criminal cases” given some subjective element involved in the “evaluation of the degrees of probability.” Thus, there is nothing in the Evidence Act which bars the adoption of an intermediate standard.
Courts have clearly demonstrated a tendency to push the standard of proof above that of ‘preponderance of probabilities’. At the same time, even if they have employed the language of ‘proof beyond reasonable doubt’, the same stems from an (apparently) false dichotomy between the civil and criminal standards of proof – an intermediate standard is possible, given the observations in Heinz India and M. Krishnan. These observations, together with the flexible nature of the civil standard of proof, may be indicative of an implicit adoption of the ‘clear and convincing’ standard in civil disputes of a more consequential nature. While this claim is merely a hypothesis at this stage, its proof is explained in the next part of this essay.
Law and Economics
Quasi-Criminal civil suits
Standards of proof are premised on the principle of equality. In adversarial civil justice systems, the court adopts a stance of impartiality, treating the parties equally and deciding in favour of the party which has better proved the case. Per Addington v. Texas (‘Addington’), according to this principle of equality, the standard of proof must be such that the plaintiff (P) and the defendant (D) equally share the ‘risk of error’. From Addington it may be extrapolated that it is the standard of proof with which courts decide for/against a party that allocates the costs of error. This is in line with the existing literature (see here and here) which supports the view that the standards of proof allocate the ‘risk of error’ between P and D. Hypothetically, in every case, a court faces the following decision-making matrix:
| The decision is in: | ||
Defendant’s favour | Plaintiff’s favour | ||
Truth is in: | Defendant’s case | No error | Type I error (CD) |
Plaintiff’s case | Type II error (CP) | No error |
Error costs emanate from wrong adjudications either in favour of the plaintiff (‘false positive’ or ‘Type I error’) or against the plaintiff (‘false negative’ or ‘Type II error’). Type I errors impose cost (CD) on the defendant; it includes the penalty imposed and the cost of the associated chilling effect. Type II errors impose cost (CP) on the plaintiff, which includes the uncompensated injury and the litigation costs incurred. Per Cheng and Pardo (pp.202-204), in civil cases, the error costs are usually similar for both parties (i.e., CD = CP). So, a court applying a minimax approach would set the standard of proof slightly above 50%, implying the ‘preponderance of probabilities’ standard (pp.204-205).
Example 1: CD = CP = -1. Then, to equally divide the ‘risk of error’ the prior probabilities should be computed as follows:
D’s expected risk = P’s expected risk
⇒ (CD)*(Probability of D’s loss) + (0)*(Probability of P’s loss) = (CP)*(Probability of P’s loss) + (0)*(Probability of D’s loss)
⇒ (CD)*(Probability of D’s loss) = (CP)*(Probability of P’s loss)
⇒ Probability of D’s loss = Probability of P’s loss
But, Probability of D’s loss + Probability of P’s loss = 1
(⸪ Sum of probabilities of all outcomes is 1)
⸫ Probability of D’s loss = Probability of P’s loss = 0.5
[Note: The equation of Expected risk is derived from that of expected value:
E(X) = μ = ∑xiP(xi)
For us, Expected payoff = expected gain + expected loss
⇒ Expected payoff = (Probability of gain)(Value of gain) + (Probability of loss)(Value of loss)]
So, when the error costs are the same for both parties, the standard of 50% probability ensures the equal treatment of parties. This justifies the ‘preponderance of probabilities’ standard in most civil suits.
Now, in a civil suit with quasi-criminal subject matter, the costs of the associated chilling effect and penalty on the defendant are very high (i.e., CD > CP). Here, to ensure equal division of the ‘risk of error’ the prior probability with which cases are decided in the defendant’s favour should be increased above the ‘preponderance of probabilities’ standard.
Example 2: CD > CP. Assume, CD = -3; CP = -1. To equally divide the ‘risk of error’ the prior probabilities should be computed as follows:
D’s expected risk = P’s expected risk
⇒ (CD)*(Probability of D’s loss) = (CP)*(Probability of P’s loss)
⇒ (Probability of D' s loss)/(Probability of P’s loss)=CP/CD=(-1)/(-3)=1/3
(But, Probability of D’s loss + Probability of P’s loss = 1)
⸫ Probability of D’s loss = 0.25; Probability of P’s loss = 0.75
In other words, if the error costs on the defendant are higher than the plaintiff, the latter must satisfy a standard of proof higher than the ‘preponderance of probabilities’ (this is shown by the requisite probability of 0.75 in the above example). Otherwise, the parties would not be treated equally, as shown in the following example.
Example 3: CD > CP. Assume CD = -3; CP = -1. But the court adopts the ‘preponderance of probabilities’ standard of 50%.
Or, Probability of D’s loss = Probability of P’s loss = 0.5
So, D’s expected risk = (CD)*(Probability of D’s loss) = (-3)*(0.5) = -1.5
And, P’s expected risk = (CP)*(Probability of P’s loss) = (-1)*(0.5) = -0.5
⸫ D’s expected risk ≠ P’s expected risk
This would violate the principle of the equal distribution of ‘risk of error’. In Heinz India and M Krishnan, Indian courts have made references to Addington’s observations on the distribution of ‘risk of error’. In other words, any such deviation from this equality principle is suboptimal because it would impose undue ex-ante costs on at least one party. Per Kaplow, the optimal standard of proof (which is determined ex-ante) strikes a balance between a low standard (which discourages malicious conduct at the cost of chilling effect) and a high standard (which encourages malicious conduct by making proof more demanding). The optimal threshold minimises chilling yet maximises deterrence.
This argument is further supported by behavioural economics. According to behavioural economists (see Schweizer, and Zamir and Ritov), judges tend to adopt a standard of proof greater than the ‘preponderance of probabilities’ due to loss aversion, omission bias, and status quo bias. These biases operate together in the following manner. Loss aversion refers to the belief that disutility from loss exceeds the utility from gain, prompting decision-makers (here, judges) to not deviate from their original reference point. This reference point is usually the status quo prior to the litigation. This leads to status quo bias, or the tendency to not opt for alternatives to the existing state of affairs based on the potential cost of regret. People anticipate a higher cost of regret if a deviation leads to a worse outcome than if they had stayed with the status quo, notwithstanding that the status quo may actually be the less favourable choice. This gets compounded with omission bias, viz., the belief that a person has a greater moral responsibility for harmful outcomes they have brought about actively than for those caused passively. In civil litigation, judges perceive dismissing a plaint as an omission or a passive action; accepting a claim and deciding in favour of the plaintiff is an active action. Conjointly, these three biases inhibit judges from accepting the plaint unless the plaintiff’s case is truly compelling. In doing so, they tend to raise the standard above that of preponderance of probabilities.
Quasi-criminal offences
Till now, this essay confined itself to quasi-criminal civil suits. What happens to quasi-criminal offences which are part of criminal statutes and prosecuted in criminal courts? Do such quasi-criminal offences demand proof ‘beyond reasonable doubt’? Illustratively, courts have held that an offence under Section 138 of the Negotiable Instruments Act is a “civil sheep” in a “criminal wolf’s clothing”. Being quasi-criminal in nature, per case law, they are held to require ‘proof beyond reasonable doubt’. Similarly, criminal contempt, being quasi-criminal in nature, is said to require proof that is ‘beyond all reasonable doubt’. I do not intend to give an exhaustive list of such quasi-criminal offences. But my main point is that courts have ruled that proof ‘beyond reasonable doubt’ is the required standard for proving quasi-criminal offences.
Such instances cut across my main point that quasi-criminal cases, in general practice, use an intermediate standard of proof. Now, I briefly comment on the true standard of proof of quasi-criminal offences. I attempt to pre-empt the argument that they are, in fact (but not in law), governed by proof ‘beyond reasonable doubt’. That proof ‘beyond reasonable doubt’ does not apply to quasi-criminal cases is justified by the following example.
Example 4: CD > CP. Assume CD = -4 (we have kept the cost slightly higher than that in a civil suit); CP = -1. But the court adopts the ‘beyond reasonable doubt’ standard of 90%.
So, D’s expected risk = (CD)*(Probability of D’s loss) = (-3)*(0.1) = -0.3
And, P’s expected risk = (CP)*(Probability of P’s loss) = (-1)*(0.9) = -0.9
⸫ D’s expected risk ≠ P’s expected risk
Thus, adopting the ‘beyond reasonable doubt’ standard also violates the principle of equal distribution of ‘risk of error’.
This argument is further buttressed by studies conducted on jurors and judges in the United States, considering no similar study has been undertaken in the Indian context. Firstly, Simon and Mahan empirically demonstrate that an inherent ambiguity plagues both the civil and criminal standards of proof. Their results show that judges and jurors tend to fixate on a standard above ‘preponderance of probabilities’ but below ‘beyond reasonable doubt’, even in criminal cases. These findings must be read along with the Law Commission’s observations on the ‘subjective’ nature of the ‘proof beyond reasonable doubt’ standard which I referred to earlier; there is nothing in law preventing the adoption of an intermediate standard even when courts speak in the language of ‘proof beyond reasonable doubt’. This, I argue, is an intermediate standard of proof in the guise of ‘beyond reasonable doubt’. Secondly, per Lowey, this ambiguity surrounding ‘beyond reasonable doubt’ persists even in serious crimes. This further casts doubts on precisely what proof is demanded by the ‘beyond reasonable doubt’ standard, even in core criminal cases. Lastly, given such ambiguity, Laudan has argued that ‘lesser crimes’ are best governed by the ‘clear and convincing’ standard of proof.
While the use of ‘beyond reasonable doubt’ in quasi-criminal cases reflects the binary approach traditionally applied to civil and criminal matters, as mentioned previously, this dichotomy is increasingly being questioned. As M. Krishnan demonstrates, there is room for an intermediate standard, such as ‘clear and convincing,’ which accounts for the gravity of quasi-criminal offences without imposing the stringent burden of criminal liability.
Behavioural economics also aligns with this conclusion. According to it, a cognitive bias influencing decision-makers is ‘extremeness aversion’ or the ‘compromise effect’ (Fisher, p.833). This translates into a tendency to prefer intermediate choice categories over extreme/end categories. This ‘extremeness aversion’, I argue, would make judges tilt towards the ‘clear and convincing’ standard instead of the ‘beyond reasonable doubt’ standard when dealing with quasi-criminal cases.
Chytanya S. Agarwal is a fourth-year B.A., LL.B. (Hons.) student at the National Law School of India University, Bangalore. He may be contacted at chytanya.agarwal@nls.ac.in.
Feature Image: Le Défenseur" (Counsel for the Defense), c. 1862/1865 by Honoré Daumier. Source: Wikimedia Commons.
Comentarios