and choices made in the research design. It is important to take these limitations into account when interpreting the results of this study. One of the limitations in this research is caused by the sample used. The sample consists of listed US companies that are part of the S&P 500. The S&P 500 contains the largest listed companies in the United States. This means that the results found with this sample do not necessarily hold for companies with different characteristics than these S&P 500 firms. This can be because the firms in this sample are the largest companies in the United States. It is imaginable that these companies are more in the spotlight than other companies. That could be a reason for the management to adopt behavior in order to maintain the reputation of a company. The situation could however be quite different for smaller firms which are less in the spotlight and therefore it might be possible that other results are found when testing a different kind of sample.
Another possible limitation one has to consider is the sample period. The period from 1999 to 2009 has been quite eventful from an accounting point of view. First there were the accounting scandals of the early 2000’s followed by a financial crisis several years later. This all led to quite some changes in financial reporting and possibly in the behavior of executives as well, as financial reporting was suddenly under public scrutiny. It could be that when these tests are performed for a different sample of firm years one may find different results because each period has its own characteristics that could influence the behavior of CEO’s and CFO’s.
A limitation of the model is that the incentive ratio used in this study splits executive remuneration in options, shares, salary and bonus. It does take into account the difference between vested and not vested options but it does not take the details of all the different remuneration contracts into account. This is a limitation to the model as the behavior of executives is influenced by their specific remuneration; these remuneration plans could contain certain aspects that are not taken into account by this model. Examples of that are possible restrictions on shares awarded to management, or specific targets that have to be met to receive a bonus.
Another possible limitation that has to be considered is the use of the “execucomp” database. Until 2005 the “execucomp” database only reports the five highest paid managers, these five do not necessarily include the CFO, this could lead to a possible sample selection, as companies where the CFO is not among the five highest paid executives might have a CFO who is less influential than companies were the CFO is among the five highest paid executives. These possibly less influential CFO’s are not included in my sample.
Appendix 2 shows the average value of shares and the average value of options held by executives. These values are rather high, certainly when compared to the mean values of these observations. These high average values are possible distorted because of a number of executives that have extremely high amounts of shares and options, some having multiple billion dollars’ worth of shares and options of their companies. These observations lead to the enormous difference between the average and the median of these values. These observations are however excluded from my actual tests, as the incentive ratios are winsorized on first and 99th percentile. These high amounts of
shares and options are normally caused by executives that were the founders of those companies and therefore have large amounts of shares in their firms. Another limitation of the study is a possible bias created by omitted variables. I use a number of control variables for firm age, leverage, sales volatility, firm size and for firm year effects. It is possible that other variables might have an influence as well. Jiang et al. (2010) for example use more control variables, like firm corporate governance, industry dummies and the volatility of cash flows.
Finally when interpreting the results of this master’s thesis one should consider that these results were found with this particular research design on this particular sample. Results could differ when one uses another research model or another sample. Therefore one should be careful generalizing this model. When the research design was made choices had to be made about what models would be used. These choices could influence the outcome of the study. For instance a linear version of the Kothari model has been used to estimate discretionary accruals. This model estimates discretionary accruals and is widely used, that however does not imply that this estimation of discretionary accruals is perfect. It could be that results are different in another model. It is therefore important to be careful when generalizing the results of this study. Taking
into account these limitations the results of this study do provide evidence for a positive relation between discretionary accruals and equity incentives for a sample of large, listed US firms. These results are supported by findings in prior literature.