RMI Research Seminars
August – September 2012
On 10 August, Princeton University’s Prof. Fan Jianqing discussed his paper on the correlation between asset returns and their volatility at RMI Research Seminar titled “Leverage Effect Puzzle”. He argued that a natural estimate of such a parameter is to use correlation between the daily returns and the changes of daily volatility estimated by highfrequency data. The puzzle is that such an estimate yields nearly zero correlation for all assets such as SP500 and Dow Jones Industrial Average that were tested. To solve the puzzle, he and his collaborators developed a model to understand the bias problem. The asymptotic biases involved in high frequency estimation of the leverage effect parameter were derived and they quantified the biases due to discretization errors in estimating spot volatilities, biases due to estimation error, and the biases due to market microstructure noises. Prof. Fan and his collaborators were able to propose novel bias correction methods for estimating the leverage effect parameter. The proposed methods are convincingly illustrated by simulation studies and several empirical applications.
Separately, on 28 September Prof. Ruey S. Tsay of University of Chicago presented at RMI’s Research Seminar his joint research entitled “Marketbased Credit Ratings” a methodology for rating the creditworthiness of public companies in the U.S. from the prices of traded assets. Their approach used asset pricing data to impute a term structure of risk neutral survival functions or default probabilities. Firms were then clustered into ratings categories based on their survival functions using a functional clustering algorithm. This allowed all public firms whose assets were traded to be directly rated by market participants. For firms whose assets were not traded, he showed how they can be indirectly rated through the use of matching estimators. During the seminar, he also illustrated how the resulting ratings can be used to construct loss distributions for portfolios of bonds. He and his collaborators concluded that their approach has the advantages of being transparent, computationally tractable, simple to implement, and easy to interpret economically.
RMI Pedagogical Research Lectures
August – September 2012
At the Pedagogical Research Lecture held from 16 to 17 August, Prof. Fan Jianqing of Princeton University talked about “Econometrics of Large Portfolios”. The proposed methods were convincingly illustrated by simulation studies and several empirical applications. During his tutorial, he covered the problems associated with large portfolios, namely, (a) Risk Assessment for Large Portfolios, (b) Testing CAPM using Large Number of Assets, and (c) False Discoveries in Mutual Fund Performance. He introduced the methods of estimating large portfolio risks and their uncertainty in risk estimation, the techniques for validating CAPM when the number of assets is larger than the length of time series, and revealed the relation between the anomaly of financial market and longshort strategies. He also addressed the estimation of the proportion of skilled, zeroalpha, and unskilled mutual funds and the persistence of fund performance. These topics reflected the recent developments of highdimensional statistics in finance. The methods of estimating large covariance matrices which play prominent roles in these problems were also addressed. The newly developed techniques were illustrated by both simulation and empirical studies.
About one month later, Prof. Ruey S. Tsay from University of Chicago, delivered two separate Pedagogical Research Lectures: (a) Principal Component Analysis and Its Extension to Volatility Analysis, and (b) Unit Roots, CoIntegration, ErrorCorrection Models and Their Applications. In the first lecture, he explained that principal component analysis (PCA) is a powerful statistical tool for dimension reduction and has been used extensively in many scientific areas. In his lecture, he showed some applications of PCA and pointed out some unsatisfactory use of PCA in the literature, especially for volatility modeling. He then proposed a generation of PCA to volatility modeling, which he called principal volatility components (PVC) analysis. The generalization was achieved by use of generalized covariance matrix and generalized lag covariance matrix. Finally, he applied PVC to construct portfolios that do have ARCH effects. Real examples were used to demonstrate the application of PVC. His second lecture reviewed properties of univariate and multivariate unitroot time series. He discussed the limiting distributions of some statistics used in unitroot and cointegration tests. For cointegrated systems, he considered errorcorrection models (ECM) and their estimation. Implications of cointegration were also addressed. Real macroeconomic variables and interest rates were used to demonstrate cointegration and the estimation of ECM models. Some applications of cointegration were also given. Finally, he also discussed the advantages and limitations in practice of using cointegration.
