[Skip to content]

Sign up for our daily newsletter
The Actuary The magazine of the Institute & Faculty of Actuaries
.

Fair values: direct or indirect?

Although the insurance industry is increas-
ingly international, there is little consistency
or transparency in the reporting of results
of the global players. The accounting frameworks for insurance business in many jurisdictions differ significantly and there is no international standard on insurance accounting to reduce the diversity of practices. Even in the EU, where financial statements need to be prepared under the Insurance Accounts Directive, it is difficult to compare accounts prepared by companies in different member states. This has an adverse impact on the credibility of the insurance industry in the eyes of the market, with the consequent impact on cost of borrowing, raising finance, and share prices.
The need therefore exists for quality financial reporting for insurers which is consistent, comparable, and transparent. Could fair values be the answer?

The story so far
The International Accounting Standards Committee (IASC) released its Issues Paper for Insurance in November 1999. This comprehensive document sets out the key aspects of insurance accounting.
Comments were due by the end of May 2000 and more than 130 were received, although admittedly about 30 of these were from the Netherlands, with identical appendices! The Faculty and Institute of Actuaries responded jointly with the Association of British Insurers and the Institute of Chartered Accountants. All the responses have now been collated, with a view to issuing a draft statement of principles in 2001.

Fair values
Much of the debate has centred around a dozen key issues, perhaps the most interesting of which is the use of fair values in insurance accounting.
Fair value is ‘the amount for which an asset could be exchanged, or a liability settled, between knowledgeable, willing parties in an arms’ length transaction’.
In the absence of a deep and liquid market, the calculation of fair value would need to use robust estimation techniques so that it could be reliably and consistently established. Issues that would need to be addressed in determining such techniques include:
– how to allow for risk and uncertainty, including discount rates;
– whether the fair value of an insurance liability is affected by the credit standing of the insurance company;
– should assumptions be based on market experience or own experience; and
– the controversial recognition of investment margins.
The UK response commented on all of these issues and supported the move to fair value based accounting. In addition, the covering letter from the Faculty and Institute included the view that any standard on the calculation of fair values ‘should set out the broad principles in this area and be neutral as to whether calculation of policyholder liabilities should be through a direct or indirect method’. It then went on to record support for the use of embedded value techniques in determining the fair value of insurance liabilities.
It is the use of direct and indirect methods in the calculation of fair value that I would like to explore further.

Background
Simplistically
The direct approach involves projecting the liability cashflows using assumptions which include an allowance for risk (sometimes referred to as market value margins). Each assumption can therefore be thought of as being made up from a best estimate, together with a market value margin. These cashflows are then discounted at a discount rate, i, to arrive at the fair value of the liability.
The indirect approach is represented by the familiar embedded value technique. Thus one starts with a measure of capital, typically the regulatory capital base, and then projects the future net cashflows to emerge from this capital base using best estimate assumptions. These future profits are then discounted at a risk-adjusted discount rate, d. The capital less the value of future profits at the risk discount rate then gives the indirect measure of the fair value. These methods are shown diagrammatically in figure 1 across.

Direct or indirect?
The controversial issue is whether the direct method should be preferred over the indirect method and, moreover, whether indirect techniques are acceptable at all as a means of calculating fair value. For example, the response of the technical staff at the US Financial Accounting Standards Board says that the ‘family of embedded value reporting schemes are not appropriate substitutes for measuring the fair value of a life insurance liability’.
However, it can be shown algebraically that, under certain conditions, the direct and indirect methods are equivalent. Two papers which explore this theme are a fair value methodology discussion note included in the response from the Institute of Actuaries in Australia, and a paper in the North American Actuarial Journal by Luke Girard. The thoughts that follow are very much in agreement with those expressed in the Australian paper.
So, if the two methods are equivalent, why the controversy?

Problems with the indirect approach
The key phrase in the previous section was that equivalence holds ‘under certain conditions’. The indirect approach, as currently implemented through embedded value techniques, does not necessarily meet these conditions:
– The capital base used in the indirect approach should be derived using the same risk factors that drive the market value margins. If the capital base is based on a regulatory capital regime that measures the risk factors differently, then equivalence will not hold.
– The determination of the risk discount rate under the indirect approach is currently rather broad brush. Indeed, in proving the equivalence of direct and indirect approaches one almost needs to ‘solve for the indirect risk discount rate’ that gives equivalence. However, given that the methods are supposed to be alternatives, determining the risk discount rate for indirect approaches in isolation is very difficult.
– The indirect approach can create or destroy spurious value if it is not implemented correctly. For example, generating value by changing the asset backing of the portfolio should not occur (the fair value of a guaranteed bond does not depend on whether it is backed by equities or gilts). The broad brush approach to the determination of the risk discount rate, and in particular its relationship with the capital base, can also cause problems when the strength of the capital base is altered.
For these reasons many rule out the use of indirect methods as a means of arriving at fair value.

Are direct approaches the answer?
The valuation drivers for direct methods are generally more transparent, and hence some of the problems above are more easily avoided using a direct approach. However, there are still a number of issues to be resolved. These include the determination of discount rates (although this is probably not as critical as it is for indirect approaches), and the market value margins for a variety of risk factors. A workable, practical approach has therefore yet to be established.

So is it the end for the embedded value?
It is worth taking a step back at this point and asking what we are trying to achieve. We are trying to estimate the fair value of a liability. Most would accept that there is an acceptable range in which this fair value should lie. Provided the regulatory capital base is a suitable proxy for risk-based capital, and the risk discount rate is ‘sensibly’ chosen, the embedded value technique will produce a measure of fair value which lies within an acceptable range. I guess the problem is where in the range! Also, problems such as the creation of value as a result of changing the asset backing could be overcome with appropriate guidance. It is therefore too soon to write off indirect methods of arriving at fair value, especially given the history of use.

In summary
There is still research to be done on appropriate direct methods and until the practical application of such methods is established it seems premature to rule out the use of embedded value techniques. Now on with that research

00_12_01.pdf