portfolio selection with skew normal asset returns quan gan discipline of finance, university of...

21
BUSINESS SCHOOL Portfolio Selection with Skew Normal Asset Returns Quan Gan Discipline of Finance, University of Sydney

Upload: matia

Post on 25-Feb-2016

42 views

Category:

Documents


0 download

DESCRIPTION

Portfolio Selection with Skew Normal Asset Returns Quan Gan Discipline of Finance, University of Sydney. Apologies. Due to the expended scope of my paper. I change the title of the paper. The previous title was: Performance of Equity REITs: Does a Skew Factor Matter?. BUSINESS SCHOOL. - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Portfolio Selection with Skew Normal Asset  Returns Quan Gan Discipline of Finance, University of Sydney

BUSINESS SCHOOL

Portfolio Selection with Skew Normal Asset Returns

Quan GanDiscipline of Finance, University of Sydney

Page 2: Portfolio Selection with Skew Normal Asset  Returns Quan Gan Discipline of Finance, University of Sydney

ApologiesDue to the expended scope of my paper. I change the title of the paper.

The previous title was:

Performance of Equity REITs: Does a Skew Factor Matter?

BUSINESS SCHOOL

Page 3: Portfolio Selection with Skew Normal Asset  Returns Quan Gan Discipline of Finance, University of Sydney

Main Contributions

3

The main contributions of the paper:1. I show that Azzalini & Dalla Valle (1996)’s

multivariate skew normal distribution is a special case of Simaan (1993)’s three-parameter model.

2. All Simaan (1993)’s results are applicable to the skew normal asset returns.

3. Combining the skew normal asset returns with the CARA utility, I obtain the closed-form certainty equivalent and skewness premium.

Page 4: Portfolio Selection with Skew Normal Asset  Returns Quan Gan Discipline of Finance, University of Sydney

Main Contributions (cont.)

4

4. I show that the skewness premium is positive (negative) when asset returns have negative (positive) skewness. The magnitude of the skewness premium is increasing in market risk aversion.

5. I find that when investors face broad investment opportunities, the economic value of considering higher moments is negligible under realistic margin requirements.

6. Implication of real estate market (high leverage is needed for exploring higher moments benefit).

Page 5: Portfolio Selection with Skew Normal Asset  Returns Quan Gan Discipline of Finance, University of Sydney

Simaan’s three-parameter model

5

Simaan (1993)’s three parameter model is a nice extension of Markowitz’s mean-variance framework.

1) it is fully compatible with the expected utility maximization;

2) it nests the mean-variance portfolio selection results; 3) it has nice geometric properties; 4) it considers information of all moments.

Page 6: Portfolio Selection with Skew Normal Asset  Returns Quan Gan Discipline of Finance, University of Sydney

Simaan’s three-parameter model (Cont.)

6

The data-generating process is written as a regression:

1. where st is a scalar perturbation with any non-elliptical distribution.

2. Conditioning on st, yt is elliptical.

Page 7: Portfolio Selection with Skew Normal Asset  Returns Quan Gan Discipline of Finance, University of Sydney

Simaan (1993) meets Engle et al. (1983)

7

Simaan (1993)’s construction is related to Engle et al. (1983)’s weak exogeneity concept.Weak exogeneity:

Information on yt|st is enough to conduct statistical inference on parameters of g(yt|st), thus information on marginal distribution g(st) can be safely ignored.

Page 8: Portfolio Selection with Skew Normal Asset  Returns Quan Gan Discipline of Finance, University of Sydney

Simaan (1993) meets Engle et al. (1983)

8

In Simaan (1993)’s construction, st is weakly exogenous to the parameter set of elliptical distribution.

Page 9: Portfolio Selection with Skew Normal Asset  Returns Quan Gan Discipline of Finance, University of Sydney

Simaan (1993) meets Azzalini & Dalla Valle (1996)

9

Azzalini & Dalla Valle (1996) proposes the multivariate skew normal distribution.

(yt|st >0) has the density:

where

Page 10: Portfolio Selection with Skew Normal Asset  Returns Quan Gan Discipline of Finance, University of Sydney

Simaan (1993) meets Azzalini & Dalla Valle (1996)

10

The multivariate skew normal is a special case of Simaan (1993)’s three-parameter model.

st follows a half normal distribution (non-elliptical).Conditioning on st, yt|st follows multivariate normal

(elliptical) distribution.

where st > 0.

Page 11: Portfolio Selection with Skew Normal Asset  Returns Quan Gan Discipline of Finance, University of Sydney

Simaan (1993) meets Athayde and Flˆores (2004)

11

The three-parameter efficient frontier has linearity property as shown in panel (b) in the next slide.

Page 12: Portfolio Selection with Skew Normal Asset  Returns Quan Gan Discipline of Finance, University of Sydney

Simaan (1993) meets Athayde and Flˆores (2004)

12

Page 13: Portfolio Selection with Skew Normal Asset  Returns Quan Gan Discipline of Finance, University of Sydney

Simaan (1993) meets Markowitz (1952)

13

The three-parameter portfolio selection problems can be written as:

Page 14: Portfolio Selection with Skew Normal Asset  Returns Quan Gan Discipline of Finance, University of Sydney

Simaan (1993) meets Markowitz (1952)

14

The solution is a nice three-fund separation which nests the classic two-fund separation result.

Page 15: Portfolio Selection with Skew Normal Asset  Returns Quan Gan Discipline of Finance, University of Sydney

Higher Moments Portfolio Selection Paradigms

15

Page 16: Portfolio Selection with Skew Normal Asset  Returns Quan Gan Discipline of Finance, University of Sydney

Certainty Equivalent of the CARA utility

16

Maximizing the expected utility is equivalent to maximizing the certainty equivalent. For the CARA utility, the certainty equivalent is:

Page 17: Portfolio Selection with Skew Normal Asset  Returns Quan Gan Discipline of Finance, University of Sydney

Certainty Equivalent of the CARA utility (normal distribution vs. skew normal distribution)

17

Combining the CARA utility with normal distribution, the certainty equivalent is:

Combining the CARA utility with normal distribution, the certainty equivalent is:

Page 18: Portfolio Selection with Skew Normal Asset  Returns Quan Gan Discipline of Finance, University of Sydney

Asset Pricing with Skew Normal Returns

18

The new asset pricing formula is:

Page 19: Portfolio Selection with Skew Normal Asset  Returns Quan Gan Discipline of Finance, University of Sydney

Skewness Premium

19

The skewness premium depends on C and risk aversion (a formal proof of the relationships is provided in my paper):

Page 20: Portfolio Selection with Skew Normal Asset  Returns Quan Gan Discipline of Finance, University of Sydney

Comparing the certainty equivalents

20

Following P´astor & Stambaugh (2000) and Tu & Zhou (2004), I use the certainty equivalents to compare economic values of different portfolio choices.

I find that under reasonable margin requirements, the economic value of higher moments are negligible.

One implication of my empirical results is that the value of higher moments might be better explored by investors with relaxed financial constraints or by those facing a small and special investment universe.

Both are relevant to real estate investors.

Page 21: Portfolio Selection with Skew Normal Asset  Returns Quan Gan Discipline of Finance, University of Sydney

More?

21

The paper is now available on the conference website and SSRN.

In the paper, I also have:

1) Bayesian algorithm of estimating skew normal distributions

2) A stochastic dominance result of skew normal distribution

3) Skew normal distribution is an unique distribution which can be nested by Simaan (1993)’s three-parameter model.

……