Alumni refresher lecture series
2009 – A tipping point
Research that informs the standard setting process
Academic research can help resolve many of the nettlesome issues that standard setters face, and play a central role in shaping global financial reporting
(pages 37-43 printed journal)
Research and standard setting
Research can inform standard setting issues for a variety of reasons. First, accounting research helps standard setters identify the issues with which they should be grappling. Second, academic research can help structure standard setters' thinking about the issues. Third, research can provide evidence on standard setting issues.
Research also provides unbiased analysis to standard setters. Academics are one of the few groups who have no stake in the outcome of any standard setting decision. Companies care about standard setting decisions because they must follow the standards. Auditors care about the standards because they have to audit the required information. Analysts want the standards to require information that they use in their own analysis. Further, academics are trained in economics and understand the role of information in capital markets. It is information economics that underlies standard setting.
In 1974, Gonedes and Dopuch published a very influential paper that maintains that accounting standards are public goods and involve externalities. Thus, it is not possible to determine the desirability of any particular accounting standard simply by looking at the relation between share prices, or equity returns, and accounting amounts - even if markets are fully efficient. Determining the desirability of any particular standard requires specifying social preferences and the inevitable trade-offs between the winners and the losers. Standard setters develop the framework and specify in it what is important in making standard setting decisions. Academic researchers then take those criteria, operationalise them in their research design, and indicate to the standard setters how particular aspects of various accounting treatments embody those criteria. Research informs standard setting, but cannot answer the question, 'What should the standard be?'
The many research questions
The good news for researchers is that there is no end to the questions to which we do not know the answers. This applies to technical agenda topics as well as cross-cutting issues (the term used by standard setters for issues that apply to many topics) that include: fair value, recognition versus disclosure, the role of uncertainty and risk, relevance and faithful representation, the role of incentives and judgement in financial reporting, the distinction between liabilities and equity, user needs, and cost and benefits.
Another important research topic is the globalisation of financial reporting. Countries such as Australia have adopted International Financial Reporting Standards (IFRS) to achieve, hopefully, some benefits; and academic research is trying to identify those benefits.
Fair value is one of the cross-cutting issues that arises in almost every International Accounting Standards Board (IASB) agenda - it is currently the focus of the Financial Accounting Standards Board (FASB) and the IASB in several joint projects. The growing academic literature on fair value considers almost every topic in financial reporting. Whenever the issue of measurement arises, fair value is considered. The IASB's current conceptual framework does not include much guidance relating to making measurement decisions, but a framework chapter on measurement is being developed. Fair value also arises in financial instruments, non-financial liabilities, and pensions. It is also relevant for revenue, insurance, and leases - again, just about every topic on the agenda.
Standard setters often look to fair value as a possible measurement basis, mainly because fair value provides information that assists in making economic decisions. Fair values are nothing more than the present value of expected future cash flows. The discount rate reflects the uncertainties inherent in those cash flows. Although fair value is not always the right measurement basis, it should be on the list to be considered.
Fair value is also consistent with the definitions of assets and liabilities. The definitions focus on the future - inflows in the case of assets and outflows in the case of liabilities. Fair value is relevant because it reflects expectations of the future adjusted for the time value of money and risk. Fair value requires taking an unbiased, market participant view rather than a management view of a favourable picture of the company.
Moreover, fair value possesses many of the qualitative characteristics of accounting information specified in the framework - relevance and faithful representation. The current measure of something will be more relevant to users making decisions than a twenty-year old, modified historical cost number. Comparability means that like things should look alike and different things should look different. Fair value would accomplish this. The mixed attribute model that we have today masks some economic mismatches and creates accounting mismatches, which is why we are then forced to use hedge accounting.
Not everyone believes fair value is the best measurement basis. An important concern is the scope for error in estimating fair value, particularly for non-traded instruments, and therefore scope for management to exercise discretion. The response of the advocates of fair value is that almost every number in the financial statements requires the exercise of judgement. Fair value puts a market discipline on the exercise of that judgement. Some are concerned about recognising holding gains and losses in earnings. They view this as an opportunity cost approach to financial reporting, which they regard as inappropriate, arguing that if users' models are designed to cope with the measures we currently report, a different view on financial reporting will be confusing to users. Others consider there is too much earnings volatility. But when does earnings volatility become 'too much'? In the current financial crisis, the concern about fair value also includes concerns about liquidity effects, pro-cyclicality, and the effects of financial reporting on bank capital.
The outcome of the IASB project would not require any greater use of fair value than applies currently. It would only clarify the meaning of 'fair value'. Let me explain some of the common sources of confusion. Fair value is an exit price from the perspective of the company. It is the price in the sale of an asset or the transfer of a liability. Fair value assumes either a sale or use in a business, both based on a market-participant perspective. It is not a liquidation notion.
Differences are sometimes made between exit price and entry price, but these are the same for a given asset or liability when they relate to the same item, on the same date, and in the same market, and when buyers and sellers have the same information. While the two prices can be different in different markets, the fair value measurement definition applies to the most advantageous market.
My own research on fair value
I will describe some of my own research on fair value. One study2 addresses fair value accounting for liabilities and own credit risk. The motivating question is, 'Do changes in credit risk affect the fair value of a liability?' Many do not think that credit risk changes should be reflected in fair value. The secondary question is, 'If we were to recognise the effects of own credit risk on liabilities, would net income be misleading?' The concern here is that when a company's financial condition deteriorates, its credit risk increases and the value of its liabilities decreases.
The first research question is whether the effect of credit risk changes on equity returns are mitigated by leverage. That is, when the company is in trouble, equity returns are negative. But is that negative return smaller when the company has more debt because the debt holders share in the fall in equity value? If so, then the presence of debt reduces the decline in equity value. The second research question is, 'How would net income, or profit or loss, differ if fair values of debt were recognised?'
On the first question, we estimate the relation between equity returns and the change in credit risk interacted with leverage to see how leverage affects the relation. We also estimate the change in debt value by looking at the financial statement disclosures of the maturities of debt - the cash payments due in each of the next five years, the following five years, and the next five years. We discount these amounts using the interest rate that is commensurate with the company's credit risk at the beginning of the year, we discount them again using the interest rate commensurate with the credit risk at the end of the year, and then we see how the amounts differ. We find that equity returns are less negative when credit risk increases, when the company has more debt. Interestingly, we find this result for most credit risk levels, not just the highest. We know from theory that the credit risk effect is greatest when the company is in severe financial distress. However, the question arises whether one can even detect this effect when the company is not in financial distress. Our findings reveal that the answer is 'yes', and confirm that equity holders gain when credit risk increases. Thus, the seemingly counterintuitive effect is not counterintuitive at all - it is an economic fact.
Regarding the second question as to how net income would differ if fair value of debt were recognised, we obtain asset and liability value estimates by inverting the Merton model. We then restate net income first to reflect all changes in fair values and then to reflect only changes in debt fair values. We find that when we consider all changes in value, as one would expect, income is lower for companies that have credit downgrades - whose credit risk has increased - and the opposite for companies with reduced risk. These findings tell us that the model is working. Also consistent with what we would expect when we only recognise debt value changes, we find that income is higher for downgrade firms and lower for upgrade firms. This anomalous effect is of concern to some. But when the company is in financial trouble, accountants write down assets. The question then becomes, 'Are those write-downs big enough to absorb this credit?' Importantly, we find that the answer is 'yes' - recognised asset write-downs exceed debt value gains for most firms. This finding tells us that the concerns about anomalous income effects are warranted because we do not recognise all assets. Since we do not write up all assets, we cannot write them down when things go bad.
The second example of my research is an examination of the relation between fixed asset revaluations and future company performance.3 The motivating questions here are: 'Are asset revaluation amounts reliable estimates of asset fair values?' and 'Do managers exercise their discretion, so as to render the fair value estimates unreliable?' Lack of reliability of fair value estimates is one concern some have about fair value. The research question we address in the study is, 'Do upward asset revaluations explain changes in future operating performance, where we define operating performance as future operating cash flows and future operating income?'
The notion behind this research question is that if asset revaluations explain future operating performance, then the revaluations must be reflecting part of the value of the asset. If the revaluation amounts were totally unreliable, or if managers were managing the revaluation numbers, we would not detect a relation. In this study, we look at UK firms with upward asset revaluations and look for incremental explanatory power in a regression of changes in realised future performance.4
We find that revaluation amounts are significantly associated with changes in future operating performance; both future operating income and operating cash flows, one, two, and three years ahead. Revaluation balances are associated with share prices and revaluation increments are associated with returns. We find that the relation is less positive for higher debt to equity firms. The interpretation from this study is that fair values of fixed assets are reliable as reflected in future operating performance, and changes in fair values are relevant to investors and reflect timely changes in asset values. We do find evidence of management discretion, but the effects of discretion do not eliminate the relevance and reliability of the revaluations.
The lessons and questions from fair value research
What have we learned from fair value research taken as a whole? The studies I have described are just two examples in a very large literature. Just about everywhere academics look, fair values are relevant and reliable enough to be reflected in investors' valuations. This finding applies to financial instruments, tangible assets, and intangible assets such as brands. We also find evidence of managers exercising discretion in estimating fair values. Nevertheless, the discretion never seems to be enough to eliminate the overall relevance of fair value amounts. Despite the large literature that exists, there are many open questions. Perhaps the biggest question is whether we can measure fair values reliably. Can we do it well enough, to justify putting the estimates in financial statements? Is fair value the right measurement basis? If it is, is fair value the right measurement basis for all assets and all liabilities, or only some? If only some, which ones? If not fair value, then what? This is a huge open question.
What information do investors need about fair values? Do they need information about the distribution of fair value estimates? Does recognition or disclosure matter? Often, companies seem to be happy disclosing numbers in financial statement notes, but not so happy about recognising those numbers. Are concerns about earnings volatility legitimate? What is 'too much' earnings volatility? What are the effects of management discretion in determining fair values? We know discretion is there, we find evidence of it, but what are the implications of it? How will the use of more fair values affect investor or management behaviour? What are the implications of incorporating more expectations about the future into financial statements today? In particular, what does profit or loss mean in such a world?
Globalisation of financial reporting
I turn now to issues related to the globalisation of financial reporting. The IASB's vision is for a single set of high-quality global standards that are used in the world's capital markets.
Why is this the vision? The most important reason is that we accountants want to improve the functioning of global capital markets. 'Improve' means increased comparability across firms, thereby reducing information processing costs. For many countries, 'improve' also means increasing the quality of accounting information; and decreasing the cost of preparing financial statements, particularly for large multinational firms currently using different standards. A single set of standards reduces information risk arising from users not understanding financial statements and so imposing a risk premium and increasing the cost of capital.
The worldwide International Financial Reporting Standards (IFRS) adoption experiment has generated considerable interest among researchers. To give two examples of research studies, consider first the stock market reaction to the adoption of IFRS in Europe.5 The first motivating question here is, 'Did investors perceive net benefits to the IFRS adoption in Europe?' The second motivating question is, 'If there were benefits, did the benefits come from convergence of standards or from perceived improvement in the quality of information?'
The main research question addressed is, 'Did the European stock market react positively to events that increased the likelihood of IFRS adoption and negatively to events that decreased this likelihood?'
We find a significant positive overall market reaction to events that increased the likelihood of IFRS adoption. We find that the reaction was incrementally negative for firms in code law countries. Prior research tells us that investors in code law countries in Europe are concerned about enforcement and implementation, so this is consistent with investors in those countries being less excited about the change. We find that the market reaction was incrementally positive for firms that had lower quality information - more pronounced for banks and firms with greater information asymmetry. These findings are consistent with investors responding more positively when they assess the change as increasing the quality of the information they would be receiving under IFRS.
However, we also find an incremental positive market reaction from firms with high-quality pre-adoption information, suggesting that investors perceive benefits from convergence. Thus, investors perceive net benefits to the adoption of IFRS in Europe; they are concerned about enforcement of the standards; and they expect net benefits associated with both increased information quality and increased convergence.
The second example relates to the possible adoption of IFRS by the US. One of the questions raised in the US is whether US Generally Accepted Accounting Principles (GAAP)-based financial information is comparable to IFRS-based financial information.6
The motivating questions here are straightforward. The first is in the study's title: 'Are international accounting standards (IAS)-based accounting amounts comparable to US GAAP-based accounting amounts?7' The second is, regardless of whether they are or are not comparable, 'Is there evidence that comparability has increased?'8
The findings of this study indicate that the US GAAP-based accounting amounts and IAS-based accounting amounts are not fully comparable. However, comparability with IAS is higher than it was with non-US domestic standards and comparability has increased over time; in other words, the widespread application of IAS by non-US firms and convergence efforts have increased comparability with US firms, but differences still remain. The study just referred to is one of several that conclude that investors viewed the adoption of IFRS in Europe positively, and the extensive literature finds that IAS-based accounting amounts are of higher quality than non-US GAAP-based amounts. This literature also finds that IAS-based accounting amounts are of comparable quality to US GAAP-based amounts reported by firms in many countries, although not to amounts reported by US firms. Again, comparability with US GAAP is increasing over time.
We also know that having a single set of accounting standards is a necessary but not sufficient condition for comparable financial reporting around the world. The IASB can write standards, and those standards may be identical to those of any other country. However, many accountants are involved in turning those standards into financial statements. The quality of the information in the resulting financial statements depends on the incentives of managers and auditors. We also know that cultures change slowly, not by fiat. It takes time for people to understand what is expected.
Finally, there is a study that I really like because it reveals my bias. This study derives equilibria from models of an economy. Sometimes such models have more than one equilibrium and there is no reason to prefer one to another. This study shows how countries that have very similar economies can end up with very different financial reporting standards. I like this study because it shows that it is possible that we did not end up with different standards around the world because we are fundamentally different. It could just be that we were not talking to each other. Thus, maybe it will be easier to bring the financial reporting world together than many people think. A key question is whether globalisation of financial reporting reduces the cost of capital, which is unobservable and hence it is very difficult to provide conclusive evidence on this question. Thus, we need several studies so that we can triangulate the findings. We would also like to know whether globalisation facilitates the allocation of capital and cross-border trading. This is part of the objective of having globalised financial reporting. Does it reduce home bias in investing? Is part of the reason for the current home bias in investing that people understand financial statements from their home country and not other countries?;
Another question: Can we remove the impediments to global standards? The literature does not explain how these impediments can be removed or which impediments are the most important. Overall, the question of whether application of IFRS results in higher quality financial reporting is still open. And we also do not fully understand the costs and benefits of globalisation. Change involves costs but do the benefits outweigh the costs?
The Global Financial Crisis
I do not believe that accountants should take the blame for the current financial crisis. The crisis is not the result of fair value accounting. We accountants are just the messengers and are simply stating the facts.
However, there has been tremendous pressure to reduce the transparency of financial reporting and the independence of standard setting as a result of the financial crisis. Bank regulators have tried to convince others that we should not permit accounting to 'tell it like it is' - they argue that when the financial statements 'told it like it was', we ended up in a financial crisis. The G20 statement on the financial crisis supports high-quality transparent global standards set by independent standard setters. This statement was extremely important. Hopefully, the statement will help offset some of the political pressure being placed on the accounting standard setters.
Also, I believe that the financial crisis has increased the focus on convergence around the world, particularly between US GAAP and IFRS. We live in a global world and we need the same financial reporting. Without global standard setters, financial reporting will end up at the lowest common denominator. It is a challenging time for all involved in financial reporting.
Many open questions remain. This is great news for academics, not so for standard setters and other accountants. For example, it is embarrassing to tell non-accountants that accountants still do not know what revenue is when, for 500 years, we have been putting revenue at the top of the income statement. Yet, the FASB and the IASB have been debating revenue for the past five years.
There are many possible research designs that can make research relevant to standard setting. I described four research designs, and I believe each provides relevant input to a standard setting issue. The researcher's task is to design research that links the research question to the motivating question, so that standard setters and others can see the contribution of the research to the issues of the motivating question. Standard setters need all the help they can get. Conducting research that contributes both to the academic literature and to standard setting has a double impact. Academic research can help resolve many of the nettlesome issues standard setters face, and play a central role in shaping global financial reporting. The slate is open and ready for our collective input.
4The technical details of this study are recorded in the full paper at: http://live.unimelb.edu.au/episode/cpa-congress-2009-70th-annual-research-lecture
A study jointly with Chris Armstrong at the Wharton School, Alan Jagolinzer, a Stanford colleague, and Eddie Riedl at the Harvard Business School.
8 The technical details of this study are recorded in the full paper at: http://live.unimelb.edu.au/episode/cpa-congress-2009-70th-annual-research-lecture