• Search

Commercial Arbitration

Overview: Economic Damages

A quick guide to calculating economic damages in commercial arbitrations

The key driver of international commercial arbitrations is almost always the claimant’s desire to recover damages – to a greater degree even than in disputes in the domestic courts. Therefore the quantum element of arbitrations usually requires three things: expert knowledge of the relevant law, expert knowledge of accounting principles (which tend to be similar whichever jurisdiction one is in), and precise quantification of the damages sought.

In the following overview, Gervase MacGregor, David Mitchell and Andrew Maclay, at accountancy firm BDO, take GAR’s readers on a whirlwind tour of the basic principles underlying the quantification of damages before highlighting some of the principal issues occurring in many of the arbitral awards forming the body of precedent in international arbitrations, as well as featuring in many arbitrations in which BDO has provided expert witness testimony.

One thing needs to be clearly stated at the outset: the authors believe that determining quantum is a fundamental aspect of any arbitration, and economic damages should always be calculated as accurately as possible. They do not subscribe to the view that, while it is the arbitrator’s job to arrive at a legally sound decision on jurisdiction and the merits, damages can be a figure that “feels about right”. Quantifying damages is no more subjective, and no less rigorous, an undertaking than reaching a decision on the merits and the authors believe there is often a disconnect between the detailed analysis supporting, say, a decision regarding the applicability of a foreign law matter with the brief paragraphs on quantum and the consequent broadbrush figure.

Conceptual framework: calculating a realistic “but for” scenario

In breach of contract cases, which probably make up the majority of commercial arbitrations, the starting point for damages is the law governing the contract being arbitrated. The principle is to put the claimant back into the position in which it would have been had the contract not been breached. In a bilateral investment treaty (BIT) case, quantum either follows the principles set out in the treaty, or the principle of reparation enshrined in the Draft ILC Articles and the Chorzow Factory case, which is “wiping out all the consequences of the illegal act and re-establishing the situation which would, in all probability, have existed if that act had not been committed”.

In light of the above, it is hardly surprising that forensic accountants and economists have developed the “but for” principle as the basis for calculating damages – damages are thus based on the difference between:

  • what would have happened “but for” the breach of contract or other intervening event; and
  • what actually happened.

The second limb is usually straightforward – for example, the profit the claimant actually made. What is more difficult is calculating what would have happened if the contract had not been breached (ie, the “but for” scenario). The “but for” scenario needs to be realistic, although one needs to remember that the calculation can only ever be an estimate – because nobody can ever actually know with complete certainty what would have happened in an alternative universe in which the breach of contract, or treaty, never actually occurred.

This sounds trite, but is actually extremely important. We have seen too many cases in which forensic accounting experts got carried away by the intrinsic logic of their calculations, in which each individual step made perfect sense, but the overall result made no sense at all. Logic and statistics are highly useful, and are tools required to persuade tribunals that one’s calculations are robust, but before finalising a valuation, the valuer needs to pause and ask: does this valuation make sense? Is it realistic? For example, just because a start-up company grew its profits from US$100,000 to US$800,000 in three years, at a compound growth rate of 200 per cent per year, does not mean that the company would have been able to continue growing at this annual rate over the next 10 years – that simply would not be realistic.

The basic calculation

The damages suffered by a claimant in a commercial arbitration are likely to be based on the profits that it would have earned, had the breach of contract not occurred, but which it did not actually earn. While in theory one can simply estimate the profits in a “but for” scenario, the calculation is likely to be more accurate if the loss of profit is broken down into its constituent parts, which are:

  • loss of gross revenue;
  • gross profit margin, as a percentage;
  • variable overheads; and
  • fixed overheads.

One tends to estimate the loss of future revenue of a business as a separate exercise – this is often the parameter that drives profitability, and it may also be the most difficult and subjective figure to estimate. On the other hand, many companies’ gross profit margins remain reasonably constant as a percentage of sales, and this constant margin may be reflected in the company’s forecasts – so in many cases it may be fairly uncontroversial to assume that the gross profit margin in the future will be the same as in the past.

When calculating overheads, it is important to distinguish between variable overheads – such as transport and distribution costs – which are linked to sales, and fixed overheads that remain the same regardless of the level of sales – such as general and administrative costs. Variable costs need to be deducted in calculating the future loss of profits, but fixed costs may not change at all whether the company is selling anything or not, and so they are generally not deducted.

So, for example, if a factory is closed for three months following a fire, employees may still have to be paid (their salary is a fixed cost). However, if they are not paid overtime, a variable cost is saved.

The fixed/variable overhead debate troubles many non-accountants, but it should be realised that all costs are ultimately avoidable and therefore in a sense variable. Timescale and the commercial aspects of the relevant business are key here.

Estimating what would have happened in the future but for

The most important part of a damages calculation is estimating what would have happened – had the breach not occurred, in an alternative future. There are three commonly used ways of doing this:

  1. The claimant’s own forecast. If the claimant had its own forecast of the future profits it expected to make, this may be the best data to use to estimate what would have happened but for the breach. However, before relying on it, one needs to consider the purpose for which the forecast was prepared, and how accurate it is likely to be. If the forecast was prepared for the claimant’s bank, which had carried out due diligence and lent the claimant money based on this forecast, and if the claimant’s actual results over the past five years have always been within 10 per cent of its forecasts, then one could feel reasonably confident in relying on it. If, on the other hand, the claimant’s forecast was limited to a single sheet of paper prepared by the directors after the breach of contract occurred, and it had never actually achieved its budget in the past, one would not feel confident in relying on that forecast without first making serious adjustments to it. 
  1. Extrapolating from the past to the future. Where a company has been operating for many years, and has a track record of always achieving a certain level of sales and profit, and where profits have grown consistently at, say, 5 per cent per year, the best estimate of future profits may be an extrapolation from past profits. This may be based on professional judgement, having taken account of past profits, or it may be based on a statistical regression line, which estimates the future based on the past. 
  1. Comparison with what actually happened to a similar company or store. If one can find a similar company operating in the same field, and one can see that the claimant’s results have been closely correlated with the results of that company in the past, the best way of estimating what would have happened may be to consider what actually happened to the comparator company. So, for example, in a supermarket loss case on which we worked, the best way of estimating the sales and gross profit of a supermarket that was closed for a 17-week period was to find some other supermarkets that had similar characteristics in terms of location, customer profile, opening hours, car park size and previous profitability, and to assume that the store would have performed in a similar way to the comparator stores, had it not been for the event causing the loss.

Frequently arising issues

Having grasped in principle how damages are calculated, we now turn our attention to some of the central issues that frequently arise in arbitrations, including many in which BDO has been involved.

Lack of trading history

An issue that arises very frequently is that the claimant may have no history of trading. This is because many claims arise very early on in a project life cycle, for example, because one party terminated the contract while the factory was still being built and before it ever started trading. As many arbitrators have become comfortable with the net present value model, following which the value of a project is equal to the cash flows expected over the project’s life, discounted at an appropriate discount rate, this may not be a problem, as the cash flows can be estimated whether or not the project has started to make profits – the issue is simply how reliable those cash flow estimates are. If the claimant was building a power station and had signed up to a 30-year power purchase agreement at a fixed price per unit of electricity generated, it may be simple to estimate the loss of revenue, and the costs may also be predictable, so the profits over the whole project are easy to estimate. If, however, the income is to be generated from some new technology that has never been commercially proven, things are a lot less certain – and it may be far more challenging to estimate future cash flows, or it may be necessary to discount for the loss of a chance or to add a risk factor to the discount rate (see below).

Arbitration panels have adopted a variety of approaches when faced with situations where the future cash flows were highly uncertain. In the case of Metalclad Corporation v Mexico, for example, the panel found that future cash flows were simply too uncertain, and awarded damages on the basis of the costs incurred by the claimant instead (ie, on a wasted costs or amount invested model). However, in CMS v Argentina, the panel was happy to rely on a discounted cash flow (DCF) valuation (see below). Whether or not the claimant has any trading history, the key issue is: can the estimate of future loss be relied upon?

There is a particular variant of this problem referred to as the “hockey stick” model – where the claimant has always made losses in the past, but the breach of contract happened just as the business was about to take off – so a graph of its historical profits shows a declining trend, whilst the forecast sweeps steeply upwards completing the hockey-stick shape, forecasting super profits. While there are instances when this may be a realistic picture, where a company has always made losses in the past, a considerable amount of persuasion will be needed to convince a tribunal that, but for the intervening breach, its future would have been golden.

How long should one claim for?

A related issue is how long the claim should go on for. If the company has been driven out of business and has gone into liquidation, normal practice is for the claim to be based on the “valuation” model discussed below. However, if the claim is based on the company using the damages recovered from the respondent to recapitalise its business and start trading again, a key question is how long that will take, and what resources it will require – for the company to ramp its business back up to its previous level of operations. The forensic accountant will require robust evidence from the claimant to support all assumptions made in this regard.


Loss of profit and valuation claims frequently involve a model to support the “but for” scenario. These models can be highly complex based on many interlinked sheets of spreadsheet analysis. Models should be clear enough for the tribunal, the lawyers and the other quantum experts to understand, by being properly designed, explained and supported by all necessary documentation. The main drivers of value should be clear from the model. It should follow good modelling practice.

In BDO’s experience, a soft copy of the model is now routinely exchanged with the other side’s expert and the tribunal, which saves time and cost in replicating the other side’s model. It is often very helpful to the tribunal if the model is constructed in such a way that the assumptions can be varied to generate alternative outcomes (for example, by setting out all the assumptions on an assumptions tab) or if the basic model can be agreed between the parties.

Unrealistic projections: capacity of the claimant to generate profits

DCFs may be all about modelling future or hypothetical cash flows; however, in one’s enthusiasm for a theoretical model, it is always important not to lose sight of reality. One needs to make sure that the overall model makes sense – a classic example is forgetting the realistic maximum capacity of a factory, so that the model calculates the future profits of a factory that would need to be twice as big as it actually is, or to have twice as much working capital, to generate the level of profits projected. A similar error would be to project a 20 per cent per annum growth rate forward over a number of years, without realising that this would mean the company would then have a market share of 120 per cent, which is clearly impossible. Indeed, no business can ever achieve long-term growth at a rate above the growth in the market – something overlooked in some claims we have seen which, if looked at rationally would mean the business would eventually be bigger than the GDP of its host country.

It cannot be stressed too much how important it is for the forensic accountant to sit back and ask: does this overall model, and does this figure for profits lost, actually make sense? Is the growth rate realistic by reference to the market? Is the company investing enough of its cash flow to generate growth in the first place? Is it really likely that competitive pressure will not chase down margin?

The use of hindsight

Should the award take account of facts that arose after the date of the claim? The traditional approach in English law is that a breach of contract has to be evaluated at the date of the breach; however, if the use of hindsight tells us that something happened after the breach that fundamentally affects the value of the claim – such as a slump in the oil price – it might enrich the claimant unfairly not to take that into account.

In Golden Strait Corporation v Nippon Yusen Kubishika Kaisha [2007] UKHL 12 (known as “the Golden Victory) the House of Lords held that achieving an accurate assessment of damages based on the losses that had actually happened was more important than considerations of contractual certainty and finality. While damages are usually assessed as at the date of breach, this is not an absolute requirement.

How one treats hindsight will probably vary according to which law applies to the claim and the underlying facts – for example, bilateral investment treaties tend to provide for valuation the day before an expropriation event, while the application of the Chorzow Factory case in the event of an “unlawful expropriation” may involve the use of hindsight.

Sunk costs

Another confusing question is whether costs already incurred by the business at the date of the breach can be claimed, in addition to lost future profits – generally they cannot. However, as we saw above in the case of a company with an insufficient trading history, a wasted costs claim can be made in the alternative to a claim for loss of future profits.

The reason is simply that a future profits claim, for example, for the output of a factory, is based on the output of the factory in the future, but must assume that the factory already exists. It would be double-counting to claim both for the costs of building the factory and for the future profits from the factory, assuming that it already exists. If the factory has not yet been built, the cash flow model used in the damages calculation has to take account of the remaining costs needed to build the factory, including interest that will have to be paid in future on any monies borrowed to build it.

Overhead recovery rates

A further question is how costs are defined, and what costs should be deducted from estimated future revenue. To an accountant there are several ways of calculating a cost. For example, the cost of an employee may be his direct salary divided by the number of days claimed – or it may be based on his hourly “charge-out rate”, which may be equal to his direct salary cost multiplied by a factor of 3 or 4, to take account of the general and administrative and other fixed cost overheads of the company, or this higher charge-out rate may be the rate that is charged to the company by a service company in the group. Some tribunals have decided one way and some the other way, and which is deemed correct in the circumstances may well depend on the particular factual matrix, as well as the persuasiveness of the legal and accounting team.

Management time 

Can the claimant recover the value of management time lost, and how should this be valued? Whilst it is normal for management time spent specifically on bringing the legal case to be claimed as part of the claimant’s legal costs, there may be a claim for management time spent that could otherwise have been spent profitably on other projects. Such a claim straddles the line between being a cost that the claimant has had to bear and being a loss of revenue that the claimant would have been able to earn if management had been able to spend the time developing new profitable activities.

Again, one needs to be careful to distinguish between the fixed and variable elements of the claim – as management would have had to manage anyway, their time is generally regarded as a fixed cost that would have been incurred in any event, so there is no loss. However, if they can show that the time would otherwise have been spent on developing a new opportunity that would have been profitable, there may effectively be a further damages claim. Under English law, the case of Aerospace Publishing Limited & another v Thames Water Utilities Ltd set out the principles that a claimant will need to have kept good records of the management time that was lost and to prove that the time lost resulted in a loss of revenue from alternative activities.


As the goal of the claim is to restore the claimant to the position it would have been in had the breach not occurred, one has to consider the claimant’s tax position, which often means calculating the loss after tax. But the situation can get very complicated – if the claimant would have had to pay tax on its profits but for the breach, but will not have to pay tax on any lump sum award or if the claimant can utilise tax losses, it would not be fair if the award was not reduced to take account of the benefit to the claimant of receiving a tax-free lump sum – and the opposite situation may occur, depending on the tax laws of the relevant jurisdiction and the tax position of the claimant.


Statistical analysis, such as regression, seasonal adjustment, and hypothesis testing may be valuable tools helping to quantify damages. However, not every quantum calculation can be forced into a particular statistical methodology. For example, if a company has traded for 20 years with profits typically increasing by 5 per cent per year, a trend line and regression formula may help; or a seasonal adjustment may be appropriate for a company that sells 70 per cent of its output in winter. But where a company made profits and losses in alternate years in the past, and the reasons for the profits and losses are not likely to be repeated in the future, statistics may simply be a distraction, and one is forced back to the professional judgement of the accountant, the lawyer and, ultimately, of the tribunal. Statistics may also be used to dress up and hide what is in effect just someone’s judgement, and tribunals would do well to bear in mind the classic quotation “there are lies, damned lies and statistics” whenever a party insists too strenuously that its figures are entirely objective or independent.

Quantification – the valuation basis

There is a second, entirely different approach to quantifying damages, on a “valuation” basis. This is the method usually used in expropriation claims, for example under bilateral investment treaties, in total extinguishment claims under compulsory purchase orders, and when a company has been put into liquidation. In these cases the “but for” scenario is that the company would have continued trading as it was, but the reality is that the company no longer exists and its value is zero.

In these cases, the calculation of the loss is based on the value of the company immediately before the breach of contract, or the act of expropriation – and it is valued in accordance with the normal ways of estimating “fair market value”: the market (or earnings) approach, the income approach, or the cost approach. As the cost approach (the net assets value of the company) is usually less relevant for ongoing businesses, this means that the most reliable valuation method may be the earnings approach or the income approach.

The International Valuation Standards Council defines “market value” as the estimated amount for which an asset or liability should exchange on the valuation date between a willing buyer and a willing seller in an arm’s-length transaction, after proper marketing and where the parties had each acted knowledgeably, prudently and without compulsion”. In reality, companies sold tend to be priced either simply at the price upon which buyer and seller agree, or by the price/earnings approach, where the price is based on a multiple of earnings – in this case, actual earnings are usually adjusted to a notional comparable level of earnings, for example by adjusting for excessive directors’ remuneration, or for depreciation. Then, this standardised level of earnings is multiplied by a price or earnings multiplier taken from comparable quoted companies and adjusted for the size and marketability of the company.

However, arbitrators on the whole have tended to prefer the income approach to valuing companies. This may be because it can be applied to any company in any industry, or because it appears to be a more objective and impartial way of valuing a company. Following this method, the valuer builds the most accurate model he or she can of the future profits and cash flows of the company over as many years as possible, often ending up with a final year in which the profits are assumed to continue in perpetuity, and then discounts the profits of each year back to the date of the loss at an appropriate discount rate.

Apart from the discount rate (which we consider below), the biggest challenge to the income approach is trying to make sure that the model for future losses is robust – that the estimates of future income and expenses make sense, that they are reasonable in the context of macro- and micro-economic expectations of the future of the relevant country and industry, and that the entire model has been consistently constructed – for example, with reasonable evidence of what capital might be required to finance the company in the future.

It may be that the future is simply too hard to predict with any degree of reliability, in which case it may be reasonable to adopt high, medium and low estimates, and then to weight these by appropriate percentages. Or it may be appropriate to adjust the forecasts on the basis of some statistical tool – for example, in a satellite case on which BDO was consulted, the claimant used hypothesis-testing to estimate the likely future lifespan of the satellite, within a 95 per cent confidence interval band, even though no engineer had physically inspected and reported on the satellite.

In the more difficult types of case, the arbitration panel may simply have to adjust the value of the company on a judgement basis – even though this may seem anathema as it seems to imply that, ultimately, valuation is an art rather than a science.


The discount rate is one of the biggest factors affecting any damages claim. If we are using the valuation income approach, we noted above how the net present value of the company is based on its future cash flows, discounted back to the present at a suitable discount rate – but any future losses also need to be discounted back to the date of loss on the time value of money basis that a dollar in your pocket today is worth more than a dollar in your pocket in five years’ time.

In the world of international arbitrations, the general methodology underlying the calculation of the discount rate has become reasonably settled and generally agreed among arbitrators; however, there may be major arguments over the parameters used to calculate it.

The discount rate is usually calculated on the basis of the weighted average cost of capital (WACC), which measures the return the market expects on a similar investment project. The WACC is based on:

  • the company’s cost of equity (see below);
  • the interest rate the company pays on its current borrowings (the cost of debt); and
  • the gearing or ratio between the two.

The cost of equity is based on the formula:

Rf + (ERP x Beta) + CRP + small company premium + specific risk premium


  • the risk free rate (Rf) is based on the rate of return on US treasury (or similar) debt, which is assumed to be risk free and thus to represent the pure time value of money, on the valuation date;
  • the equity risk premium (ERP) is the average excess return on equity compared to debt in the past. This is based on statistics gathered over the past 100 years, and is currently regarded as being between 5 and 6.5 per cent;
  • Beta represents the degree of volatility of the company’s share price movements compared to the volatility of the relevant stock market as a whole, with 1 representing perfect correlation.
  • the country risk premium (CRP) represents the risk premium associated with a particular country – these premiums can be calculated using the rates of return on credit default swaps, or by looking up Professor Damodoran’s or other similarly highly regarded tables. Alternatively, in calculating the cost of equity one may calculate the Rf by using the rate of return on debt issued by the relevant local government in the relevant local currency – which then effectively replaces the CRP;
  • the small company premium, which is based on the observation that smaller companies tend to be riskier investments than larger companies – statistics on the small company premium can be found in tables produced by Ibbotson or Duff & Phelps. A current area of controversy is whether small company premia, which are based on the observations of data from companies quoted on US stock exchanges, are relevant to emerging market countries, where all the companies tend to be smaller in any event and where the country element is already taken into account in the CRP; indeed, there is an academic debate as to whether the small company premium actually exists at all; and
  • a specific risk premium – which is a subjective additional premium that may need to be applied if the valuer thinks there is some additional risk affecting the company or project that has not been taken into account in the other risk premia above.

Contributory fault

Finally, a topic raised by the recent Yukos award nicely illustrates the balance that needs to be struck between rigorous calculation and judgement: it is how to calculate a discount for contributory fault. In the Yukos case, the Permanent Court of Arbitration in the Hague held that the Russian state had unlawfully expropriated the investments of three shareholder companies in the Russian oil company Yukos and ordered it to pay them over US$50 billion. The tribunal panel discounted its award by 25 per cent, holding that the claimants had contributed to their loss by utilising tax optimisation arrangements and the Russia–Cyprus double tax treaty.

While we cannot see any evidence that the tribunal used any specific scientific or empirical method to arrive at this discount, some commentators suggest that it is the task of forensic accountants to calculate a defensible discount for contributory fault, and that this should be tested, along with all other quantum evidence, at the hearing. The theory goes on to suggest that the experts can create successive “but for” scenarios to work out what the discount for contributory negligence should be. So, for example, in the Yukos case, the argument might have run something like this: if Yukos had not set up subsidiaries in certain tax-efficient regions, its after-tax profits would have been x per cent lower, and it would then not have been ordered to pay certain tax penalties, and the same in future years, and if this had been the case, the Russian state might not have expropriated the claimants’ shares in Yukos.

However, while we can use highly complicated statistical and analysis software to run multiple “but for” scenarios and calculate answers, this methodology risks becoming too arbitrary and remote.

At the end of the day, one needs to remember that damages calculations are ultimately estimates – since no one can know precisely what would have occurred “but for” the breach complained of – and that judgement plays a role in quantifying damages. This is a good example of when it should be applied. It is for the tribunal to use its professional judgement and estimate a figure for contributory fault, based on its own past experience, its understanding of the facts of the case and the evidence given by the witnesses. While calculating economic damages is far more precise and rigorous a process than most people unaccustomed to commercial arbitrations may assume, the quantum arrived at is not 100 per cent certain, and not every figure found in a tribunal award can necessarily be derived by a scientific method.


Comments Add your comment

Add Your Comment