Earnings management around UK open offers

Abstract

We examine the long run operating and stock price performance of UK open offer firms in the context of the earnings management hypothesis. We find that in the pre-offer period offer firms report significant improvements in their operating performance unrelated to cash flow performance. Results on return performance show that offer firms outperform various benchmarks in the pre-offer year but underperform up to four years after the offer. Regression results show that pre-offer discretionary current accruals predict the long-run post-offer return underperformance but do not predict the short-run reaction to SEO announcements. Our findings are more consistent with the earnings management hypothesis than with either the timing hypothesis or the managerial response hypothesis and suggest that investors do not take full account of the information available at the time of open offers.

Keywords:

• Earnings management
• open offers
• SEOs
• return performance
• operating performance

Acknowledgements

We have benefited from the comments of Martin Walker, Ian Garrett, William Forbes and seminar participants at the University of Stirling. We are also grateful to Weimin Liu for generously providing data to estimate the Fama–French model. We are particularly grateful to the referees and the editor for several helpful suggestions. Norman Strong gratefully acknowledges research sponsorship by the Economic and Social Research Council (grant number R000222797).

Notes

Teoh et al. define aggressive and conservative offer firms as the SEO firm quartiles with the largest and smallest discretionary current accruals in year −1.

In a related paper (Iqbal et al., 2006), we examine the earnings and return performance of a sample of 424 UK rights issues. As rights issues are primarily sold to existing shareholders, they are less comparable to SEOs used in US studies.

Details of the 44 industry groups are available on request.

Asset-scaled net earnings (ANE) is net earnings (Datastream item 625) divided by opening total assets (Datastream item 392). We require ANE of the matched firm in the pre-offer year to be at least 80% of that of the offer firm, with no upper limit.

ROS is net earnings divided by total sales (Datastream item 104) from the same period. ACF is cash flow from operations (CFO) divided by opening total assets. CFO uses Datastream's definition pre-1992 and reported CFO post-1992; Datastream's definition in terms of Datastream item codes is (where 445 is changes in stock and W.I.P., 448 is changes in debtors, 417 is changes in creditors and 1012 is other changes).

The bias arises from the potential for the Jones model to measure discretionary accruals with error by attributing extreme earnings performance to discretionary accruals (Dechow et al., 1995).

The 12-, 24-, 36-, and 48-month post-issue periods include the offer month 0.

See Lyon et al. (1999, p. 173–4) and Brown and Warner (1980, p. 251–2).

Untabulated calculations show that £1 invested in offer firms 12 months before the offer month is worth £1.13 by the offer month.

Subsequent to completing this research, a referee made us aware of a study by Ngatuni et al. (2007) that examines the long-term performance of UK SEOs and includes a subsample of open offers from the same period as we examine. Ngatuni et al. impose a survivorship bias on their results by requiring that that their sample firms, but not their matched firms, survive over their five-year post-offering period. Calculating average size- and book-to-market-matched buy-and-hold abnormal returns, they report insignificant post issue 1- and 3-year figures of −6.7% and 15.6% and significant 5-year figures of 70.16% for 132 open offer firms. To validate our own results, we applied the matching procedures of Ngatuni et al. to our own sample, but without imposing any survivorship bias. Matching on either size, size and book-to-market, or size and industry, we found mean BHARs were significantly negative at horizons between 2 and 4 years and were quantitatively similar to those in Table 5.

The compound FFAR is significantly negative for aggressive firms throughout the 48-month post-issue period, while it is (weakly) significantly negative for conservative firms at the end of 24 and 36 months post-offer.

An analysis of BHARs and FFARs over the −12 to 47 month period for issuing firms in the second and third earnings management quartiles (not tabulated) confirms the pattern of results reported here, with the degree of post-issue underperformance increasing monotonically in magnitude and significance with the degree of earnings management.

We use two-year post-issue returns in the regression following our finding in Table 5 that there is no significant underperformance on an annual basis after the second post-offer year.

CAPEX, in terms of Datastream items, is 435+438−423 for the period up to and including 1991 and is post-1991. Datastream item 435 is total new fixed assets, 438 is intangible purchases, 423 is sale of fixed assets, 1026 is net payments for fixed assets, 1029 is net payments: intangibles, and 1037 is net payments: subsidiaries etc.

The industry dummies used in the regression (in terms of LSPD codes) are; oil, gas, and mineral (125, 162, 165); building related (210, 222, 225); chemicals, pharmaceuticals, and healthcare (232, 234, 236, 370, 360); electronic (252, 253); engineering and vehicles (261, 262, 265, 268, 270); paper & packaging, publishing & printing (282, 284, 436); clothing (291, 293, 295, 297); food producers, brewers, pubs & restaurants (333, 470); distributors (412, 413, 414); hotel, leisure, and entertainment (424, 426, 428); media, and broadcasting (432, 434); retailers (440, 452, 454); computer software (487); transport (490); and other (240, 342, 346, 481, 482, 485).

Diagnostic tests show there is no heteroskedasticity, and multicollinearity is unlikely to be a problem.

The only other variable that is significant in Panel B is ΔCAPEX, its positive coefficient indicating that increases in post-issue capital expenditure reduce the extent of post-issue return underperformance.

We also examine the effect of replacing DCA₋₁ in regression (6) with performance-matched DCA₋₁. Examining whether DCA₋₁ of offer firms in excess of DCA₋₁ of performance-matched non-offer firms predicts long-run returns addresses the evidence of Xie (2001) for the US that the mispricing of discretionary accruals is a general phenomenon, not restricted to IPOs or SEOs. This substitution leaves the signs and significance of all the coefficients unchanged, with the coefficient on “excess” DCA₋₁ being −0.77 (). This means we can predict long-run post-offer returns using earnings management by offer firms conditional on the level of corporate earnings management that exists in general.

The study of Ngatuni et al. (2007) has a survivorship bias (see note 10). However, this is unlikely to have a significant effect on returns in the announcement month.