PT-2021-000393 - [2025] EWHC 2749 (Ch)

The second question is deceptively straightforward on two levels. First, there was the attractive submission of Mr Wheeler that the Court of Appeal had already done the work for me in identifying the relevant (and only relevant) legitimate interest as being the interest LCL had in timely repayment of the Loan. I have addressed that submission above; despite the persuasive way in which the argument was put, my reading of the Court of Appeal’s question requires me to consider all primary obligations to which the Default Rate applies and, in turn, the legitimate interests that LCL has in enforcing them.

Therein lies the second complexity. Some of those interests are obvious, and the protection afforded to them by the Default Rate is obviously legitimate. There is a temptation simply to assume that they are not in play and focus only on those interests that are open to question. That, I confess, was a flaw in my approach in the First Judgment: I succumbed to that temptation, rushed at the hurdle and I stumbled. I do not propose to repeat the experience.

With that in mind, one must therefore turn to the events of default. These can, I think, sensibly be grouped into categories of primary obligations. In saying that I recognise that some obligations give rise to more than one interest: factors affecting security may also affect repayment, for example; repayment cannot be divorced from non-residence requirements because the consequence of a breach of the non-residence requirement could be that a loan was unenforceable; and so forth. At the same time, addressing each clause individually risks unnecessary atomisation and duplication, while lumping everything into a broad category of “risk” can mean that some legitimate interests are ignored or confused, a point I feel I need to address in some detail in the particular context of credit risk.

It is easiest to clear the chaff out of the way first. Certain events of default have no apparent relevance to the analysis. Clause 12.2(i) dealt with defaults under what is described as the related Facility Letter. This appears to be an error, in that there is only one facility and only one Facility Letter. Clauses 12.2(n) and (o) relate to second charges and are not relevant here.

Turning to those primary obligations that do involve a legitimate interest, the first category is simple: under clause 12.2(a) there is an event of default on non-payment. A lender obviously has a legitimate interest in repayment (the Repayment Interest). As Mr Wheeler submitted, it is the sine qua non of a loan.

Second, there is clause 12.2(h): if the representations and warranties were or became untrue, that is an event of default. The analysis here is a little more involved. The representations and warranties given at inception were the basis of the whole Loan; as clause 4 of the Facility Letter makes clear, LCL’s obligation to advance funds was conditional on those representations and warranties being “true, correct and fully observed in all material respects”. Given that LCL would have had no obligation to advance funds had the representations and warranties not been true, it logically follows that it was central to the structure of the Facility Letter that they were true, and therefore also follows that LCL had a legitimate interest in them being true and correct in all material respects (the Representations Interest). The analysis is, in my view, different in respect of the provision dealing with warranties subsequently becoming “incorrect and incomplete in any material respect” by reference to circumstances existing at some later point in time. As a simple matter of causation, an event in the future cannot have been the basis of a decision in the past. Belief about what will happen in the future can be a cause of a prior decision, but while the belief may change over time, the belief at the point the decision was made is a question of fact and does not change. Where there is a subsequent change in what is represented or warranted that therefore seems to me to call for somewhat different treatment and is better addressed alongside other issues of security and credit risk.

Third, there are the events of default that relate to the security for the loan. These are clauses 12.2(g) (the security documents become unenforceable); 12.2(j) (it becomes unlawful for the CEK to perform its obligations under a Finance Document, which includes the security agreements) and 12.2(k) (CEK repudiates a Finance Document); 12.2(l) (the secured property is damaged or destroyed and the amount or timing of receipt of any insurance proceeds mean that this will have a material adverse effect on LCL); 12.2(m) (the secured property is subject to a compulsory purchase order); and 12.2(p) (death or incapacity of the guarantors).

This was a secured loan, and accordingly a lender had a legitimate interest in the security being both intact and realisable to meet any unsatisfied obligations of the borrower. As I have noted, in the context of this interest Mr Cowen raised a question about how a borrower was to be expected to prevent compulsory purchase. To my mind that is not quite the point. Compulsory purchase is an instance of the state’s power of eminent domain. Typically, once exercised the owner loses both the capital asset and the income that it can generate. The risk to a lender is therefore that the income that serviced the loan has gone and the security that provided a fallback protection is no longer available. In its place is a sum of compensation that may or may not be sufficient to repay the debt. It seems to me that a secured lender must have a legitimate interest in enforcing obligations for the protection of the security (the Security Interest).

Fourth, clause 12.2(q) makes it an event of default to breach the non-residence requirement or to fail to provide proof of address, which is obviously tied to that requirement. As an unregulated lender LCL was prohibited, under the Financial Services and Markets Act 2000 and the Financial Services and Markets Act 2000 (Regulated Activities) Order 2001 from lending to individuals where the security is their primary residence. Here, the loan was to CEK, a corporate entity. In my First Judgment I concluded that this was sufficient to mean that no interest arose. I was wrong to do so. I had considered that the argument of sham advanced in this case was very weak such that the steps taken by LCL in structuring the loan in the way that they did addressed the issue in any event. That was the wrong lens through which to view the issue. The question, I accept, is properly whether LCL was exposed to risk. As Mr Griffiths’ evidence made clear, they were so exposed and the fact that they took other measures to mitigate that risk goes, if anything, to the question of whether the Default Rate, used in this context, was extortionate. I accept that an unregulated lender has a legitimate interest in seeing a non-residence provision observed (the Non-residence Interest).

Finally, there are those primary obligations that relate to credit risk. This category covers clauses 12.2(c) (CEK defaults on other borrowing), 12.2(e) (enforcement against CEK’s property), 12.2(f) (to which I will return, but which in essence relates to unpaid judgment debts arising from obligations to third parties); and 12.2(j) and (k) (described above, Finance Document also including the Facility Letter). In principle a lender has a legitimate interest in primary obligations that go to preserve a borrower’s ability to repay the debt when due (the Credit Risk Interest).

In light of the Court of Appeal Judgment it seems to me that this is, by far, the most complex of the interests I have identified. I say that because, at least in my view, when some of the cases use the term credit risk they use it to cover two quite different concepts.

One might be described as predictive: given what we know now, what is the likelihood of the borrower (strictly, of a borrower like this one) defaulting on its payment obligations. That is the sense in which I understood the witnesses before me to have used the term, and it is certainly the sense in which I am using it by reference to the primary obligations I identify above in respect of the Credit Risk Interest. It is also the sense in which part of Bryan J’s judgment in Cargill used the term.

The second sense is descriptive: given that the borrower has defaulted, we know they were not good for their debt. I am not suggesting that such a descriptive use of the term “credit risk” is illegitimate, provided that we do in fact know that the borrower was not good for their debt, but one can see immediately that it is different. It is assessed from a different point of time (post-payment default, not pre-payment default), is therefore based on different evidence (most obviously, one knows there was a payment default) and so answers a different question (what happened, rather than what might happen). In each of Lordsvale Finance plc v Bank of Zambia [1996] QB 752, Cargill, insofar as it was referenced in the Court of Appeal Judgment, and Ahuja that is the way that the term is used: we know that a payment default has happened and that tells us about the credit risk of the borrower as regards and only as regards the defaulted obligation.

One can readily see the difficulty in equating the two different concepts by using the notation of probability. As Professor Perry has observed (Perry, “Risk, Harm and Responsibility” in Owen (Ed), “Philosophical Foundations of Tort Law” (Clarendon 1995) “The idea of risk can apply to many different types of harm or bad outcome, but what we mean by risk is fundamentally determined by the understanding of probability we take the concept to presuppose.” He is not there saying that risk and probability are synonymous; as Professor Perry makes clear, risk involves the concepts of both probability and harm. The point is that probability is integral to risk. Classically, in addressing a probability one does so by reference to another variable or variables, so the probability of occurrence x given occurrence y. That is typically abbreviated to p (x|y). To consider p (default on this debt | default on a third party judgment debt) deals with, to a greater or lesser extent, a future uncertainty. By contrast, as a predictive tool p (default on this debt | the same default on this debt) is circular.

The distinction between the two senses of credit risk is, it seems to me, critical in this case, and in particular in the application of the decisions in Cargill, Ahuja and Lordsvale to the facts before me. Specifically, the default trigger in each of those cases was limited to payment defaults. While I fully accept the logic adopted in those decisions, in my view that logic does not translate to this case well, or indeed in some ways at all. It is important to address this in some detail.

The Court of Appeal Judgment dealt with credit risk separately to the Repayment Interest. Paragraph [49] of the Court of Appeal Judgment criticised my First Judgment on the basis that: “[the Judge] does not appear to have taken any real account of Bryan J's conclusion in the Cargill case … that it is self-evident that there is a good commercial justification for charging a higher rate of interest on an advance of money after a default in repayment because a person who has defaulted is, inevitably, a greater credit risk.”

As I have indicated, the Court of Appeal was right to say that I did not rely on Cargill, or on Ahuja or Lordsvale (including the treatment of Lordsvale by Lord Hodge in Makdessi) in relation to what I have now more fully described as the Credit Risk Interest. That was not because I did not consider the conclusions reached there; nor, as I have also indicated, do I disagree with them. On the contrary, I do agree with those decisions and their application, in this case, to the Repayment Interest. As I have also noted, however, that is a separate interest as is apparent from the Court of Appeal Judgment, which deals with it separately at paragraph [52]. It is also an obvious interest, as I have accepted above. The sine qua non of the Loan.

In this case there are other primary obligations the breach of which triggers the Default Rate. In a sense they have a connection with the Repayment Interest but they are obviously free-standing – they can be breached in a way that triggers the application of the Default Rate even in the absence of a payment default. Moreover, some of those interests, specifically what I have described here as the Credit Risk Interest, seemed to me to be weaker than others, and in particular weaker than the Repayment Interest. That, in turn, seemed and still seems to me significant for reasons I have addressed: the effect of paragraph [9] of Makdessi and the cases referred to there by Lords Neuberger and Sumption is that if a clause is a penalty for any reason it is unenforceable for all purposes. As a consequence, it seemed to me at the time of my First Judgment that it was unnecessary to address the stronger interests, like the Repayment Interest, if the Default Rate was a penalty by reference to some other interest.

The difficulty I had, and still have, is how the descriptive use of credit risk is helpful, or indeed applicable at all, to a situation involving the Credit Risk Interest. Put another way, why do the descriptive observations in, say, Cargill inform the predictive task that the Credit Risk Interest requires me, in my view, to undertake? While the same or very similar issues arise in seeking to apply each of Lordsvale, Cargill and Ahuja to this case, given the terms of the Court of Appeal Judgment I start with Cargill and, so far as it forms the basis of Bryan J’s reasoning there, with Lordsvale

The application before Bryan J arose out of an earlier application for summary judgment before Teare J, reported at [2018] EWHC 2977 (Comm), to which Bryan J expressly referred for its findings of fact. Summarising those findings, Cargill sought summary judgment against Uttam for US$61.8 million plus interest under two Advance Payment and Steel Supply Agreements (the APSAs). The APSAs allowed Uttam to call upon Cargill to make advance payments of the purchase price of steel. Cargill had an option as to whether or not it made payment, but if it did so Uttam was then obliged either to supply steel (to Cargill or to an alternative buyer designated by Cargill) to that value or to refund the price. There were two APSAs in issue and evidence to suggest that the parties had been dealing with one another on similar terms for a number of years. Uttam stated it was over a decade, and no issue seems to have been taken with that.

In 2015 Uttam submitted six notices of drawdown representing the full facility under the two APSAs of US$61.8 million. The repayment dates were spread over time; no repayment was made on any of them. An email from Uttam to Cargill stated that the non-payment was a result of “huge cash flow pressure” caused by a downturn in the steel industry and increased production capacity in China. Uttam therefore requested that Cargill “rollover [sic] its amount due of US$61.8 million”. Cargill did not do so.

A number of defences to payment were subsequently advanced by Uttam, which were dismissed by Teare J who granted summary judgment. A dispute arose as to whether the default interest provision in Cargill’s favour, clause 8.12, was a penalty. That provision stated, so far as is relevant:

I pause to note two features about that provision which seem to me significant. First, a point I have already made, it is only triggered by non-payment of a sum due. That is very different to the Default Rate in this case, which can be triggered by a variety of things ranging from compulsory purchase orders to allegations of non-payment of sums due to third parties. Second, the default compensation was only payable on the overdue sum; it did not apply to any non-defaulted sums. Again, that is different to this case where there could be a non-payment default that would cause the Default Rate to apply to the repayment amount, even where the Loan was not yet due for repayment. Indeed, as I have noted, that was the position LCL adopted in respect of the alleged breach of the non-residence provision when Gunnercooke wrote to CEK on 8 September 2020.

There was insufficient time to determine the penalty question at the hearing before Teare J and it was argued at the later application before Bryan J. He found that the provision was not a penalty. It was against that backdrop that he made his observation that “a person who has defaulted is, inevitably, a greater credit risk” referred to at paragraph [49] of the Court of Appeal Judgment.

Seeking to tie Cargill to the Credit Risk Interest in this case is, in my view, problematic on a number of levels.

First, I do not consider the facts of Cargill to be analogous to this case. The question of when two cases are properly analogous has, of course, generated considerable academic debate. I will address, as precisely as I can, the more specific bases for distinction below because the detail seems to me important, but even if one takes this simply as a matter of impression the facts here are fundamentally different to those in Cargill.

That case involved the forward sale of steel between two parties with an established trading history of over a decade in sums totalling US$61.8 million over a series two contracts involving six trades. Repayment could easily have come from other sources open to Uttam, and indeed might not have involved repayment at all but, rather, the delivery of steel. By contrast, this was a one-off, short-term loan from a specialist lender in bridging finance secured on a, with all respect to the Claimants, relatively small portfolio of residential investment properties and the family home. Like some credit themed uroboros – the serpent that consumes its own tail – repayment was always going to be by way of another refinancing secured on the various properties.

Significantly, in Cargill the default could only ever be non-payment because repayment was the only primary obligation protected by the default rate provision. Non-payment is now the relevant default before me, but because the time for assessment is at formation of the agreement that does not matter. Even if hindsight were in some way relevant, at trial the focus was the Non-residence Interest, which was the primary obligation whose alleged breach was said to have triggered LCL’s right to charge the Default Rate.

The starkness of the distinction between Cargill and this case, in broad terms, can be seen from the fact that two of the claims originally before me were consumer claims. Those claims either collapsed or were unsuccessful, of course, but they were felt by the Claimants to be arguable and LCL did not apply for summary judgment or to strike them out. It is hard to see how a forward sale agreement of US$61.8 million of steel would ever reach that quite low bar: if such an agreement were pleaded as a consumer contract, it is inconceivable to me that a properly advised defendant would ever allow it to reach trial.

For completeness I should address one point of apparent distinction between this case and Cargill that seems to me not to be relevant. Bryan J was invited to compare the default interest rate under the APSAs and those that applied to other borrowing of Uttam. He declined to do so on the ground that it was “not a relevant comparator” (paragraph [64]). The latter loan was secured and backed by a personal guarantee such that:

While that was the case on the facts before Bryan J, it was not the evidence before me. The Loan was both secured and backed by a guarantee, but for reasons I will come to deal with the realisation of that security was not straightforward. It therefore seems to me that the existence of that security is not such a ground of difference between this case and Cargill as one might in principle expect.

As I noted above, the detail of the distinction seems to me important, and a more reliable guide than first impressions. My second reservation is how one translates the analysis of risk in Cargill to this case. Specifically, it seems to me difficult to apply the descriptive observation of Bryan J that“a person who has defaulted is, inevitably, a greater credit risk” to the predictive task that is necessary in assessing the Credit Risk Interest. This, I recognise, requires some developing.

Risk normally deals with uncertainty. As Lady Hale observed in Sienkiewicz v Greif (UK) Ltd [2011] UKSC 10 at [170]: “Risk is a forward looking concept – what are the chances that I will get a particular disease in the future?” As the quote itself makes clear, Lady Hale was dealing with a different context (the risk of developing mesothelioma following exposure to asbestos dust) but her observation on the concept being forward-looking is not context specific. There is also some nuance, in that one can meaningfully talk about risk of a past event to reflect epistemic uncertainty (for example, when we talk about the risk of somebody having missed their train at some point after the train has departed – they either have or they have not, but given the lack of knowledge it is appropriate to talk about the event in terms of risk). Typically, however, the uncertainty that we are discussing when we discuss risk is future uncertainty. That was certainly the case in Sienkiewicz, and in my view it is the only relevant use of the term in the context of the Credit Risk Interest.

Risk is not the same as cause. The distinction is explained by Professor Turton (“Risk and the damage requirement in negligence liability” (2015) 35 Legal Stud 75):

I have already addressed the link between probability and risk. As Professor Perry notes in his 1995 essay, the class of probability theories that seems most appropriate in private law is what are known as relative frequency theories: one identifies a reference class of sufficiently similar events in the past, identifies from that (so far as one can) the incidence of loss within that reference class and projects that forward, using inductive reasoning, to predict the likelihood of such events happening again in the future. Professor Perry was speaking in the context of tort law but nothing seems to turn on that.

The inductive logic on which relative frequency is premised is central to those events of default that represent the Credit Risk Interest. Put simply, LCL plainly has a very direct interest in the Loan being repaid – the Repayment Interest. But why should it care if third parties are repaid, even on judgment debts? All the more so where there is simply an allegation by a third party of default by CEK on sums that it owes – what business is that of LCL’s? The answer is that it tells LCL something about the probability of its Loan being repaid in due course: if CEK is defaulting on other debts, or otherwise breaching those primary obligations that relate to the Credit Risk Interest, it may be more likely in due course to default on the Loan. The strength of that link – between other obligations to other parties and default on the Loan – is the yardstick against which the question of whether the Default Rate is extortionate falls to be assessed in the context of the Credit Risk Interest.

The problem with using this aspect of Cargill in the process is that such inductive logic is not, in my view, what Bryan J was relying on when he talked about credit risk. The reasoning in Cargill on why it was legitimate to charge higher rates to a party in default referenced the decision of Colman J in Lordsvale. I will return to consider Lordsvale further, but for now what matters is the reasoning adopted by Bryan J, which is at page 763E-F of Colman J’s judgment:

Credit risk in that context is not in any sense being used predictively. Probability is not needed to predict the prospect of default on the sum to which the higher rate applies – the default has already happened. Mr Wheeler observed that there might be said to be a higher credit risk in the sense that one did not know when the default would be cured, as to which there was uncertainty. I see that argument but I do not believe that is what Colman J had in mind, because the structure of what he proposed meant that the increased interest rate could be justified for, and only for, the period of default: “The additional amount payable is … directly proportional to the period of time during which the default in payment continues.” It is a response to something that has happened and is ongoing, not an attempt to predict and allow for an uncertain future.

Moreover, were risk being used predictively one would expect there to be some knock-on impact in respect of future instalments, because the predictive credit risk in respect of them is presumably also now higher. But that is plainly, from the language I have just quoted, not what Colman J had in mind: the enhanced rate was justified only for the defaulted sums, and only for the duration of the default.

When Lord Hodge (paragraph [222]) and Lord Mance (paragraph [148]) in Makdessi agreed with the reasoning in Lordsvale they treated credit risk in the same way as Colman J – descriptively, not predictively. Lord Mance referred to a “commercial re-rating to reflect the change in risk”; and Lord Hodge stressed that he regarded Lordsvale as correct because “a default affected the credit risk that the lender undertook”. As a narrative description of what happened in respect of the borrower in question, which was the issue that Lords Hodge and Mance were addressing, it is of course right to say that the loan to the borrower had come to be a bad debt.

That is not the same thing as saying that the risk prediction made at the time that the loan was taken out was wrong. As I have noted, when risk is being used predictively it is often addressed through use of a reference class. The class remains central throughout the process. Professor Perry makes the point in his 1995 essay, but makes it particularly clearly in a later essay (Perry, “Torts, Rights, and Risk” in Oberdiek (Ed), “Philosophical Foundations of Tort Law” (Oxford University Press 2014)):

In assessing whether the prediction of risk was accurate, one does so in respect of the class as a whole. If 2% of loans across a book of 50 loans are predicted to default one would consider it unusual for any particular loan to be the defaulted obligation. But the prediction would still be accurate if the other 49 did not default. Using risk predictively, when a loan defaults it does not show that the credit risk prediction was necessarily wrong. Nor has the that risk necessarily changed. Certainly, that may be the case. As Professor Perry notes, in his explanation of epistemic risk in his 1995 essay, whenever we talk about relative frequency probability (and, in turn, risk) we are really talking about an estimate of the relative frequency and that estimate can improve with more data. But even where the estimate is right, on a 2% risk of default one in 50 borrowers will not repay. Pointing to that one in 50 who does not to prove the risk prediction was inaccurate when made would wrongly ignore all the other loans to which that prediction equally applied.

One sees the same points in Cargill. There was no uncertainty at the point that default interest started to be charged as to whether default was likely or unlikely to happen in respect of the payment in question; it had happened. There remained uncertainty around default in the sense that there were multiple repayments due under the APSAs, so at the time of the initial defaults it was uncertain whether the sums due later in time would also be the subject of default. The credit risk to which Bryan J was referring in this section of his judgment did not apply to them, however: the default interest rate in Cargill, just like in Lordsvale, only applied to the overdue payments. Of course, one would expect the first payment default to affect the (predictive) credit risk of Uttam in respect of all repayments, whether in default or not. It was never suggested by Bryan J that there was any logical incoherence in treating payments in default differently from other payments, however. Had he been using risk predictively in observing that “a person who has defaulted is, inevitably, a greater credit risk”, that distinction could not logically have been sustained.

That is in contrast to the way that Bryan J treated the term “credit risk” at paragraph [55] of Cargill. There he was considering how the market factors in the Indian steel industry at the time the APSAs were entered into might affect Uttam’s probability of later default. His conclusion that “companies like Uttam were a greater credit risk and their cost of finance increased” was forward looking and based on inductive reasoning. His description of “companies like Uttam” is the description of a reference class; it was that class that represented a higher risk. Because there was a higher probability (but still not a certainty) of default, that merited a higher interest rate to reflect the increased probability / risk for all companies in that reference class, hence the use of “their cost of financing”, not simply Uttam’s cost of financing. As Teare J found, it was that risk which subsequently matured into the (or at least a) cause of default in the particular case of Uttam, but even had that not been the case those market conditions would still have been a risk and would still have merited a higher interest charge.

The difficulty I had, and confess still have, in applying the approach adopted by Bryan J and Colman J to the facts of this case is that it only seems to me to work in the context of a known payment default. Put simply, the legitimacy of using “credit risk” descriptively depends on the description being accurate. As I have noted, it is a common feature of Cargill and Lordsvale, and for that matter Ahuja, that the higher interest rates were triggered by, and only by, payment default and applied to the defaulted sum. I accept that it does necessarily follow that if the debtor has failed to repay a debt on time that does show they are not good for that debt. Indeed, I would go further and accept that there must have been a risk, that is a predictive credit risk, that they would so default at some time between the inception of the loan and the default occurring. That seems, to me, to be Professor Turton’s point. As a matter of practice it is true that lenders charge more to borrowers they consider to have a higher credit risk, as the evidence of Mr Theophanous and Mr Griffiths in this case demonstrated. Finally, I do not need hindsight to establish any of this; where the default rate only applies to payment defaults it follows that the only context in which that higher interest rate falls to be considered as a penalty will be where there was non-payment. I know that simply from the structure of the clauses in Lordsvale, Cargill and Ahuja.

Of course, that logic says nothing about the precise scale of the risk. Both low probability and high probability risks can result in a loss, it is simply that the latter do so more frequently than the former. That is the essence of relative frequency probability. Looking backwards, from the point of loss, can obscure that important fact. It also says nothing about which, of several credit risks, caused the default. For reasons I go on to address, that is significant in this case.

What is particularly important at this stage of the analysis, however, is that whereas in Lordsvale, Cargill and Ahuja default necessarily did mean that the borrower was not good for their debt, here, at least as regards the Credit Risk Interest, that is not the case. The difficulty can be illustrated by returning to an example I have touched on from the Facility Letter.

It is an event of default under clause 12.2(f) if an unappealable judgment against CEK for a sum of not less than £20,000 is not satisfied within three Business Days. On the fourth business day after judgment, therefore, the Default Rate would apply to the outstanding amount of the Loan pursuant to clause 6.6(i). At that point, however, there has been no payment default in respect of the Loan. And there may never be such a default – CEK might repay the Loan in full on the Repayment Date. So unlike Bryan J, I cannot point to a non-payment to show that CEK has defaulted on the Loan, nor can I do so to show that there must have been a credit risk in the first place that it would do so. Less still can I say that the rendering of the £20,000 judgment was the (or even a) risk that matured into a cause of an as yet non-existent payment default on the Loan. Doubtless there were credit risks that applied to CEK in respect of the Loan, and the rendering of such a judgment might well have been one of them, but I cannot use payment default on the Loan to prove that if no such payment default has happened. I have to ascertain the risk in the normal, inductive, forward looking way.

This is not simply a philosophical diversion: maintaining conceptual clarity between the different interests that I have set out above seems, at least to me, to be highly desirable when it comes to applying the legal test. As Makdessi makes clear, (i) I must carry out the analysis at the point of formation of the agreement, when it is not known which primary obligation or obligations will be breached; (ii) where the same secondary obligation protects diverse primary obligations representing different legitimate interests in the same way that remains an indicia of penalty (a point also made in Vivienne Westwood); and (iii) if by reference to any primary obligation the provision is a penalty, it is unenforceable in its entirety.

To say that there would be a self-evident, inevitable link between payment default in respect of the Loan and the descriptive credit risk of CEK in respect of the Loan therefore seems, to me, to kick the intellectual can down the road in this case. As I have said, I accept the proposition, that if someone has defaulted under a loan that shows that there must have been a credit risk associated with them in respect of that obligation. I also accept that the increased likelihood of default may merit charging a higher, possibly a much higher, rate of interest. That does not answer the question of whether a much smaller judgment debt due to a third party under a different agreement which has not (at the time the Facility Letter was entered into, which is the relevant time for assessment) and may never cause a default under the Loan merits the same treatment.

In applying the Makdessi analysis I find it clearer to treat the Repayment Interest, which as the Court of Appeal notes is obviously a strong legitimate interest, separately to what I have described as the Credit Risk Interest, which includes what are on their face (but as I address in due course, not in substance in the context of this structure) lesser defaults on such things as obligations to third parties. The two sets of interests are essentially different in nature, and Makdessi and Vivienne Westwood require that to be addressed. For reasons I will go on to consider, while the Repayment Interest has always been quite straightforward, as it was for Bryan J in Cargill, the Credit Risk Interest is not. Bryan J was not addressing, because in this part of his judgment he did not need to address, predictive credit risk. It is the robustness or otherwise of that predictive process that makes the Credit Risk Interest complex here. What is said in Cargill (and, by extension, Lordsvale and Ahuja) does not deal with that process and so does not assist in resolving that complexity.

This leads to my second reservation. If one attempts to use Bryan J’s point outside the context in which it is made, that is if one assumes there is a link between past payment default and credit risk more generally, one overlooks critical aspects of the causal inquiry linking the two.

As I have noted, the link between default and descriptive credit risk made by Bryan J is straightforward: a person who has defaulted on a debt is, by definition, someone who did not pay that debt when it was due and so, again by definition, has bad credit in respect of it. In that sense the link is inevitable: the different propositions essentially describe the same thing, and where one is describing the same thing in different ways there is no need for some causal inquiry to link the different iterations. The same would equally be true here: when, on the Repayment Date, CEK defaulted and failed to repay the sums due under the Facility Letter it represented a bad debt in relation to that payment obligation, and in general terms we could describe that as representing a credit risk as regards that obligation.

Some, but not all, aspects of the Credit Risk Interest would also involve a payment default, albeit on other debts. Most obviously, the example I have referred to of clause 12.2(f) involves an event of default on the Loan if CEK had not paid an unappealable judgment within three business days. The difficulty in translating across Bryan J’s comment to these examples of payment default is that I do not accept the factual proposition that the link between payment default on other debts and future payment default on the Loan, which is fundamental to using credit risk in its predictive sense, is “inevitable” in this context, at least to the extent that term is taken to mean that the association will arise in all cases. Neither did Bryan J. As I have noted, he did not use his finding on the earlier defaulted payments to justify the higher interest rate on later payments, even though both obligations arose under the same or linked agreements. Nor, as I have addressed, did Colman J in Lordsvale. On the contrary, he only saw the increased rate as being appropriate on the defaulted sum while it was in default.

Indeed, far from the link being inevitable, it seems to me that were I to try to apply what Bryan J said outside of its specific, descriptive context I would risk falling prey to two, somewhat linked, errors of reasoning.

The first is the concept of a confounding factor. I accept that confounding is itself a vexed topic, on which experts often disagree, but the general principle is, I think, uncontroversial and indeed obvious. It is neatly summarised by Pearl, Glymour and Jewell (“Causal Inference in Statistics: A Primer” (Wiley, 2016)):

On the facts found by Teare J and Bryan J there appears to have been a similar confounding factor in Cargill. I have already addressed Bryan J’s observations at paragraph [55] of his judgment: in 2015, when the APSAs were entered into, it was known that “the Indian steel industry generally was experiencing difficulties” because “the market was in decline at the relevant time” such that “companies like Uttam were a greater credit risk and their cost of finance increased.”. As I have observed, in that part of his judgment Bryan J was using credit risk predictively and I respectfully agree with his approach of identifying relevant confounding factors that might impinge on the causal analysis.

The second pitfall in reasoning is the availability heuristic: because we can readily call to mind examples where past default is caused by something that will equally cause future default, we assume that this is always the case. But it is equally possible to think of cases where past default will tell us nothing about future default: the payer considered in good faith that they had a valid reason not to pay, the failure to pay was down to a cyber-attack at the payer’s bank, or the payer was suddenly incapacitated or distracted by an unexpected and traumatic event, such as a death in the family.

What is critical, then, is to identify the causal factor and ask what the probability is of it applying going forward. Typically when one is using risk to predict future default, as is the case with the Credit Risk Interest, it is not the past default that causes the future default. To the extent that the two are linked at all, it is more likely to be by reference to the factors that led to the past default. The distinction is significant, as can be illustrated with an example. Assume that debtors A and B owe two separate debts to creditor C, with a repayment instalment of £10,000 due on each of the debts on the same day. On the repayment date debtor A pays £9,000 and debtor B pays nothing. If we limit our analysis to saying that past default indicates future default, the debtor B non-payment might seem more significant, given that the magnitude of the default is greater. To stop there, however, would have parity of reasoning with governments banning the sale of ice cream in order to reduce violent crime, to return to the example of Pearl et al. We need to know what caused the non-payment in each case. If the non-payment by B is due to a dispute over whether any sums were due in respect of that repayment instalment or when while the shortfall from A is simply down to the debtor running out of money the complexion changes significantly. Now, the non-payment by B may tell us little or nothing about future credit risk, whereas the non-payment by A tells us a great deal despite its smaller magnitude: it is the sound of the falling stone that goes on to start the avalanche. The proper causal analysis is critical.

Such issues arose here even in the context of the Repayment Obligation. I ultimately found that CEK did default on the Loan, but what did that default tell me about the credit risk of CEK? At the time of the trial it was alleged that the Facility Letter and the supporting security were consumer contracts that were, by statute, unenforceable against the Claimants and that Mrs Houssein had, in any event, been induced to enter into the suite of finance agreements by undue influence exercised over her by Mr Houssein. I rejected those claims, but there is no suggestion that they were advanced in bad faith; the Claimants genuinely believed they were not obliged to pay anything. In such circumstances, past non-payment is not necessarily a good predictor of future non-payment. Likewise, disruption was caused by the death of Mr Houssein during the term of the Loan and the delays in obtaining probate over his estate. That, too, could result in a default that would say nothing about future creditworthiness. By contrast, Bryan J in Cargill was dealing with a debtor who had confessed cashflow issues that had been predictable from the inception of the APSAs. As Bryan J recognised when he was using credit risk in its predictive sense, such issues would compromise Uttam’s ability to pay any of its obligations.

Of course, the Credit Risk Interest looks well beyond the Loan itself: to return to my example, it can be triggered by any unpaid judgment debt in excess of £20,000. Showing what a default on those debts means for the Loan requires a fact specific analysis. It would include all of the factors that I have listed above. There may be other reasons that were specific to the particular contracts or obligations in issue. Certainly, I do not see how credit risk in respect of the Loan, used predictively, can be said to be “inevitably” increased by non-payment on a different obligation.

That approach of focussing on causal factors in assessing predictive risk is, it seems to me supported by the expert evidence in this case. Specifically, Mr Griffiths addressed how a lender would react to this being a third bridging loan for the Claimants, which I understood would typically be considered a risk factor:

Mr Griffiths was talking about a risk factor different to payment default, but nothing in his answer suggested to me that there would be an inevitable link between default on other obligations and likely default on the loan being considered. It seemed to me that he was speaking more generally.

I recognise that Mr Kyriacou was more cautious about how relevant such factors would be:

I took that to mean he was wary, even sceptical about the causal narratives advanced in some cases. I did not take him to be saying that a causal analysis was irrelevant.

Again, Makdessi makes it critical to carry out the causal analysis at the right point in time, which is the point when the agreement was entered into, not the time of breach and certainly not the time of trial. At the date the Facility Letter was agreed there were always potential reasons why payment default even on the Loan itself would not be a good or even any predictor of further default:

And again, the position becomes even more acute when one looks at other debts of the borrower for the reasons I addressed above.

In my view, at the point when the Facility Letter was executed any third-party onlooker considering this case objectively would see that there could be instances of non-payment, including non-payment under the Loan, that would tell you little or nothing about the risk of further default. There were, of course, other potential causes of non-payment that would. It is important to sift the wheat from the chaff.

This lengthy exegesis of causation may seem obvious, and I confess that I had assumed that it was when I prepared my First Judgment. As the Court of Appeal Judgment highlighted, that led to a lack of clarity, an error I would not want to repeat. In my view this case differs from Cargill in that Bryan J was using credit risk, in that part of his judgment, descriptively and not predictively, such that no causal analysis needed to be carried out to make the “inevitable” link that he did. If one is to use credit risk predictively, one must carry out such an analysis. While I accept that past default will correlate to future default, because causal factors like cashflow will apply to both, I do not consider that the correlation will be “inevitable”, and nor do I consider that the very different way in which “credit risk” was being used in Cargill drives me to that conclusion.

My third reservation about applying Cargill is a more straightforward one and goes to the magnitude of risk. I have already noted the link between risk and probability. Most, if not all, theories of probability measure it between 0 and 1, with 0 meaning impossibility and 1 meaning certainty. As I have indicated, if one is using credit risk predictively, which must be the case for the Credit Risk Interest, that involves an inductive approach: what do those other breaches tell us about likely payment default on the Loan? Where, between 0 and 1, do we think we land?

Answering that question involves revisiting the point made above from Professor Perry’s 2014 essay: probability and risk, in a relative frequency sense, are properties of the reference class, rather than the individual who happens to fall into it; and individuals will fall within multiple reference classes and so have multiple different probabilities associated with them.

One sees the issue here. Assume parties A, B and C all have loans with LCL on terms identical to the Facility Letter save that repayment is in instalments. Party A has defaulted on an instalment. Party B has not but is the subject of an unappealable judgment for over £20,000 in respect of other (non-LCL) borrowing. Party C has also made all payments to date but has received a letter of claim in the same amount. If the reference class is “defaulters” then the relative frequency probability for A, B and C would be the same – one looks generically at how often past default of any kind is followed by a subsequent payment default and that is the relative frequency. If the reference classes are “payment defaulters” and “non-payment defaulters” A now falls into a separate reference class to B and C and it seems likely that the probabilities will be different for the two classes. The question is ultimately a statistical one and no such statistics were before me, but as Professor Perry notes in his 1995 essay that is often the case, and does not preclude an inductive approach. Intuitively, one would expect past payment default (used as a proxy for the causal factors that underpinned it) to be the better predictor of future payment default. One would further expect that someone who fails to pay other debts would be more likely to default on payment of this one than someone who is simply alleged not to have paid their other debts, so in fact we would expect B and C to be in three reference classes: defaulters (which they share with A), non-payment defaulters (for B and C), judgment debtors on third party debts (for B alone) and alleged but unproven defaulters on third party debts (for C). And we would expect the relative frequency probabilities of future default to be different for each of those different classes.

There is an important gloss to the reference class that is addressed by Professor Oberdiek (“Imposing Risk: A Normative Framework” (Oxford University Press 2017)). It might be assumed from what I have just said that the narrower reference class will always be the more informative. That is not always the case because, as Professor Oberdiek observes: “wider reference classes are sometimes preferable to narrower ones because wider reference classes can capture causally relevant factors that are excluded from narrower classes.” I will return to this when I come to address the interest cover ratio and the prospects of refinancing in this case.

Again, the existence of different reference classes distinguishes this case from Cargill or Lordsvale, even had those cases been seeking to use risk predictively. One can accept that a party that defaults on one instalment under a loan is more likely to default in respect of a subsequent instalment of the same loan. As I say, the question is really a statistical one but even in the absence of statistical evidence it makes sense intuitively, and if a statistical association were shown to exist we would not struggle to develop a causal narrative that supported it. The party may have defaulted because they have no money or because they regarded the loan as invalid or because they felt they had a valid defence, for example a condition precedent to repayment that was unsatisfied. All of those causal factors would apply equally to all instalments.

One might be more cautious about saying the same thing about the Party B scenario I describe above. If the default on the other facility were due to credit issues then that likely would apply equally to the loan we are considering, but if it were down to an alleged invalidity of the other loan there would be no reason to assume the same would be asserted in this case. We would want to know more about the causal narrative at the very least but we would expect the probability to be lower, that is we would expect a weaker correlation. The Party C scenario is even further removed. What does the receipt of a letter of claim tell us about the borrower’s ability to pay?

That was the question I found myself unable satisfactorily to answer in my First Judgment. The structure of the Facility Letter, under which the Default Rate is triggered by very different primary obligations, means that CEK can fall into something resembling the A, B and C reference classes simultaneously. Breach of some of the obligations might say much more about the likelihood of further default than would the breach of others. In particular, it seemed to me that factors relating to third party liabilities – what I have identified here as the Credit Risk Interest – told me relatively little about the risk of future default on the repayment obligation under the Loan. The evidence suggested that such defaults merited an increase in the monthly interest rate of around 0.3% applied to the Standard Rate by Mr Theophanous to cover the Claimants’ increased risk, not the 3% of the Default Rate. Mr Griffiths thought that even that was too much. Accordingly, at the very least, the proposition that third party default represented such a significant credit risk that a 3% increase was not extortionate lacked face validity based on the evidence I had at the time. Cargill, which related only to payment default, did not help to resolve that issue.

As I go on to address, my view on that point has changed as a result of the new evidence at this hearing and what that says about the evidence before me at trial. I also benefitted greatly from the insight and pellucid clarity of Mr Wheeler’s submissions. In saying that I take nothing away from Mr Cowen, who is a rigorous and thoughtful advocate, or Mr Blakeney, who I recognise marshalled the factual and legal points admirably. As will already be apparent, on a number of points I have accepted the case they advanced. I also mean no criticism of Mr Whitfield, who acted for LCL at the first hearing. It is simply that Mr Wheeler’s crystalline submissions distilled the evidence from both hearings and tied it, rigorously and remorselessly, to the legal tests. Combined with the new evidence, they changed my view of the penalty aspects of this case for reasons I shall go on to address. The analysis is a factual one, however; it does not depend on an application of Cargill.

My final reservation about trying to apply Cargill touches on a point I have just made: Bryan J had the benefit of evidence that was not before me. At paragraph [52] he explained:

That, of course, was not this case at trial at all. As I have already addressed, Mr Theophanous’ evidence was that the Default Rate was set centrally, without reference to the creditworthiness of the particular risk or the individual borrower. The Default Rate applied to a range of defaults, not just payment default.

Bryan J continued at paragraph [72]:

Bryan J emphasised the importance of such evidence at paragraph [70], when considering the distinction between the earlier decisions of Davenham Trust Plc v Homegold [2009] 4 WLUK 368 and White v Davenham Trust Ltd [2010] EWHC 2748 (Ch).

As I noted in my First Judgment, that explanation is precisely what I did not have. I knew that the Standard Rate was increased to reflect the Houssein’s prior defaults; I knew that the Default Rate was set centrally and applied to all borrowers; I did not have, to use the words of Bryan J, “the basis and rationale” for that decision. On one level I accept that I still do not have it now. What I do have is the evidence of Mr Griffiths around how lenders treat credit risk and, critically, more information about the portfolio of properties used to secure the Loan and the issues around refinancing them. That is still not the evidence that Bryan J had in Cargill, but for reasons I go on to address in my view it allows me to understand how a reasonable lender in LCL’s position would have considered the Credit Risk Interest.

In summary, I accept the proposition that past payment default is likely to correlate with future payment default on the same or related loans and so be some evidence of credit risk in its predictive sense. That was not the point being made in Cargill, at least in that the default rate there applied only to those repayment instalments in default, not to all outstanding sums. It is irrelevant here because the repayment was structured as a single bullet repayment, so by definition there could not be default on an instalment.

I also accept the proposition in Cargill that default on a payment obligation proves that the debtor’s credit is impaired, and impaired credit can properly result in a higher rate of interest. In my view that is limited to credit risk in its definitional sense and properly falls to be considered in connection with the Repayment Interest.

I do not regard that as the most important question, however. Applying Makdessi, the whole clause becomes unenforceable if the Default Rate responds to the breach of any primary obligation in a way that is extortionate. Once Bryan J had dealt with the repayment interest in Cargill he had dealt with everything. I am not so fortunate. The Default Rate here responds to a variety of defaults and so a variety of interests; if it is extortionate in respect of any of them, the clause fails in respect of all of them. Specifically, in respect of the Credit Risk Interest the question is the strength of the association between defaults in respect of other obligations and likely default in respect of the Loan. Cargill, in my view, simply does not address that situation because Bryan did not need to do so.

Ahuja was a complicated case giving rise to a number of issues. Happily, many of them are not relevant for these purposes. The penalty issue centred on a loan agreement for £800,000 provided as part of a suite of agreements in connection with the purchase of a shopping centre. Unlike Cargill, but like this case, the loan was secured and supported by personal guarantees. The relevant provisions were summarised by HHJ Hodge QC at paragraph [122]:

Ahuja therefore involved a secured loan with personal guarantees, which makes it more like this case, but the increased interest rate was still only tied to payment default. As with Cargill, therefore, HHJ Hodge QC’s comments as to credit risk must be read in that context. For very similar reasons to those I have addressed at length in respect of Cargill I therefore do not consider that the link made between payment default and credit risk at paragraph [143] of Ahuja helps with answering the question that is central here: was the Default Rate extortionate by reference to the legitimate interests other than the Repayment Interest?

Finally I return to Lordsvale. The case involved two oil import facility agreements totalling US$230 million. The facility advances fell due for repayment but the borrower defaulted. The principal claimed was a little over $5.6 million. Again, it merits quoting the default interest provision:

Before addressing the provision as it was drafted, Colman J posited a counter-factual – what would the position be “if a loan agreement were to provide that upon the happening of a default in payment by the borrower the rate of interest were to be increased with retrospective effect” (page 763B)? He noted at page 763C that one consequence would be that “the amount of interest payable would be unrelated to the extent of default.” He concluded at page 763D-E, in respect of his counterfactual: “Such a provision would therefore have all the indicia of a penalty.” I have addressed elsewhere why I consider that it is necessary to consider each of the legitimate interests identified and for the Default Rate not to be extortionate in respect of any of them. I did not refer to this discussion in that context. While, in my view, it supports the conclusion I have reached it was not referred to by the Supreme Court in Makdessi even though other aspects of Lordsvale were. It therefore seems safer to me to proceed from Makdessi itself. What it does show is that in his discussion on credit risk Colman J was not referring to a case where “the amount of interest which would be payable would be unrelated to the extent of default”. By contrast, such an apparent discrepancy, between the Credit Risk Interest and the Default Rate, was something that troubled me in my First Judgment.

Colman J then turned to the situation where the rate of interest was to increase prospectively from the time of default. Then, he considered (at page 763E) “a rather different picture emerges.” In particular, “the borrower in default is not the same credit risk as the prospective borrower with whom the loan agreement was first negotiated.” This change in credit risk would, in his view, provide a proper basis for a “small rateable increase” in the interest rate (page 763F-G).

As I have noted, his references to credit risk are focussed on a situation where there has been a payment default and the increase in interest relates to and only to that default. For the reasons I have given, I do not find that helpful here simply because the facts of this case are quite different.

I am conscious that I was able to articulate my reasoning in respect of the other four interests in five paragraphs and that this has taken significantly longer. That is because it is my view that there is a very real concern about “double counting” credit risk on the facts of this case. If one treats payment default on the Loan as going both to the Repayment Interest and the Credit Risk Interest then one conflates the backward-looking questions of default (put another way, descriptive credit risk) and cause with the forward-looking question of risk (that is, predictive credit risk). That is entirely fair in circumstances where the only trigger to default interest is payment default, in particular where that default has already happened; it seems to me that it would not be correct here, however, when things that might never lead to a payment default would still trigger the Default Rate. Each primary obligation that triggers the Default Rate must stand or fall on its own merits; it cannot piggyback on a potential payment default that, at the time of assessment (that is, the date of entry into the agreement) had not happened and might never happen. Any other approach would, in my view, not be a proper application of Makdessi and would unjustifiably prejudice the Claimants.

The Credit Risk Interest accordingly is a legitimate interest of LCL and each of the primary obligations I have identified in respect of it seem, to me, to address some aspect of the predictive credit risk associated with the Loan. But under the terms of the Facility Letter it is structurally separate to, and so falls to be assessed separately from, the Repayment Interest. That assessment is properly something to be addressed in considering the third Vivienne Westwood question.