Behavioral Timing, Valuation and Postissue Performance of UK Initial Public Offerings

ABSTRACT

Using a sample of 1,926 UK initial public offerings (IPOs) launched from 1987 to 2007, this study introduces a new angle on testing the behavioral timing hypothesis in the context of UK IPOs via investigating relationships between the magnitude of IPOs misvaluation and postissue stock price and operating performance. IPO misvaluation is measured using (i) an intrinsic value of the firm estimated using residual income valuation model and (ii) intensity of IPO issuance activity. The findings show that stock price and operating underperformance in the postissue are directly linked to the degree of IPOs' misvaluation. Specifically, the stock price and operating performance are found to be significantly and robustly different between hot markets IPOs and cold market IPOs 3 years postissue. We also show that overvalued IPOs have lower long-run stock returns, but outperforming operating performance, than undervalued IPOs do. Our findings are broadly consistent with the behavioral explanations of the poor stock price and operating performance, supporting the U.S. results of Purnanandam and Swaminathan [2004] and Loughran and Ritter [2000].

KEYWORDS:

• Behavioral timing
• Initial Public Offerings (IPOs)
• Post-issue long-run performance
• IPOs' valuation
• IPOs' hot markets

Notes

1. Other approaches to testing the behavioral timing hypothesis use proxies to measure investor sentiment and stock misvaluations and then directly examine how these proxies impact the equity issuance activity. This presents a direct approach to empirically test the behavioral timing hypothesis (e.g., Lee, Shleifer, and Thaler [1991], Lowry, 2003]).

2. Examples of other proxies for stock misvaluations include analysts' growth forecasts (Rajan and Servaes [1997]) and earnings management (Teoh, Welch, and Wong [1998]). Teoh et al. [1998] exhibited evidence in support of a positive link between the IPOs mispricing and postissue stock price performance, using abnormal accruals and high reported earnings during the IPO-issue year as a proxy for this mispricing. Rajan and Servaes [1997] found that IPOs with highest analysts' growth forecasts significantly underperformed those with lowest analysts' growth forecasts.

3. Other explanations of the IPOs long-run underperformance include the pseudo market timing hypothesis (Schultz, 2003]) and the underwriter price support hypothesis (Ruud [1993]).

4. Other explanations to the post-IPO operating underperformance include agency hypothesis that attributes this underperformance to managers' inventing in non–value-maximizing projects (Jensen and Meckling [1976]).

5. The behavioral timing hypothesis versus the pseudo timing hypothesis in the context of UK IPOs has been studied by Gregory, Guermat, and Al–Shawawreh [2010]. However, in the present study I employed a substantially different approach. Mainly, they investigated the behavioral timing versus the pseudo timing hypothesis based on examining if abnormal returns measured in event time exist when measured in calendar time.

6. The present study substantially differs from Khurshed et al. [2003] and Coakley et al. [2004]. Khurshed et al. [2003], using a sample of sample of 415 UK IPOs launched on the official market and Alternative Investment Market over the period of 1995–1999, studied the postissue operating performance of UK IPOs. The results provide supportive evidence of deteriorating post-issue performance of firms going public on the Official List, but no evidence of underperformance is found on the Alternative Investment Market. Coakley et al. [2004] examined a sample of 568 UK IPOs launched during the time period of 1985–2000 and compare the postissue operating performance of venture-backed and non–venture-backed IPOs. Their findings exhibit a significant overall operational decline up to 5 years postissue, which is found to be mainly driven by the poor performance of IPOs launched during the 1998–2000 bubble.

7. The UK IPOs waves have been only studied by Michailides [2000] and Banerjee, Gucbilmez, and Pawlina [2013].

8. Excluding privatizations.

9. Using a moving average avoids classifying low-activity months as cold when they are actually normal.

10. Earlier theoretical roots can be found in Preinreich [1938] and Peasnell [1982], while a wide spectrum of studies has adopted recent empirical treatments of the model, such as Bernard [1994], Penman and Sougiannis [1997], Frankel and Lee [1998, 1999], Gebhardt, Lee, and Swaminathan [2001], and Bi and Gregory [2011].

11. There are a large number of forecasts missing because these are only available for the UK from 1987 and not for all the companies. Thus, many examples of earnings and dividend forecasts are missing. For example, 1-year-ahead forecasts are only available for approximately 57% of the sample firms, while 87% of firms that have a 1-year-ahead forecast also have a 2 year 1. Dividend forecasts 1 year ahead are only available for 75% of the firms which have earnings forecasts.

12. Practically, a constant payout rate could be also assumed, given the evidence on “sticky” dividends, and so assuming constant payout rate is a not unreasonable assumption and is more conservative than assuming constant payout in case if earnings are rising (Bi and Gregory 2011].

13. (1 + nominal rate)/(1 + index-linked rate) – 1.

14. As demonstrated by Gebhardt et al. [2001] and Bi and Grerory [2011], changing time horizons of mean reversion does not alter the model results. Gebhardt et al. [2001] showed similar results over horizons ranging from 6 to 21 years. Similarly, Bi and Gregory [2011] found very similar results using either 3- or 8-year mean reversion period.

15. Hall's [1992] adjustment is shown to perform better in situations of large skewness and small sample, which will suit the small size of my subsamples that are used in behavioral timing tests.

16. A full description of the factors constructions is provided in Gregory et al. [2009]. The factors are available on http://xfi.exeter.ac.uk/researchandpublications/portfoliosandfactors/.

17. Using EBITDA, which includes interest income in operating income allows for taking in account that managers might temporarily invests some of the offerings proceeds in interest-earning instruments before investing it in operating assets (Loughran and Ritter [1997]).

18. For example, Jain and Kini [1994], Mikkelson, Partch, and Shah [1997], Loughran and Ritter [1997], and Helwege and Liang [2004] reported median values.

19. The AIM is the LSE's market for smaller and growing companies. AIM started on the LSE in 1995. The market is designed to be more suitable to the needs of smaller and fast growing companies (i.e., lightly regulated compared to the main market).

20. For example, AIM accounted for 52% of total European IPOs in the year in 2005 (Pricewaterhousecoopers, 2007]).

21. The Mann-Whitney test exhibits insignificant variation among normal, hot, and cold markets (the results are not reported here).

22. I also tested the IPOs abnormal returns using calendar-time abnormal returns as a robustness check. Overall, the results are consistent with the findings drawn based on the BHAR test. The results are not reported in this study.

23. Based on the Mann-Whitney test, the degree of misvaluation is found to be significantly different between the 2 groups (the results are not reported here).

24. Difference tests, based on the Mann-Whitney test, show that the difference between the 2 groups is statistically significant (the results are not reported here).