Differential Interpretation of Public Signals and Trade in Speculative Markets. Kandel & Pearson, JPE, 1995

Transcription

1 Differential Interpretation of Public Signals and Trade in Speculative Markets Kandel & Pearson, JPE, 1995 Presented by Shunlan Fang May, 14 th, 2008 Roadmap Why differential opinions matter to asset pricing i Asset pricing research based on heterogeneous agents Evidence of agents with heterogeneous beliefs (Kandel & Pearson, 1995) Comments and Reference 1

2 Why Differential Opinions May Matter to Asset Pricing Agents with heterogeneous beliefs generate different prediction of risk Agents with heterogeneous risk aversion coefficients demand different risk premia The market prices move in line with the trading models of competing agents Agents may speculate with each other on the relative accuracy of their predictions Asset Pricing Research asset returns and stochastic discount factor (SDF), e.g. CAPM, Fama-French F Three-Factor Model term structure of interest rate optimal portfolio choice Intertemporal equilibrium model: identical agents, heterogeneous agents behavioral finance 2

3 Application Standard single agent models cannot explain: equity premium puzzle, stock market volatility puzzle, risk-free rate puzzle Aggregate consumption does not proxy for the consumption of investors Heterogeneous agents models focus on explaining the puzzles Kandel & Pearson (1995) Main idea: examine whether agents interpret t public information identically Setting: quarterly earnings announcements Data: return, volume, analyst forecasts Sample period: June 1981 to June

4 Descriptive Evidence Unexpected news versus differential opinions 4

5 Alternative Explanation VS Differential Interpretation Kim and Verrecchia (1991) k it is the dispersion i of risk tolerance coefficients i Life Cycle or Liquidity Trading Private information production Switch from a partially revealing to a fully revealing equilibrium Permanent/Temporary risk changes Empirical Test on Analyst Forecast Data: earnings forecasts of individual stock brokerage research analysts from I/B/E/S Sample Period: Sample: 2131 firms followed by no less than 10 analysts 5

6 Main Idea of the Test If analysts don t have differential interpretation of signal forecast revisions i should not move in different directions or forecasts from different analysts should get closer Empirical Test on Analyst Forecast Test: compare the frequency of inconsistencies in the forecast revisions of announcement periods and non-announcement periods 6

7 Non-announcement Period Control for private information acquisition 7

8 Results- Explicit Updates Results-Implicit Updates 8

9 Magnitudes of Inconsistencies Differences in initial forecasts Differences in revision 9

10 Comments Results do not suggest economic magnitude Forecast revisions should be deflated by prices to control for heteroskedasticity Influenced by extreme forecasts Alternative Explanations Trends in information production Information production concentrated around earnings announcements Mechanical Flips 10

11 Conclusions Agents do not seem to have identical interpretations t ti according to the evidence of volume-return relation and analysts forecast revisions Comments Short term stock return may be associated with trading volume. Market maker use trading volume information to draw inferences about better-informed investors private information on firm value. (Kim & Verrecchia, 2001) Hard to differentiate whether agents differ in interpretation or in information sets. Analysts ability to obtain information from management is different (Ke & Yu, 2006) Pre-RegFD sample, information asymmetry problem is more severe. Evidence may not apply to the post-regfd period. 11

12 Reference Detemple and Murthy(1994), Intertemporal asset pricing with heterogeneous belief, Journal of Economic Theory Dumas, Kurshev and Uppal (2007), What can rational investors do about excess volatility, WP David (2008), Heterogeneous belief, speculation and the equity premium Journal of Finance 12