Research On Asset Pricing And Portfolio Model Based On Investor Behavio

Assets pricing and portfolio selection are two important topics in finance market.Though we have developed many simple and effective theories（e.g.CAMP;Mutual fund separation theorem）from the traditional paradigm,there are still some finance phe-nomenon such as”negative”bubbles,equity premium and investor trading behavior can not be explained by the traditional paradigm.Thus,introducing the investor behavior assumptions（e.g.investors do not update their belief correctly;investors evaluate their portfolio performance using prospect theory rather than risk aversion utility）,setting up model to explain finance phenomenon.It is an important and effective extension and supplement of traditional research paradigm.In this thesis,we start up with dynamic model and setting up rigorous financial economical model to study the finance investment decisions.We find that our model can explain the anomaly in finance market.This thesis concludes three parts as follows,（1）We introduce features of investors’behavior,considering the effects of in-vestors’behavior on assets’prices.In particularly,we propose a dynamic asset pricing model which features asymmetric extrapolative beliefs and dynamic trading population in the economy.Extrapolators form their beliefs for future returns with more weight on past negative returns compared with the past positive return.The model not only gen-erates price bubbles with a large trading volume but also generates”negative”bubbles after a crash.Simulation results show that the asymmetric extrapolative beliefs and dy-namic trading population play predominant roles in the“negative”bubbles.In addition,comparing with the benchmark model,if extrapolators have prospect theory preferences rather than CARA preference then it enlarges the bubbles effect.（2）We introduce prospect theory into portfolio selection model.In particularly,we propose a dynamic portfolio selection model based on prospect theory preference,where the reference level of investors is dynamic.First,we apply dynamic programming to give the optimal policy under different market situations,then we discuss the effect of model parameters theory on the optimal investment policy,and find that the correlation of risk asset returns has effects on investment pattern.When the risk asset returns are independent,the optimal pattern of investors is”V-”shape,which has been discussed in Shi et al.^[1],Meng^[2].However,when the correlation between risky asset returns are negative,the optimal policy is inverse”N-”shape.Besides,we find that investors with a large loss aversion parameter,their optimal position in risk asset are small.The effect of diminishing sensitivity parameters on the optimal investment policy depends on the prior investment outcomes.When the outcomes is gain,diminishing sensitivity parameters,γ,is small,the optimal position in risk asset is small,however,when the outcomes are loss,diminishing sensitivity parameters,γ,is small,the optimal position in risk asset is large.At last,we study the link between our model and the disposition effect and find that asymmetric update rules,the diminishing sensitivity parameters,γ,and the correlation between risk asset returns have effect on the investors’disposition effect.In particularly,when the reference level update with a_l<a_g,the model can generate disposition effect easier,and when the diminishing sensitivity parameters,γ,is small,the disposition effect will enlarge.besides,when the correlation between risk asset returns are negative,the disposition effect will enlarge.（3）We,at last,consider a dynamic portfolio selection model based on mean-CVaR.Conditional risk value is an important measure of downside risk,ont only in theory but also in practice investment,especially after Basel agreement.Different with traditional model assumptions that the returns between risk asset are independent,in practice,the returns between risk asset is correlated.So in our model,we consider a more general situation（the returns follow a AR（1）process）.First,we find the original problem is equivalent to a loss aversion dynamic portfolio selection problem after a series of trans-formations.Seconde,we give the optimal policy of the dynamic mean-CVaR portfolio selection model.Using the numerical experiments,we find that the correlation between risk asset returns has an important effect on the effective frontier.Particularly,when the correlation between risk asset returns is positive,the effective frontier dominates the independent situation.That is,the optimal policy under independent case may be not the optimal investment policy in practice.At last,we give the distribution of terminal wealth under optimal investment policy,and find that they are positively skewed,that is,the optimal policy can control the downside risk of investment portfolio.The contributions are as follows,（1）This thesis focus on the problem that the existing literature about asset pricing dose not explain”negative”bubbles.We introduce the asymmetric extrapolative beliefs and dynamic trading population into asset pricing model.Thesis find that when intro-ducing the dynamic trading population,the asset pricing model generates”negative”bubbles,and the extend of asymmetric extrapolative beliefs has effects on the size of”negative”bubbles.The more attention investors paid on the past negative returns,the bigger size of the”negative”bubbles.（2）This thesis focus on the problem that the existing literature about prospect the-ory portfolio selection model ignores the diminishing sensitivity parameters and the correlation of returns of risky assets.Thesis propose a model that considers the fea-tures discussed above.Thesis find that the diminishing sensitivity parameters and the correlation of returns have a significant effect on the investment behaviors.Though the simulation experiments,thesis find that the correlation and the diminishing sensitivity also have a significant effect on the disposition effect.When the correlation is nega-tive,the model generates a strong disposition effect.When the diminishing sensitivity parameter is small,the model generates a strong disposition effect.（3）This thesis focus on the problem that the existing literature about dynamic mean-CVaR portfolio selection model ignore the returns of risky assets are correlated.We propose a dynamic mean-CVaR portfolio selection model,where the returns of risky assets are correlated.We find that the dynamic mean-CVaR portfolio selection model is equivalent to a dynamic loss aversion portfolio selection model.The correlation of returns have a significant effect on the investment strategy.Especially,when they are positively correlated,the effective frontier dominates.This thesis has theoretical significance for the behavior-based asset pricing and portfolio selection theory,and explains the”negative”bubble of risky asset prices and disposition effect from the perspective of investor behavior.The optimal strategy in this thesis has practical significance for investment in reality.