In the wake of cases like Enron and Worldcom in the United States, and HIH Insurance in Australia, market participants, auditors and regulators are vitally concerned with the quality and reliability of published financial reports. This is particularly so for publicly traded firms and the summary performance measures they release, such as earnings numbers.
In the more extreme cases, this naturally extends to the efficient detection of fraudulent financial reporting. Such behaviour is costly in terms of the inefficient allocation of resources, and represents a threat to the overall integrity of the capital market. However, in many cases it seems that detection of even egregious accounting fraud does not occur until well after the published accounts have been disseminated and relied on by investors and other stakeholders. Indeed, lengthy delays between the instigation of accounting fraud and the instigation of regulatory action by market regulators often draw substantial criticism and political action, as we have witnessed with Enron, Worldcom and HIH. As a consequence, there is a considerably increased demand for accounting and audit surveillance technology toolsets.
Of course, financial statement fraud is relatively rare, but it can also be viewed as an extreme case of accounting manipulation, most of which is likely to occur on a much smaller scale. Far more common are allegations that management have engaged in manipulation of the accounts that, while possibly within the boundaries of the generally accepted accounting principles (GAAP) of the country concerned, nevertheless results in a misleading picture of the company's financial performance and position. This is frequently alleged to occur via the 'judgement' that is inevitably a part of the accounting process, although it is also possible that key measures such as earnings can be manipulated by decisions that have real economic consequences as well (e.g. reduced investment creating a temporary boost in earnings). There is unlikely to be any single metric that can capture all forms of manipulating periodic results, but Capital Markets CRC (CMCRC)1 has developed some technology for gaining insights into the extent of 'earnings management': these offer surveillance capability for market participants, regulators and auditors in distinguishing between high and low quality earnings results. Moreover, the technology has the capability of gaining some insights into the quality of auditing attached to these earnings numbers. This article provides some evidence on the capability of the technology.2
There are several ways to demonstrate capability. For example, we could focus on the ability of our measures to distinguish firms that have collapsed. However, these are relatively rare, and it is certainly not clear that corporate collapses are primarily a reflection of problematic accounting. Indeed, the problematic accounting likely occurs some years prior to the ultimate collapse, at a time when such deliberate intervention in the reported results may still hold out hope of avoiding the ultimate demise of the company. Our focus in this analysis is on the timely identification of firms where it could be said that there was legitimate cause for concern about the quality of the published financial data. For this purpose, we focus on the ability of our technology to provide a timely indication of accounting 'issues' that are subsequently publicly identified by ASIC (the Australian Securities and Investments Commission) and on which they have taken action directed towards restricting and/or rectifying aspects of the financial reporting process by publicly traded firms (i.e. those listed on the Australian Stock Exchange). While these actions reflect a variety of concerns with firms' accounting, they all are expected to have some impact on reported earnings, or at least its composition.
The results show that, relative to a matched-sample (firm size and industry), the CMCRC's measure of earnings quality is able to distinguish firms subject to action by ASIC over aspects of their accounting. We provide a brief depiction of each of these cases, showing the combination of accounting and audit quality based on our measures, and describe the statistical results from the analysis.
Surveillance over the ASIC enforcement cases
The timing of accounting enforcement actions by ASIC and the allegations involved are publicly reported. Brown & Tarca (2005)3 provide a summary of 42 ASIC enforcement actions reported between 1998 and 2004. We initially exclude 19 companies due to the lack of matching firm either due to missing year, GICS-Industry or total assets size information. We further eliminate seven companies which do not have the data necessary to estimate the associated earnings quality measure. Hence, our final sample consists of 16 firms. We first calculate EQIS and AuQIS scores (our measures of earnings and audit quality respectively) as at the financial year prior to the announcement of the ASIC enforcement action.4 Higher (lower) values on these measures indicate more aggressive (more conservative) accounting and more (less) audit effort than expected for firms on average listed on the ASX.
Table 1 summarises the circumstances in each of the 16 cases together with the EQIS and AuQIS scores.
Table 1: Analysis of ASIC enforcement cases
| Company | Financial Year | Date of accounting data | EQIS score | AuQIS score | Circumstances |
| China West International Holdings Limited | 2003 | 25/02/2004 | 0.4188 | 1.2138 | Recognition group: Included in its consolidated financial statements an entity that it did not control |
| Futuris Corporation Limited | 2003 | 18/09/2003 | -0.3800 | 0.9135 | Measurement: Revenue recorded as gross sales rather than commission earned |
| Flight Centre Limited | 2003 | 24/09/2003 | -0.3007 | 0.9025 | Measurement: Recording revenue as gross price on travel products as opposed to net commission value |
| King Minerals ML | 2003 | 26/09/2003 | -0.8170 | 0.6120 | Classification: Misclassification between asset classes |
| National Telecoms Group Limited | 2003 | 03/09/2003 | -0.2521 | 1.5487 | Recognition element: Multiple element arrangements of the same contract |
| Pacifica Group Limited | 2003 | 29/03/2004 | -0.1550 | 1.1947 | Recognition element: Items recorded as deferred revenue should be written off |
| Sunland Group Limited | 2003 | 30/09/2003 | -0.1269 | 0.3706 | Recognition element: Writing off deferred expenditure directly against retained earnings |
| Stockland Trust Group | 2003 | 04/06/2004 | -0.2187 | 0.5886 | Recognition element: Goodwill arising from an acquisition was written off as an expense as opposed to being amortised |
| Tempo Services Limited | 2003 | 02/06/2004 | 0.8456 | 0.5855 | Recognition element: Recording of a receivable which had been brought to account in previous years |
| Harvey World Travel Limited | 2002 | 18/03/2003 | 0.3720 | 2.2877 | Measurement: Recording revenue as gross price on travel products as opposed to net commission value |
| Redflex Holdings Limited | 2002 | 10/07/2002 | - | - | Recognition element: Planned to capitalise written off R&D and other expenses however prevented by ASIC |
| Sabina Corporation Limited | 2002 | 05/06/2003 | -0.0950 | 0.1852 | Measurement: Overstatement of value of financial assets |
| Solution 6 | 2000 | 22/11/2000 | -0.7414 | 2.7628 | Disclosure: Description of the annualised revenue as 'forecasts' which should have been 'targets' |
| Novogen Limited | 1998 | 01/03/1999 | -0.7298 | 0.8885 | Disclosure: Explanation of change in accounting policy |
| Seven Network Consolidated | 1998 | 09/03/1999 | 0.5616 | 0.5446 | Recognition element: Inappropriate capitalisation of borrowing costs; carrying unrealised exchange loss to reserve rather than Profit & Loss statement; Recording treasury costs in year prior to which they were incurred. |
| St George Bank | 1998 | 15/10/1998 | -0.0671 | 1.0767 | Recognition element: Accounting treatment of restructuring costs |
The cases show that the CMCRC's technology provides a timely indicator of problematic accounting as evidenced by conservative or aggressive accounting at the time the accounting data are released and hence, available for analysis. This date typically precedes by some time the point at which ASIC action appears to have been launched. Moreover, in the individual case studies presented it is apparent that in many cases, the problematic accounting is accompanied by lower than expected audit fees. This raises the additional concern that audit quality is lower than expected.
A statistical analysis
In some of the cases, the resulting bias in earnings alleged by ASIC is effectively a reduction rather than an increase, or relates to disclosure and/or gross versus net revenue.5 This suggests that the absolute value of our EQIS measure, i.e. allowing for whether the accounting is either aggressive or conservative, is likely to be the most appropriate measure for flagging the likelihood of questionable accounting, at least as identified by ASIC. To assess how discriminating the EQIS score can be, we then select our matching firms so that each case is matched to a corresponding firm based on year, GICS-Industry and total asset size. We use a statistical tool called a logit regression to assess if the absolute value of EQIS is able to predict ASIC enforcement action (i.e., can these measures distinguish the accounting year subject to ASIC enforcement action from the matched control?). We find evidence that the absolute value of EQIS can reliably and significantly predict ASIC enforcement actions, and this result holds after including AuQIS scores. This evidence seems to suggest that firms with more aggressive or more conservative earnings figures are more likely to receive an ASIC enforcement action. These results are robust as to whether firms are included that are as alleged solely to have made defective disclosures (as distinct from accounting adjustments or choices).
Conclusion
We have provided a brief overview of some technology developed for measuring variation in accounting and audit quality. For a small sample of firms subject to enforcement action by ASIC, we find that the absolute value of the CMCRC's accounting quality measure (i.e. EQIS) provides a timely indicator of problematic accounting, in that it is able to distinguish these firms from a control sample at the time the accounting data are released and hence available for analysis. This date typically precedes by some time the point at which ASIC action appears to have been launched. Moreover, the individual case studies presented show that, in many cases, the problematic accounting is accompanied by some suggestions of problematic audit quality being delivered on the audit.
Of course, the power of the CMCRC's EQIS and AuQIS metrics needs much broader testing. Apart from an analysis of accounting risk, there is potential for such measures to inform broader investment analysis and funds management decision making. The current research programme at the CMCRC is directed towards these ends.
References
1 Capital Markets CRC (www.cmcrc.com) is a cooperative research centre across universities and industry, and was established with a mandate from the Australian government (along with its funding support) to undertake research into market surveillance and to work with business to deploy technology-based solutions for market surveillance world-wide.
2 Some background to the technology can be found in Friedrich, C, D Stokes and S Taylor. (2003) Applications of corporate governance technology to accounting and audit quality surveillance, The Handbook of World Stock, Derivative & Commodity Exchanges (Mondo Visione, UK), pp. lxxvii- lxxxi.
3 For example, if ASIC publicised an enforcement action at 02/08/00, then EQIS is calculated as at 30/06/00 (assuming a June financial year end). On the other hand, if ASIC publicises the action at 02/02/00, the EQIS is calculated as at 30/06/99.
4 Brown, P, and A Tarca. (2005) Achieving high quality, comparable financial reporting:A comparison of independent enforcement bodies in Australia and the United Kingdom. Working Paper, University of Western Australia.
5 See, for example, Stockland Trust Group.
