Font Size: a A A

Comparisons For Several Kinds Of Risk Measures

Posted on:2018-05-19Degree:MasterType:Thesis
Country:ChinaCandidate:Y H YaoFull Text:PDF
GTID:2359330518977283Subject:Finance
Abstract/Summary:PDF Full Text Request
Under the background of economic globalization and financial integration, the stock market is playing a more and more important role,while inevitably,it also brings large amounts of risk. Chinese financial market, with a high development rapid,affected by the invest strategies of external developed economic entities and the internal financial environment. Therefore, it is necessary to research several popular risk measures and their applicability in the Chinese stock market.Based on the previous research, this thesis is trying to analyse the power of risk discrimination of Standard Deviation, VaR (Value at Risk), CVaR (Conditional VaR),convex entropic risk measure, iso-entropic risk measure, the convex risk measure and iso-entropic risk measure with modification of relative entropy, HMCR (Higher moment coherent risk measure), and Min Alpha from the theoretical and empirical perspective.On the theoretical side, the mainly factors to be compared are convexity, the volume of information which is used to measure the risk, magnitude for these risk measures,relationship between these risk measures and stochastic dominances, and the coherent risk axioms. On the empirical side, the Spearman rank test is used to test the risk discrimination of these risk measures, coupled with the portfolio optimization models based on these risk measures are used in Chinese market. The latter is much more important, it analyses the condition of single-period, multi-period and different markets group,which constitute by "bull market" and "bear market". This research is based on the log return rate of closing price of SSE 50 index and its component stocks. Main results are as follows:On the theoretical side, Standard Deviation is not satisfied with monotonicity, VaR is not satisfied with convexity, convex risk measure, iso-entropic risk measure, HMCR and Min Alpha are coherent risk measures, which have the advantage of convexity. These risk measures can be expressed in the form of Kusuoka representation, the difference relies on the weight function. HMCR, convex entropic risk measure and iso-entropic risk measure utilize the whole information to measure the risk, which have high risk measure risk values correspondingly, while VaR and CVaR are only utilize local information, one point and a tail interval respectively. From the perspective of stochastic dominance,Standard Deviation is not consistent with first-order stochastic dominance, VaR is consistent with first-order stochastic dominance, CVaR is consistent with second-order stochastic dominance, while convex risk measure, iso-entropic risk measure, and HMCR(p=n) have consistent with stochastic dominances of almost all the orders.Therefore,this thesis speculate that these risk measures,which are satisfied with coherent axioms, convexity, utilizing more information, higher risk measure value, and higher stochastic dominance, are consistent with strong power of risk discrimination.On the empirical side, according to the Spearman rank test of these risk measures under four different confidence level, these risk measures with higher stochastic dominance are more likely to have the more power of risk discrimination, this result is also verified by the portfolio optimization, i.e. the power of risk discrimination of HMCR(p=3), Min Alpha, convex entropic risk measure, iso-entropic risk measure are higher than CVaR, VaR and Standard Deviation, this result is affected by the stock market, and more significant in the stationary period. From the result of the portfolio optimization of the other conditions, the risk discrimination become more powerful in the multi-period condition, which eliminates the influence of abnormal data in the stock market, and becomes more consistent with the theoretical expectation. In addition,according with the result of five kinds of market groups,when investors are attempt to invest according to the in-sample portfolio weights, the positive investment strategy based on several risk measures are more effective in the condition where the market is likely to turn to bear market, while the negative investment based on market index is more applicative in the condition where the market is likely to turn to bull market or the stock market has high volatility.
Keywords/Search Tags:risk measures, coherent axioms, consistence with stochastic dominance, the power of risk discrimination, Spearman rank test, portfolio optimization, investment strategy
PDF Full Text Request
Related items