Once it has been established that the defendant has made an illegitimate threat or exerted illegitimate pressure, the claimant must then show that he or she was induced to transfer a benefit to the defendant as a result or, where relevant, that he or she was induced to enter into a contract. This is a question ofcausation. It is obvious that, where the claimant has not perceived any threats made or pressure exerted by the defendant or where the claimant discounted the threats, the defendant’s conduct cannot be regarded as the operative cause of any transaction entered into with the defendant. But, where the claimant did perceive the threats or the pressure, the difficult question concerns what test of causation should be adopted to determine whether the duress was a sufficient cause for the claimant to transfer the benefit. In the context of spontaneous mistakes it was concluded that a mistake would be considered as constituting a ground of restitution only if the claimant could show that, but for the mistake, he or she would not have transferred a benefit to the defendant. But where the mistake has been fraudulently induced by the defendant, a more lenient test of causation is adopted, namely that it is sufficient that the mistake was a contributory cause of the payment without needing to show that it was the ‘but for’ cause.37 This is because, where the defendant has fraudulently induced the mistake, the defendant’s conduct makes it easier to conclude that the claimant’s restitutionary claim is stronger than the defendant’s claim to retain the enrichment. In principle, a similar argument should apply where the defendant has compelled the claimant to transfer a benefit. The key case on causation in the context of duress, Barton v Armstrong,38 has adopted such an approach, with explicit reliance on the misrepresentation cases.
In Barton v Armstrong the defendant wanted the claimant to buy his shares in a company of which the claimant was managing director. The defendant had made numerous threats to kill the claimant if he did not purchase the shares, but it appeared that the main reason for the claimant agreeing to the purchase was commercial expediency. Despite this, the Privy Council held that the threats to kill the claimant were sufficient to vitiate the deed for the transfer of the shares. Lord Cross for the majority said that the test of causation for the purposes of duress was the same as that where a contract was induced by fraudulent misrepresentation, namely that it was sufficient that the threat or misrepresentation was a reason the claimant executed the deed, but it did not have to be the predominant reason but for which the claimant would not have executed the deed. Consequently, the claimant was ‘entitled to relief even though he might well have entered into the contract if [the defendant] had uttered no threats to induce him to do so’.39 In other words, it was not necessary to show that but for the threats the claimant would not have agreed to buy the shares. It did not matter that the claimant would have agreed to buy the shares even if the threat had not been made, as long as it could be shown that, since the threat was made, this was a factor which influenced the claimant’s decision to buy the shares. As Lord Cross said, ‘in this field the court does not allow an examination into the relative importance of contributory causes’.4 The minority, Lords Wilberforce and Simon, accepted this formulation of the test of causation for the purposes of duress, though they did not agree that the test was satisfied on the facts of the case, because the trial judge’s finding of fact, namely that the threats had not even been a reason for the claimant’s decision to execute the deed, meant that the claimant had been motivated only by the commercial advantages of purchasing the defendant’s shareholding and was not influenced by the threat at all.
Although the test of causation for the purposes of duress appears to be the same as that for fraudulently induced mistakes, namely a test of operative rather than ‘but for’ cause, this test may be applied differently in two respects in the context of duress.
First, Lord Cross in Barton v Armstrong41 suggested that the burden of proof is on the defendant to show that the threats did not contribute to the claimant’s decision to transact with the defendant. If this is correct it means that, once the claimant has established that the defendant had exercised duress, there is a presumption that the duress caused the claimant to transact with the defendant. The defendant must then seek to rebut the presumption by showing that the duress did not contribute to the claimant’s decision to transact with the defendant.42 This would be a particularly difficult presumption to rebut, since the defendant would be required to show why the claimant entered into the transaction. Even so, this presumption of causation is justifiable, at least where the duress is particularly serious, for example where it involves the actual or threatened commission of a crime against the person. Where the defendant commits or threatens to commit such a serious unlawful act, this conduct can be characterized as so unreasonable that the test of causation for purposes of restitution should be weighted in the claimant’s favour.
Secondly, Barton v Armstrong itself involved duress against the person, the most serious type of duress, so it is relatively easy to conclude that the defendant’s conduct was 
sufficiently reprehensible to justify the recognition of a lenient test of causation. But it does not necessarily follow that the same test of causation should be adopted in respect of all types of duress. Consequently, for each type of duress it will be necessary to consider whether the test of operative causation should be adopted or whether an alternative test, such as that of ‘but for’ cause, is more appropriate. Indeed, in Huyton v Peter Cremer Mance J recognized that the ‘but for’ test of causation applies to economic duress. This is justifiable because the threat is less serious and so a stricter test of causation is required. This is consistent with the test for induced mistakes, since it is only for fraudulent misrepresentation that the contributory cause test applies; for negligent and innocent misrepresentations it must be shown that the representation was an operative cause of the benefit being transferred or a contract being made.
-  Banking Corporation (1988) 19 NSWLR 40.
-  See p 170, above. 37 See p 187, above.
-   AC 104. 39 Ibid, 119. 40 Ibid, 118. 41 Ibid, 120. 42 See Crescendo Management Pty Ltd v Westpac Banking Corporation (1988) 19 NSWLR40,46 (McHugh JA);Huyton SA v Peter Cremer GmbH [ 1999] 1 Lloyd’s Rep 620.
-   1 Lloyd’s Rep 620. 44 See p 187, above.
-  45 Barclays Bank plc v O’Brien  1 AC 180. See p 199, above and p 269, below.