Roger Harrington
Vice President & Chief Accounting Officer
BP p.l.c.
1 St. James’s Square
London
SW1Y 4PD
19 May 2010
International Accounting Standards Board
30 Cannon Street
London
EC4M 6XH
Dear Sir or Madam,
Direct 01932 758701
Main 020 7496 4000
Fax 01932 738216
harrrc@bp.com
www.bp.com
Re: Invitation to Comment – ED/2010/1 Measurement of Liabilities in IAS 37
We welcome the opportunity to comment on the IASB’s exposure draft ED 2010/1
Measurement of Liabilities in IAS 37 – Limited re-exposure of proposed amendment to IAS
37 (the ED). I am pleased to respond on behalf of BP p.l.c. to the invitation to comment.
We have a number of concerns about the ED and the near-final standard from which it is
inseparable:
1. The original exposure draft which was published some five years ago provoked
widespread criticism. In the intervening period, there have been several proposals and
final standards issued involving significant developments in the recognition and
measurement of both assets and liabilities. We do not believe it is appropriate,
therefore, to comment on the limited re-exposure points of the ED, as requested by the
Board, without considering the entire proposed new general standard for liabilities.
2. We are opposed to the elimination of the probability recognition criterion and believe
that such a move should not be made before a comprehensive debate has taken place
about the concepts behind the recognition of assets and liabilities. While we
acknowledge that the Conceptual Framework is a guide for the Board in its work rather
than a standard in itself, there appear to be many departures from the Framework in the
ED. We think that it is essential to bring forward the debate on that and not to change
the Framework through revisions to accounting standards. IFRS 3 (issued 2004), which
required recognition of all contingent liabilities, represented a contradiction to the
Framework and to IAS 37. IFRS 3 (2008) extended this contradiction with the
Framework to contingent assets. The ED would, we think, bring the accounting for
general liabilities into contradiction with the Framework.
3. We are also opposed to the use of expected values for all liabilities within the scope of
IAS 37. We believe that expected values may be appropriate for large populations of
similar items where statistical analysis of past events can be used as a reliable basis for
estimation, but we are opposed to the extension to independent, single items where, in
our view, this pseudo-statistical approach will be onerous to apply, of doubtful reliability,
very difficult to audit and potentially of little use to users of the financial statements. We
believe that the user is better served by the recognition of liabilities which will probably
result in an outflow of resources. This should be on a basis which is simple to
understand and be accompanied by appropriate disclosure of material contingent items
which have not been recognised. We are not aware that users of financial statements
have indicated that the proposed approach would provide more useful information.
4. We believe that the liabilities which fall within the scope of IAS 37 should be recognised
at the best estimate of the amount of expenditure required at the end of the period to
Registered in England and Wales: No. 102498
Registered Office:
1 St. James’s Square, London SW1Y 4PD
2
settle the obligation (that is, the present value of the resources to be given up). We
think this is, and should be, a cost-based measurement and therefore do not agree with
the requirement to include a profit margin, as required by paragraph B8. We are in
agreement with the alternative views of the six dissenting Board members on this topic,
for the reasons laid out in paragraphs AV2 to AV4.
5. Although we welcome the recent staff paper explaining how individual uncertain
obligations should be dealt with under the new proposals, we do think that the fact that
this is needed is evidence of the lack of clarity in the proposals of the ED and the
working draft of the standard.
In conclusion, we do not agree with the proposed replacement of IAS 37 by a draft standard
which we find inferior in concept and clarity. In respect of the ED, we do not agree with the
change in the recognition criteria and the blending of the recognition criteria and the
measurement method together. We think that this is inappropriate for individual items and
leads to confusion about the conceptual principles behind the accounting for liabilities.
If you wish to discuss any of the comments in this letter, we would be happy to do so.
Please do not hesitate to contact myself or Alison Kinnon.
Yours sincerely
Roger Harrington
3
Appendix
Question 1 – Overall requirements
The proposed measurement requirements are set out in paragraphs 36A-36F. Paragraphs
BC2-BC11 of the Basis for Conclusions explain the Board’s reasons for these proposals.
Do you support the requirements proposed in paragraphs 36A-36F? If not, with which
paragraphs do you disagree, and why?
The limited scope of the ED
It has been five years since the proposals for the replacement of IAS 37 were first
published and provoked much adverse comment. Given the length of time that has
elapsed and the developments that have occurred in many areas of IFRS since
then, we do not think it is appropriate to restrict our comments to the limited field of
the measurement of liabilities in the ED as requested by the Board.
The elimination of the probability recognition criteria
We do not agree with the Board’s proposals to eliminate the probability recognition
criterion for liabilities without having undertaken a full and transparent debate about
this issue in the context of its rethinking of the Conceptual Framework. While we
acknowledge that the Framework does not have the authority of an accounting
standard, we believe that it should be respected by the Board in its deliberations
about the development of new standards and departed from only in exceptional and
justifiable circumstances, such as those surrounding financial instruments. With
these proposals for the elimination of the probability recognition criterion in IAS 37
the Board now appears to have departed from the Framework in respect of all
liabilities and thus has changed the Framework “by the back door”.
We do not agree that a change in the guidance in existing IAS 37 was required. We
are unaware of any major issues having been raised with the IFRIC in this area.
The recent staff paper raises the issue of different interpretations being made of
what the obligating event is in the case of a provision for litigation. If this is an issue,
then we would suggest that all that is required is clarification of what the obligating
event is, not a complete change of approach. This would be consistent with the
clarification that legal fees be included in the provision in paragraph B7 of the ED.
Given the high degree of judgement involved in identifying the obligating event and
assessing the probability of settlement, the existing approach seems superior, in
our view. It provides a straightforward, bi-polar method which is simple to apply:
a) Is there an obligating event - yes or no?
b) If yes, is it more likely than not to result in a payment - yes or no?
c) If yes, what is the best estimate of the amount?
Under this method, if the answer to either a) or b) is no there is no need to make
any estimate. If it is perceived that there is inconsistency in practice in the
application, we believe that a clarification would be sufficient, rather than a change
of approach.
We think that the proposed approach will potentially create confusion. While the
guidance in paragraph 14 of the working draft is useful, it seems that consideration
of whether an obligation exists will be confused with the probability of an outflow of
resources, particularly in those cases where there is a high degree of uncertainty as
to whether an obligation exists. We believe the more practical approach is to retain
the probability criterion as a separate step in the thought process.
4
Under the proposed approach, once an obligating event is identified, the entity is
obliged to develop a series of cost scenarios, including expected timing, and assign
probabilities to each of them. For a single obligation this could be time-consuming
and the resulting probability-weighted amount may not provide useful information
and may be difficult to justify to auditors. Imagine a legal case in which the entity
thinks that on balance it will probably lose and have to pay $100 million. At present
it would provide for 100 and would be able to defend this with the auditors without
much difficulty. Under the new method it could provide for anything between 51
and, say, 75, or more. This will be almost impossible to audit unless there is a solid
statistical base of identical cases.
The proposed approach is expected to be problematic when applied to “reversions”
of liabilities i.e. where the liability has been sold to a third party yet there remains a
possibility that the obligation will revert to the prior owner in the event that the third
party fails to perform. It appears that the proposals would require a probabilityweighted estimate of the liability to be recognised, in contrast to the current
situation where generally no liability is recognised, and indeed no contingent liability
is disclosed on the basis that the likelihood is considered remote that there will be
an outflow of economic resources.
We believe that it would be onerous to estimate such liabilities on the probabilityweighted basis, and it is inappropriate to recognise such amounts in the financial
statements. It would also be helpful if further guidance was provided to explain the
criteria under which an obligation would be considered transferred. For example,
does this require agreement from the obligee that the transferor is relieved of the
obligation?
The measurement of liabilities at the lowest amount
Paragraph 36B states that the amount an entity would rationally pay to be relieved
of an obligation is the lowest amount. This may be true in general, but there are
cases where, for rational reasons, such as reasons of safeguarding reputation, an
entity may settle at a higher amount. This will often occur in the case of court cases,
where an out-of-court settlement at a higher cost may be preferable to a case which
may generate adverse publicity. We therefore do not agree that the lowest amount
is the appropriate amount to provide for when an entity expects to pay a higher
amount. The latter will usually represent the most useful information from the user’s
point of view.
Furthermore, when an entity expects to pay such an amount, we think it is
unnecessarily onerous for the preparer to run through the various scenarios
required by paragraph 36B and then recognise an amount which will not be the
amount the entity actually expects to settle at.
We therefore disagree with the way “rational” has been defined to mean lowest
amount.
We think that paragraphs 36A-36F describe a measurement at cost and that
the measurement guidance presented in B8 is inconsistent with this
We find the guidance in paragraph 36B confusing in several ways when linked to
the requirements of Appendix B.

Only paragraph 36B (a) requires the amount to be measured in accordance
with Appendix B, whereas this Appendix contains elements which could
equally usefully apply to the amounts computed in accordance with
paragraphs 36B (b) and (c);

Paragraph B15 requires the entity to include in its estimate of the amount
required to fulfil the obligation a risk adjustment which is “the amount, if
5
any, that the entity would rationally pay in excess of the expected present
values to be relieved of this risk”. This appears to put the entity into the
scope of paragraph 36B (c).
The use of expected values for single items
We are opposed to the extension of the use of expected values into new areas of
IFRS, and in particular into the measurement of single items under IAS 37. The
measurement of uncertain liabilities will always be a difficult area and involve a high
degree of judgement. We are not sure what the problem is with the current standard
and we do not believe that the proposals help in this area and indeed the proposed
guidance will certainly not be easier to apply.
We believe that expected values may be appropriate for large populations of similar
items where statistical analysis of past events can be used a reliable basis for
estimation, but we are opposed to the extension to single items where, in our view,
this pseudo-statistical approach will be onerous to apply, of doubtful reliability, very
difficult to audit and potentially of little use to users of the financial statements. We
believe that the user is better served by the recognition of liabilities which will
probably result in an outflow of resources. This should be on a basis which is
simple to understand and be accompanied by appropriate disclosure of material
contingent items which have not been recognised.
As described above, we do not believe that it is helpful for users if the entity has to
provide for a settlement which is unlikely to occur, or to under-provide for a
settlement which is likely to occur. We are not aware that users of financial
statements have indicated that this approach would provide more useful
information.
While it may be reasonably straightforward in some cases to estimate the amount
of a payment required to settle a liability, we think that it will be difficult to arrive at a
probability and timing for an individual payment scenario. It will be even more
difficult to justify the chosen probabilities and timing assumptions when challenged
by the auditors, particularly in the early stages of a legal case. We think that this will
result in time-consuming research for possibly non-existent data, and will provide
much opportunity for manipulation of provisions and therefore management of
earnings, leading to a loss of comparability between entities.
Future events
The proposals appear to change the measurement requirements relating to future
advances in technology. Whilst we agree that it is appropriate to take into account
certain advances in technology, there is little guidance provided in the draft
standard and we believe this will lead to inconsistencies in application between
entities.
Reimbursement rights
We note that the “virtually certain” threshold in the existing IAS 37 for the
recognition of reimbursement rights has been removed in the draft standard. We
agree with this removal, because this threshold is inconsistent with the Framework
which includes a “probable” threshold for the recognition of assets. However, as
with the proposals for the recognition of liabilities, the omission in the draft standard
of the probability criterion for the recognition of reimbursement rights creates a
different inconsistency with the Framework.
Question 2 - Obligations fulfilled by undertaking a service
6
Some obligations within the scope of IAS 37 will be fulfilled by undertaking a service at a
future date. Paragraph B8 of Appendix B specifies how entities should measure the future
outflows required to fulfil such obligations. It proposes that the relevant outflows are the
amounts that the entity would rationally pay a contractor at the future date to undertake the
service on its behalf.
Paragraphs BC19-BC22 of the Basis for Conclusions explain the Board’s rationale for this
proposal.
Do you support the proposal in paragraph B8? If not, why not??
We disagree with paragraph B8.
We believe that the liabilities which fall within the scope of IAS 37 should be recognised at
the best estimate of the amount of expenditure required at the end of the period to settle the
obligation (that is, the present value of the resources to be given up). We think this is, and
should be, a cost-based measurement and therefore do not agree with the requirement to
include a profit margin, as required by paragraph B8.
Paragraph B1 (a) requires the entity to take into account the outflow of resources in
estimating the amount to recognise. We do not understand why the Board believes that
resources means cost plus the hypothetical profit margin the entity would charge if it
performed this service for a third party. The Board asserts that it is this value, rather than
the cost, which has to be recognised but we find the justification of this in BC21 (d) and (e)
unconvincing.
The proposals would mean that management’s intent to either undertake the service itself
or to pay a third party to undertake the service is not taken into account in the measurement
of the liability, and we cannot understand why the Board believes that this provides more
relevant or reliable information than using a best estimate of the actual costs that are
expected to be incurred. In some cases the liability may be overstated by including a
theoretical profit margin, leading to a release to income when the liability is subsequently
settled at a lower amount. In other cases it is possible that the liability may be understated,
as sometimes the costs incurred by an entity in undertaking the service itself could be
higher than the price charged by a contractor. As mentioned above, there may be reasons
why an entity chooses not to select the lowest-cost option to fulfil an obligation, as there will
be other business factors in play.
We are in agreement with the alternative views of the six dissenting Board members on this
topic, for the reasons laid out in paragraphs AV2 to AV4.
Question 3 – Exception for onerous sales and insurance contracts
Paragraph B9 of Appendix B proposes a limited exception for onerous contracts arising
from transactions within the scope of IAS 18 Revenue and IFRS 4 Insurance Contracts. The
relevant future outflows would be the costs the entity expects to incur to fulfil its contractual
obligations, rather than the amounts the entity would pay a contractor to fulfil them on its
behalf.
Paragraphs BC23-BC27 of the Basis for Conclusions explain the reason for this exception.
Do you support the exception? If not, what would you propose instead and why?
If the Board were to maintain the measurement principles of existing IAS 37, there would be
no need for these specific exceptions. We encourage the Board not to modify those
principles.
7
If the Board decides to proceed with the proposals in the ED, then we would support, in
those limited circumstances, the Board’s proposal to provide these limited exceptions for
the reasons of pragmatism laid out in paragraphs BC23 to BC27.