When accounting standards become dangerous

In 1991 I delivered a one day course on tax and accounting in Geneva with participants from many countries. As an exercise, I had the group look at the accounts of a leading bulk chemicals company from each of the UK, Germany, Japan and the USA, to show how different their accounts looked, despite the similarities of their businesses.

Since then, IFRS have gone almost worldwide, with major increases in international comparability. (As with adoption of the metric system, the major holdout is the USA, which clings to US GAAP (Generally Accepted Accounting Principles) but there is a long term project between the IASB and the USA's Financial Accounting Standards Board to harmonise IFRS and US GAAP.

About 20 years ago, my specialised role at PwC caused me to become very involved with accounting for financial instruments. That was the first time I focused on how sometimes the IASB allows a desire for theoretical purity to result in accounting standards which may be theoretically correct, but which are potentially quite dangerous in practice.

In the last few years my voluntary role as a member of the Policy Team at the United Kingdom Shareholders' Association ("UKSA") has reawakened my involvement with accounting standards, and since last year I have been involved in looking quite closely at the standard which governs accounting for insurance contracts, IFRS 17.

With the help of a colleague at UKSA, I have realised that although IFRS 17 lays down a method of accounting that I can accept as theoretically correct, it risks stating the insurance company's shareholders funds at a figure that is significantly higher than it would be under more conservative accounting. For one insurance company I looked at, shareholders funds of about £10 billion would be restated as a deficit of £2 billion, a swing of about £12 billion.

Despite the issue looking esoteric, I decided to write a "plain English" explanation.

The article was published in the February 2021 issue of "The Private Investor" which is UKSA's magazine for members.

You can read it below. Please use the Disqus comments facility to say whether you found the explanation clear enough.

When accounting standards are correct in theory but dangerous in practice

I know that writing “IFRS 17 Insurance Contracts” will make most readers eyes glaze over. However, buried in the standard is a perfect example of how standard setters’ desire to be theoretically correct can lead to accounting standards that are dangerous in the real world.

Required common understanding

Before looking at the standard, I want to ensure that all readers share a common understanding of the basics. To simplify, I avoid the details you encounter with real insurance companies.

Assume that an insurance company’s year-end is 31 December 2020. On the morning of that day, for a price of £150, it sold an annuity contract under which it is obliged to make just two payments to the customer:

The insurance company invests the £150 in some investment assets, say quoted shares or bonds. It has no other assets or liabilities. (If you insist that it must have some share capital, assume that is an ignorable 1p!)

A Victorian might have accounted as above. However, we know that the above balance sheet is nonsense.

The obligation to pay £100 on 31 January 2021 is not the same as the obligation to pay £100 on 31 December 2070, so adding them together to show total liabilities as of 31 December 2020 of £200 is not valid.

How onerous on 31 December 2020 is the obligation to pay £100 in 50 years’ time on 31 December 2070? The slightly unsatisfactory answer is that “it depends.” What it depends on is the discount rate.

In this case “rate” is the constant periodic compound interest rate applicable throughout the entire 50-year period.

Insurance company investment assets

Accounting for investment assets, such as listed shares or bonds, where there is a proper market, is straightforward. Put them in the balance sheet at market value, referred to in accounting standards as the “fair value”. This comes from IFRS 9 Financial Instruments, paragraph 5.1.1.

The £150 used in the example is assumed to be the market value or fair value of the investments. Much more challenging is computing a balance sheet figure for the liabilities.

What IFRS 17 says about discounting long-term liabilities

“An entity shall adjust the estimates of future cash flows to reflect the time value of money and the financial risks related to those cash flows” is telling you to discount the future payment obligations. The obligations are in sterling, and fixed definite obligations. Accordingly, one would normally look at sterling gilt yields to find the discount rate.

A gilt such as Treasury 2.5% 2065 pays you £2.50 each year, and then £102.50 on the redemption date.

What you actually want is a discount rate to discount a single £100 payment due on 31 December 2070. If you know the prices of enough gilts, or look up the prices of gilt strips, you can compute what is known as the zero-coupon spot rate to discount from 31 December 2070 back to 31 December 2020. The details [of how to compute the gilt yield curve from a set of gilts' prices] would unduly lengthen the article. The key point is that normally one just discounts at the risk-free rate.

However, paragraph 36(a) tells you to reflect “the liquidity characteristics of the insurance contracts.” This is further explained in Appendix B paragraph B79:

The published application guidance for IFRS 17 tells you what this means in greater detail in paragraph BC194 below:

Our insurance company’s obligations, £100 payments due on 31 January 2021 and on 31 December 2070, are completely fixed. They cannot be accelerated, and the insurance company presumably cannot transfer them away without significant costs. They clearly fall within the definition of being illiquid.

Paragraph BC 193 goes of the application guidance goes on to explain in more detail:

Coming back to our illustrative gilt, Treasury 2.5% 2065 is of course a liquid investment. You can sell it any day you want.

The above guidance says that you cannot use the yield on such gilts arrive at your discount rate. You first have to estimate how much of the price of Treasury 2.5% 2065 consists of the value of the theoretically embedded option within the gilt to sell it.

Treasury 2.5% 2065 was quoted on 30 December 2020 (I could not find a 31 December price) at 107.05. i.e., £100 face value of the gilt would cost you £107.05.

Suppose the Government offered to issue you with a special Illiquid Security 2.5% 2065. This also pays £2.50 per year for every year until 2065 and £102.50 on redemption in 2065. There is a special condition which is that once the security is issued to you, it can never be sold; it can be inherited, but your heirs also cannot sell it; it must be held until maturity. That is what being illiquid means.

You would never pay the Government £107.05 to buy £100 face value of Illiquid Security 2.5% 2065. Why should you when you can buy Treasury 2.5% 2065 for the same price, and you can sell that any day you want. You would only pay less for Illiquid Security 2.5% 2065.

Assume you would only pay £90. The yield to maturity of Illiquid Security 2065 would be higher than the yield to maturity of Treasury 2.5% 2065, since both have identical future cash flows, but Illiquid Security 2.5% 2065 only costs £90 while Treasury 2.5% 2065 costs £107.05.

That means when you discount your insurance liabilities based on the discount rate derived from Illiquid Security2.5% 2065, you compute a lower number for the 31 December 2020 balance sheet than if you discount based on Treasury 2.5% 2065.

That is what the language of IFRS 17 and the related application guidance is telling you to do.

Why is this a problem?

Gilt securities exist in the real world, and you can compute their yield to maturity, and compute discount rates. Other people can check them, since everyone can look up gilt prices.

There is no such thing as Illiquid Security 2.5% 2065, and the Government is never going to issue it because it would cost the Government more in interest payments than normal gilts which can be bought and sold.

You may argue that if Illiquid Security 2.5% 2065 did exist, its price would be £90 (as assumed above), but there is no way of confirming that. Accordingly, the standard forces insurance companies to use discount rates that are essentially matters of judgment, in place of discount rates that are directly computable from observed market prices.

Does this matter?

The above discussion looks theoretical. However, the real-world implications are very large, because insurance company liabilities are very large.

I looked at the 31 December 2019 balance sheet of a large UK listed insurance company. Its accounts do not quantify the impact of using calculated liquidity adjusted discount rates instead of rates from the gilt yield curve.

However, by consulting the regulatory return which the company also has to file, I estimated that the impact of the adjustment was of the same order of magnitude as the entire balance sheet value of shareholders’ funds.

What should be done?

IFRS 17 is a global standard. After leaving the EU, the UK has set up its own Endorsement Board ("EB") to officially adopt international accounting standards for use in the UK. The EB cannot modify the standard; only accept or reject it in whole or in part.

However, I would like the EB to require all insurance companies using IFRS 17 to add the following additional disclosures in their accounts:

This would be very useful information for shareholders, and it is not something shareholders can ever compute for themselves. It would allow each shareholder to assess how well capitalised they think the insurance company is.

If the Endorsement Board asked for this, I am confident that insurance companies would comply, even though the disclosure is not required by IFRS 17.

Mohammed Amin MBE FRSA MA FCA AMCT CTA(Fellow)
Editor's note: although Amin is a member of UKSA’s Policy Team, he is writing in a personal capacity.

Appendix: The Regulatory Aspects

Regulation (in the financial services context) is about making and enforcing rules that limit what financial services providers are allowed to do.

A critical regulatory issue for insurance companies, as for banks, is how much capital the insurance company needs to hold, taking account of the risks it has taken on. The applicable rules in the UK are inherited from when the UK was a member of the EU, and are referred to as Solvency II.

The "Matching Adjustment" mentioned many times in her speech is the regulatory language for allowing insurance companies, for regulatory purposes, to discount their illiquid liabilities at discount rates higher than the risk free rate, as part of assessing the regulatory capital they are required to hold.

When it comes to regulation, my concern is that allowing use of the matching adjustment enables insurance companies to operate with less capital than they might otherwise be required to hold, thereby making them riskier.

Assets	£150.00
Liabilities	(£200.00)

Net assets (negative)	(£50.00)

Shareholders’ funds (deficit)	(£50.00)

Assets	£150.00
Liabilities	(£108.72)

Net assets	£41.28

Shareholders’ funds	£41.28

Assets	£150.00
Liabilities	(£195.13)

Net assets (negative)	(£45.13)

Shareholders’ funds (deficit)	(£45.13)