Accounting for Goodwill: A Critical Analysis of the Impairment-Only Approach

2.1 Goodwill Accounting and Impairment according to Current IFRS

The International Accounting Standards Board defines goodwill as “an asset representing the future economic benefits arising from other assets acquired in a business combination that are not individually identified and separately recognized” (IFRS 3, Appendix A: Defined terms). Hence, the IFRS concept of goodwill is based on the idea that an entire business unit is worth more than the sum of its parts in terms of netting its revaluated individual assets against its revaluated individual liabilities. As the following Table 2 visualizes, goodwill can, for example, come from excellent staff, high managerial ability, market power, a valuable customer or supplier base, or a good name and reputation (Gynther 1969, 247).

Not only financial accounting standards like IFRS as well as U.S. FAS, but also the International Valuation Standards incorporate the idea of a synergistic value that is the result of combining assets “where the combined value is more than the sum of the separate values” (IVSC 2022, 26). Because of, e.g. synergies, we can usually assume that the buyer of a whole business unit is willing to pay more than for all revaluated individual assets minus revaluated individual liabilities. This difference in the consideration transferred represents the phenomenon ‘goodwill’. Finance literature refers to a control premium to justify a premium price paid by the acquirer in view to obtain control over a target company (Roach 1998; Schwert 1996).

In principle, there are many different possibilities how accounting standard setters could deal with the phenomenon of ‘goodwill’ (Ding, Richard, and Stolowy 2008). On the most abstract level, they could completely ignore it, they could require that every goodwill shall be recognized as an asset, or they could choose a way in between. The International Accounting Standards Board opts for prohibiting the recognition of any internally generated, i.e. any non-purchased goodwill^[6] while requiring the recognition of every acquired goodwill, i.e. every goodwill that is purchased in a business combination (IFRS 3.32). This means, after a business combination where an acquirer has paid a consideration for acquiring a whole business, the residual payment made beyond the total value of revaluated net assets acquired in that acquisition has to be recognized in the balance sheet as acquired goodwill.

According to IFRS, after a business combination, the acquirer shall initially measure acquired goodwill as the difference between the consideration transferred and the amount of all identifiable individual assets transferred less liabilities assumed (IFRS 3.32). Both, the consideration transferred as well as the individual assets and liabilities, have to be measured at their acquisition-date fair values (IFRS 3.18, 3.32 b, 3.37). This also applies to assets and liabilities that the acquiree had not previously recognized but that, nevertheless, meet the definition of assets and liabilities in the Conceptional Framework for Financial Reporting (IFRS 3.11-3.13). If applicable, the consideration transferred has to be completed by the amounts of non-controlling interests in the acquiree and by the amounts of the acquirer’s previously held equity interests (IFRS 3.32).

For subsequent measurement, there are, again, different possibilities how accounting standard setters could treat goodwill: They could treat it as an asset that has an indefinite life and just require impairment testing; i.e. reporting entities would write-off goodwill only if there is a permanent decrease in its value (Black and Zyla 2018; Garcia, Katsuo, and Mourik 2018; Grinyer, Russel, and Walker 1990; McKinnon 1983). Alternatively, they could require that goodwill is amortized, i.e. treated like an intangible asset with a limited useful life and progressively written down over a convenient time span (Black and Zyla 2018; Garcia, Katsuo, and Mourik 2018; Grinyer, Russel, and Walker 1990; McKinnon 1983). They might also require the reporting entity to write-off it immediately at the time of the business combination. The International Accounting Standards Board has chosen the first option, i.e. they follow the so-called impairment-only approach: They prohibit amortizing goodwill but require annual impairment testing for every recognized goodwill, whether there is indication for impairment or not (IAS 36.10 b; IASB 2020). In result, the approach retained by both the IASB and also the FASB is the acquisition method, i.e. an acquired goodwill is recognized but not amortized. Instead, it is tested for impairment (Baker, Biondi, and Zhang 2009; IFRS 3.4-3.5).^[7]

For the impairment test, IAS 36 requires that goodwill is allocated to all cash-generating units of the acquirer that are expected to benefit from the synergies of the business combination (IAS 36.80). Hence, IFRS do not require further allocation to different elements of the cash-generating unit. According to IAS 36, a cash-generating unit is the smallest identifiable group of assets that generates cash inflows that are largely independent of the cash inflows from other assets or groups of assets (IAS 36.6). An impairment loss shall be recognized if, and only if, the recoverable amount of the cash-generating unit is less than its carrying amount (IAS 36.104). The recoverable amount of a cash-generating unit is, by definition, the higher amount of its fair value less costs of disposal on the one hand and its value in use on the other hand (IAS 36.18, 36.105; IASB 2020). The fair value of a cash-generating unit is the price that would be received to sell the whole unit in an orderly transaction at the date of the impairment test (IASB 2020, 28). The value in use of a cash-generating unit is the present value of the future cash flows expected from the whole cash-generating unit (IASB 2020, 28). Flow Chart 1 visualizes this valuation process for impairment testing.

Flow Chart 1:

Goodwill impairment test according to IAS 36.

Goodwill cannot be measured directly, but only as a residual (IASB 2020). The way the current IAS 36 has chosen to deal with this is firstly, to allocate an eventual impairment loss to goodwill and secondly, after the carrying amount of the goodwill is used up, to allocate the remaining impairment loss to the other assets of the unit (IAS 36.104). The impairment loss is recognized in the income statement (IAS 36.60, 36.104). If, in subsequent reporting periods, the recoverable amount of the cash-generating unit increases again, IAS 36 prohibits reversing any previous goodwill impairment loss.

Unless goodwill is impaired, a write-down of an acquired goodwill does not take place. The International Accounting Standards Board (2020) as the standard setter of the IFRS has recently confirmed its choice to follow the impairment-only approach: “A small majority (eight out of 14 Board members) reached a preliminary view that the Board should retain the impairment-only model” (IASB 2020, 3.89).

2.2 Impairment Versus Amortization in Literature

In principle, from a standard setters’ point of view, there are three general options how to treat an acquired goodwill: They could require that it is written off immediately, they could follow an impairment-only approach, or they could follow an amortization approach, i.e. they could require that an acquired goodwill is written down over a suitable time called amortization period. Historically, where standard setters chose the amortization approach, amortization periods ranged from zero to 40 years (Biondi et al. 2012).

The first option to treat an acquired goodwill, i.e. an immediate goodwill write-off after the business combination, relates to a static view on accounting that sees the determination of a liquidation market value as the central purpose of a balance sheet (Dicksee and Tillyard 1920; also see Ding, Richard, and Stolowy 2008; Garcia, Katsuo, and Mourik 2018). Ding, Richard, and Stolowy (2008) distinguish four historical phases of accounting for goodwill; a static phase or non-recognition phase, a weakened static phase, a dynamic phase, and an actuarial phase. Dicksee and Tillyard (1920) are not pure static authors, but regarding goodwill, “Dicksee thought that goodwill should be excluded from accounts whenever practicable, and that it should be written off as quickly as possible in other cases” (Carsberg 1966, 4) which relates to a static view on goodwill accounting. From this view, protecting creditors is central if the company becomes insolvent: A static balance sheet should give an indication whether the company can repay its debts also in case of insolvency (Garcia, Katsuo, and Mourik 2018). From this perspective, goodwill, whether acquired or internally generated, is not eligible for recognition in a balance sheet because it cannot be disposed in case of liquidation and, thus, does not protect creditors in such case. The option of an immediate write-off will not be explored further in this paper.

The second option to treat an acquired goodwill, i.e. an impairment-only approach, relates to an asset-liability view (also called actuarial view) that sees the determination of an economic value of equity from the shareholders’ perspective as the central purpose of the balance sheet (Ding, Richard, and Stolowy 2008; Garcia, Katsuo, and Mourik 2018; May 1957; Saito and Fukui 2016). An asset-liability view is the perspective that the International Accounting Standards Board (IASB) takes and that is incorporated in the IFRS (Saito and Fukui 2016). Goodwill is then a genuine part of this understanding of economic value of equity. In this view, goodwill is a permanent asset that can merely be affected by impairment and that is not consumed over time (Montgomery 1912). From this perspective, every amortization or even non-recognition of goodwill underestimates the economic value of the business from a shareholders’ perspective. An eventual reduction in this economic value has to be recognized in the form of impairment.

The third option to treat an acquired goodwill, i.e. an amortization approach, relates to a revenue-expense view (also called dynamic view) that sees the determination of shareholders’ periodical income as the central purpose of a balance sheet (Braun 2019; Ding, Richard, and Stolowy 2008; Garcia, Katsuo, and Mourik 2018; Richard 2017; Saito and Fukui 2016). From this view, revenues shall be matched with their costs; otherwise, a net income for a given reporting period is not determinable. The maximum distributable income has to be determined thus that, despite its distribution to shareholders, capital maintenance of the on-going concern is ensured (Ding, Richard, and Stolowy 2008). A consideration – paid to perform a business combination – that exceeds the costs of the net assets of a business is part of the costs paid to acquire the whole business. Conclusively, goodwill has to be amortized in order to match these costs with future revenues which are expected to be earned from the acquired business (Garcia, Katsuo, and Mourik 2018; Hatfield 1918; IASB 2020). From this perspective, an impairment-only approach is not adequate because it does not relate the capitalized cost that an acquired goodwill represents to future revenues.

Beyond these purpose-related arguments, prior discussion provides arguments that focus on the feasibility of these purposes. An argument against the impairment-only approach is that estimating a present value in use as well as a fair value of a whole business is all too subjective and arbitrary (Caruso, Ferrari, and Pisano 2016; Giner and Pardo 2015; Glaum, Landsman, and Wyrwa 2018; IASB 2020; Li and Sloan 2017; Ramanna and Watts 2012). As these are looking-forward values, they depend on many unobservable variables and forecasts (Martins, Sá, and Taborda 2023). This means, on the one hand, that management over-optimism can lead to an overestimation of the businesses’ economic value for the shareholders (IASB 2020). On the other hand, apart from unintended over-optimism, managers could utilize this subjectivity for deliberate overestimations. However, the IASB did not follow such arguments in its final ruling: “If estimates of cash flows are sometimes too optimistic in practice, the Board considers that this is best addressed by auditors and regulators, not by changing IFRS Standards” (IASB 2020, 3.29).

An argument for the impairment-only approach and against amortization is that, in case of amortization, the useful lifetime of goodwill is hard to estimate (Giner and Pardo 2015; Henning and Shaw 2003; IASB 2020). Thus, such an estimation is subjective and, therefore, it is arbitrary, too. Estimating the useful lifetime of goodwill can also be affected by unintended over-optimism or deliberate overestimations. In both cases, i.e. when the useful lifetime of an acquired goodwill is overestimated, annual amortization is too little and, thus, annual income as the excess of revenues over costs is overestimated. This misses the purpose of the revenue-expense view, i.e. to adequately determine periodical income.

An argument for amortization and against the impairment-only approach that the IASB (2020) mentions is the proposition that the impairment-only approach can lead to a replacement of acquired goodwill by internally generated goodwill, because “the future costs that maintain a company’s reputation and competitiveness would generate new goodwill internally rather than maintain the acquired goodwill. The acquired goodwill is continually consumed and replaced by internally generated goodwill” (IASB 2020, 3.63 c). A goodwill impairment test according to IAS 36 does not differentiate between internally generated and acquired goodwill. This is also a potential argument for quick amortization because the less annual amortization charge, the larger the probability of replacing acquired goodwill by internally generated goodwill.

A further argument for amortization and against the impairment-only approach that the IASB (2020) mentions is that amortization could reduce the costs of accounting. They relate to feedback from stakeholders who describe goodwill impairment tests as complex, time-consuming, and expensive (IASB 2020). An amortization approach could be a simple mechanism and would take pressure from expensive impairment tests.

An argument against the impairment-only approach in the current version of IAS 36 is that it leads to shielding (IASB 2020). Shielding happens when an acquired goodwill decreases but impairment does not take place because a reduction in the recoverable amount of the business is assigned not to acquired goodwill, but to unrecognized ‘headroom’. This could be unrecognized individual assets, differences between carrying amounts and recoverable amounts of recognized assets other than goodwill, and internally generated goodwill (IASB 2020). According to IAS 36, an impairment loss shall be recognized if, and only if, the recoverable amount of the whole cash-generating unit is less than its carrying amount (IAS 36.104). If the economic value of the acquired goodwill decreases but the unrecognized headroom is so large that the recoverable amount of the whole cash-generating unit does not fall below its carrying amount, no impairment takes place (IASB 2020). This problem is generally solvable by replacing the impairment test of current IAS 36 with a headroom approach that partly assigns a decrease in the recoverable amount of the whole cash-generating unit to the unrecognized headroom (IASB 2020). However, the IASB did not retain a headroom approach: “The Board concluded that the ‘headroom approach’ would reduce shielding but not eliminate it, because: (a) as discussed in paragraphs 3.43–3.46, the allocation of any reduction in total goodwill is imperfect; and (b) if the acquired business is performing poorly, better performance from other elements of the combined business could still shield the acquired goodwill from impairment. Moreover, the ‘headroom approach’ could result in recognising impairments that are, in some circumstances, difficult to understand (see paragraphs 3.45–3.46) and the approach would add cost” (IASB 2020, 3.49-3.50).

Table 3 summarizes all prior arguments regarding the impairment-only approach versus an amortization approach.

3.1 Reconstructing the Neoclassical Economic Interpretation of Impairment Tests

This paper follows literature and assigns the impairment-only approach to an asset-liability view on accounting (see above). The asset-liability view is often associated with a neoclassical economic framework (Braun 2019, Saito and Fukui 2016).^[8] Hence, it is relevant to investigate whether a neoclassical economic framework can help us assess whether the impairment-only approach contributes to the central purpose of accounting from an asset-liability point of view, i.e. to determine an economic value of equity. This section shows that a neoclassical economic framework can neither help to substantiate the impairment-only approach, nor can it help to reject it because it says too little about the value of goodwill and equity outside the neoclassical model. Furthermore, this section shows that goodwill measurement cannot adequately determine an economic value of equity and, thus, rejects the asset-liability view on goodwill measurement.

As section 2.1 illustrates with the example of IFRS, a goodwill impairment test requires determining the recoverable amount of the whole cash-generating unit, i.e. the higher amount of its fair value less costs of disposal on the one hand and its value in use on the other hand (IAS 36.18, 36.105; IASB 2020). In case of entire cash-generating units, the established concept for determining both is to apply present value techniques, especially the discounted cash flow approach (DCF approach), and to discount all expected future cash flows by a discount rate that shall represent the shareholders’ opportunity costs (Kruschwitz and Löffler 2020; Matschke and Brösel 2021; Mercer and Harms 2021, 15; Olbrich, Quill, and Rapp 2015). The core problem is that the DCF approach is based on assumptions of a neoclassical economic framework (Kruschwitz and Löffler 2020; Matschke, Brösel, and Matschke 2010; Matschke and Brösel 2021; Olbrich, Quill, and Rapp 2015).

These assumptions are clearly not realistic (Braun 2019): The neoclassical economic framework assumes that all actors are perfectly rational, they have homogeneous expectations, there are no information asymmetries between them, and markets are efficient (Hitz 2007; Schmiel and Weitz 2019). Markets are supposed to be in a stable equilibrium and free of arbitrage opportunities (Follert 2023; Ross 2005). There are no transaction costs. However, the probably most important assumption of the neoclassical economic framework is that there is a uniform and risk-free interest rate available for everyone: All market participants can lend as well as borrow any amount of money at this uniform interest rate. Thus, there are no individual differences in opportunity costs (Olbrich, Quill, and Rapp 2015). Given this set of assumptions, every asset has an unambiguous price that represents the level of indifference for every market participant (Arrow 1964; Arrow and Debreu 1954; Boulding 1935; Coase, Buchanan, and Thirlby 1938; Dean 1951; Debreu 1959; Fisher 1930; Gordon and Shapiro 1956; Graham and Dodd 1934; Hirshleifer 1958; Preinreich 1935; Quill 2020; Ross 2005; Smith 1967; Williams 1938), that is the level at which a market participant would have the same expected utility by exercising a financial transaction as by not doing so optimally. At this unambiguous equilibrium price (Follert 2023), every potential buyer of some asset is indifferent regarding buying or not. Similarly, at this price, every potential seller is indifferent regarding selling or not. This price, which represents the level of indifference for every market participant, depends on the future cash flows that the use(s) of the asset are expected to generate (Boulding 1935; Coase, Buchanan, and Thirlby 1938; Dean 1951; Fisher 1930; Gordon and Shapiro 1956; Graham and Dodd 1934; Hirshleifer 1958; Preinreich 1935; Smith 1967; Williams 1938): The future cash flows, discounted at the uniform, risk-free interest rate (discount rate), results in this unambiguous price where everyone is indifferent. The kind of asset is not specified in the neoclassical model: The model relates to whole businesses as well as to their assets and liabilities.

Taking the neoclassical economic framework as basis for goodwill measurement means understanding ‘value of goodwill’ as a residual of the value of the whole business unit over the sum of the values of its net assets within the model. We can understand this residual as the synergistic value (IVSC 2022) that cannot be assigned to the net assets and is, thus, the value of goodwill.

The following example 1 illustrates the value of goodwill within the neoclassical model in the absence of uncertainty.

3.2 The Missing Feasibility of Non-arbitrary Goodwill Impairment Tests Beyond the Neoclassical Model

In the real world, discounting future cash flows that an investment is expected to generate does not lead to the common indifference price that the neoclassical model assumes: When we just drop the assumption of the uniform, risk-free interest rate while keeping all other assumptions, the opportunity costs of all potential actors are not uniform anymore and their levels of indifference start to differ (Matschke, Brösel, and Matschke 2010; Matschke and Brösel 2021; Olbrich, Quill, and Rapp 2015).

The following example 3 highlights the effect of dropping the assumption of a uniform, risk-free interest rate.

As we see in this example, a goodwill impairment test can provide very diverse results when we just drop the assumption of a uniform discount rate. Dropping this assumption makes sense because, in reality, most market participants may be able to lend at an interest rate with little risk, but they cannot all borrow at the same rate. For instance, only AAA-rated creditors can borrow money at a AAA interest rate, but all other market participants cannot.

When we further drop the assumption of homogeneous expectations, the expected future cash flows of various market participants begin to differ (Martins, Sá, and Taborda 2023). Whether net assets or an entire cash-generating unit, the problem remains the same: Discounted future cash flows do not represent a common level of indifference anymore. Hence, also a residual between the discounted cash flow of an entire cash-generating unit and the aggregated discounted cash flow of its net assets are no unambiguous numbers anymore.

The following example 4 considers not only that opportunity costs of market participants differ, but also that they do not have homogeneous expectations.

Thus, from the perspective of A, the goodwill has a value of 19,176. From the perspective of B, its value is 7,941, and from the perspective of C, there is no positive goodwill because the immediately disposable net assets of 20,000 are worth more than the future net payment of the whole business unit.

Let us now assume that this business was acquired in the past for 30,000. Further, let us assume that there were net assets of 20,000 also in the past at the time of the acquisition (for simplification and for leaving out questions of headroom, I assume net assets to be constant over time). In case we measure goodwill in the present on the basis of the indifference prices above, the balance sheets of A, B, and C report very different impairment losses and subsequent carrying amounts of goodwill after impairment in the present (Table 6).

The central message of the examples is that, when we drop the assumption of uniform opportunity costs or the assumption of homogeneous expectations, the indifference prices can be widely different as long as we do not add reasons beyond the neoclassical model why the indifference prices should be close to each other. Beyond these two assumptions, we could also drop the other assumptions of the neoclassical economic framework and, accordingly, obtain even fewer reasons to assume similar results for different market participants.

These strict assumptions give reasons to doubt whether a goodwill impairment test based on the DCF approach could represent economic reality faithfully, because the neoclassical model proved to be unrealistic. Besides, a complete and perfect capital market would make financial accounting superfluous (Hitz 2007): Within the neoclassical economic framework, even information asymmetries are not definable because of the perfect rationality assumption and the homogeneous expectations assumption. When we drop the central assumptions of the neoclassical economic framework, there are no common indifference prices, i.e. there are no common fair values. The values of using a business unit differ among market participants, too. Furthermore, when we acknowledge that market participants have diverging expectations, someone can determine the value in use of a business unit almost arbitrarily (value in use in the sense of IAS 36^[13]): As long as we do not add reasons beyond the neoclassical model why estimated future cash flows of different market participants should be close to each other, managers have ample scope to manage reported earnings through goodwill measurement. As prior literature makes assumptions not based on theoretical reasons, but on empirical findings, such scope goes hand in hand with aggressive accounting practices (Caruso, Ferrari, and Pisano 2016; Giner and Pardo 2015; Glaum, Landsman, and Wyrwa 2018; Li and Sloan 2017; Ramanna and Watts 2012). The results of this section substantiate these empirical findings theoretically. In summary, the neoclassical understanding of value and income can neither help to substantiate the impairment-only approach, nor can it help to reject that approach because it says too little about the value of goodwill outside the model.

When we try to substantiate the impairment-only approach with alternative economic approaches beyond a neoclassical economic framework, it is hard to find theoretical reasons for the feasibility of non-arbitrary goodwill impairment tests, too. Let us look at an interpretation of markets that sees market prices for business units and assets not as a result of perfectly rational decision making, but as a result of local knowledge that is shared within social contexts like management boards. The central idea is to interpret the DCF approach, including established methods for risk-adjusting discount rates like the CAPM, as local knowledge that is subject to an ongoing evolutionary process within a social context (Campbell 1960, 1965; Knudsen 2002, 2004). For example, socially shared routines, habits, and norms are types of such evolving local knowledge (Knudsen 2002). As discounting future cash flows is not the only thinkable way of determining values; alternative approaches like valuation with multiples (Schueler 2020) are kinds of such local knowledge, too. Referring to this general idea, we can understand socially shared valuation techniques and professionals’ common opinions about them as types of evolving local knowledge.^[14] When techniques like the DCF approach are a socially shared way of determining minimum and maximum prices before selling and buying business units and assets, they can have a performative, self-fulfilling effect on market prices when market participants actually apply them, independent of whether their underlying assumptions are realistic or not (Lowe 2018; MacKenzie 2006).

However, there is a problem in the observability of evolved local knowledge about determining minimum and maximum prices. For non-arbitrary goodwill impairment testing, auditors must be able to observe this local knowledge in the specific social context (Chandler and Vargo 2011) and for the specific kind of assets or liabilities. Decision-makers of a company probably know what techniques they apply when they determine maximum and minimum prices for buying or selling several kinds of assets and whole business units. However, they are not interested in communicating them because this would weaken their negotiating position for actual transactions. Accordingly, auditors from outside can hardly observe the evolved local knowledge about determining minimum and maximum prices in sufficient detail. This opens opportunities for applying valuation techniques for goodwill accounting purposes that differ from evolved local knowledge about decision-making of the social context.

In summary, these information asymmetries lead to the result that non-arbitrary goodwill impairment testing is not feasible beyond the neoclassical model: In the real world, goodwill measurement that non-arbitrarily satisfies the central purpose of accounting from an asset-liability point of view, i.e. determining an economic value of income, is not possible. Hence, there is ample scope for reporting a biased goodwill. An alternative approach shall be developed.