Real and accrual earnings management and IPO failure risk

Abstract

This paper analyses the relationship between real and accrual earnings management activities and IPO failure risk. While the association between accrual earnings management and IPO failure has been researched in a limited setting, to date, there has been no work that analyses the impact of real activities-based manipulation on the probability of IPO failure. Based on a sample of 570 UK IPO firms that went public over the period 1998–2008, we find evidence that IPO firms manipulate earnings upward utilising real and accrual earnings management during the IPO year. We also find that IPO firms with high levels of real and/or accrual earnings management during the IPO year have a higher probability of IPO failure and lower survival rates in subsequent periods. In addition, we find that IPO firms experience a higher probability of IPO failure and lower survival rates in the post-IPO period when greater real earnings management takes place during the IPO as compared to accrual earnings management. While our work contributes to the growing literature on real and accrual earnings management around IPOs, the majority of our failed IPO events are from the Alternative Investment Market and occur during the financial crisis. Future research, therefore, should consider whether these results are generalisable to more developed firms and less turbulent economic environments.

Keywords:

• initial public offerings
• IPO failure
• earnings management
• real activities
• accounting accruals
• accounting choices

Acknowledgement

We thank Martin Walker, Igor Filatotchev, and Steve Toms for their helpful comments and suggestions. We are also grateful for the comments and feedback from the participants at the European Accounting Association-Annual Conference (2013, Paris). Mohammad Alhadab acknowledges the financial support he has received from Al albayt University, Mafraq, Jordan. We thank the associate editor and two anonymous reviewers for their constructive and helpful comments on our work. All errors remain our own.

Notes

1. Consistent with prior research (Li and Zhou 2006) we define IPO failure as those IPO firms who delisted from the stock exchange for negative reasons (involuntary delisted) within 5 years post the IPO date.

2. It is worth mentioning that Demers and Joos (2007) examine the association between several variables and IPO failure risk. However, they do not examine the association between real and accrual earnings management and IPO failure. As such, there is a significant difference between the approach of the current piece to estimate real activities manipulation and the work of Demers and Joos (2007).

3. Data are taken from Bloomberg and cover IPOs across the EU and the UK, including financial firms.

4. There are secondary markets for small firms in France, Germany and Italy. However, over the period these markets have been subjected to considerable change, fragmentation and consolidation, for example, the introduction of Euronext in 2005. AIM, however, has existed without similar disruption since 1995.

5. The revelation principle is analogous to the approach of Modigliani and Miller (1958) to set out the irrelevance of capital structure, and of Miller and Modigliani (1961) for the dividend irrelevance theorem.

6. The role of the underwriter is also an important consideration given our theoretical framework; more specifically, it suggests that the underwriters align their interests with those of the management/current owners. Owners/managers and underwriters share the common goal of the successful flotation of the firm. The failure of a flotation is clearly detrimental to the current owners/managers and equally it will damage the value and reputation of the underwriter. Thus, while underwriters have a responsibility to ensure the accuracy of the information in IPO prospectuses, they have obvious incentives to ensure a successful flotation.

7. It is worth noting that Hypothesis 1 is not new to the literature and it has been already addressed by prior research in the IPO setting (Teoh et al. 1998a, Wongsunwai 2012). However, it is necessary to examine this for consistency with the prior literature.

8. Li and Zhou (2006) define failure as delisting for negative reasons (involuntary delisting). IPO firms that delisted from the stock exchange within 5 years after the IPO date and have delisted codes between 400 and 600, excluding the following codes: 501, 502, 503 and 573 are classified as failed. Whereas codes 501, 502 and 503 denote those firms that switch from one stock exchange to another, code 573 denotes firms that choose to go private.

9. The London Stock Exchange provides information about IPOs on the Main market starting from 1998 while information about IPOs on the AIM market starts from 1995. Therefore, and to be consistent, our sample covers the period 1998–2008.

10. From those 60 delisted IPOs for negative reasons, 35 went into administration, 22 into liquidation, 2 into receivership and 1 into winding up. Further, the definition of failure is consistent with García Lara et al. (2009) who examined earnings quality for failed firms in the UK and defined failure as those firm delisted for administration, liquidation and receivership.

11. Our control sample consists of all UK active and dead firms over the sample period to avoid survivorship bias. We also repeated our analysis using 10 observations for each industry-year group and the results are qualitatively similar but this restriction leads to a large decrease in our sample size and, therefore, we follow Rosner (2003), Athanasakou et al. (2009), Iqbal et al. (2009) and Athanasakou et al. (2011) and use six observations.

12. To overcome any misspecification of the financial year end, the financial data we obtained from WorldScope are cross checked with the financial data in the prospectus and the results are qualitatively similar.

13. For those IPO listed in 2007–2008 we apply the same criteria as for all of our earlier IPOs. As the end of our study period is December 2012, we have a 5-year post IPO period to ensure that we capture all firms in our IPO sample. It is worth noting that there were 42 IPOs in 2007–2008. Out of these, just nine firms delisted and these nine IPOs were delisted for non-negative reasons. Over the following 5 years, of the 9 delisted IPOs, 4 registered as private companies, 3 were taken over 1 firm no longer met the listing conditions and reverted to being a private company; and 1 firm delisted as a company request.

14. An audit firm is classified as big N if it is considered as one of the big 4 audit firms, namely PricewaterhouseCoopers, Deloitte Touche Tohmatsu, Ernst & Young and KPMG.

15. Prestigious underwriters are those global investment banks as defined by Derrien and Kecskes (2007), while VC are those investors who hold more than 3% of a firm's shares and appear in the list of VC provided by the British Venture Capitalist Association. Specifically, data are collected from the prospectuses about all the shareholders who hold more than 3% of the total shares and then a shareholder's name is matched with a list of VC, which is obtained from the British Venture Capitalist Association.

16. This approach of incorporating the aggregate measure of real activities (a combination of two real activities) and discretionary accruals into one model is consistent with recent research that investigates earnings management (Wongsunwai 2012, Zang 2012).

17. As our proxy for size is insignificant in our regressions, we re-run the models by including just earnings management proxies and the size effect variable. The result shows that the size effect predicts the probability of failure, confirming the view that small IPO firms have a higher probability of failure as compared to large firms. However, when we add more explanatory variables into the model the size effect disappears. This suggests that there are many variables the correlate with firm size and, therefore, can be used as a proxy of size.

18. Firms at risk represent firms that are still listed on the market but there is a probability that these firms may fail in the future for any reason.

19. Compared with other hazard models, the Cox (1972) model has the advantage that no assumption is required to be made about the distribution of event dates.

20. This process is the same for all of our earnings management proxies.

21. We also re-estimate the model by adding a dummy variable for high-tech industries and the results are qualitatively similar to those reported in the paper.

22. We also report the hazard ratio which is computed as the exponentiated coefficient for each variable. For a dichotomous variable, the risk ratio is the ratio of the estimated hazard for firms with ‘1' to the estimated hazard ratio for firms with ‘0' (Jain and Martin 2005, Carpentier and Suret 2011). While the interpretation of the hazard ratio for a continuous variable is that (hazard ratio –1)*100 represents the percentage of changes in hazard for each unit increase in the variable (Allison 1995, Jain and Martin 2005). Following prior research (Teachman 1983, LeClere 2000, Jain and Martin 2005), risk ratios greater than 1, equal 1 and less than 1 are interpreted as follows; rapid time to failure, no impact on failure and slower time to failure, respectively.

23. Specifically, in Table 9 (Model 5) we find a positive coefficient of 0.571 (P < 0.05) and hazard ratio of 1.77 for the aggregate measure of real earnings management (REM_Index), while we find a positive coefficient of 0.542 (P < 0.05) and hazard ratio of 1.72 for accrual earnings management (DISACCR). It is worth noting that the magnitude of the difference in our coefficients and hazard rates for Table 9 is small but consistent with prior evidence. However, if the results of Model 5 in Tables 7–9 are taken in combination, then while the magnitude of the coefficients varies depending on model specification, it is clear that real earnings management clearly is more dominant than accrual earnings management.