The Funds Investment Style And Style Varying Analysis

Fund investment style involves fund investment strategies and embodies investment objectives. It is the core of the research on modern fund investment performance evaluation, which primarily studies the asset allocation within a given period of time. Due to the nonlinear nature of fund investment style research, researchers were not in a position to conduct hypothesis test on the parameter estimates for existing models. Moreover, the effect of collinearity has not been taken into consideration in constructing the benchmark of investment style. This dissertation introduces K-S-W simulation parameter estimation to expand the existing methodology for fund investment style research, and offers an excellent solution to the problem of testing on the estimates for parameters on the boundary of investment style. Starting from the features of the benchmark of investment style, this dissertation evaluates and compares the benchmarks of investment style through simulated empirical study, and thus constructs a return-based investment style model, which is more suitable for Chinese funds market.The empirical evidence for fund investment style in this paper suggests that closed funds are mostly big-cap balanced investment styled, while open funds are overwhelmingly big-cap growth investment styled. Furthermore, this dissertation takes sub-samples according to the bullish/bearish market conditions, and analyses the investment styles of the two kinds of funds. The findings herein shows that there is striking difference in the investment styles for bull and bear market conditions, funds preferring big-cap value investments in contrast to big-cap growth investments in a bearish market.Different from previous studies on fund investment style varying in the research paradigm of linearity on the basis of efficient market theory, this dissertation extends the investment style varying research to the nonlinear realm through constructing a Kalman Filter model with constraints, and employing nonlinear optimization theory to empirically test fund investment style varying in Chinese securities market. After comparing the estimates of the model parameters for different variations of investment styles, this dissertation finds that the constrained Kalman Filter model performs better than previous models for investment style varying. Using self-constructed investment style varying indexes and statistical tests on investment style varying, this dissertation indicates that there exists an obvious change point of investment style varying for both closed funds and open funds, and over97%of closed funds and more than57%of open funds exhibit multi-point investment style varying behavior. Drawing on the statistics of the month where the change point of investment style occurred, this paper indicates that clustering phenomena exist in adjustments of fund investment style, which fills the blank of behavioral finance on funds’herding behavior.The contents and main innovations are as follows:(1) This paper proposes a return-based fund investment style analysis model, which is more suitable for Chinese fund market, and extends the conditions for fund investment style model. For the first time, the properties of parameter estimation for structural models with different strengths are deduced--for the weak structural investment style model, the semi-strong structural investment style model and strong structural investment style model. Subsequently, this dissertation puts forward a general parameter test method for investment style varying model. Furthermore, this paper introduces a simulation parameter test method, namely K-S-W method, for the investment style varying model. This solves the difficult problem of parameter testing on investment style varying model, whose parameter is on the boundary. This method also expands the investment style analysis to the field of nonlinearity. Based on Monte Carlo simulation of data samples, this dissertation proposes an original investment style benchmark selection procedure, and constructs an investment style research benchmark for the Chinese securities market.(2) On the basis of re-conceptualizing and redefining the concept of investment style varying, this paper propose a new investment style variation index, namely Style Varying Score (SVS), which reflects the comprehensive varying degree of investment style. Comparing to former investment style varying research index, the SVS index takes into consideration the autocorrelation effects of investment style varying coefficients, thereby improving the accuracy for characterizing the investment style varying behavior. Empirical evidence shows that the SVS index is superior to other similar indexes.(3) The statistical test framework and procedures are proposed to analyze the investment style varying behavior, which include the common linear parameter test in correlation coefficient analysis, the single change point test, the CUSUM test and the multiple change point test in time series structural change. This dissertation creatively employs multiple dynamic structural change analysis in investment style varying analysis, and thus for the first time reveals the clustering phenomenon of fund’s investment style varying in terms of structural change. The research conclusions fill the research gap of behavioral finance research on fund herding behavior.(4) This dissertation introduces for the first time a strong structural Kalman Filter model with inequality constraints, which is extensively applied in engineering, into investment style varying research. This model, combined with the thinking of information set in nonlinear programming and the properties of Least Square estimation with boundary constraints, leads to a complete Kalman Filter model with inequality constraints for investment style varying research. This model remedies the deficiencies of ignoring the non-linear constraints in former fund investment style varying research, and expands the investment style research to the realm of nonlinearity, thereby making it possible to estimate the parameters of investment style varying in cases where parameters are constrained by boundaries. Besides, this new framework analyses the properties of the parameter estimation for the constrained and the unconstrained Kalman Filter investment style varying model, and with empirical evidence obtained from data sample simulation demonstrates that the Kalman Filter investment style varying model with constraints performs the best among all the existing fund investment style varying models.