Quantification of ISDS Claims: Specific Issues
As discussed in the previous chapter, the relevant standard to assess damages arising from investment treaty violations is usually that of ‘full reparation’. This is often defined by reference to the Factory at Chorzów case, which says that ‘reparation must, as far as possible, wipe out all consequences of the illegal act and re-establish the situation which would, in all probability have existed if that act had not been committed’.
The full reparation standard gives rise to two main questions. The first one relates to the need to compensate for ‘all consequences’ and identify ‘the situation which would, in all probability have existed if that act had not been committed’. To answer this, the expert will often have to identify a causal link between the alleged breaches and the claimed heads of damages. This may give rise to a debate around the level of certainty required for a tribunal to accept that certain heads of damages should be considered as part of ‘all consequences’. Expert (and factual) evidence can help the tribunal assess both the causation and the level of certainty for the different heads of damages. However, the cut-off level of uncertainty that would disqualify a claimed consequence from meriting compensation is a question of law, and a matter of judgment by the tribunal.
The second question also relates to identifying ‘the situation which would, in all probability have existed if that act had not been committed’. A discounted cash flow model can be used to determine a ‘financial value’ or ‘amount’ corresponding to this situation (as well as a value or amount for the situation that actually existed (i.e., the actual scenario)). However, that leaves open one important question: what date should be considered in making this determination? In other words, what is the appropriate ‘date of assessment’? Does one need to restore the situation that would have existed at the date of the alleged treaty breach, at the date of award or at some other point in time? To our understanding, the applicable standards under public international law do not identify this date, and in our experience different approaches have been adopted by different tribunals.
In this chapter, we first discuss the choice of the date of assessment, focusing on two possible approaches: ex post and ex ante. We then address a further question that follows on naturally and inevitably from this discussion. Because losses pre-date the award, the injured party will have waited to receive compensation. Under both approaches, it is normally accepted that they should be compensated for that delay by adding on interest. We therefore end this chapter with a discussion of the appropriate interest rate to use to bring historical damages (where applicable) forward to the date of award.
Date of assessment and the use of hindsight in damages estimation
The standard of reparation from Factory at Chorzów seeks to leave the injured party in the same position as if no breach had occurred. Translating this into economic terms, and using the language favoured by economists, that implies that a monetary award based on this standard should make the injured party indifferent between the situation where there was a breach and that where there was no breach. However, this leaves open the question of at which date the party should be indifferent.
Suppose, for example, that the alleged breach involved the expropriation of oil and gas assets, and that at the time of the expropriation oil prices were low and the market value of the assets was US$1 billion. As of the date of expropriation, assuming away other possible complications that are not germane to this discussion, an award of US$1 billion would have provided full reparation. However, some years later if oil prices are higher, then the market value of the assets might be considerably higher than US$1 billion, and an award of US$1 billion would not, at that later date, leave the injured party indifferent between breach and no breach (assuming that in the no-breach scenario it would have continued to own the assets, and that their market value would have developed as it did in actuality).
Two dates are most commonly considered for the date of assessment: (1) the date of breach (ex ante); and (2) the date of award (ex post).
The calculation of the monetary award requires one to calculate, as of either of these two dates, what the financial position of the claimant would have been in the absence of the breach (the ‘but-for’ financial position), and to compare the but-for to its actual financial position after the breach.
- If the calculation is made as of the date of breach (ex ante), then the effect on the injured party should be determined on the basis of information that is known up to the date of the breach and on the parties’ expectations about the future as of the date of the breach. In the case of expropriation of a factory, this would imply estimating the factory’s performance based on expectations about demand, prices, costs, etc.
- If the calculation is made as of the date of the award, then one can incorporate hindsight to determine how the injured party has effectively been affected by the breach. For example, in the case of expropriation of a factory, one can observe how the factory has financially performed since the breach occurred and base the award on the observed financial performance.
In either case, one uses information that was available up to the date of assessment and no other information. That is not an arbitrary rule. It reflects the underlying idea that the estimation of damages is an objective exercise that generally focuses on the notion of fair market value. The fair market value of an asset will generally reflect the information that is available to market participants, but of course cannot reflect information that is not available to them. So, the fair market value of an asset at the date of breach cannot reflect information that is only available with hindsight.
While the choice to use the date of breach or date of award is, in general, a matter of law, the final choice may give rise to very different values for the compensation. This tends to be the case when the award is dependent on a variable that is highly volatile, as in the example given above of an oil and gas firm. A concrete example is the Yukos dispute, where the value of the expropriated asset was highly dependent on the price of oil, which was around US$40/barrel as of the date of breach and US$112/barrel as of the date of award. According to the tribunal’s calculations, these figures would give rise to compensations of US$22 billion and US$67 billion, respectively.
The choice of the date also gives rise to various quantum issues, as explained below.
Use of the date of breach
As discussed above, the use of the date of breach implies that the calculation of the award is based on expectations about the future. Because expectations about the future often differ from what is eventually realised, the use of the date of breach will generally result in compensation that may leave the claimant better or worse off, as of the date of award, than in the case of no breach. For example, an event that was unexpected at the time of breach and that significantly increased the factory’s sales (e.g., a factory that produced sanitary masks and was expropriated prior to the covid-19 pandemic), would not have been incorporated in the calculation of the award, thus leaving the claimant ‘under-compensated’ (as of the date of award) because it would have been better off without the expropriation than with the award.
This means that the use of the date of breach provides the claimant with a certain fixed (as of that date) compensation that does not depend on the realisation of future events, and at least in the case of expropriation (and assuming the respondent retains ownership following the expropriation) allocates the risks of future events to the respondent, as it is the respondent that will obtain higher or lower payoffs depending on the realisations of events that occur between the dates of breach and award. To give a simplified example, consider a business that is about to engage in exploration for oil and gas at a particular location, which will either be successful, in which case the business will be worth US$1 billion, or unsuccessful, in which case it will be worth zero. The business faces uncertainty: will the exploration succeed, in which case its future value will be US$1 billion, or not, in which case it will be zero? Suppose that just as the exploration begins, the government expropriates the business and continues to operate it, and that its owners seek and obtain compensation under an investment treaty. If, as of the date of breach, geological evidence indicated a 60 per cent chance that the exploration would be successful, then the ex ante compensation would be 60 per cent of US$1 billion (i.e., US$600 million), and that figure would apply irrespective of whether or not the exploration had later proved successful.
From the perspective of a quantum expert, obtaining expectations about the future in itself is a task that often presents challenges.
It is usually relatively easy to obtain expectations of macroeconomic or general inputs on which the calculation depends. For example, for variables such as expected inflation, exchange rates or crude oil (or other commodities’) prices, there are regular published forecasts from a number of reputable bodies, as well as ‘forward prices’.
It is more difficult to obtain expectations of other more specific inputs. For example, one is unlikely to find forecasts for the sales of a specific company as of the date of breach, or of its labour costs, unless one can find and rely on internal forecasts (e.g., from a contemporaneous internal business plan). In the case of regulated assets, one needs to have an assessment of what were reasonable expectations, as of the date of breach, of the future path of regulation (e.g., the future evolution of regulated prices that largely determine the firm’s revenues), and finding or making this assessment may also (although not always) be a difficult task.
Finally, we note a caveat (as mentioned above): it is common in this context to use hindsight as a ‘sense check’. If the expert says that expectations as of the date of breach were X, but the actual outcome was Y, which is very different from X, then it is reasonable to check that this makes sense (i.e., that there is a sensible explanation for why the actual outcome differed from reasonable expectations at the time).
Use of the date of award
As discussed above, the use of the date of award allows and indeed requires the use of the information that has been observed between the dates of breach and award (i.e., ‘hindsight’). It therefore avoids the issue of over or under-compensating the claimant (as of the date of award) that is inherent to the approach of using the date of breach. However, in doing so, it introduces the ‘mirror image’ problem, in the sense that it can lead to over or under-compensation as of the date of breach, which some commentators consider inherently unattractive. It can be argued that the harm that the respondent caused was the value taken as of the date of breach and any subsequent change in value is simply luck, and there is no reason for the tribunal to reward the claimant for luck (in particular, because that implicitly assumes that the claimant would not have sold the assets at an earlier date).
It is sometimes argued that the use of the date of award is unreasonable because it ignores that, in the absence of the breach, the claimant would have borne the risks as well as reaped the rewards associated with the cash flows between breach and award. That is the argument of the classic ‘Janis Joplin’ article, which says that:
The violation did not merely deprive the plaintiff of the stream of returns that would have accompanied the asset. It also relieved the plaintiff of the uncertainty surrounding that stream. To use hindsight is to ignore the latter effect.
However, we would question (or at least qualify) the assertion that ‘The violation . . . relieved the plaintiff of the uncertainty’ surrounding its business (e.g., the uncertainty surrounding the future value of an expropriated business). If the award is calculated on an ex post basis, the uncertainty remains. In the simplified example from above, on the day before expropriation the claimant faces uncertainty: will the exploration succeed, in which case the future value of its business will be US$1 billion, or not, in which case it will be zero? On the day after the expropriation the owners still face uncertainty: will the exploration succeed, in which case the value of the future award will be US$1 billion, or not, in which case it will be zero?
In terms of risk allocation, this approach is again the mirror image of the first. In other words, contrary to the case of using the date of breach, the use of hindsight allocates the risks associated to the activity, over the period between breach and award, to the claimant (because its compensation will be higher or lower depending on the realisations of events that occur between those two dates). In the example above, the claimant’s ex post compensation will be US$1 billion or zero, whereas its ex ante compensation would be US$600 million.
From a practical perspective, the use of the date of award removes the need to assess expectations as of the date of breach. However, hindsight does not in itself remove all problems, inter alia, because one needs to assess what would have happened absent the breach. Hindsight may help shed light on that, but it is not guaranteed to. In many (though not all) cases, the observed events could have been different had the breach not occurred. For example, in the case of the expropriation of the factory, either party could argue that the factory would have performed differently (e.g., had higher or lower sales, saved more or less on costs) if it had remained in the hands of the claimant. Moreover, the use of hindsight does not remove the need to forecast the events that are expected to occur after the date of award (in both actual and but-for scenarios).
A hybrid approach
Some experts argue for a ‘hybrid approach’, which uses the date of breach as the date of assessment, but nonetheless applies hindsight. Demuth, for example, explains that:
In practice, a hybrid approach [combining ex ante and ex post approaches] can sometimes be found ‘in which all lost profits are discounted back to the date of the breach, but the practitioner would rely on all information that was available up to the date of trial’, thereby using the book of wisdom to eliminate ‘some speculation as to what the cash flows would have been’.
We make two observations in this context. First, proponents of this approach generally make the same objections to the ex post approach that we have already discussed, and the same counterarguments therefore apply. Second, it is clear that under the hybrid approach the resulting award will not reflect the fair market value of the asset (at any date). That creates issues of consistency with the principle of full reparation (i.e., Factory at Chorzów). For example, in a case of expropriation, as of whatever date one chooses to apply the principle of full reparation, the fair market value of the expropriated assets will provide full reparation.
Finally, we repeat that the choice of the date to use for the calculation of the award is a matter of law. The choice has a significant impact on the allocation of risks between the claimant and the respondent, and more directly, on the value of the final award. Economics does not provide an answer to the question of which approach should be followed, and therefore, the job of quantum experts is to be clear about the procedure that they have followed in their calculations, and to be clear about the legal and factual assumptions on which their calculations rely.
Interest rate to bring damages to the date of award
Whether one chooses the ex ante approach or the ex post approach, or some combination of the two, one needs to consider the question of what interest rate to use to bring (certain) ‘parts’ of the damages forward to the date of award.
As a starting point, we note that under either approach, some or all of the damages have occurred in the past relative to the date of award. For a given quantum of damages associated with a date in the past, the claimant has therefore been deprived of that quantum between then and the date of the award. Full compensation therefore requires that the claimant also be compensated for that deprival, by applying interest on the amount calculated as of the past date. The amount of interest to be added on must take into consideration both the amount of time between the date of breach and the date of award, and the level of uncertainty for the claimant of actually receiving this money on the date of award.
There are different approaches for determining the interest rate that shall be used to bring damages to the date of award. Below we discuss the use of interest rates that are prescribed in the contract or treaty, and those that are specific to either the claimant or the respondent.
Prescribed interest rates
In certain circumstances, the contract or relevant treaty already specifies the interest rate to use to bring damages to the date of award, or the approach to calculate this rate. For the specific case of investment treaty arbitration, different bilateral investment treaties (BITs) give rise to different principles for the determination of pre-award interest rates. For example, the BIT between the Netherlands and Egypt defines just compensation as including ‘interest at a normal commercial rate until the date of payment’ (without further specification of the meaning of a ‘normal’ commercial rate). Instead, the BIT between Georgia and Kazakhstan specifies that compensation for expropriation should include ‘interest at the London Inter-Bank Offered Rate (LIBOR)’.
Don’t be an expert who ‘jousts’
The adversarial approach to dispute resolution regularly results in cross-examination seeking to demonstrate that the opposing party’s witness is either less qualified, less reliable or simply less right than his or her counterpart. It is not unusual for experts themselves to take active part in combat, seeking to demonstrate their own impenetrable ‘rightness’ and the commensurate wrongness of his or her opposing number.
The arbitrator is not appointed to determine the winner and loser in a jousting contest. The arbitrator’s role is to ascertain and assess facts and, where necessary, analyse sometimes complex engineering-, scientific- or valuation-expert opinions about that evidence. Unlike counsel, arbitrators rarely have an expert of their own who can wade through the experts’ reports and responses in cross-examination and report on what is hyperbole and what is defensible and grounded in evidence and good practice.
Following a bout of particularly talented and evenly balanced jousting, an arbitral tribunal may well be left with two decimated sets of expert opinions and have the unappetising task of raising an award from the resulting pile of rubble.
An alternative approach might be for the expert to seek to position himself or herself as the arbitral tribunal’s trusted expert and to be affirmatively helpful to the tribunal – by taking a measured and careful approach. Such an approach, it so happens, was remarkably effective in the English Commercial Court not long ago, where the defendant’s witness was invited by the judge to run various scenarios through his valuation model, on his laptop via shared screens, during the taking of his evidence at trial. Some of the scenarios worked in the expert’s appointing party’s favour and some did not. The expert remained neutral and assisted the judge however he could without advocating for one approach over the other, unless invited to. In the judgment, the defendant’s expert’s opinion (and model) was wholly accepted, resulting in a finding that even if the claim had succeeded (which it did not) the loss would have been zero, as opposed to the £3 billion originally sought. The judgment is reported at Automotive Latch Systems v. Honeywell International Inc ( EWHC 2171 (Comms)). Paragraph 815 sums up the benefit of being the expert of trust as opposed to the champion jouster:
I have to say I did not find [the claimant’s expert] a satisfactory expert witness. He had a tendency to be argumentative and didactic in the witness box and, on occasions strayed way beyond what could conceivably have been the expertise of a forensic accountant in the opinions he was prepared to express. In marked contrast, [the defendant’s expert] was measured and careful in his evidence and as a consequence helpful to the Court in a way in which [the claimant’s expert] was not.
– Wendy Miles QC, Twenty Essex
Where law (or the relevant treaty) determines the approach to interest, the quantum expert’s job is simply to perform the calculations required by law (or at most to select an appropriate normal commercial rate). From an economic perspective, however, the appropriate choice of interest rate follows, again, from the requirement for full reparation, as we discuss in the following two approaches that are commonly considered by economic experts.
Claimant-specific interest rates
One commonly used approach by economic experts is to consider what the claimant would have done with the money, had it had access to it at the same time it would have, but for the breach (the ‘hypothetical use approach’ or the ‘claimant’s opportunity cost approach’). Here, different options can be considered for the claimant’s hypothetical use: (1) assuming the claimant had debt, it could have used the money to reduce existing debt or avoid falling into new debt. In that case, the interest rate relating to these funds corresponds to the rate on the debt that the investor could or would have avoided – most likely, its most expensive debt (or if the investor was going to incur new debt, the rate at which it would be able to secure new financing); (2) alternatively, the claimant could have either kept the money in interest-bearing deposits at the commercial deposit rates in place at the time, or found an alternative use for the money, such as investing it in its own business or in any alternative investment. There is also a theoretical question as to whether this analysis should be specific to the claimant’s financial arrangements or should start from an ‘objective’ standard. While that is mainly a legal question, we note that from an economic perspective it may be hard to explain why the claimant would have the ability to earn an above-market return. If it did have such an opportunity, then why did it not find funding to exploit it from some other source?
It is also sometimes argued that, under the hypothetical use approach, the option that yields the highest interest rate should be chosen, because, had it had access to the money, the claimant would have chosen to use it in the most ‘lucrative’ way possible (i.e., in the way leading to the highest return). For example, one could compare reinvesting the money in the claimant’s own line of business (with a corresponding rate of return equal to the cost of capital that is relevant to that business) to any alternative uses or projects that might have been available to the claimant, and choose the option providing the highest return on the understanding that the claimant would have also done so.
However, the hypothetical use approach, and in particular with respect to assuming that the claimant would have invested the money, either in its own line of business or in some alternative project, is affected by a number of issues, which may make it unsuitable to be used for choosing the appropriate interest rate from an economic (and sometimes legal or factual) perspective. First, finding and proving the ‘most lucrative’ alternative may not be straightforward.
To begin with, it can be speculative to assume that the investment would have occurred in absence of the breach. Even proponents of using the claimant’s opportunity cost as an interest rate emphasise that the difficulty of this approach is to prove the opportunity cost:
The difficulty for the claimant is to prove its lost opportunity cost. For example, if the claimant can show that it regularly placed its cash surpluses in a standard investment vehicle paying market rates, then it should be entitled to interest at such rates. A business may alternatively reinvest its earning in the business itself or pay excess cash out to its shareholders in the form of dividends. The claimant should be entitled to this amount if it can prove its lost opportunity cost. A claimant may be able to do so by producing historical financial records and through expert testimony to show the rate of its return on investment during the relevant time period.
Furthermore, it is generally not reasonable to argue that the breach prevented the claimant from engaging in profitable investments in their line of business or in any other line of business. This is because the parties in international arbitration cases are large (often publicly traded) companies, which have access to capital markets, hence it is not realistic to claim that the suffered damages have prevented them from making profitable investments. For that matter, if the claimant could have raised capital to finance these alternative lucrative investments, it would have simply done so, therefore the cost to the claimant of not having access to the damages amount would turn out to be the relevant cost of capital.
Second, and most importantly, even if the claimant can show that in the absence of the breach it would have invested in a certain activity with a certain expected rate of return, and that the breach prevented it from doing so, determining interest on the basis of the hypothetical use approach could overcompensate the claimant from an economic perspective. The reason for this is that usually, an investment with a high expected rate of return also entails high risk, due to the economic principle of aligning risk and return. However, if, as a consequence of the breach, the claimant was deprived of the opportunity of investing in a certain alternative investment as we discuss above, it was also ‘deprived’ of the risk associated with investing in that alternative investment. Therefore, using the rate of return that the claimant would have expected from the alternative investment would overcompensate it by compensating it for a risk it did not actually take. To determine a correct interest rate using the hypothetical use approach, one would have to determine or measure an appropriate reduction in the rate of return expected from the alternative investment based on the specific risk associated with the investment that has been removed. This can often be a very complex task.
Respondent-specific interest rates
Many economic experts take another route that is based on the situation of the respondent rather than the claimant: the ‘forced loan’ approach. Under this approach, the amount of the award is interpreted as a forced loan from the claimant to the respondent. The idea is that as of the moment of the loss arising from the breach, the claimant has been waiting for the respondent to pay it the amount of the award. The claimant has therefore been in the same financial position as if it had lent that amount of money to the respondent. The only difference, on this account, is that the loan was involuntary (hence, ‘forced’). Between the moment when the damages were incurred by the claimant and the date of award, the claimant was deprived of, and the respondent was able to use, that amount of money (the damages amount).
Under this approach, the appropriate interest rate is equal to the borrowing rate of the respondent because that is the rate that the respondent would have had to pay if it had wished to borrow money from the claimant in a voluntary market transaction.
The interest rate in forced loans is usually determined by several factors; for example, the respondent’s location or the currency of the damages award. In particular, the appropriate interest rate here would compensate the claimant not only for the time value of money, but also for the default risk of the respondent as the claimant was subject to this default risk while it was forced to loan money to the respondent.
Determining the interest rate based on the forced loan theory tends to be significantly simpler than the hypothetical use approach for the reasons explained above. More importantly, determining interest based on the forced loan theory respects both the standard of full reparation and the economic principle of aligning risk and return as it provides the claimant with an interest rate that reflects the risks that it bore. Because the claimant was deprived of the money by ‘lending’ it to the respondent, the only risk the claimant actually took is the risk related to not being able to ‘recover’ the loan (i.e., the risk associated with the default risk of the respondent). This risk is correctly reflected by the respondent’s borrowing rate.
It could be similarly ‘simple’ to determine the interest rate based on the hypothetical-use approach, but assuming that the claimant would have kept the money in interest-bearing deposits, or would have used it to reduce its (present or future, or both) level of debt. Moreover, the principle of aligning risk and return would arguably be respected when determining the interest rate, assuming that the claimant would have kept the money in interest-bearing deposits, because, if these interest-bearing deposits yielded a lower interest rate than a forced loan to the respondent, it would be because they were also lower risk (so the claimant would, in principle, be indifferent between lending to the respondent at the respondent’s borrowing risk or ‘lending’ to another party at a different interest rate by holding the money in deposits, as long as the interest rate reflects the risk of the other party). From an economic perspective, therefore, the claimant would be adequately compensated in any of these cases. We note, however, that choosing an interest rate lower than the respondent’s borrowing cost would imply that the respondent has been able to benefit from its wrongdoing by having been able to ‘borrow’ money (in this case, the damages amount) at a rate lower than its borrowing cost. Whether that is appropriate or not is a matter of law.
 Boaz Moselle is an executive vice president, Ruxandra Ciupagea is a senior vice president and Juan Carlos Bisso is a vice president at Compass Lexecon.
 See Factory at Chorzów, Merits, PCIJ, Series A, No. 17, 1928, p. 47.
 The situation is different in the case of the ‘hybrid’ approach that we discuss later in this chapter. Note also the caveat below concerning the use of hindsight as a ‘sense check’.
 Yukos Universal Ltd (Isle of Man) v. Russian Federation, Final Award, 18 July 2014.
 Franklin M Fisher and R Craig Romaine, ‘Janis Joplin’s Yearbook and the Theory of Damages’, Journal of Accounting, Auditing and Finance (1990), p. 155 (their example of two yearbook thefts).
 Fisher and Romaine, op cit, p. 154.
 See footnote 9.
 There is one caveat to this, which may be important in some circumstances: if the value of the business could be negative (so there are realistic scenarios where the breach will turn out to have been in the interest of the claimant), it is true that the claimant has been saved some risk. This is indeed a point for Fisher and Romaine (op cit, p. 155). Further discussion of this point would require a lengthy excursus. However, we question whether in practice it is likely to be relevant in most cases (notably, the value to an investor of its stake in a limited liability company can generally not be negative).
 Alexander Demuth, ‘Income Approach and the Discounted Cash Flow Methodology’, in John A Trenor (ed.), The Guide to Damages in International Arbitration, Third edition (Global Arbitration Review, 2018).
 We do not address here the complication that could arise if there is some realistic prospect that the investor might have sold the assets between the date of breach and the date at which full reparation is required.
 Under the ex ante approach, all of the harm is associated to the date of breach. Under the ex post approach, some of the lost cash flows would have occurred in the past (i.e., between the date of breach and date of award).
 In the latter case, this approach would be ‘claimant-specific’ in the sense that it would derive from the general circumstances facing the claimant, but it would not take into account any unique characteristics of the claimant that would differ from those of a typical market participant in those circumstances. So, for example, if one assumed that the loss of cash flows gave rise to higher levels of debt for the claimant, one would assume a cost of debt that would typically apply in that situation, even if the claimant’s actual cost of debt was lower or higher.
 The claimant’s cost of capital represents its opportunity cost of investment (i.e., the expected return that the claimant would require (and expect to obtain) from an alternative, equally risky investment).
 See J Gotanda, ‘A Study of Interest’ (2007), Villanova University School of Law Public Law and Legal Theory Working Paper Series 83, 2007, pp. 35–36.
 See J M Colon and M S Knoll, ‘Prejudgement interest in International Arbitration’, Faculty Scholarship at Penn. Law, 2007, pp. 12–13. See also M A Maniatis, F Dorobantu and F Nunez, ‘A Framework for Interest Awards in International Arbitration’, Fordham International Law Journal, Volume 41, Issue 4, 2018, pp. 829, 836–837.
 The economic principle of aligning risk and return implies that for investments of higher risk (measured as variance in the cash flows generated by the investment), investors require higher returns, and therefore these investments have higher cost of capital as opposed to investments of lower risk (e.g., fixed payments). The variance in the cash flows generated by the investment implies that the investors could either obtain a return higher or lower than their expected (required) return, thus, obtaining this required return (equivalent to the cost of capital in a firm) is not certain. Riskier investments have higher cost of capital (required returns) because investors require a compensation for bearing the risk (i.e., the variance), and the cost of capital reflects that.
 In other words, the risk in question is the risk that the award will prove unenforceable in practice, in the same way that ownership of a bond requiring the respondent to repay a certain sum may prove unenforceable in practice (i.e., the respondent may default). This approach ignores the risks around any imperfections in the investor–state dispute settlement mechanism itself.