Bayesian 접근법에 의한 자산배분 전략 모형에 관한 연구 = The Study on the Asset Allocation by Bayesian Approach

다국어 초록 (Multilingual Abstract)

The Markowitz model of modern portfolio theory aims at the objectives of investment not only to maximize the expected return but to minimize risk. The result of optimization of Markowitz's formulation is a parabolic efficient frontier, indicating the combinations of assets with highest expected return given a certain level of risk.
In the reality, however, portfolio managers face the limitation to use this kind of Markowitz model. First, portfolio managers are centered on a small segment of the investment universe. But Markowitz model requires the expected return for all of universe assets. Second, portfolio managers in the reality think the weight found in the Markowitz model is inappropriate to be implemented for client's fund managed by themselves. The limitation that Markowitz model can be used in the practical manner may be found in the estimated error of the historical average returns which are used as alternatives for the equilibrium expected return.
Black-Litterman model(1992) starts from the motivation to mitigate such limitations in the Markowitz. Their model is flexible enough to combine the market equilibrium with subjective economic views and investor's economic reasoning. Therefore, Black-Litterman model tends to have less extreme investment weight compared with the traditional mean-variance optimization model.
In order for us to see the advantage of Black-Litter model against the traditional Markowitz, the monthly rate of returns from 2001 to 2010 are collected from the industry stock index in the KRX. Among all of industries on the KRX are classified into 10 sector group. Each sector is consisted with similar related industries. The industries with the small market capital are excluded from the each sector. The rate of average return for each sector can be estimated be the weighted average rate of return of each industry's index. Based on the index average rate of return, the volatility and correlation are also estimated.
Using the Markowitz model, the optimal weight tends to be extreme result. That means that optimized portfolios tend to overweight assets with large expected return. Whereas, the assets with small expected return are underweighted. And also the optimal portfolios tend to overweight assets with negative(positive) correlation and small(large) variance. Assets with extreme returns may tend to be affected by estimation error for historical average rate of return. But using Black-Litterman, this kind of extreme return problem can be seen to be weakened. This can be seen through the implied expected return. This research also show how to allow the portfolio managers' economic view in the Markowitz model. The absolute and relative views on future economic expectation can be incorporated in the model through vector .

참고문헌 (Reference)

• 1 김진호, "평균-분산 모형과 평균-VaR 모형간 최적위험자산배분 전략 비교" 한국재무학회 15 (15): 143-172, 2002
• 2 구본일, "평균-VaR 기준과 최적 포트폴리오 선택" 한국재무관리학회 26 (26): 165-188, 2009
• 3 엄철준, "최적자산배분이론의 유용성에 관한 연구" 한국산업경제학회 16 (16): 2-26, 2003
• 4 박영석, "국내 생보사의 자산운용 현황과 전략" 한국리스크관리학회 18 (18): 25-56, 2007
• 5 Drobetz, W., "Volatility bounds for Stochastic Discount Factor on Global Stock Market" Swiss Institute of Banking and Finance, University of St 2000
• 6 Michaud, R., "The Markowitz Optimization Enigma: Is Optimized Optimal?" 31-42, 1989
• 7 He, G, "The Intuition Behind Black-Litterman Model Portfolios, In Goldman Sachs Investment Management Research" 1999
• 8 Markowitz, H., "Portfolio Selection" 77-91, 1952
• 9 Chan, L., "On Portfolio Optimization : Forecasting Covariances and Choosing the Risk Model" 12 : 937-974, 1999
• 10 Merton, R., "On Estimating the Expected Return on the Market : An Exploratory Investigation" 8 : 323-361, 1980
• 11 Jorion, P., "International Portfolio Diversification with Estimation Risk" 58 : 259-278, 1985
• 12 Black R. Litterman, "Global Portfolio Optimization" 28-43, 1992
• 13 김진호, "Conditional VaR 모형을 사용한 최적자산배분에 관한 연구: 평균-분산 모형과 비교" 한국증권학회 32 (32): 133-164, 2003
• 14 Pitts, A., "Asset Allocation and Market Opinions: the Synthesis(part 2)" 6 : 12-21, 1997
• 15 Pitts, A., "Asset Allocation and Market Opinions: the Synthesis(part 1)" 4 : 14-23, 1997