Valuation for Commutation
When assessing a commutation, how much time and cost should be spent calculating the value prior to negotiating? How sophisticated should the analysis be? What techniques are available for valuation purposes? Jonathan Broughton of EMB considers the main issues. 
Ten years ago, commutations were generally frowned upon. Most companies were highly sceptical or had a corporate policy of avoiding them altogether. Now, however, the business case for commutations is widely accepted and they are a standard way of managing legacy books.
My colleague, Peter Matthews, has been heavily involved in commutations since the mid-80’s and has written many articles about their merits. His key theme is still appropriate today: “do your homework – understand your business”. The aim of any commutation valuation work is to give your negotiator as thorough an understanding as possible of the dynamics affecting the numbers.
Of course, we should be pragmatic. If the liabilities are only in the order of $10k’s, then detailed analysis is unlikely to be cost-effective. However, as the liabilities move up to $100k’s or $m’s, then a small investment in understanding the business better and preparing more for the commutation discussions may be highly profitable. In these situations “Information is King” – especially if you are a negotiator.
This article considers when it may be appropriate for run-off operations to invest in sophisticated techniques to assist commutations and what questions should you be asking to determine the right answer for your business. Two examples of more sophisticated approaches are given. It also addresses the related issue of risk premiums and how they might be calculated.
Some options to consider
The most common commutation approach involves reconciling the incurred position between the two parties, splitting the losses between classes and then agreeing IBNR to outstanding ratios for each class and also a discount for the time value of money. Variants may involve survival ratios or IBNR to incurred ratios. Win factors may be adopted to allow for legal disputes and burn factors may allow for the effects of different layers.
For many liability types, such a methodology is still arguably the most appropriate. Indeed, using complex models can sometimes just introduce a whole layer of subjective assumptions that over-complicate the analysis for little or no gain. However, there are also many occasions where sophisticated approaches can be used to value the liabilities or reinsurance assets within commutations. They can provide important insight to allow the liabilities to be better evaluated.
So the question is, for a given commutation assessment, how sophisticated should the valuation work be. The following considerations should help steer the decision:
• Size – an obvious statement, but the bigger the deal, the greater the possible return on any investment in a valuation.
• Data – a fancy model may be the ideal option, but if there is little data available to allow it to be parameterised, then the model will be close to useless.
• Is there any gearing effect (for example high reinsurance attachment points)? – see box one.
• Is there material uncertainty in the outcome? The more uncertainty, the more likely a risk premium will be required by the cedant (see below), which lends itself to a more detailed methodology.
• Timescales – often the people valuing the deal are working to tight management deadlines. You only want to embark on building a sophisticated model if it will be ready in time, after allowing for the need to get buy-in from colleagues and for validation.
• Are there accepted benchmarks? – if yes, the case for detailed modelling recedes. For example, for many of the catastrophes of the 80’s and 90’s, the two sides should be able to agree IBNRs without any modelling.
Once you have answered these questions, you should have a good idea of whether it is worth investing in a more sophisticated commutation analysis. This may mean using a stochastic model; that is to say a model that simulates distributions of possible outcomes. Box one considers how a stochastic model allows high-attaching reinsurance to be valued.
It may, on the other hand, mean spending more time examining the detail of the data and analysing the underlying dynamics – see Box two. In this real example, alternative, cruder approaches are potentially unreliable for settlement valuation purposes.
What price certainty?
Once you have been through this process it will be easier to address another vexed issue that frequently crops up in commutations: whether both sides should be prepared to strike a deal at discounted best estimate reserves, and what principles should guide these negotiations.
Statistically, a best estimate valuation weights all potential outcomes by their probabilities, so you might think it a fair basis. I would argue otherwise. One side is gaining certainty (the reinsurer/debtor) whilst the other is taking on more risk (the cedant/creditor). This certainty has a price – the risk premium. The higher the volatility, the higher the risk premium.
Let’s look at risk premium from a capital perspective. The regulator requires firms to hold capital above best estimate reserves in order to cater for adverse experience. If uncertainty is reduced, then so does the capital requirement. The value of releasing the capital early is the risk premium.
The risk premium can also be viewed as the peace of mind or extra stability that having reinsurance brings. Putting this in capital terms allows the appropriate level of risk premium to be assessed. Stochastic models, which look at distributions of possible outcomes, are the best way to do this.
All live companies in the UK will now have undertaken an individual capital assessment (ICA) for the FSA, usually on a stochastic basis. If a usable ICA model exists, then it should be relatively straightforward to support commutation valuations on a stochastic basis.
In theory, all UK companies in run-off should also be prepared to undertake an ICA. In practice, many have not done so. The trigger for doing so may be a major forthcoming event, such as a Part VII transfer. Before long most such exercises will require a detailed stochastic analysis.
Do these principles about risk premiums apply to a company in run-off? To the extent that the capital varies with the uncertainty in the underlying business, then definitely yes. Where the capital is fixed (no option of topping it up) at a level well below where the regulator would allow a capital release, the theory still holds. For the outwards business, reinsurers should be prepared to pay a risk premium. For the inwards business, other factors such as credit risk may dominate the negotiations. However, even here, the run-off entity should be prepared to pay some part of its remaining capital to remove those inwards risks with the greatest uncertainty.
Knowledge is power No one model will be appropriate for valuing all commutations. Managers will consider the trade-off between the time and money (often surprisingly modest) required to invest in sophisticated techniques and the potential for increased profitability that comes with increased knowledge. Examples include more detailed data analyses, stochastic calculations, estimating risk premiums or a combination of these. Such techniques, if applied appropriately, should put those negotiating in a far more powerful position.
A Stochastic ApproachA stochastic model that looks at a whole distribution of possible outcomes can be extremely helpful when, for example, the commutation involves excess of loss reinsurance of individual losses and where the value of each claim varies considerably. Let’s take UK asbestos. Assume that you reinsured a public liability insurer for £250k excess of £50k for a stretch of years.
To simplify matters, there is no indexation, no inflation of claims, and no discounting and each loss gets allocated to one year. If the analysis estimates that 100 future claims would arise from your years at an average value of £40k, then a straight application of the frequency and average costs produces no reinsurance claim at all.
However, if the severity is instead represented by a distribution of possible outcomes, it may be seen that around 30% of the claims reach the reinsurance retention. The diagrams below show how the reinsurance may be valued.
Alternative last sentence: The diagram below shows how this approach recognises the value of the reinsurance.
Although the stochastic methodology will require distribution parameters to be estimated, this is relatively easy; both sides of the commutation can look at whether the fitted distributions are reasonable. Alternative non-stochastic methodologies do exist. However, the stochastic approach is flexible and transparent, allowing more complicated features, such as indexation or alternative allocation methodologies, to be incorporated.
This is how many live reinsurance companies price their contracts, using a frequency and severity model that allows both to vary stochastically. Capital can then be considered so that the technical price is set according to company objectives. Run-off commutations involving such excess of loss reinsurance contracts increasingly use similar techniques. This more sophisticated approach allows the true value of a contract to be estimated.
Assetta
Project Assetta** is an initiative designed to help estimate the value of insurance policies potentially exposed to US asbestos direct liability.
Companies with the need to do the valuation work as thoroughly as possible are participating. This includes some of the largest players in the London Market, as well as overseas companies with much smaller asbestos direct exposures.
Typically, a review of a US asbestos direct book of business (often an aggregate actuarial reserve review) will not create assured level numbers that are reliable for negotiation purposes. It is important to better understand for each assured the make-up of the underlying filings using the best available data.
This will allow invalid claims to be identified and removed, and the future frequency and severity of filings to be estimated by disease type. Assured specific information such as date of use of asbestos in products, the relevant state legal issues and the consideration of all policy terms and conditions should all be taken into full account.
Often there are big numbers at stake and there is material sensitivity to the assumptions chosen, so it is important not to cut corners. Other methods, even ones that appear fairly sophisticated (for example using tiers and burning costs) can produce unreliable numbers. These are therefore not generally suitable for commutation purposes, as the insurer would be exposed to anti-selection by assureds who are using more sophisticated models.
Assetta will produce appropriately sophisticated commutation valuations for a liability type where potential valuation uncertainty is great. In this example, the sophistication is in the detailed data analysis, rather than the use of a stochastic model.
** Project Assetta is a joint initiative between EMB Consulting LLP and Navigant Consulting (UK) Ltd.
This article appeared in Run Off Business Magazine in September 2006