The Case for Deprival Value
At the beginning, we should note that although IASB/FASB now loosely talk of ‘the measurement objective’ this is conceptually confusing, as issues related to accounting measurement cannot be considered independently of the context of some accepted underlying objective for financial accounting and reporting, something with which the IASB and FASB are currently grappling as part of their conceptual framework (CF) revision. Major concerns with the current FASB/IASB approach to the CF include whether there is any likelihood of agreement on a single objective, given the lack of homogeneous users and uses. For example, the dilemma of ‘stewardship’ and other legally based uses (e.g., deriving from contracts and the need for ‘settling up’, which generally do require ‘one number’ to achieve closure) vs ‘economic decision making’ uses (where, as Geoff Whittington argues here, several numbers may be useful), still bedevils standard setting now as it has done for decades. And even if the relevance of a particular measure were agreed, different criteria of reliability appropriate for these different purposes might tip the balance.1
Naively perhaps—as based on my own auditing experiences of long ago—the following ideas are based on several behavioural premises:
- •
Information that is useful for and used by managers is more likely to be reliable than information that is only produced to satisfy external reporting requirements.
- •
However, this has to be tempered by the fear that management may manipulate data: hence bodies like the FASB/IASB seek ‘objective statistics’ (e.g., based on the original ‘myth’ of fair value (FV) derived from financial economics (e.g., Bromwich, 2007; Hitz 2007; Dean, 2008).
- •
It is important to triangulate management-prepared values against external market evidence wherever available—and especially now given recent experiences (post-2001) with Enron, WorldCom, etc., and even more recently, given the matters emerging in the aftermath of the GFC.
- •
That said, the major question now is whether, in the current markets, even the Level 1 FVs are now reliable. It is worth contemplating this in the light of the recent developments regarding FASB's new FSP 157-4; and also FAS 115-2, FAS 124-2 regarding impairments.
This leads me, with my accounting historian's hat on, to draw on the wisdom of the elder statesman of double-entry bookkeeping, Luca Pacioli. In 1494 he advised, in his instructions for drawing up the opening balance sheet: ‘distinguish clearly each item . . . assigning the usual value to each. Set the price higher (“fatter”) rather than lower (“leaner”), so that if you believe it is worth 20, attribute 24 etc. so that you can more easily obtain a profit.’2
So Pacioli does not hesitate to recommend a choice of measurement basis on the criterion of what he regards as most useful, and most likely to have desirable behavioural consequences for managing the business: His bias towards overestimating the ‘usual value’ suggests that target pricing, not historical costing, is where he starts from (Macve, 1996). Deciding on the objective(s) of accounting is therefore crucial, before one can seriously discuss measurement choices.
In this context, I must lament that, while Plantin et al. (2008) explore potential adverse behavioural consequences of FV (relative to historical cost [HC]) for financial institutions—at least when ‘short-tem’ horizons dominate decision making—unfortunately they do not fully understand normal HC accounting practice (which is actually ‘lower of HC and recoverable amount’ (e.g., Solomons, 1961), so the policy implications remain unclear.
LIABILITIES AND FAIR VALUE
Attention also needs to be given to an area not so often discussed, namely how useful is accounting for liabilities under FV?
- 1
In the case of liabilities that are financial instruments, if they are traded then FV works reasonably well (subject to the issues about transaction costs that Geoff Whittington has discussed); but where they are not traded, the paradoxes of Hicks's Income No. I (has value changed?) vs Hicks's Income No. II (has maintainable income changed?) make deciding how most usefully to report earnings intractable—for example, Bromwich et al. (2009). Changes in credit risk (e.g., Macve, 1984; Horton and Macve, 2000; IASB, 2009b) can also have counterintuitive consequences for earnings if these are measured as ‘change in FV’, unless the complementary falls in asset values are also recognized. Barth et al. (2008) find that in practice for a majority of ‘ordinary’ U.S. firms downward asset revaluations3 do outweigh the debt revaluation effect to give an overall value-relevant negative net effect on equity—but by definition any reported asset devaluations cannot include what (in addition to falls in previously unrecognized upward asset revaluations) may be the biggest impact for previously successful firms, that is, the fall in the value of their unrecorded internal goodwill as their credit risk rises.
- 2
In the case of liabilities representing contractual business obligations, such as deferred revenue for long-term contracts (e.g., Macve and Serafeim, 2009), there is widespread unease that using FV could often give a ‘Day 1’ profit. The latest IASB/FASB DP on Revenue Recognition (December 2008) is therefore against using FV as the Boards' members are ‘uncomfortable’. Although this discomfort seems intuitively very wise, it is surely a new CF ‘concept’ that has not been exposed before?
- 3
The issues get even more complex with pension and life insurance liabilities: Should we be accounting on the basis of immediate transfer (FV / ‘CEV’?) or ‘settlement over term’ (i.e., a PV of future cash flows measure) (e.g., Horton et al., 2007).4
A ROLE FOR DEPRIVAL VALUE?
Given this background of difficulties with FV, we should consider whether there is a role for deprival value (DV):
- •
This takes us back to the 1970s debates on ‘inflation accounting’.
- •
ASB has been brave enough to continue advocating it in its Principles (as did the SolomonsGuidelines, 1989, and AARF, 1998).
- •
It appears superficially similar to traditional ‘lower of HC and recoverable amount’: but HC per se cannot be decision relevant (cf. Penman, 2007).
The mechanics of DV are often expressed diagrammatically in a decision tree:
So DV = lower of replacement cost (RC) and ‘recoverable amount’ (i.e., the higher of NRV and PV), or equivalently DV = min [RC (max NRV, PV)]. A ‘mirror’ diagram gives ‘relief value’ for liabilities, so RV = higher of ‘replacement liability’ and what might be labelled ‘obligation satisfaction’ (i.e., the lower of ‘transfer payment’ and PV of amounts to settle the liability directly)—see, for example, Macve and Serafeim (2009) for fuller explanation.5
Clearly, unless prices are extremely volatile, or a serious commercial mistake has been made, DV/RV at initial recognition (i.e., when purchase of an asset, or inception of a contract obligation to supply goods or services, has just taken place) will normally be equal to ‘HC’, that is, what the asset has just cost, or what consideration has just been received for future contract performance.
But even though it is effective in getting across the message that normally RC gives an upper bound on asset value (even thought the asset is expected to yield a higher PV or NRV) because of the normal market opportunity to replace, there can be serious dangers in over-reliance on this simple pedagogical tree (which I would label as the ‘kindergarten’ version of DV) and on the ranking tables frequently derived from it (e.g., Van Zijl and Whittington, 2006;6Weetman, 2007). For example, for depreciating assets over the subsequent course of their useful life, Baxter'sDepreciation (1971, pp. 34–6) showed that RC needs interpreting, not necessarily as the second-hand purchase price, but more generally as ‘the adverse consequence of deprival on the present value of all future cash flows, given that the asset now has to be replaced earlier than planned’. (Similar complexities arise with long-term liabilities.) This does require management estimates of optimal actions so, like PV, the resulting DV might be criticized as ‘too subjective’. However, as markets get deeper, RC and NRV become ever more readily observable and, as Baxter demonstrated, DV must always be bounded by these market prices so the only subjectivity lies in fixing where on the spectrum in between them (and including them) DV lies. So DV (and equivalently RV) tends towards FV in ‘perfect markets’ but copes much more adequately and simply than FV does with typical real-world market imperfections.
FENDING OFF THE POLITICIANS?
Now I turn to the current question (discussed above in Graeme Dean's introduction) of how far professional standard setters need FV as objective market evidence to fend off both the politicians' responses to the current crisis and also industry (e.g., banks') special pleading. Several instances recently should give us cause for concern.
The recent threats of political interventions in the crisis (e.g., from the President of France—see also Dean above) have had the effect of strengthening ‘technical’ accounting calls, for example, from the accounting and auditing profession, to rally round the standard setters and support FV to demonstrate FASB/IASB's independence. But as already noted, even for financial instruments, FV has many problems that have been identified long before the current crisis, especially with respect to the consequences for income measurement. However, to put all this in perspective, and as Andrew Lennard observes in his presentation here, the business world does not only consist of banks!
Unfortunately the U.S. standard setter's contribution to the measurement debate, beyond the area of financial instruments, has generally been weak, given the longstanding SEC opposition to anything but HC in the primary financial statements (e.g., Zeff, 2007). So it is perhaps no surprise that U.S. and Canadian standard setters generally are dismissive of DV, as they have previously had little reason to debate measurement bases for other assets (or liabilities). More worrying is that the IASB now seems to be going down the same path, which is much more surprising given that, for example, Australia, the U.K. and the Netherlands are familiar with revaluations of real property and other long-term operating assets.7
IN CONCLUSION
I think it is clear that it is impossible to prove that any individual measurement approach is Pareto superior to others for external users—ideally they probably need a range of alternative measures in order to triangulate the information they receive from various sources. But deprival value/relief value does have the advantage over the others of being a rule for identifying which is the most appropriate value measure in the particular economic circumstances of the business: and therefore is superior for internal use by managers—so on my behavioural guideline, if it is more useful for managers, it is more likely to be reliable information when reported externally. However, logically it is not a measurement basis per se but a rule requiring consideration of the various bases to determine which is the relevant outcome in each situation, that is, if the best decision between alternative actions is made. But of course these various bases (which do include RC, PV, NRV and their variants) do not include HC (as this measure from the past is irrelevant to current economic decisions, except insofar as it can serve as a satisfactory cheaper proxy for the relevant current value); and usually will not include FV, which is defined as a price but not an actual amount that a business would receive.8 So if managements reported DV and were to explain their balance sheet values by reference to what current external (market) benchmark prices (for both buying and selling) were available, that could be useful to external users.
My discussion has not addressed the ‘aggregation problem’ (what FASB/IASB now label the ‘unit of account’ problem), that is, what is the relevant asset to value? Edey (1974) discusses this in relation to DV, but the problem is common to all the measurement bases, including FV. This matter has been discussed by other panellists.
It is a major matter of concern that standard setters' ingrained, exclusive focus on assets and liabilities has almost driven out any direct conceptual debate on income measurement / earnings (other than tinkering with presentation). But earnings / profit is what users seem primarily interested in for ordinary companies (e.g., Penman, 2007). And the Boards' members have now wisely stated that they need to be comfortable with the profit patterns resulting from measurement choices (in the IASB/FASB December 2008 Discussion Paper, Preliminary Views on Revenue Recognition in Contracts with Customers). Hence what is best for the balance sheet may be different for the income statement and these two different approaches may then need to be reconciled, perhaps along the lines of the IAS 39 treatment for the ‘Available for Sale’ category of financial instruments. This line of enquiry should be explored further (e.g., Horton and Macve, 1996; see also now a recent paper by Kothari et al., 2009).
POSTSCRIPT
After the date of the EAA symposium, IASB (2009a) published on 28 May its Exposure Draft on Fair Value Measurement. The ED itself does not refer to DV, but there are the following two paragraphs in the Basis for Conclusions:
BC 65 Some advocate measuring assets using an approach known as deprival value (also called ‘value to the business’). Deprival value represents the loss that an entity would suffer if it were deprived of the asset being measured. Deprival value is the lower of the replacement cost of an asset (ie the amount the entity would need to pay to replace the asset) and its recoverable amount. The asset's recoverable amount is the higher of its net realizable value (the amount that can be obtained by selling the asset, net of selling expenses) and its value in use (the present value of the future net cash flows from continued use of the asset within the business and its ultimate disposal).
BC 66 Deprival value is similar to the cost approach described in paragraph 38(c) of the draft IFRS in that replacement cost is integral to both approaches. Using deprival value, the asset's replacement cost is reduced when it exceeds the asset's recoverable amount. Using the cost approach described in the draft IFRS, the asset's replacement cost is adjusted for obsolescence factors to reflect the service capacity of the asset. The primary difference between deprival value and the cost approach is that deprival value is based on entity-specific information, such as the entity's estimate of the cost to replace the asset, whereas the cost approach uses market participant assumptions. Although the two approaches may produce similar results in some circumstances, the entity-specific focus of deprival value is not consistent in concept, nor sometimes in practice, with the market-based focus of fair value.
The discussion in BC, paras 65–6, is clearly inadequate and potentially misleading. Although its initial definition of DV in BC, para. 65 is correct, it then gives only what I have called above the familiar ‘kindergarten’ version of DV, which does not address the generally more complex versions applicable to long-lived assets and liabilities (e.g., Baxter, 1971).
However, as Andrew Lennard and Geoff Whittington have already explained here, while FV is defined as an exit price (BC, para. 28), BC para. 66 partly addresses this weakness by calling on the ED's ‘cost approach’ to FV. Thus, when an in-use valuation premise is appropriate, ‘exit price’ becomes ‘replacement cost’ because ‘[i]n effect, the market participant buyer steps into the shoes of the entity that holds those specialized assets’ (BC, para. 61). ‘Given the “economic principle of substitution”, fair value cannot exceed current replacement cost for the asset's services’ (BC, para. 63).9 So replacement cost will (rightly) continue to be used for in-use tangible assets, except presumably where an ‘income approach’ (i.e., using discounted cash flows) provides a lower value for the recoverable amount (BC, para. 62).
All this conceptual twisting and turning is effectively a roundabout way of taking into account the principle of DV while appearing not to retreat from the basic exit price concept of FV that was first, and mistakenly, put forward by FASB.10
Moreover, even the IASB's discussion of the ‘kindergarten’ approach to DV is incorrect. In BC, para. 66 there is an attempt to distinguish the Board's in-use / cost approach to FV (i.e., normally RC), on the grounds that ‘deprival value is based on entity-specific information, such as the entity's estimate of the cost to replace the asset, whereas the cost approach uses market participant assumptions’. This (and even more so the following sentence quoted above) has the tone of a desperate last-ditch attempt to defend a metaphysical doctrinal nicety that must surely be invisible to the practical accountant, business manager, investor or other user of the accounts. How is an entity itself to ‘estimate the cost to replace the asset’ without referring to market prices? And if market prices are not readily available then how is its estimate going to differ from the estimate that it considers would be made by the (probably hypothetical—Bromwich, 2007) ‘market participant buyer’ who ‘steps into the shoes of the entity that holds those specialized assets’ (BC, para. 60)?
Now consider again the common case of an in-use, depreciating, tangible asset where the entity's analysis may show that the optimal course of action on deprival is acceleration of the planned replacement date rather than buying a second-hand asset in equivalent condition, as that would cost more—that is, where cost measures the negative impact on the present value of all future cash flows, which is the fundamental DV approach (e.g. Baxter, 1971). If (as assumed in the ED) the ‘market participant buyer’ shares the same information about market opportunities as the entity it will surely also conclude that this is the optimal strategy. It would not therefore be prepared to pay the entity any more than its own equivalent calculation of the DV, so both would arrive at the same RC, which is cheaper than the current second-hand price. Of course as a market gets deeper (so there are relatively fewer hypothetical and relatively more actual market participants) then the more will the second-hand price be driven down by competition to the DV-driven estimate of RC. But until then, on the ED's own principles, the current second-hand market price itself cannot be the unchallenged arbiter of RC.
While the BCs then go on to discuss the general principles applicable to valuation of liabilities at FV, there is no corresponding discussion of ‘relief value’ to mirror that of DV in BC, paras 65–6.
As anyone, outside the IASB and its staff, who has a hope of understanding these convoluted theoretical arguments in the ED and its BC is likely to be much more familiar with the existing academic and professional literature on DV (none of which is cited in the BC, beyond stating that ‘some advocate . . . ’) it is not clear why IASB considers itself and FASB to have what economists would call a ‘comparative advantage’ in proposing an alternative but apparently effectively equivalent concept. Sticking to DV as the basic theoretical principle, and then arguing for any practical adaptations considered to be necessary (e.g., to enhance reliability or reduce implementation cost), would be a much more understandable and incrementally constructive approach, which would also have a sound basis in economics for deciding when to use an ‘exit-price’ based approach and when to use a ‘cost’ or ‘income’ based approach.
