Nicki L. Tillinghast
34 rue François Bonvin
75005 PARIS, FRANCE
tillinghastn@yahoo.fr
(Comment Letter on IASB Discussion Paper, Reducing Complexity in Reporting Financial Instruments)
The fair value of financial liabilities
28 September 2008
Dear Sir/Madam,
The Discussion Paper, Reducing Complexity in Reporting Financial Instruments, makes a number of important
statements on its perception of the effect of credit risk in the measurement of liabilities:
1
§3.72: "[...] the recognised amount of a financial liability does not represent nor is it intended to represent
the face amount of an obligation, even if a cost-based method is used. For example, if two entities whose
credit standing differs issue an unsecured note promising to pay CU1 million in five years, the proceeds will
be different. Although the two entities have the same obligation in five years, the liability amount each
recognises will be different until the issue discount is fully accreted."
Readers will agree that they would trade promissory notes given by the Bank of England at the present value of
money for their "face amount" but would probably only buy and sell promissory notes issued by others at lesser
values. Efficient trading in this market would establish stable values for the spot prices at which the promissory
notes could be exchanged between the creditors prior to their maturity: the fair (asset) value of promissory notes
on the creditors' market.
However, if they accepted taking over liability for any promissory note themselves, as a debtor trading on the
debtors' market, readers would always want to receive the exchange value of the Bank of England's promissory
notes on the creditors' market. This is because they would need to invest the proceeds, the transfer price or
current exit value for the liability, so as to enable them to pay the CU1 million due, and they cannot expect to
achieve a higher return than the risk-free return offered by the Bank of England.
The risk that future credit defaults might occur reduces the fair (asset) value of a promissory note but it does not
reduce the current exit (transfer price) value of the liability assumed. The probability of credit default has a
different effect on the measurement of a liability than it does on the measurement of an asset. As the Discussion
Paper on Insurance Contracts pointed out, risk margins also increase the value of a liability but decrease the
value of an asset.
When the "issue discount is fully accreted" the issuer will, provided the credit risk has not already materialised,
indeed account for the liability correctly at CU1 million. If credit default has occurred it will of course account for
the liability at CU0 million. In the meantime, when the "issue discount" is partially "accreted" the issuer will not
be accounting for the liability at its fair value as a liability. This is because, as the issuer may default on its
promise, the "issue discount" reflects a credit spread over the risk-free interest rate. However, accounting for
the liability by accreting the "issue discount" just amortises a capital of CU1 million over five years. It therefore
treats this capital as a risk-certain amount and not as a capital at risk.
The fair (current exit) value of a liability obviously takes account of credit risk—just as does the fair (spot
exchange) value of an asset—but not in a way that gives them equal values. Only if there is an embedded
derivative in the promissory note which enables the debtor to buy back its debt at its fair (asset) value of
exchange on the creditors' market is it possible to demonstrate that the fair (current exit) value of the combined
instrument equals the fair (spot exchange) value of the corresponding asset.
2
§3.73: "[...] a decrease in the probability that the financial liabilities will be settled creates a gain to earnings
in the financial statements of a borrower."
The settlement value of a financial liability is the contractual liability stipulated in the financial instrument, i.e. it is
the amount the debtor owes to its creditors. This liability is entirely independent of any credit default occurring
unless the contract stipulates what lesser amounts, in that event, would be due to the creditors.
On the other hand, "a decrease in the probability that the financial liabilities will be settled" would cause
creditors to trade the corresponding assets at a lower spot exchange (fair) value in anticipation that the debtor
might be less able than before to meet its contractual obligations to pay the cash flows at their settlement dates.
As previously, however, no transferee debtor could accept a price for assuming the contractual liability that
would not enable it to obtain a risk-free return on its investment sufficient to meet the contractual liability
transferred. This price is invariant whatever the respective probabilities of default involved—it could be neither
higher nor lower. This is easy to prove using probabilities. Even if the transferee itself issues junk bonds and
invests in junk bonds, its accepting the liability transferred at a lower price would just increase the probabilities
of credit default even further.
3
§3.74: "The following points explain why a decrease in the probability that a financial liability will be settled
is associated with a gain in the financial statements of a borrower: (a) The liability is a contract between two
entities. Generally, when circumstances change that result in one entity incurring a loss, it might be
expected that the other party will have a gain. That leads to a conclusion that, when a lender recognises a
loss, the borrower should recognise a gain."
Trading on the creditors' market may procure capital gains and losses reflected in the cash flows on the spot
exchange asset (fair) values. One creditor's gain would in effect be another creditor's loss on that market.
The short-term cash flows on the creditors' market do not change the contractual cash flows due by the debtor;
they simply change which creditors benefit most or least from the issued contract.
This does not lead to the conclusion that the debt issuer has a gain when a creditor makes a cash-flow loss in
its short-term trading on the creditors' market.
4
§3.74: "The following points explain why a decrease in the probability that a financial liability will be settled
is associated with a gain in the financial statements of a borrower: [...] (b) A financial liability’s fair value on
initial recognition reflects its credit risk. It seems inconsistent to include credit risk in the initial fair value
measurement of a financial liability but not in the subsequent measurement of the financial liability."
This is obviously untrue. A financial liability on recognition equals the cash proceeds creditors pay for the debt
instrument; they would not pay for it if the cash proceeds were not its fair value equivalent and those cash
proceeds are deterministic amounts quite unaffected by any future credit defaults on the instrument.
The expected return on the investment is identical, whatever the credit risk involved. This is because the
nominal, contractual return offered must exactly offset the credit risk. Gamblers betting CU1 on an "efficient
market" on a "head" arising on tossing an unbiased coin would not bet unless they got a return of at least CU2
in that event. A return of CU3 in that event would bankrupt the house, for on an efficient market there is no limit
to the number of bets that may be placed. The fair value of the bet is always CU1, irrespective of the credit
risk—for example, if the Bank of England's betting machine always returned CU1 irrespective of a "head" or
"tail" being tossed.
5
§3.74: "The following points explain why a decrease in the probability that a financial liability will be settled
is associated with a gain in the financial statements of a borrower: [...] (c) The apparent gain does not occur
in a vacuum. The reason why a borrower is unable to pay is that it has suffered losses or expects to have
shortfalls in profits. If those losses are fully recognised in the financial statements of the borrower, the
amount of the losses is likely to exceed the amount of gain arising from a decrease in fair value of the
liability. However, not all of the losses or shortfalls are recognised in financial statements. For example,
losses arising from decreases in value of unrecognised intangible assets are not recognised. The gain on
the liability might provide a signal to users of the borrower’s financial statements that unrecognised losses or
shortfalls have been incurred."
The apparent, and unjustified, gain obviously does "occur in a vacuum" of the accountants' own creating.
6
§3.74: "The following points explain why a decrease in the probability that a financial liability will be settled
is associated with a gain in the financial statements of a borrower: [...] (d) Equity holders of an entity are not
required to make any additional investment to cover losses incurred by the entity except to the extent that
the equity holders have a binding obligation to do so. However, when the credit risk of an instrument
increases, the lender might suffer a loss. Therefore, the apparent gain to the borrower can be seen as an
allocation of deficits from the owners of the borrower to the lender."
The shareholders limited liability would enable them not to make up any loss suffered by a lender. However, any
loss suffered by the lender would necessarily entail that the shareholders' equity was already reduced to zero.
As the creditors with the most subordinated rights, the shareholders could not make any gain as a result of the
lender suffering a loss.
In all events, there is can no doubt that the understanding of the fair value of a financial liability in the Discussion
Paper, Reducing Complexity in Reporting Financial Instruments, is based on the misconception that future cash
flows may be treated as if they are risk-certain and therefore not subject to credit risk. In the paper entitled "The
effect of credit risk in the measurement of liabilities", which accompanies my comment letter on the IASB
Discussion Paper, Preliminary Views on Amendments to IAS 19 Employee Benefits, this question is
discussed extensively with respect to these misconceptions using a simple probability distribution for credit risk.
I most earnestly suggest that accountants wishing to make pronouncements about fair values, including the fair
values of financial instruments and the "fair value option" in IAS 39, spend a little time trying to understand
elementary probability theory and the derivation of expected (probability-weighted) cash flows. To that end, the
abovementioned paper, "The effect of credit risk in the measurement of liabilities", might be helpful.
I should also like to repeat the conclusion in the corresponding comment letter here: "Whereas accountants
distance their standards and discussion papers from an 'actuarial textbook' (Discussion Paper: §7.9 and on
Insurance: §34), their concepts must not stray from the realm of science. To put fair value accounting in an
Alice-in-Wonderland world runs the risks of misleading financial reporting in such a mark-to-market accounting
environment and further Enron-like experiences."
Yours faithfully,
Nicki L.Tillinghast