Corporate Strategy, Conformism, and the Stock Market

Transcription

1 Corporate Strategy, Conformism, and the Stock Market Thierry Foucault (HEC) Laurent Frésard (Maryland) November 20, 2015

2 Corporate Strategy, Conformism, and the Stock Market Thierry Foucault (HEC) Laurent Frésard (Maryland) November 20, 2015

3 Real Effects of Stock Price Informativeness Hayek (1945)... Fama & Miller (1972): (an efficient market) has a very desirable feature. In particular, at any point in time market prices of securities provide accurate signals for resource allocation; that is, firms can make production-investment decisions....

4 Real Effects of Stock Price Informativeness Hayek (1945)... Fama & Miller (1972): (an efficient market) has a very desirable feature. In particular, at any point in time market prices of securities provide accurate signals for resource allocation; that is, firms can make production-investment decisions.... Stock price informativeness matters because stock prices are used as a source of information by real decision makers: 1. Firms Managers, Firms creditors, Regulators, central banks, etc.

5 Real Effects of Stock Price Informativeness Hayek (1945)... Fama & Miller (1972): (an efficient market) has a very desirable feature. In particular, at any point in time market prices of securities provide accurate signals for resource allocation; that is, firms can make production-investment decisions.... Stock price informativeness matters because stock prices are used as a source of information by real decision makers: 1. Firms Managers, Firms creditors, Regulators, central banks, etc. Price informativeness in secondary markets matters for the real economy

6 Real Effects of Stock Price Informativeness Hayek (1945)... Fama & Miller (1972): (an efficient market) has a very desirable feature. In particular, at any point in time market prices of securities provide accurate signals for resource allocation; that is, firms can make production-investment decisions.... Stock price informativeness matters because stock prices are used as a source of information by real decision makers: 1. Firms Managers, Firms creditors, Regulators, central banks, etc. Price informativeness in secondary markets matters for the real economy 1. Fast growing theoretical and empirical literature on this topic (see Bond, Edmans, and Goldstein (2012) for a survey).

7 Real Effects of Stock Price Informativeness Hayek (1945)... Fama & Miller (1972): (an efficient market) has a very desirable feature. In particular, at any point in time market prices of securities provide accurate signals for resource allocation; that is, firms can make production-investment decisions.... Stock price informativeness matters because stock prices are used as a source of information by real decision makers: 1. Firms Managers, Firms creditors, Regulators, central banks, etc. Price informativeness in secondary markets matters for the real economy 1. Fast growing theoretical and empirical literature on this topic (see Bond, Edmans, and Goldstein (2012) for a survey). 2. Mostly focused on the relationship between stock price informativeness and investment (e.g., Chen, Goldstein, and Jiang (2007), Bakke and Whited (2010), or Foucault and Fresard (2012, 2014))

8 Our contribution: The Conformity Effect If managers use information in stock prices then they have an incentive to take actions to enhance stock price informativeness:

9 Our contribution: The Conformity Effect If managers use information in stock prices then they have an incentive to take actions to enhance stock price informativeness: 1. These actions can have real effects as well

10 Our contribution: The Conformity Effect If managers use information in stock prices then they have an incentive to take actions to enhance stock price informativeness: 1. These actions can have real effects as well 2. Here: they can affect product market choices by firms

11 Our contribution: The Conformity Effect If managers use information in stock prices then they have an incentive to take actions to enhance stock price informativeness: 1. These actions can have real effects as well 2. Here: they can affect product market choices by firms Why? Differentiation versus Learning 1. Differentiation (uniqueness) is valuable (e.g., local monopoly) higher average cash-flows. 2. BUT the stock market produces more information about common strategies because: 2.1 More scope for cross market learning by market makers (or managers) 2.2 Economies of scale in the production of information

12 Our contribution: The Conformity Effect If managers use information in stock prices then they have an incentive to take actions to enhance stock price informativeness: 1. These actions can have real effects as well 2. Here: they can affect product market choices by firms Why? Differentiation versus Learning 1. Differentiation (uniqueness) is valuable (e.g., local monopoly) higher average cash-flows. 2. BUT the stock market produces more information about common strategies because: 2.1 More scope for cross market learning by market makers (or managers) 2.2 Economies of scale in the production of information Conformity effect: 1. Firms differentiate less when managers learn from stock prices than when they don t. 2. The correlation among fundamentals is higher when managers learn from stock prices than when they don t.

13 Theory and Tests A manager must choose among two strategies common strategy followed by several public firms unique strategy Abandonment option: Before implementing the strategy, the manager can decide to abandon it based on private information and stock market prices. Trade-off: 1. Expected payoff of the unique strategy is higher with perfect information. 2. But the value of the abandonment option is smaller because stock market information is less precise for unique strategies. We derive and test new specific predictions for IPO firms 1. Transition from private to public changes informational environment because newly public firms can learn from their own prices 2. Increase incentives to choose the unique strategy

14 Related literature

15 Related literature Conformism in managerial decisions Reputation concerns (Scharfstein and Stein (1990)) or herding (Bikhchandani et al.(1998)) push towards conformism. Our paper: Conformism is a way to enhance stock price informativeness

16 Related literature Conformism in managerial decisions Reputation concerns (Scharfstein and Stein (1990)) or herding (Bikhchandani et al.(1998)) push towards conformism. Our paper: Conformism is a way to enhance stock price informativeness Learning from stock prices Most of the literature focuses on the effect of stock prices on investment (I -Q regressions) Our focus on product differentiation is novel

17 Related literature Conformism in managerial decisions Reputation concerns (Scharfstein and Stein (1990)) or herding (Bikhchandani et al.(1998)) push towards conformism. Our paper: Conformism is a way to enhance stock price informativeness Learning from stock prices Most of the literature focuses on the effect of stock prices on investment (I -Q regressions) Our focus on product differentiation is novel Finance and product markets Many studies look at the impact of industry structures on asset prices (e.g., Peress (2010) or Hou and Robinson (2006)) We focus on the opposite: stock prices = product market structures

18 Timeline of the baseline model t=1 t=2 t=3 t=4 The manager of firm A chooses a strategy SA {Su,Sc} Unique: Su Common: Sc: already chosen by n firms A strategy can be Good (G) or Bad (B). Stock market opens: order flows ({f1,,fn,fa}) and stock prices ({p1,,pn,pa}) are realized With probability γ, the manager of firm A receives a signal (sm) about the type (G or B) of his strategy He observes his stock price (pa). Abandonment Option: o Pay $1 and Go on with strategy SA (I=1) or o Abandon it (I=0) at zero cost. Firms cash flows (r(s,ms,ts)) are realized.

19 Payoffs rs (, m, G) rs (, m ) 0 A S A S S A A A rs (, m, B) r( S, m ) 0 A S A S S A A A Manager of firm A has chosen strategy S A at date 1 0 Date t=3: Exercise option Date t=4: Payoffs

20 Payoffs: Assumptions A.1: For identical types, uniqueness yields higher payoffs: λ(n) = r(s u,1,t S ) r(s c,n+1,t S ) > 1

21 Payoffs: Assumptions A.1: For identical types, uniqueness yields higher payoffs: λ(n) = r(s u,1,t S ) r(s c,n+1,t S ) > 1 A.2: The abandonment option has value. The NPV of a good strategy is positive but the NPV of a bad strategy is negative.

22 Payoffs: Assumptions A.1: For identical types, uniqueness yields higher payoffs: λ(n) = r(s u,1,t S ) r(s c,n+1,t S ) > 1 A.2: The abandonment option has value. The NPV of a good strategy is positive but the NPV of a bad strategy is negative. Additional more technical assumptions: 1. A.3: Information is useful: The expected net present value of both strategies is negative in the absence of additional information when the abandonment option can be exercised. 2. A.4: A good strategy with one entrant is better than a bad strategy with no entrant: r(s c, n + 1, G ) > r(s c, n, B)

23 Value of a Strategy Let I = 1 if the manager implements her strategy and I = 0 if she abandons it.

24 Value of a Strategy Let I = 1 if the manager implements her strategy and I = 0 if she abandons it. Value of firm A at date 3: V A3 = I E (NPV (S A, m SA ) p A, s m ) where NPV (S A, m SA ) = r(s A, m SA, 1) 1 is the firm s NPV if it implements its strategy and m SA is the resulting number of firms following strategy S A (1 if S A = S u or n + 1 if S A = S c ).

25 Value of a Strategy Let I = 1 if the manager implements her strategy and I = 0 if she abandons it. Value of firm A at date 3: V A3 = I E (NPV (S A, m SA ) p A, s m ) where NPV (S A, m SA ) = r(s A, m SA, 1) 1 is the firm s NPV if it implements its strategy and m SA is the resulting number of firms following strategy S A (1 if S A = S u or n + 1 if S A = S c ). Value of firm A at date 1: V A1 = E (I NPV (S A, m SA )) = E (Max{0, E (NPV (S A, m SA ) p A, s m )}

26 Value of a Strategy Let I = 1 if the manager implements her strategy and I = 0 if she abandons it. Value of firm A at date 3: V A3 = I E (NPV (S A, m SA ) p A, s m ) where NPV (S A, m SA ) = r(s A, m SA, 1) 1 is the firm s NPV if it implements its strategy and m SA is the resulting number of firms following strategy S A (1 if S A = S u or n + 1 if S A = S c ). Value of firm A at date 1: V A1 = E (I NPV (S A, m SA )) = E (Max{0, E (NPV (S A, m SA ) p A, s m )} No financing constraints or agency conflicts The manager chooses I and SA to maximize the firm value at dates 3 and 1.

27 The Stock Market (date 2) 3 types of investors Continuum of risk-neutral speculators Liquidity traders with agg. demand uniformly distributed for firm j on [ 1, 1] Risk neutral dealers

28 The Stock Market (date 2) 3 types of investors Continuum of risk-neutral speculators Liquidity traders with agg. demand uniformly distributed for firm j on [ 1, 1] Risk neutral dealers Speculators: Fraction π c (π u ) receives a perfect signal s i {G, B} about S c (S u ) Each speculator can buy/sell one share of each firm Trade of speculator i in firm j: x ij (s i (S j )) { 1, 0, +1}

29 The Stock Market (date 2) 3 types of investors Continuum of risk-neutral speculators Liquidity traders with agg. demand uniformly distributed for firm j on [ 1, 1] Risk neutral dealers Speculators: Fraction π c (π u ) receives a perfect signal s i {G, B} about S c (S u ) Each speculator can buy/sell one share of each firm Trade of speculator i in firm j: x ij (s i (S j )) { 1, 0, +1} Trading: as in Kyle (1985): Order flow in stock j is: f j = total demand from speculators and noise traders in firm j (speculators aggregate demand is +/ π c or +/ π u ). Absorbed by dealers who observe flows in each market Dealers are competitive, risk neutral, and observe order flows in all stocks: they set prices equal to their expectation of the final payoff of the firm given Ω 2 = {f 1, f 2,..., f n, f A }.

30 The stock market equilibrium p A2 (Ω 2 ) = E (V A3 (I (Ω 3, S A ), S A ) Ω 2 ) p j2 (Ω 2 ) = E (r(s c ), m Sc, t Sc ) Ω 2 ) for j {1,..., n} Feedback effect 1. The price of A and established firms depends on I (Ω 3 ) 2. I (Ω 3 ) depends on the stock price of all firms, in particular A. 3. Jointly determined in equilibrium (fixed point) A stock market equilibrium is a set {x ij ( ), p j2 ( ), I ( )} 1. The trading strategy of each speculator maximizes his expected profit 2. Policy I ( ) max. firm value at date 3 3. Dealers set their prices such that they break-even (correctly anticipating x ij ( ) and I ).

31 Benchmark: No Learning from stock prices The decision to exercise the abandonment option (I ) is only driven by private managerial information. Hence, the value of the option only depends on the quality of managerial information γ. The manager should implement his strategy at date 2 if he learns that the strategy is good, and should do nothing otherwise. The expected value of firm A at date 1 with strategy S A is therefore V A1 = γ 2 (r(s A, m SA, G ) 1) Result: As r(s u, 1, G ) > r(s c, n + 1, G ) (Assumption A.1), the manager chooses the unique strategy.

32 Stock market equilibrium with learning from stock prices 1/2 Case 1: Firm A chooses the common strategy at date 1. Let f max (f min ) be the largest (smallest) order flow realization across all stocks. Stock Price of firm A at date 2 here here Manager s decision at date 3 if no private info: Abandon Do not abandon

33 Stock market equilibrium with learning from stock prices 2/2 Case 2: Firm A chooses the unique strategy at date 1. Stock price of firm A at date 2 f A here f A here 0 1 f A here 1 Manager s decision at date 3 if no private info: Abandon Do not abandon

34 introduction Model Prediction Data Empirical Findings Conclusion Stock Price Informativeness Let π (n, πc ) = 1 (1 πc )(n+1). Cross asset learning effect: if πu < π (n, πc ) then the stock market is more informative about the type (good/bad) of the common strategy than about the type of the unique strategy. Stock market informativeness about the unique strategy (Πu = 0.3L Stock market informativeness about the common strategy (Πc = 0.1L Number of established firms 10

35 Stock Price Informativeness vs. Uniqueness The value of firm A at date 1 if it chooses the common strategy: V A1 (S c ) = [ γ + (1 γ) π(n, π c) ](r(s c, n + 1, G ) 1). 2

36 Stock Price Informativeness vs. Uniqueness The value of firm A at date 1 if it chooses the common strategy: V A1 (S c ) = [ γ + (1 γ) π(n, π c) ](r(s c, n + 1, G ) 1). 2 The value of firm A at date 1 if it chooses the unique strategy: V A1 (S u ) = [ γ + (1 γ)π u ](r(s u, 1, G ) 1). 2

37 Stock Price Informativeness vs. Uniqueness The value of firm A at date 1 if it chooses the common strategy: V A1 (S c ) = [ γ + (1 γ) π(n, π c) ](r(s c, n + 1, G ) 1). 2 The value of firm A at date 1 if it chooses the unique strategy: V A1 (S u ) = [ γ + (1 γ)π u ](r(s u, 1, G ) 1). 2 Trade-off: 1. The unique strategy has a higher payoff holding the type of each strategy identical (e.g., both are good). 2. BUT: The stock market is less informative about the type of the unique strategy the value of the abandonment option is lower with this strategy.

38 Conformity effect (if π u < π(n, π c )) Cash Flow Gains from differentiation (λ(n)) The common strategy maximizes the value of firm A The unique strategy maximizes the value of firm A λ = (γ + (1 - γ) π(n, π c )) / (γ + (1 - γ) π u ) Managerial Information (γ)

39 Conformity effect (π u > π(n, π c )) If π u > π(n, π c ), no conformity effect. However, even if π u > π c, the case π u < π(n, π c ) obtains for n large enough due to cross-market learning. Still true if production of information about a strategy is endogenous: in equilibrium, π u < π(n, π c ) for n large enough because speculator can amortize the cost of information production about the common strategy over a larger number of stocks.

40 How can we test the conformity effect?

41 Private vs Public Firms π u = 0 for private firms (no stock price), but π c (n, π c ) > 0. Thus, compare two identical firms A but one is private and the other public. 1. The drop in the value of the abandonment option with the unique strategy is larger for the private firm because the drop in stock price informativeness is larger for the private firm. 2. The set of parameters for which the private firm chooses the common strategy is larger. 3. The conformity effect is stronger for private firms, other things equal. Testable Implication: On average, firms should become increasingly differentiated after going public.

42 Conformity effect: Private vs. Public Firms Cash Flow gain from differentiation (λ(n)) The common (resp. unique) strategy is optimal if A is private (resp public) The common strategy maximizes firm value whether A is public or private The unique strategy maximizes firm value whether A is public or private Managerial Information (γ)

43 Testing the model s prediction Methodological challenges 1. Measurement of uniqueness 2. Observation of firms strategic choices when private and public

44 Testing the model s prediction Methodological challenges 1. Measurement of uniqueness 2. Observation of firms strategic choices when private and public Issue 1: Let A,B (k A ) be a measure of product differentiation between A and B B is one public established peer of A (among a set J A ) k A is the ownership status of A (private or public) Assume: of A,B (k A ) means a move from S c to S u relative to B Prediction: Γ A,B = A,B (public) A,B (private) 0 (not negative!!!).

45 Testing the model s prediction Methodological challenges 1. Measurement of uniqueness 2. Observation of firms strategic choices when private and public Issue 1: Let A,B (k A ) be a measure of product differentiation between A and B B is one public established peer of A (among a set J A ) k A is the ownership status of A (private or public) Assume: of A,B (k A ) means a move from S c to S u relative to B Prediction: Γ A,B = A,B (public) A,B (private) 0 (not negative!!!). Issue 2: we do not observe A,B (private, λ A ). But A,B (public) is observed over time Let τ = 0, 1,..., 5 be public age Assume: A,B (private) = A,B,τ=0 (public)

46 Testing the model s prediction We then estimate Γ {A,B} by estimating: {A,B,τ} (public) = α AB + η A τ + ɛ AB for a large number of pairs (A, B). η A measures the change in diff. between A and B over time after an IPO relative to their differentiation at the time of the IPO (α AB = {A,B,0} ). We have η A A,B (public) A,B (private)

47 Empirical Tests

48 Measure of Differentiation and IPOs Product differentiation ( i,j ) Text-Based Network Classification (TNIC) of Hoberg and Phillips Measure of product similarity between pairs of firms (ρ i,j (0 100%)) based on 10K text TNIC includes only firms with ρ i,j > 21% Varies over time (period ): i,j,t = 1 ρ i,j,t Return co-movement (β i,j ) Returns of more unique firms should be less related to returns of other firms r i,t = β 0 + β m,t r m,t + β i,j,t r j,t + ν i,t Annual estimation (weekly returns) for all firm-pairs Higher differentiation lower β i,j,t IPO firms IPO data from Jay Ritter (restrict to filters) Sample of 1,231 IPO firms

49 Unit of Observation: Firm-Pairs B 21 A B 1A B 2A B 24 B 22 B 23 B NA «Treated pairs»: TNIC Peers (age > 5 years) of firm A at event (IPO) date «Counterfactual»: TNIC Peers (age > 5 years) of firm B 2A at event (IPO) date (matched such that Δ BB Δ AB)

50 Econometric specification: Diff-in-Diff spirit We stack treated and untreated pairs in a large panel i,j,τ,t = η 0 τ + η 1 (τ Treated i,j,τ,t ) + α i,j + δ t + βx i,j,τ,t + ε i,j,τ,t 1. τ {0, 1,..., 5} years post IPO. 2. Pair fixed effects α i,j capture fixed characteristics (e.g. λ i ) 3. Year fixed effects δ t capture calendar-time effects 4. ε i,j,τ,t is clustered within pairs η 0 measures change in diff. for ALL pairs η 0 + η 1 measures change in diff. for Treated pairs (going public firms) η 1 measures the relative change in diff. for IPO firms η 1 > 0 = average IPO firm moves from S c to S u

51 Summary Statistics Panel A: Firm level τ=0 IPO firms Established Peers Peers of Peers N 1,231 2,678 2,961 Age Δ i,j # of Peers log(a) MB Panel B: Pair level τ=0 Treat=1 Treat=0 all N 31,427 90, ,195 Δ i,j Age i Age j log(a) i log(a) j MB i MB j Panel C: Pair year level All τ Treat=1 Treat=0 all N 139, , ,745 Δ i,j Age i Age j log(a) i log(a) j MB i MB j τ

52 Main empirical finding - Differentiation increases after IPOs Dependent Variable: Product Differentiation ( i,j) (1) (2) (3) (4) (5) (6) controls >3yrs M 5 M size 0.147*** 0.142*** 0.142*** 0.130*** 0.135*** 0.146*** [62.907] [57.389] [57.331] [50.524] [70.914] [55.948] x Treat 0.027*** 0.026*** 0.014** 0.035*** 0.021*** [4.824] [4.736] [2.501] [6.461] [3.763] log(a) i log(a) j 0.024*** 0.014** 0.020*** 0.070*** [ 3.718] [ 2.068] [ 3.771] [ ] MB i MB j 0.007*** 0.004** 0.007*** [ 3.350] [ 2.044] [ 4.350] [ 0.825] Pair Fixed Effects ( i,j) Yes Yes Yes Yes Yes Yes Year Fixed Effects ( t) Yes Yes Yes Yes Yes Yes Obs. 633, , , , , ,986 Adj. R

53 Main result - dynamics Event-time (tau) Treat=1 Treat=0

54 Cross-sectional tests (model-based) We show theoretically that the conformity effect Decreases with γ (manager private information) Increases with π(n, π c ) (price informativeness of peers) This is UNIQUE to our model and story! Empirical proxies for γ Intensity of insider trading (Insider i ) Performance of insider trades (InsiderAR i ) Empirical proxies for π(n, π c ) Probability of Informed Trading (PIN j ) Ability of stock price to forecast earnings (BPS j ) Market reaction to earnings surprises (ERC j ) Coverage by analysts (Coverage j ) Add interaction terms (i.e. triple-differences)

55 Managerial Information and Price Informativeness Dep. Variable: Product Differentation (Δi,j) Managerial Information Peers' Price Informativeness φ: InsiderAR i Insider i PIN j BPS j ERC j Coverage j (1) (2) (3) (4) (5) (6) 0.142*** 0.142*** 0.146*** 0.125*** 0.136*** 0.139*** [57.44] [57.31] [56.43] [38.13] [50.98] [55.62] τ x Treat 0.020*** 0.018*** 0.093*** 0.083*** 0.266*** 0.143*** [3.57] [2.79] [3.30] [7.74] [ 8.32] [6.84] τ x Treat x φ 0.022*** 0.009** 0.022** 0.036*** 0.073*** 0.061*** [5.40] [2.16] [ 2.30] [ 5.78] [9.65] [ 5.39] Control Variables Yes Yes Yes Yes Yes Yes Pair Fixed Effects Yes Yes Yes Yes Yes Yes Year Fixed Effects Yes Yes Yes Yes Yes Yes Obs. 633, , , , , ,745 Adj. R

56 Alternative Explanations Conclusion: IPOs are positively correlated with subsequent differentiation and the effect varies with firm characteristics as predicted by our model. Caveat: We cannot discard the possibility that other mechanisms omitted variables drive our findings. However these variables must be: 1. time-varying since we control for pair fixed effects. 2. Unrelated to firm size and market to book ratios since we control for these variables. 3. Strongly correlated with our proxies for managerial information and price informativeness to explain our cross-sectional findings.

57 Alternative Explanations The financing channel: 1. Innovation is a way to differentiate but innovation requires financing 2. IPOs relax financial constraints and allow firms to finance new innovations 3. IPOs are positively related to differentiation To control for this channel, we add controls for financing conditions (availability of credity ratings, dividend payments, and a text-based measure of financing constraints) and innovation (the ratio of R&D over sales, the ratio of advertising expenses over sales, and the number of new patents over assets). Our empirical findings are not affected by the inclusion of these controls

58 Alternative Explanations Dep. Variable: Product Differentiation (Δi,j) Managerial Information Peers' Price Informativeness φ: InsiderAR i Insider i PIN j BPS j ERC j Coverage j (1) (2) (3) (4) (5) (6) τ 0.142*** 0.142*** 0.147*** 0.125*** 0.136*** 0.139*** [57.44] [56.57] [56.08] [37.78] [50.67] [54.88] τ x Treat 0.014** 0.011* 0.082*** 0.072*** 0.279*** 0.138*** [2.47] [1.70] [3.30] [6.48] [ 8.57] [6.46] τ x Treat x φ 0.021*** 0.010** 0.020** 0.032*** 0.074*** 0.062*** [5.06] [2.37] [ 2.10] [ 4.90] [9.64] [ 5.30] Control Variables Yes Yes Yes Yes Yes Yes Financing Controls Yes Yes Yes Yes Yes Yes Innovation Controls Yes Yes Yes Yes Yes Yes Pair Fixed Effects Yes Yes Yes Yes Yes Yes Year Fixed Effects Yes Yes Yes Yes Yes Yes Obs. 619, , , , , ,151 Adj. R

59 Time-varying financing conditions Dep. Variable: Product Differentiation (Δi,j) φ: D ( ) D ( ) TED Spread IPO Volume (1) (2) (3) (4) τ 0.142*** 0.142*** 0.141*** 0.142*** [57.35] [57.40] [57.10] [57.21] τ x Treat 0.024*** 0.023*** 0.013* 0.024*** [4.21] [4.11] [1.95] [3.27] τ x Treat x φ 0.010* 0.026*** 0.033*** [1.71] [2.94] [3.65] [0.043] Control Variables Yes Yes Yes Yes Pair Fixed Effects Yes Yes Yes Yes Year Fixed Effects Yes Yes Yes Yes Obs. 633, , , ,745 Adj. R

60 Conclusion How do stock markets on real decisions when managers learn information from stock prices?

61 Conclusion How do stock markets on real decisions when managers learn information from stock prices? 1. Directly: Stock prices cause a decision (e.g., investment).

62 Conclusion How do stock markets on real decisions when managers learn information from stock prices? 1. Directly: Stock prices cause a decision (e.g., investment). 2. Indirectly: Managers make decisions to shape the informativeness of their stock price.

63 Conclusion How do stock markets on real decisions when managers learn information from stock prices? 1. Directly: Stock prices cause a decision (e.g., investment). 2. Indirectly: Managers make decisions to shape the informativeness of their stock price. We provide theory and evidence for the second channel by considering differentiation decision by firms. 1. Conformity effect: Managers prefer not to differentiate because it lowers their ability to learn from stock prices.

64 Conclusion How do stock markets on real decisions when managers learn information from stock prices? 1. Directly: Stock prices cause a decision (e.g., investment). 2. Indirectly: Managers make decisions to shape the informativeness of their stock price. We provide theory and evidence for the second channel by considering differentiation decision by firms. 1. Conformity effect: Managers prefer not to differentiate because it lowers their ability to learn from stock prices. Several possible extensions 1. Learning externalities: firms might inefficiently coordinate on projects with low expected payoffs 2. Other decisions than product market differentiation.