509a8e7c2fedc8a316561405d38243d0.ppt

- Количество слайдов: 20

CEO Overconfidence and Corporate Investment Malmendier and Tate JF(2005) 1

I. Introduction The investment distortions and the sensitivity of corporate investment to cash flows. 1. Traditional explanations Ø Agency problems(Jensen and Meckling (1976) and Jensen(1986)): managers over-invest to reap private benefits. An influx of cash flow enables managers to invest more and increases investment distortions. Ø Asymmetric information (Myers and Majluf(1984): managers restrict external financing in order to avoid diluting the shares of their company. Cash flow increase investment, but reduce the distortion. n 2

I. Introduction 2. 本文提出的解釋 Ø Building on Roll (1986) and Heaton(2002), this paper argues that one important link between investment levels and cash flow is the tension between the belief of CEO and the market about the value of the firm. Ø Overconfident CEO : over-invest if having sufficient internal cash flows : under-invest if having insufficient internal funds 3

I. Introduction Measures of overconfidence 基本想法 : the overexposure of typical CEOs to the idiosyncratic risk of their firms. Risk averse CEOs should exercise their granted stock options early given a sufficiently high price. 1. Identify a benchmark for the minimum percentage inthe-money at which CEOs should exercise their options for a given years following the vesting period. 2. Hold options all the way to expiration or habitually increase holdings of company stock. n 4

I. Introduction n 作者 apply these overconfidence measures to a panel data set (options and stock holdings of CEOs of 477 firms between 1980 to 1994) 1. CEOs who excessively hold company stock options do not earn significant abnormal returns over the S&P 500 on average. 2. Investment-cash flow sensitivity is significant higher for “late exercisers” or “stock purchasers” than for their peers. Besides, this sensitivity is strongest for CEOs of equity-dependent firms. 5

I. Introduction n Policy implications The traditional provisions ( timely disclosure of corporate accounts or high-powered incentives) may not suffice to address managerial discretion. Refined corporate governance structures, involving a more active board of directors or constraints on the use of internal funds, may be necessary to achieve first-best investment levels. 6

II. The Model basics: Consider a firm with existing assets A and s shares outstanding. At t=1, cash flow C is realized. The CEO chooses the level of investment I that generates expected return R(I). CEO is overconfident such that he overestimates future returns by percentage Δ. 1. To finance I, the amount of cash financing is c, the amount of debt financing is d, and s’ is the number of new shares. The maximization problem of the CEO is eq 1~4. Let (I*, c*, d*) be the solution to the CEO’s maximization problem. IFB is the first-best level of investment and is the level of investment that satisfies. n 7

II. The Model n Model results and implications 1. Lemma 1: rational CEO invests at the first-best level. The overconfident CEO over-invests. Perceived financing costs mitigate over-investment if the CEO is cash constrained. 2. Proposition 1: 若 CEO為 rational或公司可運用資 金 C+D 超過 , 則 CEO所選擇的投資金額 I 與 公司的現金流量 C無關 ; 反之若 CEO為過度自 信 , 且公司可運用資金少而必須發行新股時 , 則投資金額隨現金流量而增加. 8

II. The Model Ø Predictions: First, the investment of overconfident CEOs is more sensitive to cash flow than the investment of rational CEOs. Second, the investment-cash flow sensitivity of overconfident CEOs is more pronounced in equity dependent firms. 9

III. Data and Measures A sample of 477 large publicly traded US. firms from years 1980 to 1994(必須在該段樣本期間 至少出現過 4次 ). 1. Stock ownership and set of option packages (exercise price, remaining duration and the number of underlying shares) for the CEO of each company, year-by-year. 2. Various corporate accounts from the COMPUSTAT database. Ø 表 1為公司與 CEO資料的敘述統計量. n 10

III. Data and Measures n Overconfidence measures: Overconfident managers expose themselves to too high idiosyncratic risk of their firms (postpone option exercise or buy additional company stock). 1. Measure 1: Holder 67. Consider the status of each individual option package in the sample at the end of vesting period. 以五年為準 (此為樣本內所有 option package至少有部份可執行的凍結期期限 ), 計算每一個 package 價內的程度 , 並以 67%為 門檻 (Hall and Muphy(2002))(67% in-the-money during the fifth year as the threshold). 若在第五年 中有達到此門檻 , 則 CEOs應至少執行其部份之 11

III. Data and Measures Holder 67 is constructed as follows: Ø 先決定子樣本 : CEOs who at lease twice had options that were valued above threshold during the fifth year. Ø 然後 identify the first instance (if any) at which the CEO failed to exercise such an option during or before the fifth year. Ø From this point in time onward, classify the CEO as overconfident if he subsequently exhibits the same behavior at least one more time during his tenure as CEO. (robust check: 50%~150%) 由於子樣本限制之故 , 樣本觀察數由 3, 728減為 1, 058. 12

III. Data and Measures 2. Measure 2 : Longholder. Focus on the expiration date of option packages rather than the end of the vesting period. A CEO is classified as overconfident (for all of his year in the sample) if he ever holds an option until the last year of its duration. Ø Over 85% options that are held until their final year are in-the-money and the median is 253%. Thus this measure alleviates the dependence on calibrated threshold for rational exercise. In addition, it can circumvent the sample restriction of the HOLDER 67 measure. 13

III. Data and Measures 3. Measure 3: Net Buyer. Exploit the tendency of some CEOs to purchase additional company stock despite their already high exposure to company risk. Ø 首先決定子樣本 : CEOs who keep their positions as CEO for at least 10 of the 15 years in the sample. Ø Identify CEOs as overconfident if they were net buyers of company equity during their first five years in the sample. 14

IV. Test 1: Overconfidence and Investment Empirical specification: test prediction 1 that the sensitivity of investment to cash flows increases in overconfidence, 迴歸式為 eq 5. 1. 其中 , Qit-1 為 M/B, Cit為 cash flows, Xit 為控制變 數 , Δit為過度自信測度 , 並考慮現金流量 與其它解釋變數的交叉項. The null hypothesis is that β 8 is equal to zero. 2. 迴歸結果在表 5, 考慮了多種迴歸方式 , 皆得 到符合 prediction 1的一致性結論 : β 8顯著為正. n 15

IV. Test 1: Overconfidence and Investment Three baseline regression (to demonstrate the effects of Q and cash flow on investment): these regressions confirm the stylized facts of the investment-cash flow sensitivity literature: cash flow has a large amount of explanatory power beyond Q for investment. Ø Next the authors estimate eq 5 using the holder 67 as a proxy for Δ. 3. To further distinguish the overconfidence effect on investment decisions from inside trading, the authors split Holder 67 into late exercisers who lose money on Ø 16

IV. Test 1: Overconfidence and Investment at least one of the options they hold beyond the threshold and late exercisers who always profit(若 holder 67測度中隱藏有內部訊息的意涵 , 而 之前迴歸所觀察到過度自信造成投資之現金 流量敏感度提高的現象是因為內部訊息的話 , 那麼此現象便在 Hold and Lose 67變數上便 不會產生. Ø 表 6顯示 : the estimated coefficient of Hold and Lose 67 interacted with cash flow is positive, significant and similar to the coefficient on Holder 67 in table 5. 17

IV. Test 1: Overconfidence and Investment 4. Table VII (VIII) gives the results of estimating eq. 5 using the Longholder (Net Buyer) variable as the proxy for Δ. 5. Overall, overconfidence increases the sensitivity of investment to cash flow under any measure. 18

V. Test 2: Overconfidence and Financial Constraints Predictions 2: overconfidence should matter most for firms that are equity-dependent (equity-dependent firms必須發行新股融資 , 此時經理人過度自信 將造成投資對現金流量的敏感度更高 ). 1. Construct the Kaplan-Zingales index of financial constraint. 即根據 eq 6由公司的財務資料來量化 融資限制的程度. 2. Then separate the sample into quintiles based on the lagged value of the Kaplan-Zingales index and estimate eq 5 on each quintile. n 19

V. Test 2: Overconfidence and Financial Constraints 表 8(以 Longholder作為 Δ的 proxy, 因其樣本數較 多 , 否則分組後的樣本數可能過少 )顯示 : the effect of overconfidence on the sensitivity of investment to cash flows is significant only for the top quintile of the KZ index (0. 4990, t=3. 52). 3. Apply several other measures of equity dependencw as substitute for the KZ index. These include firm age, firm size, dividend payment and S&P long-term debt ratings. Ø 20