• No results found

CEO optimism & Insider trading : do more optimistic CEO engage in more insider trading and how do they react after an EPS announcement?

N/A
N/A
Protected

Academic year: 2021

Share "CEO optimism & Insider trading : do more optimistic CEO engage in more insider trading and how do they react after an EPS announcement?"

Copied!
23
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

CEO optimism & Insider trading: do more optimistic CEO

engage in more insider trading and how do they react after an

EPS announcement?

Pim van Diemen 10003243 MSc Finance

Abstract

This thesis shows that optimistic managers engage in more insider trading compared to their peers. After an EPS announcement they react (negatively) surprised by the returns because of their optimistic beliefs and they tend to rebalance their portfolio. CEOs with outstanding far in the money stock options react different to EPS announcement compared to other optimistic CEOs. Using two measures to determine optimism and using both continuous and discrete methods to determine insider trading on the US S&P 500 index firms, I find empirical support for this prediction. Investors and shareholders should be aware of these

value-destroying actions so they can control the management more. These results are an interesting complementation of previous studies because they create a bigger picture of how optimistic managers act and react to certain situations.

(2)

Table of Contents

1. Introduction………....3

2. Hypothesis and Data………..……….……...6

2.1 Hypothesis………....6 2.2 Data………..………...6 2.2.1 LongHolder………...6 2.2.2 HighForecast………....7 2.2.3 Insider Trading………...8 2.3 Control Variables………..………...9 3. Method………..………...10

4. The Empirical Analysis and Results………..….11

4.1 Summary Statistics………....…….11

4.2 Regression Analysis………...14

4.2.1 LongHolder & HighForecast………...14

4.2.2 Optimistic Dummy………...15

4.2.3 Bad forecast Dummy………..17

5. Conclusion……….19

6. Appendix………...20

7. Bibliography……….22

Statement of originality:

Hierbij verklaar ik, Pim van Diemen, dat ik deze scriptie zelf geschreven heb en dat ik de volledige verantwoordelijkheid op me neem voor de inhoud ervan.

Ik bevestig dat de tekst en het werk dat in deze scriptie gepresenteerd wordt origineel is en dat ik geen gebruik heb gemaakt van andere bronnen dan die welke in de tekst en in de referenties worden genoemd.

De Faculteit Economie en Bedrijfskunde is alleen verantwoordelijk voor de begeleiding tot het inleveren van de scriptie, niet voor de inhoud.

(3)

1. Introduction

Stock options are the largest component of an executive’s compensation, therefore it makes up for a significant part of the main income for executives (Hall and Murphy, 2003). However, executives buy and sell shares of their firm on the open-market and this is also a significant part of their compensation (Roulstone, 2003). Executives have knowledge about the firm that outsiders don’t know. But trading on this information is illegal. Over the past decades, the stock returns of executives are significantly higher than the market (Lakonishok and Lee, 2001). This suggests that insiders use private information to strategically time their insider transactions to maximize their own wealth. There are a few different stock price influencing events that managers know of before the market does on which they could trade, these are merger announcements, share repurchase programs, Seasoned Equity Offerings and earnings announcements.

The stock prices of firms that are takeover targets increase up to approximately thirty percent (Agrawal and Jaffe, (1995). Because of these high returns after a merger

announcement, there should be an incentive for insiders to increase their insider purchases and to at least delay their insider sales prior to the announcement. Keown and Pinkerton (1981) and Jain and Sunderman (2014) have studied insider trading activity around merger announcements. The research of Jain and Sunderman (2014) is more recent, but it is limited to the Indian market which is an emerging market. That means that there is weaker corporate governance and control which can lead to a variety of market failures. Their methodology uses the same measure for abnormal returns as the research of Keown and Pinkerton (1981). As for the methodology is concerned, there is an alternative measure for insider trading: the “Abnormal Amihud” (Diether et all., 2009; Anderson et all., 2012). Other researches on insider trading and merger announcements are Jaryaraman et all., (2001) and Kedia & Zhou (2014). These studies look into the options market and how they react around merger announcements.

Other corporate events that influences the share price of the company are share repurchase program announcements and seasoned equity offerings (SEOs). A share

repurchase program is a positive signal to the market and a SEO is a negative signal. An SEO is done when a firm is in need of cash because it is in distress or it is in need of cash to finance a corporate vision. A firm would probably engage in an SEO when the firm thinks that his share price is overvalued so it has to offer less shares to get the required funds. Karpoff and Lee (1991) find significant insider selling prior to SEOs, which suggest that

(4)

insiders are exploiting their private information about an upcoming SEO. A study by

Gombola et al. (1997) find that insiders sell more stock in the months after the announcement of a SEO. This is also a strategically timed transaction, but it is not illegal as the information is not private anymore. A share repurchase program is a positive signal to the market. When a stock repurchase program is announced, stock prices of the announcing firm increase

(Comment & Jarrel, 1991). This creates an incentive for insiders to buy stock right before the announcement.

An other stock price increasing/decreasing event is earnings announcements. Ke et all (2003) show that insiders are selling stock before a negative earnings announcement up to two years’ prior of the event. Stock sales by insiders increase three to nine quarters prior to a break in a string of consecutive increases of quarterly earnings. This paper suggests that insiders anticipate bad news and that they use their private knowledge to benefit from the situation.

CEO optimism and overconfidence have been shown theoretically and empirically to explain important corporate decisions, including investment, financing, dividends and mergers. Optimism above or below the interior optimum face a higher probability of forced turnover (Campbell et all, 2011). Their paper argues that there is a “value-maximizing level” of optimism. Optimism below (above) the value-maximizing level of optimism leads the CEO to underinvest (overinvest), which could lead to higher forced turnover of the CEO. However, little is known about the insider trading activities of overconfident and optimistic managers. The results from the paper of Malmendier and Tate (2005, 2008) show that you can make a distinction between “normal” and “optimistic” managers. These optimistic managers tend to overestimate the future returns of their company. They show that optimistic CEOs display higher investment-cash flow sensitivities, are more acquisitive and are less likely to rely on equity financing relative to their peers. An interesting question to ask is if these optimistic managers are also engaging in more optimistically insider trading. There has been written a lot about insider trading and its existence. It could be contributing to the literature if I can establish a link between insider optimism and insider trading. If I can show that more optimistic managers engage in more insider trading than their peers, investors can take into account that firms with optimistic managers are doing more shareholder-value destroying actions.

My goal is to link insider trading with CEO optimism. I’m using earnings

(5)

you can have more observations per firm, hugely increasing my dataset compared to merger announcements or SEO’s as these are far less common.

I assess CEO optimism with two different measures. The first measure is based on CEO choices regarding his own options exercise and it follows the rationale of Malmendier and Tate (2005, 2008). The idea is that CEO’s that are optimistic about the future will keep their stock options until they are “late” (close to expiration) while their options are far in the money. Options that are kept late while they are far in the money are taken as an indication for optimistic beliefs about the company’s prospects. The fraction of late option exercise is used as a proxy for CEO optimism. The second measure is based on the earnings forecasts that are voluntarily released by each firm. If a forecast released by the firm exceeds the ex post earnings it is classified as “optimistic”. The fraction of “optimistic” forecasts is used as another proxy for CEO optimism. The idea behind this is that optimistic CEOs should be more likely to overestimate the earnings of the company and should thus be more likely than peers to release higher earnings forecasts.

An CEO may want to rebalance his portfolio after a negative earnings announcement: if a CEO is optimistic and does a lot of insider trading, he will try to reverse or compensate his portfolio when his expectations didn’t match the outcome. In this thesis I’m trying to find out if a CEO tries to rebalance his portfolio after a negative earnings announcement.

With this research I want to contribute to the existing literature about insider trading and stock pricing. Insider trading destroys shareholder value, so investors should take CEO optimism into when constructing their portfolio. This thesis should provide insights in how optimistic CEOs are acting and reacting around announcements and how they manage their personal portfolio. This is an important topic of research, because the personal portfolio of a CEO is a large component of its main income (Hall and Murphy, 2003). The results of this thesis show that optimistic managers engage in more insider trading compared to their peers and they rebalance their portfolio after the announcement. Really interesting results given that more optimistic CEOs also experience a higher turnover (Campbell et all. 2011) and receive a lower total compensation compared to their peers. Optimistic managers are being “punished” for being optimistic and this thesis shows that optimistic behavior should be punished because being optimistic leads to more insider trading. Which is a value-destroying action.

(6)

2. Hypothesis and Data

2.1. Hypotheses

In this section I will go over the predicted effect of CEO optimism on insider trading. The intuition is as follows: An optimistic CEO believes that good outcomes are going to be more likely than they really are and thus he overestimates the probability of a successful outcome and the firm’s future prospects. When he engages in insider trading based on his beliefs I predict that he will be trading more aggressive to negative earnings announcements because his expectation exceeds the actual outcome and he gets surprised by the actual outcome of the announcement. I predict that optimistic CEO’s engage in more insider trading post-announcement. They are want to create a portfolio that perfectly suits their beliefs, so when the earnings are not as high as suspected, they have to rebalance their portfolio.

2.2. Data

As for the data on the measure of CEO optimism I use the approach of Otto (2014). The first approach uses the exercise of stock options of CEOs to determine their optimism. The measure identifies a CEO as optimistic if he holds on to his own stock options longer than a CEO with unbiased beliefs. The intuition behind this approach is that a risk-averse CEO is expected to reduce his exposure to company-specific risk by exercising his stock options early if they are sufficiently deep in the money (Hall and Murphy, 2002; Huddart and Lang, 1996). So holding options that are late in the options life and that are far in the money is considered evidence for optimistic beliefs about the company of the CEO.

The second measure for optimistic CEO beliefs is based on the forecasts of the earnings per share (EPS) and the actual EPS. Optimistic CEO’s tend to overestimate their future performance and so they should release forecasts that exceed the actual performance of the company relative to their peers. There is a problem though, data on management forecasts is very limited to UvA students, so it is impossible to construct a solid dataset on

management EPS forecasts. However, Wong & Zang (2009) have found a relationship between optimistic management forecasts and analyst forecasts. They argue that higher CEO optimism results in a higher analyst forecast error. CEOs exert influence on analysts’

forecasts via their communication with these analysts (Wong & Zang, 2009). Their result is that a higher CEO optimism results in a higher analyst forecast. Analysts go along with optimistic CEOs on basis of their cost/benefit analysis. Even when they know that a CEO is optimistic, they tend to release a biased forecast because the costs are small and they are at risk of losing access to the management, which is their biggest source of analyst information.

(7)

It is possible to construct a dataset on analyst forecasts and with the relationship that is established by Wong & Zang it is possible to link this to management EPS forecast and CEO optimism. CEOs do not publish or announce EPS forecasts themselves. But it is unlikely that they allow a statement to be made when they strongly disagree (Otto, 2014). For both

measures I use firms listed in the S&P 500 for at least 3 consecutive years. This gives me 612 firms to work with.

2.2.1 LongHolder

The data on the exercise of options by CEO’s comes from the Thomson Reuters insider filings database. Corporate insiders have to fill out forms and submit them to this database. Corporate insiders, which are individuals who have access to nonpublic insider information are required to file Forms 3, 4, and 5 for transactions involving their companies' stock . Form 4 indicates changes in an insider’s ownership position. So I focus on this form. A change in ownership position could be a purchase, sale, exercise or any other transaction that causes a change in ownership. I start with all the Form 4 observations between January 2005 and December 2014 for all the S&P500 firms that have been listed for at least three consecutive years and I focus only on the exercise of stock options. In only keep observations with the cleanse indicators R, H, C, L, or I, because these observations are the most accurate. I also discard observations when the following items are missing: the person ID that

identifies each CEO, the transaction date, the exercise date and the expiration date of the options. The time to expiration is calculated as follows: the difference between the expiration date of the options and the transaction date. The transaction date is the date of exercise. I merge this database with stock data from the CRSP database so I can calculate the

“moneyness” of the option on the day of exercise as the difference between the closing price and the exercise price divided by the exercise price.

For each observation I make an optimism dummy that takes the value of 1 when the option is exercised within two years of its expiration date (time to expiration <730) and at least 20% in the money (value of moneyness > .20). Otherwise, it takes the value of zero. Finally, I

average the value of the optimism dummy for each CEO across all observations within a given firm. This dataset contains data on 289 firms. The mean of the LongHolder variable is .193 with a standard deviation of .0006

(8)

2.2.2 HighForecast

The second measure of CEO optimism is based on the comparison between analyst EPS forecasts and realized EPS by the firm at the end of a fiscal period. Almost all of the companies in the S&P 500 index provide analyst EPS forecasts during the sample period.

I started my dataset with data from the Thomson Reuters Institutional Brokers

Estimate System (I/B/E/S) Database. Downloading all the analyst EPS forecasts for S&P 500 firms between January 2005 and December 2014. I only keep forecasts for the common stock of each firm and drop observations if the announcement date falls after the end of the fiscal period for which the announcement was made, or after the date on which the actual EPS were announced. Furthermore, I drop observations if information on the EPS that were eventually realized is not available. I assign a dummy to each forecast and realization pair. When the forecast exceeds the realization, the dummy will get a value of one. When the forecast is the same or lower than the realization, the dummy will have a value of zero. So, the dummy indicates whether a forecast was optimistic or not. I calculate the equally weighted average of the observations across all years. This is to measure a CEO's average optimism across all years. The mean of the HighForecast variable is 0.3888 and the standard deviation is 0.0004

2.2.3. Insider trading

I use the same firms in my event study as I have used for the optimism measures. Intuitively, if the firm i’s earnings announcement contains stakeholder-relevant news, firm i employees who have access to relevant information can trade on or share the information with outsiders, resulting in abnormal trading activity in firm i prior to firm i’s earnings announcement. My event study starts 20 days prior to the announcement date (t) and it ends 20 days after the announcement date (-20, t, +20). I will compare the 20 days prior to the announcement date with the 20 days after the announcement date (-20, t) (t, +20). A CEO has his expectations about the earnings announcements that are coming, and when these

expectations were optimistic, he has to reverse his portfolio to get the position that he wants. This risk shifting is an interesting subject to research. The abnormal returns found in the 20 days prior to the announcement date are a proxy for insider trading and I will compare this with the abnormal returns found in 20 days after the announcement to see if there are some significant changes in abnormal returns. The event window is cut into two parts, which gives me three windows to work with: a pre-event window (-20,t) (t,0), a post-event window (0,t) (t,+20) and both (-20,t) (t,+20).

(9)

I measure insider trading with the standard cumulative abnormal return (CAR) model. Earnings announcements can be found in the Thomson Reuters Institutional Brokers Estimate System (I/B/E/S) Database. Daily stock data can be found on CRSP and the abnormal returns are calculated by the following market model regression:

!"# = &" + ("!)# + *"#

With Rit being the return on the stock of firm i on day t and (Rmt) being the return of the NYSE. εit is the value of the abnormal return for firm i on day t and is in this regression the variable of interest. εit is estimated with pre-event return data, starting 30 days prior to the event. Portfolio cumulative abnormal returns between event days t1 and t2 (CAR(t1,t2)) are calculated by summing daily abnormal returns.

The significance of the portfolio cumulative abnormal returns is assessed with a t-statistic:

+ ,-! = ,-!. +/, +1 ) ∗ 3.

Where m is the number of days of the event period and Sp is the standard deviation of the estimated daily abnormal returns. The mean of the t statistic is .0369 with a standard deviation of .0037.

2.3 Control Variables

Control variables for the firms include Firm Size (log of total assets) as larger firms are likely to have greater liquidity and transparency, reducing the ability to profit from informed trading. Leverage (long-term liabilities divided by total assets), ROA (income before extraordinary items divided by total assets), Market-to-Book (market value of equity divided by book value of equity), are also included. The bid-ask spread (daily bid-ask spread averaged over the previous year) and Trading volume (log of the average daily trading volume of the previous year) are also included. These variables have influence on the liquidity of stocks which can be used to facilitate insider trading. More liquid stocks are easier to sell on the open market, so it would be less noticeable for insider traders to buy/sell these shares.

(10)

3. Method.

In order to determine whether a CEO is optimistic or not we have the variables LongHolder and HighForecast. Previous studies have shown (Malmendier & Tate, 2008 & Otto 2014, Wong & Zhang 2009) that both these variables are good proxies for CEO optimism. In my research, these two variables are key in assessing CEO optimism and I am going to use them individually and complementary to research the influence of CEO optimism on insider trading. The first set of regressions will have both variables of interest (LongHolder & HighForecast), regressed on the cumulative abnormal returns with the control variables, so the regressions will look like this:

,-! = & + (/4567859:;< + (18=7ℎ?5<;@AB+ + (C,56+<59BC + * (1)

With LongHolder and HighForecast being the variables of interest as they show what their individual influence is on the abnormal returns on the underlying stock prices.

To robust my results, I want to combine both LongHolder and HighForecast to assess CEO optimism. It has been discussed that some external factors can influence both variables, but that these variables are nonetheless good proxies for CEO optimism. To minimize the influence of external factors on CEO optimism I am creating a dummy variable for each firm. A firm is considered as optimistic when both LongHolder and Highforecast are in the highest tercentile/quantile of the sample. When this is the case the number 1 is assigned to the

dummy variable optimistic. Otherwise it is zero. The idea behind this approach is that the firm that has 1 assigned for both LongHolder and HighForecast (the highest number possible) is considered the most optimistic firm of all the firms in the sample. So I’m basically

comparing the 33% / 25% most optimistic companies with the other 67% / 75% of the sample. These regressions will be more discrete. But this way it is easier to compare firms within the sample. Influence of external factors in one of the key variables HighForecast and LongHolder is a lot less existent now. It is not possible to just add both HighForecast and LongHolder and use the highest tercentile as the mean of HighForecast is much higher than the mean of LongHolder which will skew the optimistic dummy towards firms with a high value of HighForecast. The regressions with the optimistic dummy will look like this:

(11)

It is also important to measure how CEO’s react when the firm is performing worse than expected. Do optimistic CEO’s rebalance their portfolio when they find out that their expectations were optimistic?

I have created a dummy variable that has the value of 1 when the value of the analyst EPS forecast exceeds the actual EPS realized and has zero otherwise. By doing this I am creating a sample that only has “optimistic” CEO forecasts and therefore I can compare the results to the “non-optimistic” sample, when the value of the dummy variable is zero. In my dataset, 2519 events are “optimistic” and 3343 events “non-optimistic”.

To see how optimistic CEOs behave after an unexpected negative forecast I have created the dummy variable unexpected which is 1 when the lagged value of the predicted EPS is smaller than the actual value & the lagged value of the actual EPS is greater than the actual value of the EPS. The idea behind this is that I want to see what happens when an optimistic manager predicts an increase in EPS but it turns out to be a decrease in EPS. The results from this dummy can be interesting when researching how optimistic managers react when they are surprised by a negative event that decreases the value of their stock and/or options. Do they engage in more insider trading to rebalance their portfolio? If so, do they adjust their portfolio after the announcement or before? I’m hoping to find an answer to these questions

4. Empirical Analysis and Results

4.1. Summary Statistics

Table 4.1 presents the summary statistics for the final sample of 181.153

observations. The average cumulative abnormal return is close to zero with a high standard deviation. The HighForecast variable has a higher mean and median, and a lower standard deviation compared to the LongHolder variable. Furthermore, more info can be found on the control variables. Not every firm in the sample has granted options to their CEO and firms are not obliged to announce their earnings forecast, in Table 4.2 you can see that there is a difference in number of firms in the sample for both LongHolder and HighForecast. The average number of announcements per firm is 9 and the minimum number of announcements is 3. There are 4541 events in the sample, and a total number of 514 firms. The optimistic dummy divides the two (three) lower tertiles (quartiles) from the highest quartile, which is called the optimistic group in my method. The number of announcements in the highest

(12)

tertile (quartile) is 1514 (1135), spread over 171 (128) firms. In Table 4.2 you can see that the cumulative abnormal return post-announcement is always smaller than pre-announcement. Table 4.3 shows the mean and standard deviation of the different groups that are created by using tertiles and quartiles of the sample. The table shows that the number of optimistic managers drastically decreases when comparing the last tertile with the last quartile because the conditions to be addressed as optimistic are higher. The number of observations in the last segment goes from 8020 to 2858.

Table 4.1: Descriptive statistics

n Mean Median Std. Dev.

Lower Quantile

Upper Quantile

Cumulative Abnormal Return 181153 -0,0008 0 0,1652 -0,1 0,1

HighForecast 160273 8 0,4 0,1615 0,27 0,53 LongHolder 136043 0,192 0,08 0,2503 0 0,31 Return on Assets 160185 0,0593 0,05 0,0603 0,02 0,09 Leverage 160185 0,2077 0,18 0,1554 0,09 0,3 Bid-Ask Spread 181153 0,056 0,01 0,09166 0,01 0,04 Book/Market Ratio 181153 2985,873 2364,01 2079,426 1474,2 3877,33 Trading Volume 181153 3461873 1865600 4054554 815100 4259450 Firm Size 160185 9,26 9,163 1,4592 8,209 10,172

(13)

Table 4.3: Optimism dummy

Lower Tertile Middle Tertile Upper Tertile First Quartile Second Quartile Third Quartile Fourth Quartile

LongHolder 0 0.1148 0.5109 0 0.04686 0.1968 0.5690 (Std. Dev) (0) (0.0057) (0.00237) (0) (0.0036) (0.00070) (0.0032) HighForecast 0.2239 0.4180 0.5842 0.1813 0.36680 0.4798 0.6027 (Std. Dev) (0.0007) (0.0037) (0.0085) (0.0007) (0.0006) (0.0003) (0.007) #N of observations 16971 11601 8020 11894 3320 7642 2858

Note: Table 4.3 shows the mean of the three (four) groups that have been created by creating tertiles (quartiles) of both LongHolder and HighForecast. The upper tertile and the Fourth quartile are the variables of interest, these represent the most optimistic groups of CEOs

Table 4.2: HighForecast & LongHolder

Non-optimistic CEOs Optimistic CEOs Non-optimistic CEOs Optimistic CEOs LongHolder = 0 LongHolder > 0 HighForecast = 0 HighForecast > 0

#N of firms Mean #N of firms Mean #N of firms Mean #N of firms Mean

CAR 283 -.00450376 231 .00064341 44 -.02639 657 -.0006145

CAR Pre-announcement 283 -.00213323 231 .00283416 44 -.01112 657 .00151

(14)

4.2. Regression Analyses

4.2.1. LongHolder & HighForecast

This section presents the results of regression analyses concerning the effect of the LongHolder and the HighForecast variables on the cumulative abnormal returns. Table 4.3 shows the basic regression output. I have created three sample periods: 20 days’ pre-announcement, 20 days’ post announcement and all 40 days. Both LongHolder and

HighForecast are positive and significant different from zero at the 95% confidence interval for the whole sample (t-20, t, t+20) and the pre-announcement period (t-20, t). The post-announcement regression shows that the LongHolder variable is no longer positive, but significantly negative. This means that managers who own options that are close to expiration in the firm have a negative effect on the abnormal returns. This is an interesting result,

because normally you will find a positive or non-significant result. You can conclude from this result that LongHolder CEOs follow the market after the announcement and

HighForecast CEOs still engage in insider trading. It looks like LongHolder CEOs are adjusting their portfolio after an announcement. The performance of their stock options is lower than what they have expected so these CEOs are shifting their portfolio towards their new beliefs. Their actions have a negative effect on the abnormal returns. Which could mean that they sell the stock of their firm after the announcement to rebalance the portfolio that they want. LongHolder CEOs have options that are far in the money and close to expiration. When they believe that the company is performing worse as they thought before they can exercise their options and sell them on the market. HighForecast CEOs may not have vested stock or stock options, so even when they want to sell options because of their new beliefs, they may not be able to because they don’t own them yet. This can explain the difference between the LongHolder and the HighForecast variable in the post-announcement period.

The control variables are all significant at the 1% level, except for the variable Volume in the post-announcement period which is not significant and Return on Assets, which is significant at the 5% level.

(15)

Table 4.3: HighForecast and LongHolder

This table presents the regression results regarding the effect of CEO optimism on the cumulative abnormal return (insider trading). The sample period is from 2000 to 2015. LongHolder is the optimism measure based on the CEO's option exercise behavior. HighForecast is the optimism measure based on the EPS forecast behavior. Leverage is long term liabilities divided by total assets. Volume is the logarithm of the average daily trading volume of the previous year. Firm size is the logarithm of the total assets. Bid-Ask spread is the difference between the closing Bid and Ask price. Book/Market ratio is the book-value of equity divided by the market value of equity.

(t-20,t,t+20) CAR CAR Pre-announcement (t-20,t) CAR Post-announcement (t,t+20) LongHolder 0.002 0.004 -0.002 (2.38)* (7.95)** (4.20)** HighForecast 0.029 0.014 0.010 (6.00)** (5.23)** (3.85)** Book/Market Ratio 0.000 0.000 0.000 (11.71)** (8.12)** (7.82)** Bid-Ask Spread 0.001 -0.011 0.014 (0.25) (3.77)** (5.10)** Leverage 0.020 0.006 0.012 (7.30)** (3.35)** (7.97)** Return on Assets 0.025 0.036 -0.010 (3.44)** (8.23)** (2.50)* Volume 0.003 0.002 0.000 (9.41)** (8.94)** (0.96) Firm Size 0.004 0.002 0.001 (11.43)** (11.87)** (8.36)** Constant -0.123 -0.067 -0.030 (15.78)** (14.96)** (6.89)** R2 0.0033 0.0035 0.0019 N 159,876 159,876 159,876

Standard errors are in parenthesis. Significance at the 1% and 5% level is denoted by * , **

4.2.2. Optimistic Dummy

Table 4.4 shows the regression results of the optimistic dummies on the abnormal return periods as described in section 4.2.1. Optimistic Dummy 1 is the dummy variable that divides the highest tertile of optimistic firms with the two tertiles of less optimistic firms and Optimistic Dummy 2 is the dummy variable that divides the highest quartile of optimistic firms with the rest. Again, both dummy variables are positive and significant at the 1% level, except for the post-announcement period, where the results are negative at the 1%

(16)

optimistic dummy, but the difference is very small. Post-announcement, the effect of the optimistic dummies on the abnormal returns become significantly negative. As explained in 4.2.1 this could be due to optimistic managers who are adjusting their portfolio based on the earnings announcement. It seems like the (negative) LongHolder effect is greater than the (positive) HighForecast effect, at least for the most optimistic managers, because the

HighForecast effect on the abnormal returns was positive post-announcement (see table 4.3). All the control variables are significant at the 1% level except for Volume in the

post-announcement sample which has no significance and Return on Assets, which is only significant at the 5% level.

Table 4.4: Optimistic dummy regression results

This table presents the regression results regarding the effect of the optimistic dummies on the cumulative abnormal return (insider trading). The sample period is from 2000 to 2015. Optimistic Dummy 1 is the dummy variable that captures the highest tertile of optimistic CEOs. Optimistic Dummy 2 captures the highest Quartile of optimistic CEOs from the same sample. Leverage is long term liabilities divided by total assets. Volume is the logarithm of the average daily trading volume of the previous year. Firm size is the logarithm of the total assets. Return on assets is the income before extraordinary items divided by total assets.

Cumulative Abnormal Return

Cumulative Abnormal Return Pre-announcement Cumulative Abnormal Return Post-announcement Optimistic Dummy 1 (4.91)** 0.005 (10.32)** 0.006 (2.99)** -0.002 Optimistic Dummy 2 0.005 0.007 -0.003 (3.23)** (7.70)** (2.90)** Book/Market Ratio (11.32)** 0.000 (11.36)** 0.000 (7.74)** 0.000 (7.81)** 0.000 (7.63)** 0.000 (7.63)** 0.000 Bid-Ask Spread (0.26) 0.001 (0.35) 0.002 (3.80)** -0.011 (3.61)** -0.011 (5.13)** 0.014 (5.08)** 0.014 Leverage 0.020 0.020 0.005 0.005 0.013 0.013 (7.34)** (7.38)** (3.05)** (3.15)** (8.33)** (8.29)** Return on Assets (3.60)** 0.026 (3.84)** 0.028 (7.98)** 0.035 (8.49)** 0.037 (2.01)* -0.008 (2.18)* -0.009 Volume 0.003 0.003 0.002 0.002 0.000 0.000 (9.28)** (9.38)** (9.22)** (9.41)** (0.47) (0.44) Firm Size 0.004 0.004 0.002 0.002 0.002 0.002 (11.97)** (11.83)** (12.19)** (11.89)** (8.83)** (8.91)** Constant -0.094 -0.094 -0.052 -0.051 -0.020 -0.021 (15.17)** (15.12)** (14.31)** (14.19)** (5.99)** (6.03)** R2 0.00 0.00 0.00 0.00 0.00 0.00 N 159,876 159,876 159,876 159,876 159,876 159,876

(17)

4.2.3 Bad forecast dummy.

In this section the results are showed when a CEO had a “bad forecast”. The HighForecast variable is created by comparing forecasted EPS with the realized EPS. But the forecast is addressed as “bad” when the new forecast predicted a rise in EPS compared to the last actual EPS when the realized EPS was lower than its lagged value (thus an EPS decreased over the year while a rise was predicted). 795 events are addressed as “bad forecasts” in my sample, with a total number of 19.532 observations. Table 4.5 shows the regression results. Both LongHolder and HighForecast are positive and significant for all three sample periods. Both optimistic dummies are positive and significant, except for optimistic dummy 2 in the Pre-announcement sample. With a t-statistic of 0.92 the variable is highly insignificant. The influence of the optimistic measures on the abnormal returns are smaller post-announcement compared to pre-announcement. The interesting difference between regressions on the bad forecast sample and the original sample is that the optimistic dummies and the LongHolder variables are no longer negative in the Post-announcement period. All the variables are significant and positive, which shows a positive relationship between optimism and insider trading after a bad forecast. This seems counterintuitive, but it can be explained by the fact that the market reacts surprised too. Investors rely on forecasts and when firms announce that they do not meet their forecast, the value of the shares drop. In the previous sample it was possible that the actual EPS announcement exceeded the forecast and the market responded positively. But the (optimistic) CEO did respond negatively because he was expecting an even higher EPS. After a bad forecast. This can not be the case anymore and thus explaining the positive relation between the abnormal returns and CEO optimism.

(18)

Table 4.5: Bad Forecast regression results

Cumulative Abnormal Return Cumulative Abnormal Return Pre-announcement

Cumulative Abnormal Return Post-announcement LongHolder 0.053 0.024 0.026 (10.00)** (7.77)** (8.93)** HighForecast 0.069 0.055 0.015 (7.73)** (10.32)** (3.01)** Optimistic Dummy 1 0.051 0.029 0.013 (19.25)** (18.11)** (8.14)** Optimistic Dummy 2 0.028 0.002 0.016 (7.99)** (0.92) (6.34)** Book/Market Ratio 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 (7.71)** (6.82)** (5.51)** (4.94)** (5.09)** (3.68)** (10.04)** (8.70)** (8.15)** Bid-Ask Spread 0.090 0.017 0.018 0.039 -0.005 -0.003 0.061 0.033 0.033 (5.57)** (1.14) (1.22) (4.14)** (0.60) (0.40) (6.75)** (3.98)** (3.97)** Leverage 0.022 0.061 0.062 0.006 0.023 0.023 0.035 0.050 0.050 (2.36)* (7.48)** (7.58)** (1.15) (4.80)** (4.87)** (6.98)** (11.38)** (11.47)** Return on Assets -0.081 -0.005 -0.005 -0.000 0.051 0.052 -0.082 -0.052 -0.053 (3.43)** (0.22) (0.22) (0.01) (4.10)** (4.15)** (6.86)** (4.74)** (4.80)** Volume 0.008 0.007 0.008 0.005 0.004 0.005 -0.000 0.000 0.000 (6.48)** (6.89)** (7.37)** (7.77)** (7.12)** (7.80)** (0.68) (0.64) (0.77) Firm Size 0.008 0.006 0.006 0.004 0.006 0.006 0.003 0.000 -0.000 (7.18)** (6.31)** (6.53)** (6.72)** (10.99)** (11.79)** (4.85)** (0.16) (0.14) Constant -0.241 -0.198 -0.201 -0.147 -0.133 -0.139 -0.046 -0.024 -0.022 (12.08)** (10.96)** (10.98)** (12.50)** (12.58)** (13.00)** (4.47)** (2.57)* (2.35)* R2 0.02 0.02 0.01 0.02 0.03 0.02 0.02 0.02 0.02 N 15,845 19,532 19,532 15,845 19,532 19,532 15,845 19,532 19,532

(19)

5. Conclusion

CEOs having optimistic symptoms are engaging in more pre-announcement insider trading. Post-announcement they try to rebalance their portfolio, as they realize that their portfolio was too optimistic. CEOs who have stock options that are far in the money and close to expiration are exercising their options and they use them to rebalance their new portfolio of stocks. Using discrete measures to determine optimistic CEO insider trading yields the same results as using the continuous measures LongHolder and HighForecast. After a bad forecast, the negative influence of CEO optimism on abnormal returns disappears, which is

counterintuitive, but in fact the manager chooses to follow the market when he readjusts his portfolio. Optimistic managers engage in more aggressive insider trading compared to their peers. This destroys firm value and investors should take this into account when composing their portfolio. However, the effect on stock returns by the insider trading of optimistic CEOs are very little and therefore it is doubtful that it is worth it for an investor to research CEO optimism before making a portfolio. The most important conclusion for a CEO is that he/she should be aware of the fact that being optimistic may not be a good thing. Optimistic CEOs have a higher forced turnover, a lower total compensation and they are engaging in more insider trading compared to their peers.

This research is limited to the US market. It could be interesting to use a more diversified sample. There are more ways to determine insider trading, they could be

implemented for robustness. Are the other proxies for insider trading giving the same results? Another limitation of this thesis is that it does not show if optimistic CEOs benefit from their trading. Are they benefitting more than their peers? Are they outperforming the market? These questions are still unanswered while they are important in the discussion about CEO optimism. Another limitation is that the control variables of the study are mostly limited to market control variables. There are no control variables for CEO characteristics which could influence the results. In further research it is possible to see if LongHolder CEOs decide to exercise their options right after EPS announcements (or earnings announcements) to see if they use their stock to rebalance their portfolio. Data on exercise dates is available on WRDS. Another interesting addition is to research how (optimistic) CEOs react to earnings

announcements when comparing their compensation portfolio. Are CEOs that are having more variable compensation engaging in more insider trading because they are more dependent on variable compensation? Do optimistic managers have other biases regarding their behavior compared to their peers? These questions are still mostly unanswered.

(20)

Appendix

Stata code used for the LongHolder variable. In green, comments to clarify:

use "/Users/Pim/Documents/MSc Thesis/data longholder.dta", clear

*obtained database from Thomson Reuters Insider Filings database

encode ticker, gen (ticker1) encode cleanse, gen (cleanse1) drop ticker

rename ticker1 ticker drop if cleanse1 ==. drop if personid ==. drop if trandate ==. drop if tdate ==. drop if ticker ==. drop if acqdisp == "A" keep if acqdisp == "D"

*A is acquired, D is disposal, so we have to keep “D” only

destring formtype, replace keep if formtype == 4

*Form 4 focuses on insider position

drop if num_deriv ==. drop if xprice==. gen dummy = 0

gen time_expiration = tdate – trandate

*Variable that determines the length between selling date and expiration date

drop if time_expiration < 0 drop if time_expiration > 73000 drop if time_expiration ==.

duplicates drop personid tdate, force

*Dropping duplicates and entry errors (negative time_expiration and time_expiration >200 years)

joinby ticker xdate using "/Users/Pim/Documents/MSc Thesis/stockpricedata.dta", unmatched (master)

*Merge the database with stock price data collected from CRSP

gen dummyd =0

replace dummyd=1 if derivative == "CALL" replace dummyd=1 if derivative =="OPTNS" replace dummyd=1 if derivative == "PUT" replace dummyd=1 if derivative =="EMPO" replace dummyd=1 if derivative =="ISO"

*Limit the dataset to options and option grants

drop if dummyd==0 drop dummy

rename prc price drop if price ==. gen dummy = 0

(21)

gen moneyness = (price-xprice)/xprice

*Difference between closing price and exercise price, divided by exercise price

replace dummy=1 if time_expiration < 730

*The option is exercised 2 years before the expiration date at most

gen dummy2 = 0

replace dummy2 = 1 if moneyness > 0.2

*The options exercised are far in the money

gen dummyoptimism =0

replace dummyoptimism=1 if dummy==1 & dummy2==1

*Optimism dummy =1 if both requirements are met (exercised 2 years before expiration date at most and far in the money)

bysort ticker: egen optimism=mean (dummyoptimism)

*Calculate the mean of the optimistic dummies for each firm

gen dup = trandate drop if dup < 14175

(22)

6. Bibliography

Agrawal, A. and Jaffe, J.F. (1995). Does Section 16b deter insider trading by target managers? Journal of financial economics, Vol. 39, 295-319.

Amihud, Y., 2002. Illiquidity and stock returns: Cross-section and time series effects, Journal of Financial Markets Vol. 5, 31-56.

Anderson, R. C., Reeb, D. M., and Zhao, W., (2012). “Family-controlled firms and insider trading: Evidence from short sales”, Journal of Finance 67, 351-386.

Chen, H.-C., Chen, S.-S., Huang, C.-W. and Schatzberg, J. D. (2014), “Insider Trading and Firm Performance Following Open Market Share Repurchase Announcements.” Journal of Business Finance & Accounting

Comment, R. and Jarrell, G. A. (1991). “The relative signaling power of dutchauction and fixed-price self-tender offers and open-market share repurchases.” Journal of Finance Vol 46, 1243–71.

Cziraki, P., Lyandres, E., and Michaely, R., (2016). “What Do Insiders Know? Evidence from Insider Trading Around Share Repurchases and SEOs.” Available at SSRN:

http://ssrn.com/abstract=2732969

Diether, K. B., Lee, K-H., and Werner, I. M., 2009. “Short-sale strategies and return predictability”, Review of Financial Studies 22, 575-607.

Gombola, M., Lee, H.W. and Liu, F. (1997). “Evidence of selling by managers after seasoned equity offering announcements.” Financial management, Vol. 26, No. 3, 37-53.

Graham, J., Harvey, C., Rajgopal, S., 2005. The economic implications of corporate financial reporting. Journal of Accounting and Economics 40, 3–73.

Hall, B., Murphy, K., 2002. Stock options for undiversified executives. Journal of Accounting and Economics 33, 3–42.

Hall, B., Murphy, K. (2003). The trouble with stock options. Journal of economic perspectives, Vol. 17, Number 3, 49-70.

Huddart, S., Lang, M., 1996. Employee stock option exercises: an empirical analysis. Journal of Accounting and Economics 21, 5–43.

P. Jain., M. Sunderman,, (2014),"Stock price movement around the merger announcements: insider trading or market anticipation?", Managerial Finance, Vol. 40 Iss 8 pp. 821 – 843 Jayaraman, N., Frye, M. B. and Sabherwal, S. (2001), “Informed Trading around Merger Announcements: An Empirical Test Using Transaction Volume and Open Interest in Options Market.” Financial Review, 36: 45–74

Karpoff, J.M. and Lee, D. 1991, “Insider trading before new issue announcements.” Financial Management, Vol. 20, 18-26.

(23)

Keown, A., Pinkerton, J. (1981), “Merger Announcements and Insider Trading Activity: An Empirical Investigation”, The Journal of Finance, Vol. 36, No. 4 (Sep., 1981), pp. 855-869 Kyle, A. S., 1985. Continuous auctions and insider trading, Econometrica Vol. 53, 1315-1335.

Lee, D.S., Mikkelson, W.H. and Partch, M.M. (1992). “Managers’ trading around stock repurchases.” Journal of finance, Vol. 47, 1947-1961.

Malmendier, U., Tate, G., 2005. CEO optimism and corporate investment. Journal of Finance 60, 2661–2700.

Otto, C. (2014), “CEO optimism and incentive compensation.” Journal of Financial

Economics, 114 (2): 336-404

Roulstone, D.T. (2003). The relation between insider-trading restrictions and executive compensation. Journal of accounting research, Vol. 41, 525-551.

Sivakumar, K., & Waymire, G. (1994). Insider Trading Following Material News Events: Evidence from Earnings. Financial Management, 23(1), 23-32.

Wong, M.H., & Zhong, X. (2009). CEO optimism, Analyst Rationality, and Earnings Forecast Bias. Available at SSRN: http://ssrn.com/abstract=814085

Referenties

GERELATEERDE DOCUMENTEN

This paper examines immigrant wage growth taking selective out-migration into account using administrative data from the Netherlands.Addressing a limitation in the previous

Each question was related to one of the frames, for example: “Does the story contain information about the indictment?” (court case), “Does the story provide personal

That’s'where'I'grew'up,'that’s'my'Liverpool.'All'these'places'I’ve'been'dragged'to,'you'know'what'

Als onderdeel van een groter geheel onderzoekt deze studie aan de hand van een 2 (winstframe vs. volslanke endorser) tussen-proefpersonen design bij 146

Finally, the supporters were asked about their perceptions of possible sources of support that families with a child accused of witchcraft received and the possible additional

Instellingen die in het groene gebied liggen hebben een switch die lager is dan de afspraak voor 2015. De instellingen in het 

The data presented herein have shown that the surface topography of calcium phosphate ceramics reg- ulates mesenchymal stromal cells (e.g., human bone marrow mesenchymal stromal

Wanneer er geen interactie tussen de punten zou zijn, zou het verwachte aantal punten in een cirkel om een specifiek punt... rechtevenredig zijn aan de oppervlakte van