Publications
Detailed Information
Empirical Research on Common Preferred Stock Spread Index and its Application on Investment Strategy : 우선주식과 보통주식 수익율 스프레드 지수를 이용한 금융투자 실증 연구
| DC Field | Value | Language |
|---|---|---|
| dc.contributor.advisor | 장우진 | - |
| dc.contributor.author | 이창주 | - |
| dc.date.accessioned | 2019-10-21T01:57:11Z | - |
| dc.date.available | 2019-10-21T01:57:11Z | - |
| dc.date.issued | 2019-08 | - |
| dc.identifier.other | 000000157199 | - |
| dc.identifier.uri | https://hdl.handle.net/10371/161929 | - |
| dc.identifier.uri | http://dcollection.snu.ac.kr/common/orgView/000000157199 | ko_KR |
| dc.description | 학위논문(박사)--서울대학교 대학원 :공과대학 산업공학과,2019. 8. 장우진. | - |
| dc.description.abstract | 최근 다양한 금융위기 이후, 시장의 위험을 측정 하고 안전한 수익을 낼 수 있는 금융 상품의 개발의 토대가 될 수 있는 금융시장 분석의 중요성은 더욱 강조 되고 있다. 특히 다양한 금융 시장의 특성을 활용한 지표들을 중심으로 시장의 상태와 미래 수익률 과의 관계를 설명 하였는데 대표적인 위험 지표로는 TED와 VIX가 있고 시장 가치 평가 지표로는 Price to Earning ratio, Price to Book ratio, CAPE (Cyclically Adjusted Price to Earning ratio), Price to Operational Earning ratio 등이 있다. 사전의 연구들은 이러한 지표들을 통하여 현재 주식시장의 상태를 측정하고, 정확한 측정을 통하여 미래 수익률을 설명 하였다. 많은 연구들이 기존에 존재하고 있는 시장 지표들을 변형하고 발전 시켜 미래 수익률을 보다 더 잘 설명 하는 지표를 개발 한 것과 달리 본 논문에서는 그동안 잘 쓰이지 않았던 주식 시장의 우선주를 활용하여 시장의 미래 수익률을 설명 할 수 있는 지표를 개발하고 이를 검증 하였다. 먼저 본 논문에서는 주식 시장에서의 우선주를 가지고 있는 본주들의 누적 수익률과 우선주들의 누적 수익률의 편차를 이용하여 CPS-index(Common Preferred Spread index)를 개발하였다. 이 CPS-index는 현재 주식시장에서 우선주를 가지고 있는 보통주 들이 우선주들에 대비하여 과거보다 얼마나 더 올랐는지 혹은 떨어졌는지를 알려준다. 본 논문에서는 CPS-index 가 높을 때 즉 본주들이 우선주보다 더 많이 올랐을 때를 시장이 고평가 되어 있다고 판단하고 CPS-index 가 낮을 때 즉 본주들이 우선주보다 더 적게 올랐을 때를 시장이 저평가 되어 있다고 판단하여 이를 미래 장기 수익률과의 상관관계 분석을 실시하였다. 그 결과 CPS-index는 미래 장기 수익률과 매우 높은 음의 상관관계를 가지고 있었고, CPS-index가 높을 때 공매도 하고 CPS-index가 낮을 때 매수하는 투자전략을 제시하여 실증적으로 이 투자 전략을 활용 시 높은 수익률을 낼 수 있다는 것을 보여주었다. 또한 granger causality test 와 신경망 분석을 활용하여 CPS-index가 미래 수익률을 보다 정확하게 예측 할 수 있다는 것을 보여주었다. 두번째로는 CPS-index를 기존의 미래 수익률을 설명할 수 있는 지표들과 비교 분석하여 기존 지표들과 함께 사용하였을 때 미래 수익률을 보다 정확하게 설명 할 수 있는지를 보여주었고, 또한 parameter tuning을 통하여 어느 정도의 과거 데이터를 활용해야 보다 정확한 CPS-index를 만들 수 있는지 제시하였다. 이 연구를 통하여 CPS-index를 기존의 지표들인 Price to Earning ratio, Price to book ratio, Price to Operational Earning ratio 등과 같이 활용하였을 때 시장의 미래 수익률 기존의 지표들만 활용하였을 때 보다 월등히 더 잘 설명 할 수 있음을 보였고 이를 통하여 CPS-index의 시장 평가 지표로의 가능성을 확인하였다. 마지막으로 우선주와 우선주를 가지고 있는 본주를 활용한 Pairs Trading 방법을 제시하였다. 기존의 방법인 같이 움직이는 두 주식을 활용한 Pairs Trading 방법과 달리 우선주와 본주를 활용하면 pair를 찾는데 필요한 시간과 계산을 줄일 수 있고, Pairs Trading 운영기간 동안 두 주식이 다른 방향으로 움직이는 위험을 줄일 수 있다고 판단하였다. 본 논문에서는 각 금융시장 상황에 따른 최적의 Pairs Trading 방법을 찾는 것이 중요하다고 생각이 들어 이를 검증하는데 초점을 맞추었고, 그 결과 우선주와 본주를 활용한 Pairs Trading 방법이 기존의 Pairs Trading 방법보다 수익률은 좋지 않았지만, 위험 대비 수익률인 Sharpe Ratio는 월등히 좋다는 것을 검증하였다. 결과적으로 본 논문에서는 기존에 연구가 진행되지 않았던 주식 시장에서의 우선주를 가지고 있는 본주와 우선주의 움직임을 분석하여 현재 주식시장의 상태를 측정하고 장기 미래수익률을 설명 할 수 있는 지표를 개발하여 투자자들이 보다 안전하게 투자할 수 있는 투자 전략을 제시였고, 안전한 투자전략인 Pairs Trading에 적용하여 pair를 찾는데 필요한 시간과 돈을 절약하고 위험 대비 수익률을 높일 수 있는 방법을 제시하였다. 이를 통하여 우선주의 관한 연구가 활발히 진행될 것이며 일반 투자자들이 쉽게 이해하고 활용할 수 있는 투자 지표와 투자방법을 제공하였다. | - |
| dc.description.abstract | After various financial crises, the importance of financial market analysis for financial risk management has been emphasized. To minimize the risk of losing money from unforeseen financial crises, it became critical to develop a market index that could both evaluate the current financial market and explain the future market return. In addition, it became imperative that the investment strategy could continue to make profit during the financial crises. Many researchers have tried to evaluate the financial market and explain the future market return with various risk index such as the VIX (CBOE volatility index) index, and TED index (the spread between three-month LIBOR interest rate and three-month US treasury bill interest rate) and valuation ratios such as Price to earning ratio, Price to book ratio, CAPE( Cyclically Adjusted Price Earning Ratio) and price to operational earning ratio. Previous studies attempted to explain future market returns better by adapting existing indexes and ratios. However, in this dissertation, we introduce a new market index called; Common Preferred Spread index (CPS index), which was empirically tested to confirm that it could not only evaluate current stock market condition but also has explanatory power for the future market return. First, this dissertation explains the CPS-index using the spread return between common and preferred stock pairs and shows that CPS-index has explanatory power for long term market return. We observed that common stocks are more sensitive to the market condition than preferred stocks so the CPS-index tends to oscillate according to market conditions. The future realized market return increases when CPS-index is low, and vice versa. We present that there is an inverse relationship between CPS-index and the future realized return of S&P500 index. By comparing the fitting validity of statistical models including correlation analysis and linear regression between the future realized return and each of CPS-index, VIX index, TED index, CAPE ratio and S&P500 index, we confirm the superior power of CPS-index to explain the future realized return. We developed a trading strategy based on CPS-index and assessed how to enhance the predictive power of CPS-index through stepwise regression, Granger causality test and neural network method. Second, we conduct an empirical analysis on CPS-index in comparison with currently existing valuation ratios such as the price to earning ratio, price to book ratio, and price to operational earning ratio. The multivariate regression method is applied to test whether adding CPS-index as an independent variable significantly increases the explanatory power of regression for the future market return. According to the test results of every multivariate regression model, the CPS-index as an independent variable has the most market predictability power among other benchmark independent variables of regression. In addition, we also discovered the optimal parameters to use the CPS-index. Lastly, a new Pairs Trading strategy is proposed, using common stocks and preferred stocks. Unlike the traditional method of Pairs Trading, which is based on two different stocks that were moving together in the past, a common stock and its preferred stock as a pair are used in Pair Trading. This new method could reduce the risk of losing money from the traditional method of Pairs Trading. We explain through the test results of every portfolio that this new method of Pairs has the highest Sharpe Ratio. | - |
| dc.description.tableofcontents | Chapter 1. Introduction 1
1.1 Resarch Motivation and Purpose 1 1.2 Theoretical Background 5 1.3 Research Overview 10 Chapter 2. Development of Common Preferred Spread Index 12 2.1 Introduction 12 2.2 Related Literature 16 2.3 Method 20 2.3.1 Spread between Common Stocks and their Preferred Stocks 20 2.3.2 CPS Index and Future Realized Market Return 22 2.3.3 Multivariate Regression Analysis 26 2.3.4 Stepwise Regression and Granger Causality Test 28 2.3.5 Neural Network and Prediction 29 2.4 Data 30 2.5 Empirical Results 31 2.5.1 Correlation and Univariate Regression Analysis 33 2.5.2 Investment Strategy with CPS-index 42 2.5.3 Multivariate Regression Analysis Results 47 2.5.4 Stepwise Regression and Granger Causality Test Results 52 2.5.5 Neural Network Prediction Results 56 2.6 Conclusion 61 Chapter 3. Empirical Analysis of Common Preferred Spread Index 64 3.1 Introduction 64 3.2 Method 67 3.2.1 Empirical Analysis of Spread between Common Stocks and their Preferred Stocks 67 3.2.2 Empirical Analysis of CPS Index and Future Realized Market Return. 69 3.2.3 Empirical Analysis of Multivariate Regression Analysis 72 3.2.4 Optimal Starting Point of CPS Index 75 3.3 Data 78 3.4 Empirical Results 79 3.4.1 Empirical Results of Correlation and Univariate Regression 80 3.4.2 Investment Strategy with CPS Index and other Valuation Ratios 86 3.4.3 Parameter tuning and Granger Causality Test 97 3.5 Conclusion 105 Chapter 4. Empirical Analysis of Pairs Trading Using Preferred Stocks. 108 4.1 Introduction 108 4.2 Background and Literature Review. 111 4.3 Methodology 115 4.3.1 Pairs Formation 115 4.3.2 Trading Strategy and Periods 117 4.3.3 Excess Return Computation 118 4.4 Empirical Results 121 4.4.1 The Whole Periods 122 4.4.2 Pre Crisis 125 4.4.3 Subprime Crisis 128 4.4.4 European Crisis 131 4.4.5 Post Crisis 134 4.5 Conclusion 137 Chapter 5 Concluding Remarks 140 5.1 Summary and Contributions 140 5.2 Limitations and Future Work 146 References 147 Abstract (in Korean) 151 List of Tables Table 2.1 Pearson Correlations for each index. 31 Table 2.2 Univariate regression for the CPS, VIX, TED and S&P500 index 36 Table 2.3 Average future realized return of r^h and r^l. in equations (2.9) and (2.10), respectively. 45 Table 2.4 Sharpe Ratio for the r_(i,avg)^h and r_(i,avg)^lfor different time horizon. 45 Table 2.5 r^h and p^h for the different upper threshold and for the different time horizon 46 Table 2.6 r^h and p^h for the different upper threshold and for the different time horizon 46 Table 2.7 Step-wise regression for CPS index. This table summarizes step-wise regression results in equation (2.11). 48 Table 2.8 Variance Inflation Factor for each index in equation.(2.11) 49 Table 2.9 C_p and stepwise regression results for the multivariate regression in equation (2.11) 49 Table 2.10 F-test for multivariate regression adding CPS-index to the model. 50 Table 2.11 Step-wise regression for CPS index. 54 Table 2.12 Granger Causality test of CPS-index for the i month ahead realized return τ is the maximum time lag in equation (2.12) 55 Table 2.13 Neural network for the CPS, VIX, TED and S&P500 index. This table summarizes the test set of neural network results 55 Table 2.14 Confusion matrix for the sign of prediction periods in Neural Netwrok with CPS-index 59 Table 3.1 Pearson Correlations for each index. 79 Table 3.2 Univariate regression for the CPS, PER, PBR, CAPE and S&P500 index. This table summarizes single regression results in equation(3.8). 83 Table 3.3 Variance Inflation Factor for each index in equation.(3.12) 86 Table 3.4 Adjusted R-square for multivariate regression using the CPS, PER, PBR, CAPE and S&P500 index as independent variables and the dependent variable is i month ahead realized return in equation (3.9). 90 Table 3.5 Adjusted R-square for multivariate regression using the CPS, PER, PBR, CAPE and S&P500 index as independent variables and the dependent variable is i month ahead realized return in equation (3.10). 91 Table 3.6 Adjusted R-square for multivariate regression using the CPS, PER, PBR, OPER and S&P500 index as independent variables and the dependent variable is i month ahead realized return in equation (3.11) 92 Table 3.7 Multivariate regression for the CPS, PER, PBR, OPER and S&P500 94 Table 3.8 F-test for multivariate regression adding CPS-index to the model 96 Table 3.9 Adjusted R-square for univariate regression using the m-year CPS index as independent variables and i month ahead realized return as dependent variable is in equation (3.18) 99 Table 3.10 Adjusted R-square for univariate regression using the 7-year CPS, PER, PBR, OPER and S&P500 index as independent variables and i month ahead realized return as dependent variable is in equation (3.18). 100 Table 3.11 Adjusted R-square for univariate regression using the 8-year CPS, PER, PBR, OPER and S&P500 index as independent variables and i month ahead realized return as dependent variable is in equation (3.18). 101 Table 3.12 Granger Causality test of 7-year CPS-index for the i month ahead realized return τ is the maximum time lag in equation (3.19) 102 Table 4.1 Excess return of Pairs Trading for whole periods 122 Table 4.2 Excess return of Pairs Trading for Pre-crisis periods 125 Table 4.3 Excess return of Pairs Trading for Subprime-crisis periods 128 Table 4.4 Excess return of Pairs Trading for European-crisis periods 129 Table 4.5 Excess return of Pairs Trading for Post-crisis periods 134 List of Figures Figure 2.1 Plots of CPS,S&P500,C, and P 33 Figure 2.2 Scatter plot of CPS and 1, 2, 3, 4 year ahead realized return 35 Figure 2.3 Correlation of CPS and i-month ahead realized return 41 Figure 2.4 R-square of CPS included model and excluded model 51 Figure 2.5 Accuracy rate of confusion matrix of neural network with CPS-index 60 Figure 3.1 Correlation of Each index and i-month ahead realized return 82 Figure 3.2 Adjusted R-square of each univariate regressions and multivariate regressions 93 Figure 3.3 Adjusted R-square of each univariate regressions and multivariate regression 98 | - |
| dc.language.iso | eng | - |
| dc.publisher | 서울대학교 대학원 | - |
| dc.subject | Preferred Stocks | - |
| dc.subject | Common Stocks | - |
| dc.subject | Pairs Trading | - |
| dc.subject | Predictive Index | - |
| dc.subject | Market valuation Index | - |
| dc.subject | Spread Return | - |
| dc.subject | Common preferred spread index | - |
| dc.subject.ddc | 670.42 | - |
| dc.title | Empirical Research on Common Preferred Stock Spread Index and its Application on Investment Strategy | - |
| dc.title.alternative | 우선주식과 보통주식 수익율 스프레드 지수를 이용한 금융투자 실증 연구 | - |
| dc.type | Thesis | - |
| dc.type | Dissertation | - |
| dc.contributor.department | 공과대학 산업공학과 | - |
| dc.description.degree | Doctor | - |
| dc.date.awarded | 2019-08 | - |
| dc.identifier.uci | I804:11032-000000157199 | - |
| dc.identifier.holdings | 000000000040▲000000000041▲000000157199▲ | - |
- Appears in Collections:
- Files in This Item:
Item View & Download Count
Items in S-Space are protected by copyright, with all rights reserved, unless otherwise indicated.