Disclosed Values of Option-Based Compensation – Incompetence, Deliberate Underreporting or the Use of Expected Option Life?

Abstract

New accounting standards require firms to value the costs of option-based compensation (OBC). Earlier research has documented that firms in the US generally underreport the values of OBC by manipulating the model inputs used for valuation purposes. This paper examines the information on and values of OBC disclosed by Danish firms. The results show that many firms fail to provide the information required on OBC. However, this does not seem to be a deliberate attempt to hide information, but rather is the result of firms not paying enough attention to the information requirements. Similarly, when studying the disclosed values of OBC, there is no clear evidence of underreporting. For example, there is no evidence that firms use manipulated values for the Black–Scholes (Merton) model inputs in their valuations. Furthermore, firms determine the expected option life in a way that is generally consistent with the guidelines provided by the accounting standards. The only exception is when options are granted to the board of directors, as this led a few firms to underreport option values in a way that cannot be explained by an appropriate adjustment of the expected option life. These findings differ from those of the US, but are consistent with the more limited use of OBC and the lower level of attention paid to these values in Denmark. Furthermore, the financial press in Denmark and powerful Danish institutional investors seem to have a disciplinary effect on Danish firms such that most of the firms provide sufficient and accurate information even though the official consequences of not doing so are very limited.

Acknowledgements

The authors wish to thank the editor, two anonymous referees, Trevor Buck, Peter Ove Christensen, Martin Conyon, Thomas Hemmer, Bjørn N. Jørgensen, Peter Løchte Jørgensen, Ole Sørensen, David Yermack, and participants in the Workshop on Executive Compensation at Copenhagen Business School, the 2007 Loughborough University (ESRC) Conference on Executive Pay in London, the 2006 meeting of the Danish Center for Accounting and Finance (D-CAF), the 2006 Skinance Conference, the 2007 Ryde Workshop on Empirical Finance, and participants in seminars at Clarkson University, Copenhagen Business School, San Diego State University, and the University of Aarhus for many helpful comments. Excellent research assistance was provided by Simon Krantz Kristensen. Financial support from the Danish Center for Accounting and Finance (D-CAF) is gratefully acknowledged. An important part of this paper was performed while Toke K. Hjortshøj was affiliated with the School of Economics and Management, University of Aarhus. The views expressed in this article are those of the authors and not necessarily those of Sparinvest A/S.

Notes

Aboody et al. (2004, 2006) and Hodder et al. (2006) are among the few studies that examine the effect of the firms' corporate governance and this is only done by including a corporate governance index.

Updated frequently, the Rules have required information on OBC since the version that became effective as of October 1999. The main change that came into effect in January 2002 was the requirement that firms provide their own valuations of OBC.

The current (May 2008) exchange rate is DKK 100 = EUR 13.40 = USD 20.81.

This critique is voiced, for example, in several articles in the financial press. The CSE's passive role in connection with the lack of information on OBC is also discussed, for example, in Bechmann and Jørgensen (2002).

The Exposure Draft for IFRS 2, ED 2 Share-Based Payment (November 2002) and Bechmann (2003) provided some ideas on how the inputs could be estimated.

The following is mainly based on Rose and Mejer (2003) and Edling et al. (2007), both of whom provide a description of the Danish corporate governance system from an international perspective. However, several other articles and reports contain similar descriptions. A list of further references is available upon request.

These two pension funds generally favor the use of OBC. However, they suggest that the values should be reasonable and insist that firms provide complete information on OBC to shareholders.

For example, Bechmann (2008) finds that the ratio of remuneration for top executives relative to the remuneration for blue-collar workers is around 10 to 15 in Denmark. These numbers are much smaller than similar numbers from the US, where the reported ratios can be as high as 400 (see Daines et al., 2005). Bechmann (2008) also provides evidence that the remuneration of top executives is on the low side compared with other European countries.

The results described in this section are primarily based on work by Bechmann and Jørgensen (2004), whereas the US results primarily are from Murphy (1999). However, further information can also be found in various remuneration reports, for example, in the Towers Perrin Worldwide Total Remuneration Report.

See, for example, the so-called Nørby report, www.corporategovernance.dk.

The influence of size is often discussed and analyzed in the financial disclosure literature (see, for example, Watts and Zimmerman, 1990; Lang and Lundholm, 1993; Botosan, 1997).

To ensure high quality data, we have been in contact with the firms when information was missing or when we were unsure of how the information should be read. In many cases, this helped us to correct mistakes due to, for example, typos, etc. However, we also encountered firms that were not able to clarify information on their OBC.

The Danish Center for Corporate Governance assisted us by providing some initial information on ownership structure. Similarly, Johannes Raaballe, University of Aarhus, helped with information on the use of dual class shares.

In effect, this leads to the removal of one firm from 2002 to 2005. However, the following results are not sensitive to whether this firm is included or not. The firms that used other valuation models had to be excluded due to a lack of information on OBC.

The number of tranches varies from firm to firm with an average (median) of 5.7 (4.0) tranches. Similarly, the tranche size also varies with an average (median) tranche-level option overhang (number of options granted in the tranche divided by number of shares outstanding) of 0.7% (0.3%). The minimum tranche-level option overhang is 0.01% and the maximum is 6.7%.

We will only present the marginal effect given the mean values. The results are, however, qualitatively similar when we apply the medians instead. These results are available upon request.

Valuations at the tranche level provide similar results. Here, the mean across all 578 individual trances is 0.83.

In relation to these results, it is also interesting to examine if there are any differences between the values disclosed in accounting reports and the values disclosed in company announcements. One reason to believe that this is the case is that the valuations in company announcements are values for new option grants; hence, they relate directly to what, for example, executives are paid at the cost of shareholders, whereas the accounting reports simply disclose the value of all outstanding options. Furthermore, for grants in 2005, the values disclosed in company announcements provide the values that should be given. Hence, one could expect more underreporting for these values. The evidence, available upon request, does not point to this being the case, i.e. there are no significant differences between valuations disclosed in company announcements and valuations disclosed in accounting reports.

All observations from 2002 have to be excluded when including the lagged values of the dependent variable because they are only available from 2002 and onwards.

There are also examples of firms that have used indexed options where the exercise price is adjusted for changes in different indices. However, as they do not provide any details on how this feature has been taken into account, we have left them out of our analysis. Similarly, there are cases with knock-in features, performance criteria and other quite exotic features, but, interestingly, the firms do not appear to try to take these features into account in their valuations, even though doing so would nearly always lower the values. As a result, we have also decided to leave these options out of our analysis, but the conclusions remain unchanged even if they are included.

Details on the estimation of the interest rates (and the following model inputs) are provided in Appendix A.

In connection with the valuation, 62% of the observations provide some information on dividends, but only 42% of the firms provide information on the dividend yield (dividend rate) normally used when the Black–Scholes (Merton) model is applied to price options taking dividends into account. The remaining 58% provide information on dividends in monetary terms per share, but how this is used in the valuation of the options is not explained. In the following, we have assumed that the latter firms actually adjust this to a dividend yield, as described in Appendix A.

The few firms that explain how the volatility has been obtained all use historical volatility, generally based on a year of weekly or daily stock returns. The fact that only historical volatility is used is consistent with the fact that the options market in Denmark is not very developed, generally preventing the calculation of usable implied volatilities.

More precisely, there are examples where firms use volatilities that are either very large or very small compared with the historical volatilities. In one extreme case, a firm used a volatility of 5% even though our estimated volatility was more than 35%.

Even though OLS regressions are not appropriate in this case, we perform the same regressions using OLS with robust standard errors in order to check robustness. In results available upon request, we show that the conclusions are unchanged if OLS regressions are considered instead.

In the regressions, the number of observations per firm varies across firms and across years, but we have checked and confirmed that the findings are not in any way caused by the firms with relatively more observations.

However, note that the coefficient is positive but insignificant in Models 5–6.

As discussed in Section 4.3.2, a significant fraction of the options have exercise prices that increase over time. The IFRS 2 does not mention that EOL should be adjusted in such cases, but everything being equal, an increasing exercise price should lead to earlier exercise and, hence, a shorter EOL should be used. We have examined this issue in the regressions by including a dummy, which turns up negative but insignificant for programs with increasing exercise prices.

We have examined the robustness of this result by using several other measures for the magnitude of the OBC. However, there is no evidence that larger programs lead to a shorter expected option life.

While this is not explicitly mentioned, the IFRS 2 does state that, ‘Other factors that knowledgeable, willing market participants would consider in setting the price shall also be taken into account …’ (Section B7, IFRS 2). We consider time varying exercise price to be just such a ‘factor’.