The Global Practices of IFRS Reporting
4.3 Reporting Choices by IFRS-Adopting Firms
4.3.1 Measurement Choices
4.3.1.1 The Relevance of Measurement Choices Under IFRS
As described above, databases represent reported IFRS numbers that are standardized and adjusted. In addition to these adjustments, re-searchers must also understand the accounting choices that are part of a firm’s reporting practice and determined by its management before the reporting. Management choices in the measurement of assets and liabilities impact balance sheet and income statement and are shaped by a diverse set of incentives. The more discretion the accounting standards provide management with, the greater is the role of these incentives.
IFRS tend to offer a relatively high level of reporting discretion, not least because they lack a consistent measurement basis (Wüstemann and Wüstemann,2010). Chapter 6 of the IFRS Conceptual Framework introduces different measurement bases, with some of these being hardly compatible with each other. Most importantly, the framework leaves it to the design of individual standards whether a cost basis or a fair value (current value) basis is adopted.
In the absence of an overriding valuation principle, the choice of the measurement basis is part of the political process in standard setting, and the outcome varies substantially across different standards. Different standards are thus requiring different measurement bases. For example, a standard such as IFRS 9 (or formerly IAS 39) is heavily exposed to
28Cairns (2006) provides an early overview of the measurement bases used in different standards. Nobes (2015) presents a more recent update which considers the replacement of IAS 39 by IFRS 9.
lobbying by the banking industry, which generally opposes a fair value basis, at least for loans (Hodder and Hopkins,2014). Contrary to the public perception from the controversial debates around measurement issues of the standard (Laux and Leuz,2009), IFRS 9 puts relatively little emphasis on the fair value basis for financial instruments as compared to other standards (e.g., a fair value basis is dominant in IAS 41 for certain assets held by agricultural firms).
Even for the fair value basis, IFRS standards rely on very differ-ent approaches. A first difference is the frequency of remeasuremdiffer-ent.
Revaluation models for property, plant, and equipment (IAS 16) and intangible assets (IAS 38) require an infrequent revaluation (up to every five years). Under this approach, current values are only infrequently used to adjust the cost basis and the amortization scheme. In contrast, full fair value models like the one for investment property (IAS 40) or financial instruments (IFRS 9) require a frequent use of current values at each reporting date. A second difference comes from the income effect of fair value changes. These income effects are shown either in profit or loss (e.g., for investment property under the IAS 40 fair value model) or in other comprehensive income (e.g., for assets under the IAS 16 and IAS 38 revaluation model). This accounting treatment can vary within the very same standard and for the very same asset, depending on its intended use. For example, IFRS requires fair value through profit or loss for some financial instruments (e.g., equity securities held for trading) and allows for fair value through other comprehensive income for others (e.g., equity securities not held for trading).
It is essential for the interpretation of accounting performance to understand the valuation basis of assets and liabilities. The coverage of this information in databases varies substantially across different types of assets and liabilities and across industries. In particular, it is helpful to distinguish between (1) financial instruments held by financial institutions and (2) financial and nonfinancial assets held by nonfinancial entities.
4.3.1.2 Evidence on Financial Instruments Held by Financial Institutions Evidence on measurement choices by financial institutions comes from the use of the measurement categories under IFRS 9 (and formerly IAS 39). Specialized databases like BvD Bankfocus or S&P Market Intelli-gence (formerly SNL Financial) typically capture these measurement categories in detail. Therefore, we can rely on relatively broad evidence regarding the measurement of financial instruments by financial insti-tutions. We ignore nonfinancial assets and liabilities in this discussion because they play a negligible role on the average bank balance sheet.
Fiechter and Novotny-Farkas (2017) provide the most comprehensive overview of fair value measurement by global banks in the literature.
Based on hand-collected data for 907 bank-years from 46 countries over the period from 2006 through 2009, they find that on average 9.4%
(5.9%) of bank assets (liabilities) are measured at fair value through profit or loss. Banks voluntarily elected the IAS 39 fair value option for 3.2% of assets and 2.7% of liabilities; fair value through profit or loss is mandatory for the remaining parts (especially the trading portfolio and other derivatives). In addition, banks measure 7.7% of their assets at fair value through other comprehensive income, amounting to a total fair value ratio of 17.1% for assets (fair value through other comprehensive income is not a feasible measurement basis for financial liabilities). These numbers are largely at par with comparable evidence. For example, Bischof and Daske (2016) report an average fair value ratio of 23.3% for assets (with a median of 14.3%) using a European sample of 320 banks over the period from 2006 through 2010. Here, fair value through profit or loss accounts for 15.3% of assets (4.7% under the fair value option) and fair value through other comprehensive income for 8.0% of assets.
Figure 4.1 and Table 4.2 update this evidence for the post-2010 period until 2017, right before IFRS 9 becomes effective, which would mitigate the comparability of the time-series data. We retrieve the data from BvD Bankfocus for a balanced sample of 1,188 banks from
Panel A.Financial Assets at Fair Value through Profit or Loss
Panel B.Financial Liabilities at Fair Value through Profit or Loss
0%
1%
2%
3%
4%
5%
6%
7%
8%
9%
10%
11%
2011 2012 2013 2014 2015 2016 2017
FVtPL (Total) FVtPL (EU) FVtPL (Non-EU) FV Option (Total) FV Option (EU) FV Option (Non-EU)
0%
1%
2%
3%
4%
5%
6%
7%
8%
9%
10%
11%
2011 2012 2013 2014 2015 2016 2017
FVtPL (Total) FVtPL (EU) FVtPL (Non-EU) FV Option (Total) FV Option (EU) FV Option (Non-EU)
Figure 4.1:Fair value measurement of financial instruments: Evidence from
inter-national banks.
Notes: The figure reports the proportion of financial assets (Panel A) and financial liabilities
(Panel B) that are measured at fair value through profit or loss under IAS 39 (scaled by total assets). The fair value through profit or loss (FVtPL) category includes the trading portfolio, all derivatives not designated for cash flow hedges, and instruments for which
the fair value option (FV Option) is elected. The sample for Panel A includes 1,188 banks
with 8,316 bank-year observations (thereof 5,915 from the EU). The sample for Panel B includes 780 banks with 5,460 bank-year observations (thereof 4,592 from the EU). The sample covers the period from 2011 until 2017 (right before the IFRS 9 adoption). Data is retrieved from BvD Bankfocus (in July 2020).
Table4.2:Thechoiceoffairvaluethroughprofitorlossforfinancialinstruments:Evidencefrominternationalbanks MeanStandardDeviationP1P10P25P50P75P90P99 TotalSample FVtPLAssets8.0%12.5%0.0%0.0%0.3%2.4%10.7%22.8%58.1% thereof:Level157.4%37.2%0.0%0.4%18.2%68.4%93.3%98.3%100.0% thereof:Level231.5%34.6%0.0%0.0%1.2%15.3%58.7%92.0%100.0% thereof:Level310.4%21.7%0.0%0.0%0.1%1.9%7.1%34.8%99.9% FVtPLLiabilities6.9%12.5%0.0%0.0%0.0%1.2%7.8%21.1%59.3% thereof:Level19.2%22.1%0.0%0.0%0.0%0.0%4.0%31.4%100.0% thereof:Level283.5%30.1%0.0%23.7%84.5%100.0%100.0%100.0%100.0% thereof:Level36.8%21.2%0.0%0.0%0.0%0.0%0.7%12.0%100.0% EU-Sample FVtPLAssets7.4%13.0%0.0%0.0%0.1%1.2%8.6%23.4%56.1% thereof:Level165.9%34.7%0.0%6.1%36.9%82.6%95.2%98.4%100.0% thereof:Level225.1%31.3%0.0%0.0%0.8%8.5%43.1%81.2%100.0% thereof:Level37.8%17.7%0.0%0.0%0.3%1.9%5.2%16.9%93.8% FVtPLLiabilities6.4%12.1%0.0%0.0%0.0%0.6%6.6%21.2%55.7% thereof:Level19.1%21.6%0.0%0.0%0.0%0.0%4.8%31.2%100.0% thereof:Level284.7%28.1%0.0%37.3%85.0%100.0%100.0%100.0%100.0% thereof:Level36.2%19.7%0.0%0.0%0.0%0.0%0.8%11.3%100.0% Continued.
Table4.2:Continued MeanStandardDeviationP1P10P25P50P75P90P99 Non-EUSample FVtPLAssets9.4%11.2%0.1%0.7%1.9%5.9%13.0%20.8%61.7% thereof:Level143.1%36.8%0.0%0.0%4.9%39.4%78.6%97.3%100.0% thereof:Level242.0%37.3%0.0%0.0%4.4%33.6%79.4%99.2%100.0% thereof:Level314.9%26.5%0.0%0.0%0.0%1.4%15.3%60.1%100.0% FVtPLLiabilities9.4%14.2%0.1%0.7%1.7%4.7%10.6%20.6%81.8% thereof:Level19.4%23.2%0.0%0.0%0.0%0.0%3.2%31.9%100.0% thereof:Level281.3%33.4%0.0%6.2%82.3%100.0%100.0%100.0%100.0% thereof:Level37.8%23.6%0.0%0.0%0.0%0.0%0.3%15.6%100.0% Notes:Thetablereportsstatisticsfortheproportionofassets(FTtPLAssets)andliabilities(FVtPLLiabilities)thataremeasuredatfair valuethroughprofitorlossunderIAS39(scaledbytotalassets).Thefairvaluethroughprofitorlosscategoryincludesthetradingportfolio, allderivativesnotdesignatedforcashflowhedges,andinstrumentsforwhichthefairvalueoptioniselected.Thetablealsoreportsthe fractionoftotalfairvalueassets(andfairvalueliabilities)forwhichthefairvalueismeasuredatLevel1,Level2,andLevel3ofthe IFRS13fairvaluehierarchy.Thesampleincludes8,316bank-yearobservations(thereof5,915fromtheEU)forassetsand5,460bank-year observations(thereof4,592fromtheEU)forliabilitiesandcoverstheperiodfrom2011until2017(rightbeforetheIFRS9adoption).Data isretrievedfromBvDBankfocus(inJuly2020).
36 countries (i.e., 8,316 bank-year observations) for the asset data and 780 banks (i.e., 5,460 bank-year observations) for the liability data. On average, our sample banks use fair value through profit or loss for 8.0%
of assets and 6.9% of liabilities. The use of fair value through profit or loss is slightly more widespread outside the European Union (9.4%
versus 7.4% for assets and 9.4% versus 6.4% for liabilities). The sample average is lower than the evidence from the pre-2010 period suggests.
This is consistent with the time trend that Figure 4.1 displays. The graphs report average fair value ratios per year for the same sample that we use in Table4.2. Following the discussions around the financial crisis, the use of fair value through profit or loss is on a clear decline in IFRS reporting practice for both financial assets (Panel A) and liabilities (Panel B) and both inside and outside the European Union.
However, Panel A suggests that the downward trend for financial assets is driven by portfolio composition and thus mandatory fair value measurement rather than banks’ voluntary choice. Especially after the financial crisis, many banks started to reduce their trading activities for which IAS 39 mandated fair value through profit or loss. In contrast, the use of the fair value option for financial assets remains relatively stable over time, which is consistent with continuing demand for fair value information by investors. Panel B shows a different pattern for financial liabilities. The downward trend in fair value measurement is supported by financial institutions using the fair value option much less frequently for financial liabilities. It is harder to come up with a demand-side explanation for this trend because prior evidence points to investors perceiving the own credit risk effect from liability fair values reported under IAS 39 as useful information (Fontes et al., 2018;
Schneider and Tran,2015). At least compared to assets, it appears more costly to explain the potentially counterintuitive results in footnotes and conference calls (Bischof et al., 2014; Gaynor et al.,2011). This phenomenon needs explanation, and the revised IFRS 9 rules under which the own credit risk effect is separated and transferred to other comprehensive income offer a potentially interesting setting to explore this question.
It is also insightful to look at the distribution of fair value ratios.
Table 4.2 reveals that the distribution is highly skewed. A few banks
use fair value through profit or loss for a relatively large proportion of their portfolio (22.8% at the 90th percentile and 58.1% at the 99th percentile). However, the fair value ratio is almost negligible for many more banks (2.4% at the median, 0.3% at the 25th percentile, and 0.0%
at the 10th percentile). The distribution is similar for financial liabilities.
The skewness in the distribution points to a few banks with specific business models driving the average of the fair value ratio upwards.
For most banks, fair value through profit or loss does not really play a meaningful role. Some cross-country patterns that are mainly shaped by a fair value tradition in local GAAP (e.g., in Denmark; Bernardet al., 1995) also play a role and are visible in the Fiechter and Novotny-Farkas (2017) data.
In addition to the magnitude of fair value measurement on the balance sheets of financial institutions, Table4.2 and Figure4.2 provide evidence on the type of fair values, that is, the fair value levels under the IFRS 13 fair value hierarchy. Level 1 fair values (i.e., mark-to-market accounting) are prevalent for financial assets with an average proportion of 57.4% (65.9% inside the European Union and 43.1%
elsewhere). McDonoughet al. (2020) report a similar ratio of almost 50% for a small hand-collected sample of 36 IFRS-adopting banks. The high proportion should not be misinterpreted. In fact, many small banks with small trading portfolios of highly liquid securities strongly influence the sample average. The Level 1 ratio is thus negatively correlated with bank size and the magnitude of the fair value portfolio. McDonoughet al.
(2020) make a similar observation and note that this is a systematic difference to U.S. banks, where the average proportion of Level 1 fair values is much lower (less than 10%) and positively correlated with bank size. For large IFRS-adopting institutions with many complex securities in the trading books, mark-to-model accounting (i.e., in Levels 2 and 3) thus continues to play a larger role than the sample average might suggest.
At the other end of the hierarchy, mark-to-model fair values at Level 3 are relatively rare with an average proportion of 10.4% (7.75% in the European Union and 14.9% elsewhere). Again, the distribution is highly skewed, with the median firm using Level 3 fair values for 1.9% of its fair value assets only. For financial liabilities, the use of Level 3 fair
Panel A.Level 1 and Level 3 Fair Values of Financial Assets
Panel B.Level 1 and Level 3 Fair Values of Financial Liabilities
0%
10%
20%
30%
40%
50%
60%
70%
2011 2012 2013 2014 2015 2016 2017
% Level 1 Total
% Level 1 EU
% Level 1 Non-EU
% Level 3 Total
% Level 3 EU
% Level 3 Non-EU
0%
2%
4%
6%
8%
10%
12%
14%
16%
2011 2012 2013 2014 2015 2016 2017
% Level 1 Total
% Level 1 EU
% Level 1 Non-EU
% Level 3 Total
% Level 3 EU
% Level 3 Non-EU
Figure 4.2:Fair value levels: Evidence from international banks.
Notes: The figure reports the proportion of financial assets (Panel A) and financial liabilities
(Panel B) that are measured at Level 1 and Level 3 fair values (scaled by total fair value assets and total fair value liabilities, respectively). The sample for Panel A includes 1,152 banks with 8,064 bank-year observations (thereof 5,040 from the EU). The sample for Panel B includes 738 banks with 5,166 bank-year observations (thereof 3,381 from the EU).
The sample covers the period from 2011 until 2017 (right before the IFRS 9 adoption). Data is retrieved from BvD Bankfocus (in July 2020).
values is even less frequent. The distribution shows that more than half of our sample banks do not even use them at all (both in the European Union and elsewhere). At the 75th percentile, the average proportion of Level 3 fair values for financial liabilities is as low as 0.7%. Therefore, in summary, it is a misconception that fair value accounting and internal fair value estimates dominate banks’ balance sheets (Laux and Leuz, 2010).
4.3.1.3 Evidence on Assets and Liabilities Held by Nonfinancial Entities Empirical evidence on the choice of the measurement basis is less readily available for the nonfinancial industry. Standard databases such as Compustat Global do not cover the use of valuation options for assets and liabilities (McDonough et al., 2020), including financial instruments (Gebhardt, 2012). Therefore, we have to rely on hand-collected evidence that is naturally coming from diverse samples and settings. Table4.3 provides an overview of several studies that examine measurement choices by nonfinancial firms. Five standards are most frequently studied: IAS 16 (Property, Plant, and Equipment), IAS 38 (Intangible Assets), IAS 39 (Financial Instruments), IAS 40 (Investment
Property), and IAS 41 (Agriculture).
Even though the evidence comes from different periods and different regions (see Table4.3 for details), the picture is relatively consistent.
Fair value is the dominant valuation basis for agricultural products under IAS 41 (Cairnset al.,2011; Huffman,2018). The dominance of fair value measurement comes from consumable biological assets for which amortized cost measurement is very rare,29 whereas cost basis remains important for bearer biological assets.30The difference between consumable and bearer biological assets is consistent with the greater
29Strictly speaking, IAS 41 does not introduce a fair value option but requires fair value measurement for consumable biological assets unless their fair value cannot be reliably estimated.
30Consumable biological assets are cultivated for sale (e.g., crops) and can thus be characterized as commodities. Bearer biological assets are self-regenerating and thus continuously contribute to the production of agricultural output (e.g., trees in a plantation); see Huffman (2018) for details. Since 2014, bearer biological assets are accounted for like property, plant, and equipment in the scope of IAS 16 once their growth period is completed.
Table4.3:Theuseofthefairvalueoption:Aliteraturereview %Firms%AssetsSample UsingMeasuredatSizeSampleSampleSample StudyOptionFVtPL(#Firms)IndustryPeriodOrigin IAS16(Property,Plant, andEquipment) Cairnsetal.(2011)2.5%N/A447All2004–2006Australia,UK ChristensenandNikolaev (2013)3.3%N/A1,508All2005–2006Germany,UK KvaalandNobes(2010)8.2%N/A207All2005–2006Australia, France, Germany, Spain,UK KvaalandNobes(2012)7.1%N/A212All2008–2009Australia, France, Germany, Spain,UK IAS38(IntangibleAssets) Cairnsetal.(2011)0.0%N/A195All2004–2006Australia,UK ChristensenandNikolaev (2013)0.0%N/A1,397All2005–2006Germany,UK Continued.
Table4.3:Continued %Firms%AssetsSample UsingMeasuredatSizeSampleSampleSample StudyOptionFVtPL(#Firms)IndustryPeriodOrigin IAS39(Financial Instruments) Gebhardt(2012)12.3%3.6%154Non-Financial2009–2010Europe KvaalandNobes(2010)18.1%N/A155Non-Financial2005–2006Australia, France, Germany, Spain,UK KvaalandNobes(2012)19.3%N/A161Non-Financial2008–2009Australia, France, Germany, Spain,UK IAS40(Investment Property) Cairnsetal.(2011)82.9%N/A41All2004–2006Australia,UK ChristensenandNikolaev (2013)47.3%N/A275All2005–2006Germany,UK Israeli(2015)67.4%N/A86RealEstate2005–2010France, Germany, Italy,Spain Continued.
Table4.3:Continued %Firms%AssetsSample UsingMeasuredSizeSampleSampleSample StudyOptionatFVtPL(#Firms)IndustryPeriodOrigin KvaalandNobes(2010)33.0%N/A88All2005–2006Australia, France, Germany, Spain,UK KvaalandNobes(2012)34.4%N/A93All2008–2009Australia, France, Germany, Spain,UK Mülleretal.(2015)85.3%80.0%245RealEstate2003–2012Europe QuagliandAvallone (2010)52.6%N/A76RealEstate2005–2007Europe IAS41(Agriculture) Cairnsetal.(2011)82.4%N/A17All2004–2006Australia,UK Huffman(2018)72.1%N/A183All2003–2014Global(35 countries) Notes:Thetablesummarizesevidencefromprioraccountingliteratureontheuseofafairvalueoptionbynon-financialfirmsorfornon- financialassets.ThefairvalueoptionincludestherevaluationmodelunderIAS16andIAS38aswellasthefairvaluemodelunderIAS39, IAS40,andIAS41.%FirmsusingFVOptionreportstheproportionofsamplefirmswhichelectthefairvalueoption.%AssetsMeasured atFVtPLreportstheaverageproportionofassetsthataremeasuredatfairvaluethroughprofitorlossbythesamplefirms.Thetablealso reportssampledetailsforeachstudy.
availability of fair values for commodities (i.e., consumable assets), for which external markets typically exist, as compared to the availability of fair values for in-use production equipment (i.e., bearer assets), which are rarely traded.
The fair value basis is also an option for most investment property under IAS 40. Unlike IAS 41, IAS 40 offers an explicit accounting option, which is the choice between the fair value model and the cost model. The IAS 40 evidence thus reflects the outcome of an actual management choice. The preference for the fair value model that likely reflects investor demand is particularly strong if a company is specialized in real estate business (e.g., Quagli and Avallone,2010). The cost model tends to be more prevalent in more diverse samples (see Table4.3) that include less specialized firms where investment property is just one part of the business model, potentially even a minor one (e.g., Christensen and Nikolaev,2013).
Across different business models, the use of the fair value model for investment property is substantially more common than the choice of the revaluation model for property, plant, and equipment (in the scope of IAS 16) and, even more so, intangible assets (in the scope of IAS 38). The evidence is very consistent and shows that there is a very small fraction of firms (less than 10% of the sample population) that chooses the revaluation model for at least one class of property, plant, and equipment and hardly any firm at all that chooses the model for intangible assets (0 out of 1,397 sample firms from Germany and the United Kingdom according to Christensen and Nikolaev, 2013). These firms’ choices indicate the existence of a market equilibrium in which a fair value basis for these highly illiquid nonfinancial assets appears to be prohibitively costly (relative to a cost basis).
Finally, nonfinancial firms invest in financial instruments, and their choice of the IAS 39 and IFRS 9 fair value options is fully equivalent to financial institutions (Gebhardt, 2012). The proportion of firms voluntarily electing the fair value basis varies between 12.3% in a larger European sample (Gebhardt,2012) and 19.3% in a smaller global sample (Kvaal and Nobes,2012). These rates suggest that demand for fair value information is not confined to the specific business model of financial institutions.
4.3.1.4 Key Determinants of Measurement Choices
Heterogeneity in the reporting practices under IFRS has two dimensions.
The evidence suggests that the choice of the measurement basis varies across different firms (even if holding the business model and the asset portfolios constant) and across different asset classes (within the same firm). Both asset and firm characteristics thus help explain valuation choices.
The evidence on asset characteristics is relatively clear. The net benefit of choosing the fair value basis varies with the availability of fair value estimates. Fair value estimates are readily available for assets that can be traded on active markets. Costs of estimating fair values increase if no such sales markets exist (e.g., Barker and Schulte, 2017; Müller et al., 2015). If observable market prices do not even exist for similar assets, management has to set up potentially complex valuation models, often with support from external consultants and greater diligence in the external audit (Goncharovet al.,2014). The higher costs of obtaining the fair value estimates tend to be correlated with lower reliability, because managerial discretion becomes greater if market evidence cannot be used to support and verify the internal estimates (Christensen and Nikolaev,2013). Even if market prices exist, they fail to provide reliable evidence for the asset’s value if the asset primarily generates value through internal use and its illiquidity stems from a management decision. Investors do not perceive a sales price as useful information about these kind of “in-use assets” (Huffman, 2018).
Therefore, the less frequent use of the fair value basis for illiquid assets is also attributable to the lower demand for this kind of information.
The evidence on firm characteristics is more nuanced and varies across countries. One important factor, which often explains the mea-surement choice, is the pre-IFRS reporting practice under local GAAP (e.g., Kvaal and Nobes,2010; Stadler and Nobes,2014). For example, firms from the United Kingdom, which had a similar option under UK GAAP before, use the revaluation option for property, plant, and equipment under IAS 16 more often than German firms, which were confined to the cost model under local German GAAP (Christensen and Nikolaev, 2013). Similarly, almost all real estate firms from the
United Kingdom continued to use the fair value model under IAS 40, which they were required to use under UK GAAP before (Danbolt and Rees,2008). In contrast, there is substantial within-country variation among German real estate firms, which were all used to applying the cost model under local German GAAP (Müller et al., 2015). Therefore, pre-IFRS differences in reporting practice across countries tend to per-sist in the IFRS period and even more so where IFRS permit different accounting options. This observation points to strong incentives for firms to avoid switching costs from changing their accounting methods.
These switching costs tend to be lower when a firm is using fair values for internal performance measures.
Apart from this, individual measurement choices, such as the use of a fair value option, are often associated with a firm’s commitment to transparency (e.g., Müller et al.,2015). Firms that benefit more from transparency are more likely to commit to more costly measurement on a timely fair value basis. A firm’s commitment to transparency originates from a host of factors, such as capital structure, exposure to capital market pressures, intensity of internal monitoring through effective corporate governance, institutional characteristics, and strength of local market supervision. These factors are rarely IFRS-specific and closely related to the reporting incentives that shape the overall transparency of firms (e.g., Burgstahleret al.,2006; Bushman and Piotroski,2006).