Risk disclosures on the second tier markets of the London Stock Exchange
We would like to thank Matt Bamber, Yu-Hsuan Wu and the staff of the Accounting and Finance Division, Leeds University Business School, for their helpful comments and suggestions.
Abstract
The identification, management and disclosure of risks have been the subject of recent legislation, directives and reporting standards issued across a number of international jurisdictions. To inform the disclosure debate, this paper provides a detailed analysis of the risk warning disclosures of initial public offering (IPO) companies and the factors that drive such disclosures. We find that risk disclosures of IPO companies contain a greater proportion of forward-looking information but a lower proportion of information on internal controls and risk management than the disclosures of listed companies. We find evidence that such disclosure has increased across time but that larger directors’ shareholdings are associated with a reduction in risk disclosure.
1. Introduction
Risk disclosure has received increasing attention among regulators across the globe over the last 15 years. Regulators promote such disclosure in the interests of fair markets, and this has become more pertinent given the indignation of both investors and the public at large in the wake of recent financial scandals, such as Enron and WorldCom. Investors require risk information to allow them to adequately diversify their portfolios (Beretta and Bozzolan, 2004). A number of authors note that empirical academic research into current risk disclosure practices is limited (Solomon et al., 2000; Linsley and Shrives, 2006), and such research is required as a basis for any regulation or recommendations with respect to best practice. Linsley and Lawrence (2007) report that the level of readability of risk information contained in financial reports is ‘difficult or very difficult’ and call for further research into disclosure formats. A 1999 report by the Institute of Chartered Accountants in England and Wales (ICAEW, 1999b) argues that the risk disclosure of listed companies should be of a standard produced in initial public offering (IPO) prospectuses, and in this paper we aim to inform the debate on risk disclosure via a detailed analysis of the risk disclosures of IPO companies.
Our contributions are threefold. First, we compare the risk disclosures of IPO companies with those of listed companies (reported in extant research). To fulfil our first aim, we synthesize the frameworks for the analysis of risk disclosure employed in the studies of Beretta and Bozzolan (2004), Linsley and Shrives (2006) and Abraham and Cox (2007) and extend this to incorporate the framework suggested by Beattie et al. (2004) for analysing financial report narratives (see Section 3.2 for full details), and we use this framework to undertake an analysis of the risk disclosure of a sample of over 400 IPO companies. We then compare our IPO sample disclosures with those of listed companies as reported in the extant literature.
Second, against a background of increasing regulation of risk reporting, we examine the extent of any changes in disclosure practices across time. Research that has been undertaken on listed companies has tended to be characterized by small samples (n ≈ 80) of listed companies drawn at one point in time.1 Our study is based on a larger sample (n = 420) of IPO companies drawn over a 13 year period (1991–2003), which allows us to analyse the characteristics of risk disclosure across time. Our sample size allows us to select a relatively homogeneous sector (Information Technology (IT)) on which to base a robust regression analysis of changes in the quantity of risk disclosure across time. To provide a context for this analysis, in Section 2 we provide a discussion of the evolution of risk reporting requirements during our sample period and we discuss our results within the context of the changes in risk reporting regulations.
Third, we analyse the factors that drive the voluntary disclosure of risk by IPO companies. The precise role of regulation in relation to disclosure has been the subject of increasing debate for a number of years, given that disclosure might be of benefit to a company. In the presence of information asymmetry between firm insiders and investors, investors demand an information risk premium (Barry and Brown, 1985, 1986), which increases the cost of capital and lowers firm value. Any reduction in information asymmetry – for example, via the disclosure of information on key risks – would reduce a firm's cost of capital and increase its value. In addition, risk disclosure reduces the cost to analysts of obtaining information about the firm and will increase the number of analysts following the firm (Lang and Lundholm, 1996), further reducing information asymmetry. Healy and Palepu (2001) call for further research into the reasons why firms engage in voluntary disclosure.
We undertake both a content analysis and a regression analysis. Our content analysis reveals that the risk disclosures of IPO companies contain a greater proportion of forward-looking information but a lower proportion of information on internal controls and risk management than the disclosures of listed companies. Linsley and Shrives (2006) argue that information relating to future risks will be more useful to stakeholders than information relating to past risks, and IPO company disclosures suggest that listed companies might move the balance of information provided even further towards future risks. Both the UK Turnbull guidelines and the US Sarbanes–Oxley Act emphasize the importance of the disclosure of internal control systems for the management of risks (see Section 2 for details). We find that the risk statements produced in IPO prospectuses fall far short of providing adequate disclosure in relation to internal control systems and risk management and would provide a poor benchmark for listed companies in this respect.
Separate statements on corporate governance are provided in IPO prospectuses, which disclose that many companies set up internal control procedures to oversee performance around the time of listing on the stock market. This suggests that many companies only implement such procedures when obliged to do so and, as such, regulators may have a role to play in providing safeguards to investors. However, we acknowledge that closely held companies may have considered such systems superfluous prior to listing.
Our content analysis also shows that IPO companies are more focused on external risks, whereas listed companies are more focused on internal risks, which might be expected given the differing characteristics of the companies being examined. In our regression analysis, we explore the voluntary disclosure of risks along two dimensions: (i) disclosure versus non-disclosure and (ii) the quantity of disclosure. We find that voluntary disclosure of a risk statement is more probable the greater the extent of information asymmetries faced by a firm, which may suggest that disclosure of information on key risks reduces such asymmetries. We also find that risk disclosures are less likely where firms use alternative signals of firm quality, which suggests that risk disclosure is not preferred by all firms as a means of reducing information asymmetry.
We separately examine the quantity of risk disclosure across a relatively homogeneous sector: the IT sector. We find that to some degree the quantity of disclosure is driven by expediency (firms that would be expected to face a greater variety of risks report more risks), but we also find some evidence that larger directors’ shareholdings reduce the quantity of disclosure. Two arguments are possible here. One is that directors’ shareholdings directly mitigate information asymmetry and reduce the need for additional disclosure. The second is that directors with larger shareholdings have more influence and reduce the amount disclosure to avoid passing on information to competitors. This latter argument is supported by the finding that increased risk disclosure was unpopular among managers in a study of finance directors by Tillinghast-Towers Perrin (1998). Finally, we find that risk disclosure increased across our sample period (1991–2003) against a background of increased regulatory emphasis, despite no significant changes in the regulations affecting risk reporting in the IPO sector. It would appear that increased regulatory emphasis that stops short of mandatory requirement is effective in achieving increased disclosure.
UK IPO companies identify all key risks in a dedicated ‘risk warning’ statement. Risk disclosure may occur outside the risk warning statement; however, all significant risks are cross-referenced to the risk warning statement. Researchers examining the risk disclosure of listed companies (see, inter alia, Linsley and Shrives, 2006) have prefaced their analyses by considering how risk disclosure might be identified, a problem that is circumvented by focusing on the risk statements of IPO companies where all key risks are identified and categorized as such by the company (and/or the company's advisors).2
The remainder of this paper is structured as follows. In Section 2, we provide a discussion of regulatory developments in relation to risk reporting. Section 3 provides a content analysis of risk disclosure and Section 4 provides a regression analysis of risk disclosure. Finally, Section 5 provides a concluding discussion.
2. Regulatory background
A framework for the voluntary disclosure of business risk was established in the UK in 1993 when the Accounting Standards Board introduced the Operating and Financial Review (the equivalent of the US ‘Management Discussion and Analysis’). Subsequent calls for increased risk disclosure have come from both professional bodies and academic studies. The ICAEW (1997, 1999b) published two steering group discussion papers on the subject of the disclosure of business risk in annual reports, recommending increased disclosure of the key risks a company faces. Solomon et al. (2003) report that institutional investors would like firms to make further disclosure of key risks to assist with investment decisions. On 10 May 2005, the UK Accounting Standard Board issued Reporting Standard 1, ‘Operating and Financial Review’, the primary aim of which was to improve the availability of information available to (existing) shareholders to allow them to fully evaluate the investment they have undertaken. This statutory reporting requirement was subsequently repealed in November 2005, and publication of the Operating and Financial Review, and the information on key risks contained therein, once again became voluntary. Abraham and Fox (2007) argue that increased regulation in relation to risk disclosure simply leads to formulaic disclosures and an attendant reduction in the amount of useful information on key risks available to investors. Disclosure theories indicate that regulation may be superfluous given the benefits that would accrue to a company from the information asymmetry resolution that the disclosure of key risks implies (see Healy and Palepu, 2001).
While the Operating and Financial Review deals with business risk disclosure, the Combined Code on Corporate Governance, issued by the Hampel Committee in 1998, places emphasis on internal systems and the process of describing and managing risks. Following publication of the Combined Code, the ICAEW was asked to establish a working party (known as the Turnbull Committee) to provide guidance to assist companies in the implementation of the requirements of the Combined Code relating to internal control. The Turnbull guidance (1999) underlined the responsibility of Boards to ensure that risks are effectively identified and managed. Companies are required to adopt a continuing system of internal control that analyses all risks to the business, rather than just narrow financial risks. Despite this widening in scope, an ICAEW briefing guide on Turnbull (ICAEW, 1999a) recognizes that the Turnbull guidance focuses on ‘significant’ risks, since the identification of too many risks may cloud the ability of a company to identify and manage its significant risks. An update of the Turnbull guidance was issued by the Financial Reporting Council in October 2005, Internal Control: Revised Guidance for Directors on the Combined Code. This report emphasizes the need for the assessment of risks and the devising and monitoring of an internal control system, at all levels of a company.
The identification and evaluation of risks is also a key aspect of the Sarbanes– Oxley Act (2002) in the USA, which was passed in response to the financial scandals of Enron and WorldCom. Of particular interest from the perspective of this paper is Section 404 of the Act, which requires a company to provide, in its annual report, an internal control report stating management's responsibilities for establishing and maintaining adequate internal controls and procedures for financial reporting; management's conclusions about the effectiveness of the company's internal controls and procedures; and a statement from the company's auditor on the management's evaluation of internal control and procedures.
Finally, financial risks are the subject of UK Financial Reporting Standard 13 (FRS 13), effective for all accounting periods ending on or after 23 March 1999. The increased use of derivative instruments has required the provision of quantitative and qualitative information in respect of market risks in both the USA and UK (via Financial Reporting Release #48 and FRS 13, respectively). These are primarily confined to the reporting of an explanation of the role of derivative instruments in managing the risks a firm faces and the risks of losses on derivative instruments arising from adverse changes in equity prices, interest rates, exchange rates and commodity prices.
In summary, the regulators have focused on (i) business; (ii) internal control; and (iii) financial risks. Abraham and Cox (2007) follow this regulatory framework for analysing risk disclosure but other research uses alternative frameworks. Our examination of the disclosure of risk by IPO companies incorporates relevant dimensions of alternative frameworks; however, we give precedence to discussing risk within the context of regulation.
3. Content analysis of risk disclosure
3.1. Sample
A preliminary analysis suggests that risk warnings are over five times more likely to appear in the prospectuses of the second tier market IPOs (the Unlisted Securities Market (USM) and the Alternative Investment Market (AIM)) and, therefore, we focus our analysis on USM and AIM companies. There were 625 IPOs involving second tier market listings, excluding introductions and investment trusts, for the period 1 January 1991 to 31 December 2003, and we have prospectuses for 420 of these companies which constitute the sample for our study. All data are taken from the prospectuses of the 420 IPO companies.
In Panel A of Table 1, we show how risk disclosure varies across time and the breakdown of our sample by industrial sector each year. As can be seen from Panel A of Table 1, 88 per cent of our sample companies over the period 1991–2003 provide at least one risk warning in the prospectus. Two factors are particularly significant in Panel A of Table 1; first, risk warning disclosures increase across time such that only 42 per cent of USM companies (1991–1994) make a risk disclosure but for the years 2000 onwards, effectively all companies make a risk disclosure. Second, the industrial breakdown of the sample shows a significant reduction in the percentage of the sample made up of Services and IT companies in the years 2002 and 2003 with a commensurate rise in the percentage of Consumer Goods (2002 and 2003), Resources (2002) and Financial (2003) companies in those years.3
| Panel A: Time series properties and industry breakdown of risk declarations | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1991–1994 | 1995 | 1996 | 1997 | 1998 | 1999 | 2000 | 2001 | 2002 | 2003 | Total | ||
| (i) Sample and risk declarations by year | ||||||||||||
| N | 12 | 11 | 62 | 52 | 26 | 32 | 96 | 60 | 32 | 37 | 420 | |
| N with risk | 5 | 9 | 43 | 41 | 20 | 28 | 95 | 60 | 31 | 37 | 369 | |
| % with risk | 42 | 82 | 69 | 79 | 77 | 88 | 99 | 100 | 97 | 100 | 88 | |
| (ii) Industrial breakdown of 420 sample companies (% of sample total each year) | ||||||||||||
| Resources | 8% | 0% | 2% | 2% | 8% | 0% | 3% | –8% | 13% | 0% | 4% | |
| Industrials | 17% | 0% | 16% | 4% | 8% | 6% | 4% | 12% | 19% | 14% | 10% | |
| Consumer Goods | 0% | 18% | 11% | 8% | 8% | 9% | 5% | 18% | 25% | 19% | 12% | |
| Services | 58% | 55% | 56% | 60% | 42% | 50% | 51% | 47% | 38% | 35% | 50% | |
| Utilities | 0% | 0% | 2% | 0% | 0% | 0% | 0% | 0% | 0% | 3% | 0% | |
| Financials | 0% | 18% | 3% | 13% | 15% | 16% | 7% | 0% | 0% | 22% | 8% | |
| Information Technology | 17% | 9% | 10% | 13% | 19% | 19% | 29% | 15% | 6% | 8% | 16% | |
| (iii) Industrial breakdown of 369 risk disclosure companies (% of disclosing total each sub-period) | ||||||||||||
| Period | 1991–1999 | 2000–01 | 2002–03 | |||||||||
| N | 146 | 155 | 68 | |||||||||
| Resources | 3% | 5% | 6% | 4% | ||||||||
| Industrials | 8% | 7% | 16% | 9% | ||||||||
| Consumer Goods | 10% | 10% | 22% | 12% | ||||||||
| Services | 53% | 49% | 35% | 48% | ||||||||
| Utilities | 1% | 0% | 1% | 1% | ||||||||
| Financials | 11% | 5% | 12% | 8% | ||||||||
| Information Technology | 14% | 24% | 7% | 17% | ||||||||
| Panel B: Risk declarations by sector | |||||
|---|---|---|---|---|---|
| Total Issues | Mean (median) market capitalization (£’000) | Risk statement | No risk statement | Mean (median) number factors | |
| Resources | 17 | 17 972 (10 798) | 94% | 6% | 15.73 (15.00) |
| Industrials | 40 | 13 788 (9739) | 85% | 15% | 10.38 (10.00) |
| Consumer Goods | 49 | 20 866 (11 059) | 94% | 6% | 10.85 (11.00) |
| Services | 208 | 18 249 (10 873) | 85% | 15% | 10.05 (9.00) |
| Utilities | 2 | 239 983 (as mean) | 100% | 0% | 14.5 (as mean) |
| Financials | 35 | 12 452 (6934) | 89% | 11% | 8.45 (8.00) |
| Information Technology | 69 | 19 106 (14 317) | 91% | 9% | 10.74 (10.00) |
| Total | 420 | 18 839 (10 928) | 88% | 12% | 10.41 (10.00) |
- Note: The number of risk warnings is based on 420 companies for which we have prospectuses. The average number of risk factors is based on companies with risk statements only.
This analysis suggests that three sample periods are appropriate for undertaking our content analysis of risk disclosure: (i) 1991–1999 when the disclosure of risk had a greater voluntary element; (ii) 2000–2001, a period when almost 100 per cent of companies made risk disclosures with a relatively higher percentage of Services and IT companies; and (iii) 2002–2003, a period when almost 100 per cent of companies made risk disclosures with a relatively lower percentage of Services and IT companies. In Panel A of Table 1, we also therefore provide a breakdown of the industrial structure of the 369 companies that make risk disclosures for these three subperiods, which shows that the percentage of risk-declaring companies in the sample increases for the IT sector between 1991/1999 and 2000/2001 from 14 to 24 per cent (other sectors remain roughly the same other than Financials, which shows a decline) and as expected the percentage of risk-declaring companies in the sample decreases for both the IT and Services sectors between 2000/2001 and 2002/2003 and increases for Financials, Consumer Goods and Industrials.
In Panel B of Table 1, we show how risk disclosure varies across industry. There is little variation in the number of companies making risk disclosure statements across industrial sector. The number of companies making risk disclosure statements is highest in the Resources and Consumer Good sectors (94 per cent) and lowest in the Services and Industrials Sectors (85 per cent). There is greater variation in the number of separate sources of risk identified across sectors, with Financial sector companies identifying the lowest number of risks (mean = 8.45, median = 8) and the Resources sector the largest number (mean = 15.73, median = 15).
3.2. Content analysis
Our analysis focuses on key risks, in line with the Turnbull guidelines, and we extract our information firstly from the risk warning statement, and secondly from the remainder of the prospectus where attention is drawn to additional risk disclosure in the risk warning statement.
We now turn to a detailed analysis of the content of IPO risk declarations. Beattie et al. (2004) suggest a framework for analysing narratives in annual reports in which narratives are not just categorized by topic, but also by type. Accordingly, we undertake analyses based on both topic and type and we discuss each in turn.
By reference to the studies of Beretta and Bozzolan (2004) and Linsley and Shrives (2006), we analyse risk topics along the following dimensions: (i) internal risks; (ii) external risks; and (iii) corporate development. We add to this list the following categories: (iv) third-party risks; (v) information risks; (vi) ongoing claims and disputes; and (vii) ‘boiler plate’ disclosures.4 As discussed in Section 2, Abraham and Cox (2007) categorize risks according to regulatory guidelines and, therefore, risk topics are segmented into (i) financial (risks with a financial topic such as exchange and interest rate risks); (ii) internal control (risks that refer to internal control systems); and (iii) business (risks that relate to the business, such as competition). To facilitate the re-categorization of risks disclosed along the internal control/financial/business dimension, we further subdivide each of our seven risk categories to give 51 narrower risk categories.
By reference to the studies of Beattie et al. (2004), Beretta and Bozzolan (2004) and Linsley and Shrives (2006), we analyse the types of disclosure along the following dimensions: (i) time orientation; (ii) financial/non-financial; (iii) quantitative/qualitative; (iv) economic sign; and (v) risk management strategies. We deal with financial versus business versus internal control risks as part of our analysis by topic and the financial/non-financial dichotomy as part of our analysis by type. We commence with an analysis of content by topic. We present the results of our analysis of risk declarations by topic in Table 2. Panel A deals with the external environment, Panel B with the internal environment and Panel C with the other risk categories.
| Panel A: Risk factors related to the external environment | ||||
|---|---|---|---|---|
| Risk factor | Proportion of companies with risk factor | No. of factors | ||
| 1991–1999 | 2000–2001 | 2002–2003 | ||
| N | 146 | 153 | 67 | |
| A. The regulatory environment | ||||
| 1. Regulation of the operating environment | 0.36 | 0.50 | 0.55 | 167 |
| 2. Product/business approval | 0.13 | 0.18 | 0.24 | 62 |
| 3. Protection for proprietary technology | 0.16 | 0.42 | 0.37 | 112 |
| 4. Taxation | 0.16 | 0.05 | 0.19 | 44 |
| 5. Government financial support | 0.03 | 0.05 | 0.07 | 17 |
| 6. Movement of capital restrictions | 0.01 | 0.00 | 0.00 | 1 |
| 7. Legal framework | 0.05 | 0.08 | 0.00 | 19 |
| 8. Capital market regulations | 0.01 | 0.00 | 0.00 | 2 |
| 9. Non-specific changes to the regulatory environment | 0.18 | 0.47 | 0.43 | 127 |
| B. Environmental and technological factors | ||||
| 10. Environmental | 0.05 | 0.05 | 0.03 | 18 |
| 11. Technological | 0.14 | 0.39 | 0.18 | 91 |
| C. Political, macroeconomic and sociological factors | ||||
| 12. Sociological | 0.01 | 0.00 | 0.00 | 1 |
| 13. Economic/political factors that do not relate specifically to trade in/with overseas markets | 0.14 | 0.07 | 0.16 | 42 |
| 14. Economic/political factors that do relate specifically to trade in/with overseas markets | 0.23 | 0.31 | 0.39 | 108 |
| D. The competitive environment | ||||
| 15. Existing competitive environment | 0.19 | 0.39 | 0.46 | 119 |
| 16. Potential competitive environment | 0.37 | 0.68 | 0.49 | 191 |
| E. Demand uncertainties | ||||
| 17. Potential demand | 0.73 | 0.92 | 0.81 | 300 |
| 18. Existing demand | 0.23 | 0.15 | 0.18 | 69 |
| Total number of risk factors relating to the external environment | 3.18 | 4.71 | 4.57 | 1487 |
| Panel B: Risk factors related to the internal environment | ||||
|---|---|---|---|---|
| Risk factor | Proportion companies with risk factor | No. of factors | ||
| 1991–1999 | 2000–2001 | 2002–2003 | ||
| A. Company size and history | ||||
| 19. Company trading/operating history | 0.16 | 0.52 | 0.45 | 134 |
| 20. Company size | 0.02 | 0.00 | 0.01 | 4 |
| B. Product, processes and systems | ||||
| 21. Product/service | 0.10 | 0.51 | 0.58 | 131 |
| 22. Equipment and systems | 0.14 | 0.26 | 0.06 | 64 |
| 23. Production difficulties and delays | 0.07 | 0.14 | 0.07 | 37 |
| 24. Facilities/site | 0.03 | 0.03 | 0.03 | 12 |
| 25. Other operational | 0.01 | 0.00 | 0.00 | 1 |
| 26. Yields/operating gearing | 0.02 | 0.01 | 0.01 | 5 |
| 27. Factors peculiar to franchisors | 0.01 | 0.00 | 0.00 | 1 |
| 28. Factors peculiar to investment companies | 0.05 | 0.07 | 0.06 | 23 |
| C. Human resources | ||||
| 29. Key employees | 0.64 | 0.82 | 0.88 | 279 |
| 30. Other human resource issues | 0.08 | 0.18 | 0.21 | 54 |
| D. Other risk factors related to the internal environment | ||||
| 31. Revenues dependent on performance | 0.06 | 0.01 | 0.03 | 13 |
| 32. Financing | 0.05 | 0.14 | 0.24 | 44 |
| 33. Dividends | 0.01 | 0.03 | 0.06 | 11 |
| 34. Articles of association | 0.01 | 0.00 | 0.00 | 1 |
| 35. Insurance | 0.07 | 0.10 | 0.13 | 34 |
| Total number of risk factors relating to the internal environment | 1.53 | 2.85 | 2.84 | 850 |
| Panel C: Risk factors related to (i) corporate development; (ii) third-party businesses; (iii) information; (iv) ongoing claims and disputes; (v) ‘boiler-plate’ disclosures | ||||
|---|---|---|---|---|
| Risk factor | Proportion companies with risk factor | No. of factors | ||
| 1991–1999 | 2000–2001 | 2002–2003 | ||
| Risk factors related to corporate development | ||||
| 36. Expansion of current operations | 0.30 | 0.26 | 0.24 | 100 |
| 37. Acquisitions/Investments | 0.16 | 0.18 | 0.15 | 62 |
| 38. Financing growth | 0.18 | 0.70 | 0.88 | 192 |
| 39. Ability of systems to cope | 0.03 | 0.11 | 0.19 | 35 |
| Total number of risk factors relating to the corporate development | 0.68 | 1.25 | 1.46 | 389 |
| Risk factors related to third-party businesses | ||||
| 40. Suppliers | 0.09 | 0.15 | 0.10 | 43 |
| 41. Partners/collaborations | 0.08 | 0.14 | 0.12 | 41 |
| 42. Distributors/sales teams | 0.09 | 0.05 | 0.06 | 25 |
| 43. Outsourced manufacturing | 0.03 | 0.02 | 0.07 | 13 |
| 44. Patent infringements/security breaches | 0.08 | 0.08 | 0.06 | 28 |
| 45. Reliance on third-party technology/ infrastructure | 0.06 | 0.19 | 0.12 | 46 |
| 46. Reliance on negotiations between third parties | 0.04 | 0.00 | 0.00 | 6 |
| 47. Other risks involving third parties | 0.03 | 0.07 | 0.13 | 25 |
| Total number of risk factors relating to third-party businesses | 0.51 | 0.71 | 0.67 | 227 |
| Risk factors related to information | ||||
| 48. Reliability/adequacy of data | 0.06 | 0.03 | 0.09 | 19 |
| 49. Reliability of forecasts | 0.17 | 0.14 | 0.09 | 53 |
| Total number of risk factors relating to information | 0.23 | 0.17 | 0.18 | 72 |
| 50. Risk factors related to ongoing claims and disputes | 0.08 | 0.02 | 0.12 | 23 |
| 51. ‘Boiler plate’ disclosures | 1.75 | 2.07 | 2.81 | 760 |
| ‘Catch all’ | 0.03 | 0.08 | 0.21 | 30 |
| Shares are listed on AIM | 0.63 | 0.59 | 0.73 | 231 |
| Small company shares are less liquid | 0.05 | 0.31 | 0.49 | 87 |
| Share prices can go up or down | 0.51 | 0.52 | 0.67 | 200 |
| Stipulation of listing regulations | 0.01 | 0.00 | 0.00 | 2 |
| Market price may not reflect underlying value | 0.10 | 0.14 | 0.25 | 54 |
| Investment may not be suitable for all recipients of prospectus | 0.42 | 0.42 | 0.45 | 156 |
| Total risk disclosures | 7.97 | 11.77 | 12.64 | 3811 |
| The information reported in this table is based on a sample of 366 2nd market IPO companies for the years 1991–2003 who declare risk statements. | ||||
There were 3811 disclosures of risk by topic for 366 IPO companies, which included risk warning disclosures in their prospectuses (369 less 3 companies where the risk disclosure was incomplete). Across the three subsample periods the mean disclosures per company are: 1991/1999, 7.97; 2000/2001, 11.77; 2002/2003, 12.64. The average number of disclosures therefore increases across time. Of 3051 ‘non-boiler plate’ disclosures, approximately 49 per cent of the total relate to the external environment, 28 per cent to the internal environment, 13 per cent to corporate development, 7 per cent to third-party businesses, 2 per cent to information, and 1 per cent to ongoing claims and disputes.
Since a key dimension of our analysis is changes in risk disclosure across time, we document the risk disclosures with significant changes across the subperiods. Between the periods 1991/1999 and 2000/2001, there was an average increase in the number of disclosures of 47.6 per cent, and between 2000/2001 and 2002/2003, an average increase of 7.39 per cent and we use these as our benchmark for determining major increases. Focusing on risk factors that have a minimum of 100 disclosures, four risk topics show a larger than average increase between 1991/1999 and 2000/2001 followed by a decline between 2000/2001 and 2002/2003 (#3 in Table 2, Protection for proprietary technology; #9, Non-specific changes to the regulatory environment; #16, Potential competitive environment; #19, Company trading/operating history). Two risk topics show a larger than average increase between 1991/1999 and 2000/2001 followed by a further larger than average increase between 2000/2001 and 2002/2003 (#21, Product/Service; #38, Financing growth). In unreported results, we find that these patterns cannot be fully explained by changes in the industrial categorization of our sample.5 This suggests that either there has been a fundamental change in the nature of businesses and/or their objectives, or that there have been changes to the external environment, or both.
We now re-categorize risks disclosed in accordance with the various listed company frameworks and compare the disclosures made by IPO companies in their risk disclosure statements with those of listed companies. For this purpose we re-categorize risks by topic as (i) Financial,6 (ii) Internal Control, (iii) Strategic,7 and (iv) Operating risks,8 with (iii) and (iv) both being a subset of business risks. Information risks and ‘boiler plate’ risks are excluded from this analysis.
Abraham and Cox (2007) find that the mean word count for Internal Control risks is about 26 per cent of the mean word count of total risks versus 16.5 per cent for business risks and 60.5 per cent for financial risks. Excluding risk management policy disclosures, Linsley and Shrives (2006) report that 94 per cent of risks can be categorized as either strategic, operations or financial risks, and of these risks 55.4 per cent are strategic, 25.5 per cent are operations and 19.1 per cent are financial.9 Of 2953 risk topics considered in our analysis, 15.4 per cent are financial risks, 1.2 per cent are internal control risks, 31.0 per cent are operations risks and 52.4 per cent are strategic risks. Our figures are roughly in line with the analysis of listed companies undertaken by Linsley and Shrives but differ from the results produced by Abraham and Cox. The difference in the results of Linsley and Shrives and Abraham and Cox can be explained by the introduction of UK FRS 13, effective for all accounting periods ending on or after 23 March 1999 (see Section 2) and suggests that FRS 13 considerably increased the amount of financial risk reporting by UK companies. The risk statements of certain IPO companies disclose the use of forward contracts to deal with exchange rate risk, but otherwise FRS 13 is of limited applicability within the IPO setting and, hence, the similarity between our results and those of Linsley and Shrives.
Our analysis of internal control risks differs significantly from that of Abraham and Cox (2007). Whereas roughly 26 per cent of the risk disclosures of listed companies relate to internal controls, only 1 per cent of the key risks disclosed in the risk statements of IPO companies address internal control. Both the UK Turnbull guidelines and the US Sarbanes–Oxley Act emphasize the importance of the disclosure of internal control systems for the management of risks and clearly the risk statements produced in IPO prospectuses fall far short of providing adequate disclosure in relation to internal control systems and would provide a poor benchmark for listed companies in this respect. That said, IPO companies make separate statements on corporate governance. A typical statement will emphasize that the directors intend to comply with corporate governance best practice. It will usually go on to state that audit and remuneration committees have been established and that the audit committee will be responsible for ensuring that the financial performance of the group is properly monitored and that the remuneration committee will be responsible for reviewing the performance of executive directors and for setting their pay. These statements tend to be forward looking rather than an evaluation of systems already in place and the corporate governance statements are typically much shorter than risk statements. The fact that many IPO companies disclose that they set up internal procedures to oversee performance around the time of listing tends to suggests that companies only act when obliged to do so, albeit that closely held companies may feel no need for significant internal control procedures prior to listing. We consider risk management strategies separately under our analysis of disclosure types.
Beretta and Bozzolan (2004) report for their sample of Italian listed firms that 50.9 per cent of the risks relate to company characteristics (internal risks) versus 35.9 per cent related to corporate strategy and 13.2 per cent related to the environment (external risks). Based on Panels A, B and C of Table 2, the proportions of internal to corporate development to external risks are 31.2 to 14.3 to 54.5 per cent. Given the slightly different emphasis between corporate development and corporate strategy (corporate development relates to future strategy rather than current strategy), we focus on the internal/external dichotomy. This evidence suggests that more mature, listed companies have relatively more internal risks (relatively fewer external risks) than those that are coming to a second tier market for the first time, which would be expected.
We now move on to an analysis of risk disclosure by type. For this analysis, we draw directly on extant literature to provide a framework and we compare the ‘types’ of disclosure made by IPO companies with those made by listed companies. We consider each of the types of disclosure listed earlier (time orientation, financial/non-financial, quantitative/qualitative, economic sign, risk management strategies) in turn.
As we are dealing with the categorization of risks at the time of flotation, the time orientation is primarily the future; however, we find that a small number of risks (relating to ongoing claims and disputes, and outstanding patent and licence applications) are characterized by a partly historical time orientation. Linsley and Shrives (2006) argue that information relating to future risks will be more useful to stakeholders than information relating to past risks; however, based on past evidence (Kajüter, 2001; Woods and Reber, 2003; Beretta and Bozzolan, 2004), they hypothesize that directors of listed companies will be reluctant to make disclosures relating to the future. Contrary to their hypothesis, they find that 26 per cent of risk disclosures among listed UK companies relate to past risks against 35 per cent which relate to future risks (39 per cent are non-time-specific), and because these findings compare favourably with prior studies, this suggests that disclosure of forward-looking information in relation to risk has improved. However, our analysis of IPO company disclosures suggests that listed companies might move the balance even further towards providing relatively more information on future risks faced and in this respect the risk disclosure practices of IPO companies offer a useful benchmark.
We explore further the financial/non-financial dichotomy based on all risk disclosures of the 366 IPO companies with complete risk statements.10 On average 31 per cent of risk disclosures are financial and 69 per cent are non-financial. This proportion varies across industrial sectors with a low of 24 per cent of risks being financial in the IT sector and a high of 44 per cent of risks being financial in the Financial sector. There is no evidence of a monotonic increase or decrease in the proportion of financial risks across time periods, although all sectors, other than the Financial sector, declare the highest proportion of financial risks in the final subperiod (2002–2003). We discussed above the limited relevance of FRS 13 to our sample of IPO companies; however, an increased emphasis on financial risks within the UK regulatory setting may make companies more aware of financial risks and may account for an increased disclosure of these risks in the last period. We can compare these results with results for the financial/non-financial dichotomy in studies of listed companies. Beretta and Bozzolan (2004) report for their sample of Italian listed firms that 34.1 per cent of disclosures are financial, 12.1 per cent are non-financial and 53.8 per cent are not disclosed. On samples of UK listed companies, Linsley and Shrives (2006) report that 26.8 per cent of risk disclosures are financial against 73.2 per cent non-financial,11 and Abraham and Cox (2007) report that 60.5 per cent are financial against 39.5 per cent non-financial. Again our IPO analysis is in line with that of Linsley and Shrives and we discussed earlier the probable reasons for a discrepancy with the results of Abraham and Cox, where FRS 13 was deemed to influence the results.
In respect of the quantitative/qualitative dichotomy, we find that across our sample a quantity tends to be specified for only three types of risk: (i) Where a company is dependent on one or a few major customers or contracts (27 per cent of 51 companies which specify this risk attach a quantity). (ii) Keyman insurance (of 279 companies who declare a dependence on key personnel, about 8 per cent have indicated the amount of Keyman insurance taken out in respect of these personnel, which ranges from a minimum of £150 000 to a maximum of £2m). (iii) Large shareholdings (because shareholdings over 3 per cent are stated in the statutory declarations, all prospectuses quantify this risk; however, 27 companies also specified in their risk declaration that a large proportion of the shares would be in the hands of a significant shareholder(s) who would thereby be able to exert significant influence over decisions to be voted on by shareholders). Linsley and Shrives (2006) report that 5 per cent of risk declarations by their sample of UK companies are monetary against 95 per cent that are non-monetary, and this low proportion of quantitative disclosure accords with our analysis of IPO companies. For their Italian sample Beretta and Bozzolan report that 15.2 per cent of risks disclosed are quantitative against 31 per cent qualitative and 53.8 per cent not disclosed, again showing a predominance of qualitative risks reported.
We turn now to the economic sign of risk declarations. Risk declarations across our sample relate either to increased volatility or to a potential reduction in income/increase in costs and, therefore, we might categorize risks according to whether they express (i) uncertainty (upside and downside) or (ii) solely downside. The former would include statements, for example, that the sector is at an early stage of development or that sector demand in cyclical; the latter would include concerns over, for example, key employees leaving, competition from other businesses and inadequate financing being available for expansion, all of which would be expected to have a negative impact upon revenues. We find that across our sample, 9.45 per cent of risks express the potential for upside against 90.55 per cent that express downside. Our results for IPO companies are in contrast to those reported for listed companies, where greater emphasis is given to risks with a positive economic sign. Linsley and Shrives (2006) identify 25 per cent of disclosures as ‘good news’, 21 per cent of disclosures as ‘bad news’ and 54 per cent of disclosures as ‘neutral news’. Beretta and Bozzolan (2004) identify 10.3 per cent of disclosures as having a positive sign, 4.8 per cent as having a negative sign, 0.4 per cent as ‘equal’ and 84.5 per cent as not disclosed. The guidelines on risk reporting repeatedly stress the importance of information on key risks and the relative usefulness of disclosures on ‘good’ and ‘bad’ risks is a useful line to pursue in further research. Linsley and Shrives and Beretta and Bozzolan do not provide adequate examples of specific good news disclosures12 and this, too, would be useful information.
For the final piece of our analysis of disclosure type (i.e. risk management strategies), we turn to the extent to which risk statements included an element of mitigation. Across our sample, we found that only around 5 per cent of the time risks were mitigated, and overwhelmingly this was in relation to the potential loss of key employees where mitigating factors were Keyman insurance and share option and other schemes to persuade employees to stay with the company. This is in accordance with our findings in relation to internal control where we found that IPO risk statements were deficient relative to the risk disclosures made by listed companies. Beretta and Bozzolan (2004) report that 16.2 per cent of risk disclosures relate to information concerning risk management programmes, which they consider to be low, although this figure compares favourably with the IPO disclosures.
In summary, the major findings of this section are that IPO companies’ key risks statements are relatively deficient in their provision of information in relation to internal controls and risk management, which are a key feature of risk reporting guidelines for listed companies. IPO companies tend to disclose, via a separate corporate governance statement, that systems for overseeing the performance of the company are established around the time of listing. The fact that such systems were not already in place suggests that regulation may have a role to play in protecting investors’ interests, albeit that many companies may have considered such systems superfluous where companies were closely held prior to listing. IPO companies provide a much greater focus on forward-looking risks than listed companies, which would be expected to improve the quality of the risk information provided to company stakeholders. IPO company risk disclosures tend to include a higher proportion of risks related to the external environment, whereas listed company risk disclosures tend to include a higher proportion of risks related to the internal environment.
4. Regression analysis of risk disclosure
In this section, we fulfil the request of Healy and Palepu (2001) for further research into the reasons why firms engage in voluntary disclosure. We are not aware of any other studies that have undertaken such an analysis in relation to risk disclosures of newly listing firms.13 Table 1 suggests that risk disclosure has increased across time alongside increased emphasis on, and guidelines relating to, risk disclosure and a key dimension of the empirical analysis in this section is the extent to which disclosure has increased across time. In Section 4.1, we discuss the firm-specific factors that are likely to influence the degree of risk disclosure and the empirical analyses follow in Sections 4.2 (logit analysis of the factors, which determine the decision to voluntarily disclose a risk statement) and 4.3 (ordered probit analysis of the factors, which determine the extent of risk disclosure).
4.1. Factors affecting the disclosure of risk
Any reduction in information asymmetry, for example, via the disclosure of information on key risks, will reduce the firm's cost of capital and increase its value (see, inter alia, Barry and Brown, 1985, 1986; Lang and Lundholm, 1996). Jorgensen and Kirschenheiter (2003) argue that voluntary disclosure is more likely the more uncertain investors are about the variance of a firm's future cashflows. We use the following variables to capture information asymmetries: size (variable = mktcap), age (variable = age), book to market value14 (variable = tngmkt) and sector (variable = ind) of the issuing company. Smaller, younger companies with low book to market values are more likely to suffer from greater information asymmetry and, therefore, are more likely to include risk warnings in their IPO prospectuses. We allow for the possibility that the variables related to the age, size and book to market of companies do not fully reflect sector information asymmetry differences by including industrial sector dummy variables.
Leland and Pyle (1977) develop a model of financial market equilibrium under asymmetric information (between firm insiders and capital providers) in which entrepreneurs signal firm quality via their equity stakes. Other signals of quality are nominated advisor (NOMAD) or reporting accountant reputation and institutional investors with block holdings might also provide evidence of monitoring (Stoughton and Zechner, 1998). Signals of quality mitigate the effect of information asymmetry and, therefore, reduce the information risk premium, which in turn will reduce the need for risk warning disclosure. Leone et al. (2003) find that underwriter and reporting accountant reputation and venture capital backing are all negatively related to the extent of IPO proceeds disclosure. The variables we use to capture signals of quality are the proportion of equity held by the directors in the aftermarket (variable = dirpost), the employment of higher quality advisors (variables = nomad and repacc), and the proportion of equity held in blocks by institutional investors (variable = instblock).
Although the reduction of information asymmetry might favour risk warnings, disclosure of proprietary information might undermine the company's competitive position (Verrecchia, 1983; Darrough and Stoughton, 1990; Wagenhofer, 1990; Feltham and Xie, 1992; Darrough, 1993; Newman and Sansing, 1993; Gigler, 1994). When the ICAEW (1997) put forward proposals for companies to include a Statement of Key Risks in their annual reports, a study of managers’ reactions found that increased risk disclosure was unpopular, with the fact that this would undermine the competitive position of the disclosing company being the primary objection cited by responding Finance Directors (Tillinghast-Towers Perrin, 1998). Leone et al. (2003) argue that proprietary costs are likely to be higher where firms face high growth rates. In support of this hypothesis, they find that high-technology IPO companies make relatively less voluntary disclosure. We assign a company high-tech status where it is in a sector associated with a high degree of scientific/technological input (variable = hitech).
It remains possible that risk warning disclosures are simply indicative of an IPO firm being a higher risk, or lower ‘quality’ firm. If we define a ‘quality’ firm as one associated with lower ex ante aftermarket volatility, then quality is related to risk. Many variables might proxy for both information asymmetry and the expected volatility of returns. We identify two risk variables that are unrelated to information risk and yet are likely to give rise to higher post-IPO volatility. Ceteris paribus, it is expected that if risk warnings declarations are driven by risks other than information risk, firms with lower levels of liquidity, as proxied by the percentage of equity in public hands (variable = pereq) and higher levels of gearing (variable = gearing) are more likely to include risk warning declarations in their IPO prospectus.
NOMADs have reputation capital to protect, from both association with an IPO which subsequently performs poorly, and from association with legal proceedings resulting from the inadequacy and/or inaccuracy of information provided to investors. In the USA, prestigious underwriters often compensate clients at the threat of litigation (Ibbotson et al., 1988), despite the fact that the greatest part of compensation claims are met by insurers of the directors and officers (Alexander, 1993). We acknowledge that on average NOMADs operating on the AIM provide a lower level of certification than a traditional investment bank acting as an underwriter and have less legal exposure.15 Risk disclosure will serve three purposes for the NOMAD. First, in the event that certain of the risks disclosed come to fruition, investors are less likely to view the NOMAD to the issue unfavourably where they have been forewarned. Second, the share price reaction to events of which investors have been unambiguously forewarned will be less than where information has been less clearly communicated; hence, the NOMAD is less likely to be associated with a company that performs poorly in the secondary market. Third, risk warnings will reduce the threat of litigation.
If NOMADs use risk warnings to protect their reputation capital, then we would expect that the greater the reputation capital of the NOMAD at stake, the greater the incidence of risk warnings. If, on the other hand, the reputation of the NOMAD acts as a monitoring device, as argued above, we would expect to see the reverse. As higher quality firms select higher quality advisors (see, inter alia, Carter and Manaster, 1990), it is also possible that the variable nomad will proxy for firm quality.16
To summarize, we argue in this section that the extent of risk warning disclosure might be driven by (i) the extent of issuing firm information asymmetry; (ii) the extent to which alternative signals of quality (monitoring) exist; (iii) the extent to which the IPO firm might suffer proprietary costs; (iv) the expected aftermarket volatility of returns; and (v) the self-interests of NOMADs to protect their reputation capital. We now undertake an empirical analysis to determine the extent to which each of these factors determines the existence and extent of risk warning disclosure.
4.2. Logit analysis of the factors that affect disclosure
We use a logit model to analyse the factors that determine the propensity to carry a risk warning disclosure. The sample for this analysis consists of 195 IPO companies for the period 1991–1999 (i.e. the period in which disclosure contained a voluntary element since for the years 2000 et seq. risk statements tended to be ubiquitous). The dependent variable (risk) is defined as the propensity to carry a risk warning. Risk takes a value of 1 where a prospectus carries a risk warning, and a value of 0 otherwise.
The selection of firm-specific explanatory variables is explained in Section 4.1. We add to the model a variable time, which takes a value of 1 if the IPO occurred in the period 1991–1995 and thereafter increases by a value of 1 each year.17Time should have a significant positive coefficient in the presence of firm-specific determinants of risk declarations if the increased regulatory focus on risk had an impact on risk disclosure practices. The model to be tested is therefore:
(1)Full variable definitions and the expected signs on each of these variables are provided in Table 3. Variable statistics and a correlation matrix for the variables are provided in Table 4.
| Name | Description | Expected sign |
|---|---|---|
| Dependent variables | ||
| Logit analysis of 195 sample companies across all sectors for the period 1991–1999 | ||
| risk | A dummy variable which takes a value of 1 where the IPO prospectus carries a risk warning, and 0 otherwise | |
| Ordered probit analysis of 68 companies in the IT sector for the period 1991–2003 | ||
| risknumber | A dummy variable which takes a value of 1, 2 or 3 where the number of separate sources of risk disclosed is, respectively, 7 or less, between 8 and 10, and 11 or greater | |
| risklength | A dummy variable which takes a value of 1, 2 or 3 where the length of the risk statement is, respectively, 0.75 pages or less, between 1 and 1.25 pages, and 1.5 pages or greater | |
| Explanatory variables | ||
| Information asymmetry | ||
| tngmkt | The book value of tangible fixed assets returned in the latest accounts for a full 12 month period prior to issue divided by the market value of the company at the issue price | – |
| mktcap | Natural logarithm of market capitalization at the issue price, in constant 1998 prices | – |
| age | The natural logarithm of 1 plus the age of the company at flotation (in years). The measure of the age of a company is from the date at which the current trade commenced within the company, as stated in the IPO prospectus | – |
| ind | A dummy variable which takes a value of 1 for firms within the Industrials sector, and 0 otherwise | – |
| Monitoring/signals | ||
| nomad | The natural logarithm of the IPO market share of the nominated advisor (sponsor) over the period 1991–1998 | – |
| repacc | A dummy variable which takes a value of 1 where the reporting accountant is a member of the ‘Big Six’ group, and 0 otherwise. The Big Six for the sample period are the following Chartered Accountants and any merger between them: KPMG, Coopers and Lybrand, Ernst & Young, Price Waterhouse/ PricewaterhouseCoopers, Touche Ross/Deloitte Touche, Arthur Andersen | – |
| dirpost | The percentage of the IPO firm owned by directors immediately following the IPO | – |
| instblock | The percentage of shares to be held in blocks of 3 per cent or over by institutional investors immediately following the IPO | – |
| Proprietary costs | ||
| hitech | A dummy variable which takes a value of 1 where a company belongs to the Information Technology plus the narrower sector classifications of (i) Pharmaceuticals and Biotechnology and (ii) Telecommunications, and 0 otherwise | – |
| Nominated advisor (sponsor) reputation capital | ||
| nomad (as above) | As above | + |
| Other risks | ||
| pereq | The percentage of the IPO company offered for sale at the IPO | – |
| gearing | Immediate post flotation debt as a percentage of debt plus the market value of equity | + |
| Panel A: Variable statistics | ||||||||
|---|---|---|---|---|---|---|---|---|
| Risk (n = 146) | No risk (n = 49) | |||||||
| Mean | SD | Minimum | Maximum | Mean | SD | Minimum | Maximum | |
| tngmkt | 0.12 | 0.26 | 0.00 | 2.05 | 0.20 | 0.25 | 0.00 | 1.02 |
| mktcap | 9.14 | 0.90 | 6.94 | 11.99 | 9.36 | 0.80 | 6.93 | 11.06 |
| age | 2.17 | 1.02 | 0.00 | 5.46 | 2.54 | 0.99 | 0.69 | 4.72 |
| dirpost | 0.40 | 0.22 | 0.00 | 0.97 | 0.49 | 0.24 | 0.00 | 0.90 |
| repacc | 0.36 | 0.48 | 0.00 | 1.00 | 0.43 | 0.50 | 0.00 | 1.00 |
| instblock | 0.09 | 0.16 | 0.00 | 0.82 | 0.10 | 0.15 | 0.00 | 0.53 |
| nomad (%) | 0.33 | 0.43 | 0.00 | 3.53 | 0.43 | 0.31 | 0.00 | 0.94 |
| hitech | 0.21 | 0.41 | 0.00 | 1.00 | 0.16 | 0.37 | 0.00 | 1.00 |
| pereq | 0.36 | 0.21 | 0.01 | 1.00 | 0.31 | 0.16 | 0.07 | 1.00 |
| gearing | 0.07 | 0.11 | 0.00 | 0.65 | 0.11 | 0.14 | 0.00 | 0.85 |
| Panel B: Correlations | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| risk | tngmkt | mktcap | age | dirpost | repacc | instblock | nomad | hitech | pereq | |
| risk | 1.000 | |||||||||
| tngmkt | –0.128 | 1.000 | ||||||||
| mktcap | –0.109 | –0.047 | 1.000 | |||||||
| age | –0.158 | 0.236 | 0.100 | 1.000 | ||||||
| dirpost | –0.171 | –0.154 | –0.168 | 0.200 | 1.000 | |||||
| repacc | –0.059 | 0.124 | 0.369 | 0.097 | –0.081 | 1.000 | ||||
| instblock | –0.032 | 0.022 | 0.055 | –0.084 | –0.379 | 0.188 | 1.000 | |||
| nomad | –0.107 | 0.020 | 0.322 | –0.051 | –0.143 | 0.205 | –0.027 | 1.000 | ||
| hitech | 0.053 | –0.210 | 0.159 | –0.061 | –0.020 | 0.058 | 0.066 | –0.092 | 1.000 | |
| pereq | 0.107 | –0.012 | –0.235 | –0.231 | –0.417 | –0.002 | 0.230 | 0.094 | –0.121 | 1.000 |
| gearing | –0.148 | 0.000 | 0.051 | 0.043 | –0.015 | –0.008 | –0.075 | 0.029 | –0.148 | –0.062 |
- SD, standard deviation.
Results for the logit analysis (Model 1) of the factors, which predispose a firm to disclose a risk warning, are presented in Table 5. We used Hendry's general to specific methodology and we present only the parsimonious version of the model, which contains variables significant at a 10 per cent level or greater.
| Predicted sign | Coefficient β | SE β | Wald | χ2p-value | |
|---|---|---|---|---|---|
| Constant | 6.347 | 2.371 | 7.165 | 0.007 | |
| tngmkt | – | –1.781 | 0.655 | 7.403 | 0.007 |
| mktcap | – | –0.425 | 0.226 | 3.553 | 0.059 |
| dirpost | – | –3.599 | 0.985 | 13.339 | 0.000 |
| instblock | – | –2.595 | 1.332 | 3.797 | 0.051 |
| gearing | + | –2.615 | 1.418 | 3.400 | 0.065 |
| time | + | 0.388 | 0.162 | 5.730 | 0.017 |
| N | 195 | ||||
| Max-rescaled R2 | 0.207 | ||||
| Prediction ratio | 0.737 |
- Estimation of equation (1) based on 195 second tier market IPOs for the period 1991–1999. See Table 3 for variable definitions. The predictive power of the cumulative probit model is measured by the generalized R2 measure proposed by Cox and Snell (1989) adjusted to achieve a maximum value of 1 (Nagelkerke, 1991).
We undertake a robustness check by re-estimating Model 1 using companies listed on the AIM market only (i.e. the 12 companies listed on the USM are omitted from this analysis). We do not tabulate the results of this analysis but we report that the coefficient on the variable nomad is negative and significant at the 5 per cent level, that mktcap and time both fail to be significant (time has a p-value of 0.1312) but otherwise all variables shown in Table 5 are significant at the 5 per cent level.
The variable time has a positive coefficient and is significant at the 5 per cent level across the full sample, but it fails to be significant in the robustness check. This suggests that the significant increase in the disclosure of risk in the period 1991–1999 is driven largely by the increase between 1991 and 1995, although there is some evidence of an increase in disclosure between the period 1995 and 1999.
Risk warnings are more prevalent among firms with low book to market values (tngmkt) and small firms (mktcap). Therefore, risk warnings appear to be used to resolve information asymmetry between the issuing company and investors; however, as we acknowledged in Section 4.1, variables that proxy for information asymmetry might also proxy for ex ante expectations regarding the volatility of a firm's shares. The variable gearing represents financial risk and we find that this is negatively related to the incidence of risk warnings, and is significant. We had predicted that if risk per se, and not just information risk, were a determinant of risk warnings, this variable would be significantly positively related to risk warnings. However, we can interpret the significant negative relationship between gearing and risk within the context of information asymmetries. IPO firms with higher debt levels have been vetted by creditor banks. Therefore, a higher gearing level might also act to reduce information asymmetry by providing the guarantee of the creditor banks. Alternatively, it might be that only the highest quality firms feel confident about undertaking an IPO with relatively higher debt levels, and these firms are less inclined to make risk disclosures.
We find that risk warning disclosure is significantly reduced where monitoring or other signals of firm quality are present. The variables dirpost and instblock are both significant and negatively related to the incidence of risk warnings. The sign on the variable nomad is negative and significant in the robustness check on AIM companies (which is not reported in Table 6), which refutes the hypothesis that quality NOMADs are more inclined to include risk warnings to protect their reputation capital. However, the negative sign is as predicted under our monitoring hypothesis and under the hypothesis that the variable nomad is a proxy for firm quality. The fact that few risk warning disclosures appear on the first tier market suggests that the selection of the market of listing might also operate as signal of firm quality. Therefore, the use of signals of firm quality reduces the incidence of risk warnings. This suggests that in certain cases managers prefer alternative means of resolving information asymmetry to the exclusion of risk warnings, and, hence, risk warnings are either more costly, less effective or disadvantageous in some cases.
| Variable | Predicted sign | Coefficient β | SE β | Wald | χ2p-value |
|---|---|---|---|---|---|
| Panel A: Number of sources of risk (dependent variable = risknumber) | |||||
| mktcap | – | 0.429 | 0.161 | 7.138 | 0.007 |
| age | – | –0.376 | 0.184 | 4.184 | 0.041 |
| time | + | 0.215 | 0.080 | 7.156 | 0.007 |
| Intercept 3 | –4.946 | 1.648 | 9.005 | 0.003 | |
| Intercept 2 | –3.851 | 1.617 | 5.667 | 0.017 | |
| N | 68 | ||||
| Max-rescaled R2 | 0.273 | ||||
| Prediction ratio | 0.741 | ||||
| Panel B: Length of risk disclosure (dependent variable = risklength) | |||||
| mktcap | – | 0.399 | 0.161 | 6.150 | 0.013 |
| dirpost | – | –1.177 | 0.705 | 2.782 | 0.095 |
| time | + | 0.310 | 0.084 | 13.549 | 0.000 |
| Intercept 3 | –5.515 | 1.734 | 10.120 | 0.001 | |
| Intercept 2 | –4.474 | 1.702 | 6.911 | 0.009 | |
| N | 68 | ||||
| Max-rescaled R2 | 0.324 | ||||
| Prediction ratio | 0.768 | ||||
- Estimation of equation (1) (in the text) on 68 IPOs in the Information Technology sector for the period 1991–2003. See Table 3 for variable definitions. The predictive power of the cumulative probit model is measured by the generalized R2 measure proposed by Cox and Snell (1989) adjusted to achieve a maximum value of 1 (Nagelkerke, 1991).
Finally, none of our industry dummies are significant and we conclude that the factors driving risk disclosure are captured by firm-specific variables and a time dimension.
4.3. Ordered probit analysis of the extent of risk disclosure
We undertake an ordered probit analysis of the extent of risk statements over the full sample period (1991–2003) across 68 IPOs in the IT sector,18 a sector that is characterized by greater homogeneity in terms of the types of risks declared than the other broad (1 digit) industrial sectors. The IT sector contains only two 2 digit subsectors: (i) IT Hardware and (ii) Software and Computer Services. Less than 5 per cent of our sample IT companies belong to the former subsector.19
The variable risknumber takes a value of 1, 2 or 3 where the number of separate sources of risk disclosed is, respectively, 7 or less (n = 20), between 8 and 10 (n = 24) and 11 or greater (n = 24). All explanatory variables are as for the logit analysis in Section 4.2.
In addition to the number of separate sources of risk disclosed, we also examine the length of the risk statement declaration, with the length being given to the nearest 0.25 of a page. There is a correlation of +0.88 between the length of the risk statement and the number of separate sources of risk disclosed; however, we also create a variable risklength, which takes a value of 1, 2 or 3 where the length of the risk statement is, respectively, 0.75 pages or less (n = 24), between 1 and 1.25 pages (n = 22) and 1.5 pages or greater (n = 22).
We find that only three variables are significant determinants of the number of sources of risk disclosed across the IT sector. Two of the variables are firm specific with younger and larger companies making greater disclosure by this measure. The third significant variable is time, which is positively related to the number of sources of risk disclosed. This result is confirmed when the analysis is repeated employing only ‘non-boiler plate’ disclosures.20
We again find that three variables are significant determinants of the length of risk disclosures with larger companies and those with lower directors’ shareholdings making greater disclosure. Time is again positive and significant.
The primary conclusion to draw from this analysis is that the extent of risk disclosure via both measures has increased across time alongside increased regulatory emphasis, even though none of this regulation was aimed specifically at IPO companies. This provides evidence that increased regulatory emphasis which stops short of mandatory requirement is effective in achieving increased disclosure.
It seems likely that smaller firms make less disclosure as they simply have a more limited business and, therefore, it is unlikely that there would be a correlation between the variance of returns and the extent of risk disclosure, although this remains an interesting avenue to pursue in future research. We provide some evidence that where directors hold a large proportion of shares in the IPO company, the extent of disclosure is less. By holding larger blocks of shares, directors signal their confidence in the future of the firm thereby mitigating the need to reduce information asymmetry via disclosure. It is also possible that where directors hold larger blocks of shares, they hold more influence over the final prospectus format and they are less keen to depart with what they deem to be proprietary information. A survey cited earlier of the Finance Directors of the FTSE 500 companies undertaken by Tillinghast-Towers Perrin (1998) suggested that considerations of commercial confidentially would limit the value of risk statements.
5. Concluding discussion
The aims of this paper were threefold. The first aim was to compare the disclosures of IPO companies with those of listed companies reported in the extant literature. A 1999 report by the ICAEW calls for greater uniformity in the presentation of key risks such that the average investor can readily gather and analyse the requisite information, envisaging the provision of information on risk in a systematic way. UK IPO companies provide risk disclosures in the form of a statement of key risks and the extent to which risk disclosure should be of a standardized format, possibly in the form of a statement of key risks, remains a concern for regulators. IPO risk disclosures are certainly easier to identify. We find that the IPO risk statements tend to be forward looking, whereas listed companies disclose more information related to the past, but that IPO disclosures also contain less information relating to risk management and internal controls than is provided by listed companies. Separate statements on corporate governance are provided in IPO prospectuses, which provide some information on internal controls, but these tend to be formulaic. Many IPO companies disclose that they set up internal procedures to oversee performance around the time of listing, which suggests that companies only act when obliged to do and, hence, regulators may have an important role to play in providing safeguards to investors (we acknowledge that prior to listing closely held companies may have considered such systems superfluous). IPO risk statements focus almost exclusively on downside risks rather than ‘good’ risks. We suggest that the extent to which investors are interested in ‘good’ versus ‘bad’ risks remains a useful avenue for future research. IPO disclosures are also more externally rather than internally focused, which again contrasts with listed companies, although this is likely to be a function of the maturity of the company and the sector in which it operates.
The second aim of the paper was to determine the extent to which risk disclosure increased across time given the increased emphasis on risk by various regulatory bodies. We find evidence that risk disclosure has increased across time although none of the risk regulation was aimed specifically at IPO companies. It would appear that increased regulatory emphasis which stops short of mandatory requirement is effective in achieving increased disclosure.
The third aim of the paper was to investigate the factors that determine the voluntary disclosure of risks in an IPO setting, which we analyse along two dimensions; (i) disclosure versus non-disclosure and (ii) the quantity of disclosure. We find that risk disclosure is more likely where IPO firms are associated with information asymmetry but that risk disclosure is not more likely where IPO firms are associated with either liquidity risk or financial risk, and, therefore, we conclude that while risk disclosure may be related to the extent of information risk, it would seem unrelated to risk per se. We find that various signals of quality (directors and institutional shareholdings and the employment of higher quality advisors) reduce the incidence of risk warnings. This supplanting of the role of risk disclosures by alternative means of reducing information asymmetry suggests that increased risk disclosure is not perceived to be beneficial in all cases. Reasons for this might be either the concerns of managers regarding the fact that risk warnings might give away competitive advantage, or that in certain cases risk disclosures are more costly to provide or less effective than other means of resolving information asymmetry.
