Title: Asset Pricing with Heterogeneous Agents, Incomplete Markets and Trading Constraints
1Asset Pricing with Heterogeneous Agents,
Incomplete Markets and Trading Constraints
- Tsvetanka Karagyozova
- 25 October 2007
2Outline
- Motivation/Contribution
- Model
- Calibration
- Results
- Robustness
- Conclusion
Asset Pricing with Heterogeneous Agents
3Motivation
- How do individuals make decisions in the presence
of risk? - Parsimonious vs. realistic models
- Can heterogeneous preferences account for some of
the puzzles recorded in empirical studies?
Asset Pricing with Heterogeneous Agents
4The Problem
- In equilibrium, what determines the differential
in expected real returns across assets? - How do we define risk? What are the sources of
risk? - CAPM (Sharpe, 1964, Lintner, 1965, Mossin, 1966)
- CCAPM (Lucas, 1978, Breeden, 1979)
Asset Pricing with Heterogeneous Agents
5Problems with the consumption capital asset
pricing model
- Fails to account quantitatively for the
historical high average rate of return and
volatility of stocks - The Equity Premium Puzzle (Mehra
Prescott,1985). - The Risk Free Rate Puzzle (Weil, 1989)
- The Excess Volatility Puzzle of stock returns
(Shiller, 1981 Leroy Porter, 1981 Singleton,
1980) - No satisfactory alternative model of stock
pricing.
Asset Pricing with Heterogeneous Agents
6Issues
- Are models of aggregate savings behavior
omitting some crucial element? - What is the cost of the business cycle?
- Kocherlakota, 1996
Asset Pricing with Heterogeneous Agents
7Key Assumptions
- Constant relative risk aversion utility function
- Generalized Expected Utility (Epstein and Zin,
1989, 1991) - Habit Formation (Constantinides, 1990 Heaton
1995, Campbell and Cochrane, 1999) - Keeping up with the Joneses (Abel, 1990 Gali,
1994) - Complete asset markets (Constantinides Duffie,
1996 Lucas, 1994, Telmer, 1993, Marcet
Singleton, 1999) - Costless asset trading (Lucas Heaton, 1995
Luttmer, 1999 McGrattan Prescott, 2003)
8Contribution
- Allowing for agent heterogeneity is instrumental
in accounting for the stylized facts of financial
markets. - Positive models of human behavior provide
insights into decades-old economic puzzles.
Asset Pricing with Heterogeneous Agents
9Model Features
- DSGE model
- Endowment economy
- Two assets in the economy risky and risk-free.
- Heterogeneous agents
- Preferences
- Income
- Incomplete markets.
- Trading constraints
10Model Features
- Heterogeneous Preferences
- The consumption of stockholders is more volatile
and more highly correlated with the excess return
on the stock market (Mankiew and Zeldes, 1991) - An attempt to introduce elements of prospect
theory (Kahneman Tversky, 1979) into a standard
asset pricing model. - Barberis, Huang Santos, 2001 extension of the
prospect theory to a dynamic model
11Asset Pricing with Heterogeneous Agents
12Consumer Preferences Type A agent
- Type A agent (non-investors) maximizes
- is the consumption of Type A agent at time
t - ? is the subjective discount factor
- ? is the coefficient of relative risk aversion.
13Consumer Preferences Type B agent
- Type B agent (investors or stockholders)
maximizes - consumption of Type B agent at time t
- the risky asset holdings of Type B agent at
time t - Xt1 the gain or loss on risky asset holdings
between t and t1. - zt measures gains and losses prior to time t a
state variable. - b0 exogenous scaling factor.
- The second term measures the utility from
fluctuations in financial wealth the prospect
theory term in the utility function.
14The Model Gains and Losses
- Gains and losses refer to changes in financial
wealth. - The investor only cares about changes in the
value of the risky asset. - Changes in financial wealth are determined once a
year. - Pt1 the price of the risky asset price at t1
- Pf,t the price of risk-free asset at t
15The Model Prior Gains and Losses
- Zt historical benchmark level for the price of
the risky asset. - StB (Pt -Zt ) gt 0 ? the investor has realized
gains in his financial wealth in time t - Define ztZt/Pt.
- The house money effect Thaler and Johnson
(1990) - The extent to which an economic agent is loss
averse depends on his prior stock market
performance.
16The Model Utility from Gains and Losses
- ztlt1 (prior gains or neither gains nor losses)
- StBPt Rt1-StBPt Rf,t
Rt1 ztRf,t - v(Yt1, StB,zt)
for - StBPt Rf,t(zt -
1)?StBPt (Rt1-ztPt Rf,t) Rt1ltztRf,t - where ? gt1
- zt gt1 (prior losses)
- StBPtRt1-StBPtRf,t
Rt1 Rf,t - v(Yt1,StB,zt) for
- ?(zt)(StBPtRt1-StPtRf,
t) Rt1 lt Rf,t - where ?(zt) ? k(zt -1) and k gt 0,
17Utility from Gains and Losses
18The Model The dynamics of the benchmark level, zt
- Assumptions
- The benchmark level, zt, responds sluggishly to
changes in the price of financial asset over
time. - where
- fixed parameter set in such a way that in
equilibrium, the median value of zt 1. - h a measure of investor's memory.
19Budget Constraints
- Budget Constraints
- where
- bond holdings of agent i for i A,B in
period t - income of agent i for i A,B in period t
-
20Short-Sale and Borrowing Constraints
- No Ponzi schemes
- Borrowing Constraint
- where
- Short-sale constraint
21Endowment and Dividend Processes
- Aggregate endowment normalized to 1.
- where dt is the per-share real dividend
- Population normalized to 1.
22Market Equilibrium
- Endogenous variables
- The equilibrium distributions of
- Consumption
- Asset holdings
- Asset prices
23First Order (Kuhn-Tucker) Conditions
- Bond holdings
- Either
- or
- Equity holdings Agent A
- Either
- or
24FOC, Equity Holdings, Agent B
- Either
- or
- where
- for zt 1
- for zt gt1
for
for
25Market-Clearing Conditions
- Consumption
- Where p is the proportion of Type A agents in the
population and is the aggregate endowment
(income). - Bonds zero net supply
- Stocks normalized to 1
26State Variables
- Exogenous
- At every t, given by ln(ytA) ln(ytB) ln(dt )
- The income of Type i agent for i A, B follows
- ln(yti) ?i ?i ln(yt-1i) eti
- eti ? Niid(0,(se t )2)
- The aggregate dividend
- ln(dt) ?i ?i ln(dt-1) et
- eti ? Niid(0,(se t )2)
- Endogenous
- The elements of wealth Bt-1i , St-1i ,
- Prior investment outcomes zt-1
27Model Calibration
28Model Calibration Income and Dividend Processes
- Incomes PSID data 1968-1997
- Panel survey data, annual
- Stockholders 35 of population
- Non-investors 65 of population.
- Income processes of Type A and B agents
- Dividends Net dividends, NIPA tables 1929-2006.
- Solution algorithm Parameterized Expectations
Approach (Marcet, 1989 Marcet Singleton, 1999)
29Income and Dividend Processes (Cont.)
30Results
Asset Pricing with Heterogeneous Agents
31Excess Stock Return Volatility
32Consumption Processes
33Income Processes
34Equilibrium Distribution of z
35Prior Stock Market Performance and Agent Bs
Consumption
36Income, Stock and Bond Holdings
37Income, Stock and Bond Holdings
38Scatter plot
39Robustness Prospect Theory
Asset Pricing with Heterogeneous Agents
40Robustness Prospect Theory
Asset Pricing with Heterogeneous Agents
41Robustness Borrowing Constraints
Asset Pricing with Heterogeneous Agents
42Robustness Investors memory
Asset Pricing with Heterogeneous Agents
43Conclusion
- In equilibrium, individuals hold different
portfolios. - Individuals trade on financial markets to smooth
their consumption. - Time-varying risk premium of stocks over bonds.
- Low risk free rate and low correlation of
consumption with stock returns. - Accounting for loss aversion and prior stock
market performance are instrumental in matching
the empirical moments of asset returns.