Outline In-Class Experiment on Security Markets with Insider Information Test of Rational Expectation Hypothesis I: Plott and Sunder (1982) Can market be used to disseminate information? (or does price reflect insider information?) Test of Rational Expectation Hypothesis II: Plott and Sunder (1988) Can market be used to aggregate diverse information? (or does price reflect aggregate information?) Field Application at HP: Kay-Yut Chen, Senior Scientist, HP Lab
Induced Preference
Controls on Cash and Security Flow
Type of Traders and Dividend Rate Markets 1-3: Two types of trader (I and II) Markets 4-6, 10-11: Three types of traders (I, II, and III) Four traders in each type of traders Markets 7-9: One type of traders but there are 12 of them Initial endowment is 2 except in markets 5_S, 6-7, and 9- 10, it is 4
Design of Markets
Asset Type Single AssetState-Contingent Claims Uniform DividendsSeries C: 7,8,9 Diverse DividendsSeries A: 1,2,3,6,10,11 4,5-last few periods Series B: 4,5 –Periods 1-9
Single Security vs. Contingent Claims Single Security (e.g., Market 3) A type II trader yielded a dividend of 230 if the state was X, 90 francs if the state was Y, and 60 francs if the state was Z. Contingent Claims (e.g., Market 4) The contingent claims markets had 3 different securities x, y, z. Let’s focus on Type I trader. The x securities yielded a positive dividend of 70 if x occurred and zero otherwise. The y securities yielded a positive dividend of 130 if y occurred and zero otherwise. The z securities yielded a positive dividend of 300 if z occurred and zero otherwise. A portfolio of one of each type of security is equivalent to one security in the single security markets.
Design of Markets
Hypotheses Rational Expectation (RE) Hypothesis (Null) Traders behave as if they are aware of the pooled information of all traders in the system. That is, they behave as if they know the state with certainty Prior-Information (PI) Hypothesis Determine posterior probability EV maximizers Maximin (MM) Hypothesis Determine posterior probability Traders will not purchase unless the price is below the minimum they could possibly receive given their prior information
Price and Allocation Predictions: RE vs. PI vs. MM
Design of Markets
Price and Allocation Predictions: RE vs. PI. vs. MM
PI: Prices of Contingent Claims in Market 4
Actual versus Predicted Prices at the last occurrence of each state
Market 10: RE did not work well
Market 1: RE did not work well
Market 2: RE did not work well
Market 4: RE Worked Well
Market 8: RE Worked Well
Market 9: RE Worked Well
Security Transfer Measure
Actual and Predicted Allocations at the End of Each Market
Actual and Predicted Profit Distributions
Efficiency
|Actual – Predicted Efficiency|
Summary Behaviors in Series A (single security with diverse preferences) are only partially captured by RE model (e.g., Market 10) If the markets are complete (as in Series B) or is preferences are identical (Series C), the RE model provides a reasonably accurate description of behaviors (Market 4-CC, Markets 8 and 9).