Commitment to Overinvest and Price Informativeness James Dow Itay Goldstein Alexander Guembel London Business University of University of Oxford School Pennsylvania European Central Bank, 15-16 May, 2006 1
The Feedback Effect Prices in financial markets affect resource allocation. Justification: Markets aggregate information from many participants and thus provide new information to decision makers (managers, providers of capital). Related Evidence: Ability of markets to generate predictions: Roll (1984), Wolfers and Zitzewitz (2004). Managers learning from prices: Luo (2005), Chen, Goldstein, and Jiang (2006). Feedback effect via capital constraints: Baker, Stein, and Wurgler (2003). Theoretical Implication: Prices affect and reflect firm value at the same time. 2
Theoretical Literature There is a small literature analyzing the theoretical implications of the feedback effect from prices to the real economy. Early Contributions: Emphasize the increase in efficiency: Leland (1992), Khanna, Slezak, and Bradley (1994), Dow and Gorton (1997), and Subrahmanyam and Titman (1999). Recent Contributions: Goldstein and Guembel (2005): Feedback generates manipulation of stock price. Bond, Goldstein, and Prescott (2006): Feedback leads to no equilibrium/multiple equilibria. Bond and Eraslan (2006): Feedback creates a motive for trade. 3
In this Paper: We study the incentives to collect information in the presence of feedback, and identify a limitation inherent in the allocational role. Suppose that previously announced investments are always cancelled when prices suggest they have slightly negative NPVs. The value of the security becomes insensitive to information about investments fundamentals. Speculators don t acquire the information, and the allocational role of prices breaks down. 4
Two Main Implications More information collected during booms than during busts. o This amplifies shocks to fundamentals. o Consistent with analysts activities in the business cycle. Firms may optimally commit to overinvest i.e., invest when, according to information in the price, projects have slightly negative NPVs. o Cancellation fee (break-up fee in M&A). o Leave financial slack in the firm to allow empire-building managers to make interim inefficient investments. 5
The model Dates t {1, 2} Investment Opportunity (Decision at t=2): - Requires investment I; yields uncertain payoff R ω. - ω {l, h} with equal probability; R h > I > R l 1 - Unconditional expected payoff is: R 2 ( R l + R h ) 6
Financial Market (t=1): - One speculator who can learn ω at cost c. - Noise trade is either -n or n with equal probability. - The speculator, if informed, submits orders -n or n. - A market maker observes the set of orders but not the identity of the traders who submit them. He then sets the price equal to the expected firm value contingent on the information contained in the order flow. - Modification of Kyle (1985): price both reflects and affects firm value Feedback Effect. 7
Prices Two buy orders (probability 0.25): market maker infers that the speculator traded on positive news and sets a price p bb. Buy and a sell order (probability 0.5): MM cannot tell whether the speculator bought and the liquidity trader sold, or vice versa, and sets a price p sb. Two sell orders (probability 0.25): MM infers that the speculator traded on negative news and sets a price p ss. 8
Investment Policies Consider investment policies that are monotone in the order flow. Never Invest (NPV = 0) Type I Conditional Investment (Reject unless Certain) Invest when price is p bb and don t invest when price is p sb or p ss. Type II Conditional Investment (Accept unless Certain) Invest when price is p bb or p sb and don t invest when price is p ss. Always Invest (NPV < 0 by assumption) 9
Interim Efficient Investment Policy Investing when p=p bb is always good; Investing when p=p ss is always bad. When ex-ante NPV is negative, I > R, it is interim efficient not to invest when p=p sb. Type I Conditional Investment is interim efficient. When ex-ante NPV is positive, I < R, it is interim efficient to invest when p=p sb. Type II Conditional Investment is interim efficient. 10
Feasibility of Type I Conditional Investment Policy Suppose the firm does not invest except when there are two buy orders. Then, the speculator does not acquire costly information in equilibrium and the firm therefore never invests. Intuition Suppose the speculator acquires information, and then buys on good news and sells on bad news. We will go through each possibility for order flow and price to see if he can make a profit that justifies acquiring the information. 11
Intuition cont d In case of two buy orders or two sell orders, there is no uncertainty about the project. The price fully reveals the information, and the speculator makes no profit. This is a standard feature of traditional market microstructure models. In case of two opposite orders, there is uncertainty about the project, which is the source of profit in traditional market microstructure models. However, the speculator cannot profit on his information in this case, because the firm does not invest, and thus its value is no longer sensitive to the private information acquired by the speculator. 12
Implications If interim efficient rule is followed, information will be collected if and only if the ex ante NPV of the investment is positive. This amplifies shocks to fundamentals: o When NPV is negative, there is no collection of information, and as a result, efficiency of investment decisions is harmed, and amount of investments decreases. This seems consistent with patterns of analysts activities. The firm may want to commit to overinvest. 13
Ex-Ante Efficient Investment Policy Suppose that the investment has ex ante negative NPV: R < I, but has positive NPV if the real option is accounted for: 1 1 ( I ) + ( R I ) > 0 n assume that c ( V V ) h l 4 R h 2. Also, 4 (the cost of producing information is sufficiently low). Then, Type II Conditional Investment is ex-ante most efficient. This policy is interim inefficient, as it entails investment when p=p sb, and thus the NPV is slightly negative. However, it enables production of information, and thus is ex-ante efficient. It requires commitment to overinvestment. 14
Commitment Managerial Learning Suppose that managers interests are aligned with those of shareholders. Also, managers have enough capital to finance the investment. Managers learn from the price before they make the investment decision. Managers can commit by setting a cancellation fee F to be paid out when the investment is cancelled: Break-up fee in mergers and acquisitions. n When c 4 ( R R ) and 1 3 ( I ) + ( R I ) > 0 h l 4 R h 4, then it is optimal for the firm to commit to a fee F = I R when it rejects the investment project. Otherwise it is optimal for the firm never to invest. 15
Commitment Financial Slack Suppose that managers are empire builders and want to invest as much as possible (Jensen (1986)). Shareholders can design the financial structure of the firm in a way to allow investment when p=p sb, but not when p=p ss. Let E denote the initial cash holdings in the firm. In order to invest, the firm needs to raise external finance max{0,i-e} from uninformed investors, who learn from the price about the expectation that the project succeeds. The firm must offer a competitive return (assumed zero) to the investors. 16
Commitment Financial Slack Cont d For E < I R, the firm is unable to raise external finance and never undertakes the investment project. n When c 4 ( Rh Rl ) c E [ I R +, I R ) n l, the shareholders will choose retained earnings 2. The speculator will produce information and the n investment policy is type II conditional investment. If c ( R R ) shareholders will choose E < I R and the firm will never invest. > 4, the Our point on the optimality of retained earnings is related to Myers and Majluf (1984). Interestingly, the underling mechanism is almost opposite. h l 17
Empirical Implications Commitment Commitment occurs when the economy is in an intermediate state. Firms that can learn more from the market are more likely to use commitment devices such as break-up fees or financial slack. o One would expect speculators to have information about demand factors / strategic factors, not on technological factors. Firms with commitment devices will exhibit informative prices. Two measures of informativeness are used in Chen, Goldstein, and Jiang (2005): o Price non-synchronicity: Durnev, Morck, and Yeung (2004). o Probability of informed trading: Easley, Kiefer, and O Hara (1996). 18
Investment Frequency Bust Intermediate Boom Interim Efficient Plan Equilibrium Investment When different from interim efficient Frequency=0.75 (Type II Investment policy) Frequency=0.25 (Type I Investment policy) 0 * R h I ** R h R h 19
Conclusions Prices perform important role affecting real investment decisions. When investments are cancelled too often following negative information in the price, securities become insensitive to information about investment fundamentals. As a result, speculators will not produce information. Mechanism amplifies shocks to fundamentals. Firms may decide to commit to overinvest, in order to encourage information production. Possible commitment devices are: cancellation fee, financial slack, etc. 20