IFRS standards use too many different terms to describe how likely it is that an event will occur. That is a clear conclusion of KASB Research Report No. 39 / AASB Research Report No. 2 Accounting Judgements on Terms of Likelihood in IFRS: Korea and Australia, issued in 2016 by the Korea Accounting Standards Board (KASB) and Australian Accounting Standards Board (AASB). That report summarises joint research by KASB and the AASB into how preparers of financial statements and auditors interpret various terms used in IFRS Standards to denote how likely an event is. The report calls those terms ‘terms of likelihood’.
Those terms play an important role in IFRS Standards becomes many of those standards use those terms in setting criteria for recognising assets or liabilities or for disclosing information.
In a future post, I will discuss how to take forward the recommendations in the report. In the meantime, this post summarises the research report under the following headings:
- key findings
- recommendations in the report
- nature of the research
- questions asked in the research
- detailed findings on the 13 terms
- other findings
I encourage researchers to investigate how IFRS standards use language and how IFRS standards are translated.
Here is my summary of the key findings listed in the report:
- Korean and Australian accounting professionals differ in how they interpret terms of likelihood and in how they rank different terms of likelihood.
- Some respondents interpret terms of likelihood differently in different contexts. For example, some respondents interpret the term ‘probable’ as referring to a lower probability when considering liability recognition than when considering asset recognition. The report acknowledges that further research would be needed to assess whether differences in interpretation lead to different financial reporting outcomes.
- Respondents attach the same meaning to some different terms of likelihood, for example ‘probable’ and ‘likely’.
- Some terms of likelihood (for example, ‘virtually certain’) communicate their meaning more efficiently than others (for example, ‘possible’).
- Korean accounting professionals interpret some terms of likelihood differently in different languages. Thus, there may be a translation issue that should be addressed.
- Some terms of likelihood cannot be translated into distinct Korean terms. For example, ‘probable’ and ‘likely’ are translated into a single Korean expression. Similarly, ‘virtually certain’ and ‘reasonably certain’ are also both translated into a single Korean term.
In response to an explicit question in the survey, about 40% of Korean respondents (both preparers and auditors) said they find terms of likelihood in IFRS difficult or very difficult to understand. In Australia, though, this was the case for only 3% of preparers and 6% of auditors.
Recommendations in the report
Here is my summary of the report’s recommendations. The International Accounting Standards Board (IASB) should:
- when developing a standard, give considerable attention to how people might interpret and translate terms of likelihood in different jurisdictions, particularly if differences in interpretation or translation could cause material differences between financial statements.
- use fewer different terms of likelihood in standards, perhaps limiting terms used to a limited set.
- be aware, in revising the Conceptual Framework for Financial Reporting, that many preparers and auditors factor in their own level of ‘conservatism’ when applying IFRS Standards’.
(The Board completed that revision in 2018, after the KSB and AASB issued the report.)
- consider developing principles and guidance on terms of likelihood (perhaps including examples) for consistent application across the standards.
- explicitly seek input on translation and interpretation issues in different jurisdictions when undertaking standard-setting outreach and consultation.
Nature of the research
The report lists (in Appendix A) 35 terms of likelihood used in IFRS Standards. The research focussed on 13 of those terms.
The 13 terms studied did not include 2 important terms sometimes used in IFRS Standards: ‘expected’ and ‘more likely than not’. I will discuss those 2 terms in a future post.
The research carried out a survey asking respondents to say:
- for each of 16 paragraphs within IFRS Standards, which probability point (on a scale from 0% to 100%) corresponds to the ‘term of likelihood’ used in that paragraph. The paragraphs covered all 13 terms of likelihood studied. I summarise those paragraphs in a separate post today. https://accountingmiscellany.com/saying-how-likely-something-is-research-methodology
- for each of the 13 terms, what range of probabilities (on a scale from 0% to 100%) corresponds to that term.
The survey was based on a questionnaire available in either English or Korean. 504 Korean accounting professionals responded to the survey (183 auditors and 86 preparers for the Korean version; 144 auditors and 91 preparers for the English version) and 208 Australian accounting professionals (88 auditors and 120 preparers) responded to the English version.
Detailed findings on the 13 terms
I summarise the main findings using 2 figures reproduced from the report and a table:
- Figure 15 summarises the point estimates respondents provided for the meaning of the 13 terms in the context of the 16 paragraphs from IFRS Standards. It shows the means of the percentages provided by respondents.
- Figure 17 presents the ranges attributed by respondents to each of the 13 terms of likelihood.
- Following figures 15 and 17 are my comments on what those figures show.
- Table 1 below is drawn from table 9 of the report, which clusters most of the 13 terms studied into one of 5 groups, labelled A to E.
Comments on what figures 15 and 17 show
Virtually certain / substantially all / highly probable / reasonably certain / reasonably assured:
- Both Australian respondents and Korean respondents viewed ‘virtually certain’ as denoting a probability of around 90%.
- For Australian respondents, ‘substantially all’ had a slightly lower mean (90%) and slightly narrower range than those for ‘virtually certain’ (92%). For Korean respondents, the gap from ‘virtually certain’ (mean 90%) down to ‘substantially all’ (mean 85%) was slightly more than for Australians.
Comment: the term ‘substantially all’ does not denote the likelihood of an outcome. Instead, as the report acknowledges, this term ‘refers to the proportion of something’. (The questionnaire gave respondents the example of ceasing to capitalise borrowing costs when substantially construction activities.) Perhaps for this reason, the range was wider (and for Korean respondents, much wider) for ‘substantially all’ than for ‘virtually certain’.
- Both Australian respondents and Korean respondents viewed ‘highly probable’ as denoting a slightly lower probability than ‘virtually certain’ (around 3% to 9% lower). For ‘highly probable’, the range was wider than for ‘virtually certain’, for both Korean and Australian respondents. Both top and bottom of the range for ‘highly probable’ were lower for Australian respondents than for Korean respondents.
- Koreans viewed ‘reasonably certain’ (mean 90%) as a synonym for ‘virtually certain’ (also 90%), well above ‘highly probable’ (mean 86%) and even further above ‘reasonably assured’ (79%). In contrast, Australians seemingly view ‘reasonably certain’ (mean 81%) as a near synonym for ‘highly probable’ (83%) and for ‘reasonably assured’ (76%). Moreover, for Australians, ‘reasonably certain’ and ‘reasonably assured’ had an almost identical range, but for Koreans the two ranges did not overlap at all.
- For 3 terms denoting probabilities above 50%, Korean respondents produced means noticeably higher than Australian respondents did: reasonably certain (90% v 81%); probable (71% v 62%); reasonably possible (65% vs 57%)
Probable / likely
- Australians attributed roughly the same probability to ‘probable’ (mean 62%) as to ‘likely’ (64%). In contrast, the Korean means were slightly higher for ‘probable’ (71%) and slightly lower for ‘likely’ (58%), with those 2 means differing by 13 percentage points.
(There may be a typo in the Korean range reported for ‘likely’ because (a) it is identical to the range given for ‘probable’; (b) the low point is above the mean and median; and (c) the mean and median are 15 percentage points below the mean and median for ‘probable’.)
- To the best of my knowledge, the IASB never intended ‘likely’ to denote a threshold different from ‘probable’. The IASB used the term ‘likely’ rather than ‘probable for a few years in the early 2000s. This was mainly because of concerns that the term ‘probable’ put some readers off, by appearing to suggest excessively technical and rigorous statistical precision.
- Paragraphs BC81 of the Basis for Conclusions on IFRS 5 Non-current Assets Held for Sale and Discontinued Operations and BC44 of the Basis for Conclusions on IFRS 15 Revenue from Contracts with Customers discuss whether the term ‘probable’ has the same meaning in IFRS Standards as in US Generally Accepted Accounting Principles (US GAAP), and how that term relates to some other terms denoting likelihood. The report does not mention this discussion.
Possible / reasonably possible
- Both Australians and Koreans place ‘reasonably possible’ well above ‘possible’, but for Koreans the difference in means is greater (25%) than for Australians (14%). Indeed, Koreans place ‘reasonably possible’ (mean 65%) even above ‘likely’ (58%).
- For both Australians and Koreans, even the low point of the range for ‘reasonably possible’ lies above 50%. I found that result surprising and disturbing. The context provided to respondents was the requirement to disclose a sensitivity analysis showing the effect of reasonably possible changes in a risk variable. Above such a high probability threshold, such effects may well be reflected in recognition and measurement already; if so, there would be no effect to disclose, and setting such a high probability threshold could nullify the disclosure requirement.
- For Australians, the range ‘reasonably possible’ overlaps greatly with the range for ‘possible’. But for Koreans the bottom of the range for ‘reasonably possible’ lies above the top of the range for ‘possible’. This suggests that many Australians see these 2 terms as synonyms but that all or most Koreans see ‘reasonably possible’ as denoting a higher probability than ‘possible’.
- For ‘possible’, the top end of the range lies well above 50% for Australians but below 50% for Koreans. The bottom end of the range lies close to the top of the range for ‘unlikely’ (just above for Australians but just below for Koreans).
Unlikely / highly unlikely / extremely unlikely / remote
- Australian respondents attributed much higher percentages than Korean respondents did to 2 terms denoting probabilities below 50%: unlikely (28% v 12%); highly unlikely (24% v 15%).
- Oddly, the Korean mean for unlikely (12%) is slightly lower than the Korean mean for ‘highly unlikely’ (15%). But this unexpected result may be a statistical anomaly, because both the top and bottom of the range are higher for ‘unlikely’ than for ‘highly likely’.
- For ‘unlikely’, almost all of the lower end of the range overlaps with the whole of the range for ‘highly unlikely’ for Australians. This suggests that some Australians may see them as near synonyms. On the other hand, there is almost no overlap in these 2 ranges for Koreans.
- For both Australians and Koreans, the range for ‘unlikely’ is much wider than the ranges for ‘highly unlikely, ‘extremely unlikely’ and ‘remote’.
- The ranges for ‘remote’ and ‘extremely unlikely’ overlap almost completely with each other—and for Koreans also with the range for ‘highly unlikely’. But for Australians, the bottom of the range for ‘highly unlikely’ is a little above the top of the range for ‘remote’ and ‘extremely unlikely’. It appears that both groups see ‘remote’ and ‘extremely unlikely’ as synonyms of each other and that Koreans see ‘highly unlikely’ as yet another synonym; but Australians seem to attribute a higher probability to ‘highly unlikely’ than to the other 2 terms.
- Both Australian and Korean report a median of 10% for ‘highly unlikely’ and 5% for ‘remote’. In each case, the means are rather higher, suggesting skewed distributions (‘highly unlikely’: 24% in Australia and 15% in Korea, ‘remote’ around 10% for both Australians and Koreans.)
(For most other terms investigated, the medians were close to the means.)
Some minor caveats
The overall messages from the research report are clear:
- IFRS Standards use too many different ‘terms of likelihood’; and
- Different people using IFRS Standards interpret those terms in different ways.
I agree with those messages. I will explain in a future post what I think is a viable way for the IASB to address those messages.
I do have some reservations about some of the detail in the findings:
- Some of the 16 paragraphs used to elicit respondents’ point estimates of the meanings of ‘terms of likelihood’ did not, in fact, refer to a probability threshold that has a direct and unequivocal accounting consequence. I comment on those paragraphs in a separate post today. https://accountingmiscellany.com/saying-how-likely-something-is-research-methodology
- In my view, it is not clear what the questionnaire meant by ‘range’. I also comment on that problem in my separate post today.
Grouping the terms
Table 1 is drawn from table 9 of the report. It shows the 13 terms in (approximately) descending order of probability and clusters most of the 13 terms studied into one of 5 groups, labelled A to E. The report states that each term within any one group seemed to have a similar meaning, not differing at the 1% significance level from any other term within that group. The report suggests that there might only be a need for one term within each group.
For ease of comparison, because the report did not identify a Korean group corresponding to Australian group C, I relabelled Korean groups C and D as D and E in table 1.
|Virtually certain||Group A||Group A|
|Substantially all||Group A||Group B|
|Highly probable||Group B||Group B|
|Reasonably certain||Group B||Group A|
|Unlikely||Group D||Group D|
|Highly unlikely||Group D||Group D|
|Extremely unlikely||Group E||Group E|
|Remote||Group E||Group E|
Comments on the grouping
- For Australian respondents, ‘substantially all’ is in Group A and ‘reasonably certain’ is in Group B. But for Korean respondents, ‘substantially all’ is in Group B and ‘reasonably certain’ is in Group A.
- ‘Reasonably assured’, ‘reasonably possible’ and ‘possible’ do not fall into any group.
- For my comments above on figures 15 and 17, I grouped the terms into 4 clusters, covering: (1) groups A and B, and ‘reasonably assured’; (2) group C; (3) ‘reasonably possible’ / ‘possible’; and (4) groups D and E.
I summarise below some other findings of the research, on:
- ‘Probable’ v ‘no longer probable’
- Communication efficiency
The research investigated whether respondents apply the term ‘probable’ in the same way (‘symmetrically’) to both assets and liabilities. Table 2 summarises the results (extracted from table 8 in the report.)
|Probable||Recognition of liability IAS 37||62%||71%|
|Probable||Recognition of asset IAS 38||64%||75%|
|Remote||Recognition of asset IAS 16||10%||10%|
|Remote||Disclosure of liability IAS 37||11%||27%|
Note to table 2: Korean uses the same term to translate both instances of ‘probable’ but uses 2 different terms to translate the 2 instances of ‘remote’.
The report states that there are statistically significant mean differences in the interpretation of ‘probable’ at the 0.05 level and ‘remote’ at the 0.01 level. The report also says that it does not assess whether differences are economically significant.
In my view, only one difference in table 2 could have an economically significant effect: the difference in Koreans’ interpretation of remote (10% for recognition of an asset but 27% for disclosure of a liability). And maybe one driver of this difference in interpretation was the use of 2 different Korean terms to translate the 2 instances of ‘remote’.
‘Probable’ v ‘no longer probable’
Respondents gave a point estimate around 20 percentage points higher for an event that was considered ‘probable’ than for an event that was considered ‘no longer probable’:
- probable: mean of 62% (Australians) and 71% (Koreans)
- no longer probable: mean of 42% (Australians) and 48% (Koreans)
The standard deviations of the percentages reported by respondents were almost twice as large for ‘no longer probable’ as they were for ‘probable’. The report suggests a linguistic explanation for this difference: that this difference in probabilities may be because using negative expressions makes it harder for readers to determine the intended meaning.
I agree that using negative expressions often does make things harder for readers. Nevertheless, I suspect that the main cause of this difference might be a technical uncertainty: whether derecognition decisions should reflect ‘stickiness’: requiring more evidence for derecognition than for recognition.
The report reviews briefly whether the terms communicate their meaning efficiently enough to achieve consensus in interpretation between individuals. It estimates this using 2 measures:
- the range between the means of the lower and upper numerical probabilities (depicted in figure 17, reproduced above).
- the standard deviation of the estimates. This shows a pattern broadly similar to figure 17.
As figure 17 shows, the range is smaller for the terms covering probabilities near the ends of the probability scale (such as ‘virtually certain’ and ‘remote’) and is wider near the middle of the scale (for example, ‘probable’ and ‘possible’. This result does not seem surprising.
Because English is not used widely in Korea, Korea uses a translation of IFRS, known as K-IFRS. Table 1 in the report lists the Korean translations of the 13 terms studied in the reported.
The report makes several interesting comments on translation issues:
- Korean translates: ‘probable’ and ‘likely’ into a single Korean term; ‘virtually certain’ and ‘reasonably certain’ into a single term; and ‘highly unlikely’ and ‘extremely unlikely’ into a single term.
- Korean uses 2 different expressions in translating the English term ‘remote’. The report suggests that ‘the absence of direct equivalence of expressions between the two languages might suggest a lack of equivalence between the underlying concepts of the two languages.’
- Korean responses based on the Korean version of the survey differed in some respects from Korean responses based on the English version. (The report does not explore whether the existence of the Korean version may have influenced Korean responses to the English version—or indeed vice versa.)
This very useful report confirms that IFRS Standards use too many different terms to denote levels of probability and that there are differences in how different people interpret some of those terms. Using so many different terms:
- increases the risk that companies will apply the standards inconsistently; and
- makes it unnecessarily difficult to apply IFRS Standards and to translate the standards from English into other languages.
In a future post, I will discuss how I think the IASB could reduce the number of terms used and also make the terms used more understandable. Update: that post is now at Saying how likely something is: way forward – Accounting Miscellany
There is scope for a lot more research on how IFRS standards use language and on how IFRS standards are translated. I encourage researchers to investigate these topics.
Sign up for updates
To get new posts on this blog, please go to the home page https://accountingmiscellany.com and sign up for email alerts.
This post also appears on my blog Language Miscellany at https://languagemiscellany.com/2022/07/saying-how-likely-something-is
If language interests you, please have a look at Language Miscellany!