Are seasoned equity offerings made in response to weak operating performance?

Abstract

This article examines the operating performance of firms surrounding Seasoned Equity Offerings (SEOs) and finds that weak operating performance by issuing firms begins during a 2-year period prior to issue. This is in contrast to the stylized facts that a seasoned equity issue initiates a period of weak performance. Our findings suggest instead that an issue is more likely a reaction to a period of weak performance that is already well under way. Consistent with previous studies, we find that weak performance continues after the issue despite the evidence of favourable macroeconomic conditions.

Notes

¹Ritter (2003) summarizes the evidence on stock returns as follows. Relative to carefully selected benchmarks SEO firms experience stock returns of 72% in the year prior to offer followed by an announcement effect of −2%, and 5-year abnormal returns of approximately −30% following the issue. Studies of operating performance include Hansen and Crutchley (1990), Healy and Palepu (1990), Cheng (1995), McLaughlin et al. (1996), Loughran and Ritter (1997), Teoh et al. (1998a, b), Hertzel et al. (2002). In general, these studies find that the operating performance of issuing firms peaks around the time of an SEO and then declines over the subsequent 4 years. Similarly, Hertzel et al. (2002) find poor post issue performance for firms making private placements.

² Other studies which examine operating performance in conjunction with an equity issue include Hansen and Crutchley (1990), Healy and Palepu (1990), Cheng (1995), Teoh et al. (1998a, b), Hertzel et al. (2002) and Dittmar and Thakor (2007).

³ See Masulis (1980) and Bayless and Jay (2003).

⁴ In comparison, Hansen and Crutchley (1990) examine a 7-year period from 1975 to 1982, Loughran and Ritter (1997) study an 11-year period from 1979 to 1989 and Dittmar and Thakor (2007) examine the 10-year period 1993 to 2002.

⁵ In contrast, in the sample studied by Loughran and Ritter (1997) contained nearly one quarter of issues occurred in a single year.

⁶ In addition to Loughran and Ritter (1997), see McLaughlin et al. (1998b), Alderson and Betker (2000), Lewis et al. (2001), Korajczyk and Levy (2003) and Huang and Ritter (2004).

⁷ See McLaughlin et al. (1996, 1998b), Korajczyk and Levy (2003), Huang and Ritter (2004), Leary and Roberts (2005) and Kayhan and Titman (2007).

⁸ See, for example, Jain and Kini (1994), McLaughlin et al. (1998a), Korajczyk and Levy (2003), Eberhart et al. (2004), Huang and Ritter (2004), Leary and Roberts (2005) and Mitton (2006).

⁹ Friday et al. (2000) use industry measures that are constructed using similar methodology. Patel et al. (1993) and McLaughlin et al. (1996) also evaluate SEO firm performance relative to the industry. Others including McLaughlin et al. (1998a) compare SEO performance to a portfolio of firms selected in part based on industry membership.

¹⁰ Similar macroeconomic measures have been studied by, among others, Eckbo et al. (2000), Korajczyk and Levy (2003) and Huang and Ritter (2004). We also examined percent change in real per capita consumption seasonally adjusted from the US Department of Commerce (Eckbo et al., 2000), but found this variable to be highly correlated with real GDP when measured on an annual basis.

¹¹ Other papers that study property, plant and equipment include McLaughlin et al. (1996, 1998b), Korajczyk and Levy (2003), Huang and Ritter (2004), Leary and Roberts (2005) and Kayhan and Titman (2007).

¹² See, for example, Hovakimian et al. (2004), Leary and Roberts (2005) and Kim and Weisbach (2008).

¹³ See, for example, Spiess and Affleck-Graves (1995), Brav et al. (2009) and Jegadeesh (2000).

¹⁴ See also Bayless and Jay (2003).

¹⁵ Including year −3 in the prior period yields results that are qualitatively identical to those we report below.

¹⁶ An important exception is Dittmar and Thakor (2007).

¹⁷ Studies that subtract control firm performance from SEO firm performance include McLaughlin et al. (1996, 1998a) and Friday et al. (2000). Pindyck and Rubinfeld (1998) describe the econometric implications of this procedure.

¹⁸ We trim the top and bottom 1% as recommended by Barber and Lyon (1996) as a way to eliminate the excessive influence of outliers and to mitigate the problems of skewness in our logistic regressions (papers that use this procedure include McLaughlin et al., 1996; Hovakimian et al., 2004).

¹⁹ Because we find that performance is weak in the prior period compared to the pre-issue period, inflating reported earnings in the prior period would cause the coefficients on earnings and profits variables to be reduced in absolute value. In the absence of earnings management, the coefficients in Table 4 would likely be more negative than those we report because performance in the prior period would be even weaker than what we currently find.

²⁰ The p-value for this variable is 0.078 in specification (1) indicating significance a 10% level.

²¹ Recent papers that discuss the influence of deviations from target leverage ratios on equity issuance include Hovakimian et al. (2004) and Welch (2004).

²² Lower values of the AIC measure indicate a more desirable model.

²³ For a summary of this literature, see Ritter (2003). For an exception see Dittmar and Thakor (2007).

²⁴ In other experiments we included additional firm characteristics including book-to-market and size, and we substituted book leverage for market leverage. The coefficients and significance levels for EPS in these specifications were similar to those reported in Table 4 but the coefficients for the additional firm characteristics variables were generally not significant. Because some previous studies report size effects (see, e.g. Brav et al., 2000), we also divided our sample into size terciles and re-ran the specifications reported in Table 4. The coefficients and significance levels for EPS showed little variation across the three size categories of firms. Similar to Kim and Weisbach (2008), we split our sample into issue size terciles and bifurcated the sample by period using 1987 as the point of division. These experiments failed to reveal any significant differences across subsamples from the results reported above. As a result, these and the other experiments described above are not reported in order to save space.

²⁵ See, for example, Barberis et al. (1998), Daniel et al. (1998), Teoh et al. (1998a, b), Hong and Stein (1999) and Baker and Wurgler (2002).

²⁶ Also consistent with this view, Loughran and Ritter (1997) and Kim and Weisbach (2008) report increased capital spending following issue.