Research On Efficiency Evaluation Models And Applications For The Multi-period Portfolio

The extant literature on the evaluation of portfolio focuses mostly on that of static portfolio,while,both theoretically and empirically,rarely on dynamic one.By far,the limitations can be boiled down to four categories:(1)The commonly used framework for the evaluation of portfolio,which takes risk as input and return as output,is not fit for the evaluation of the multi-period portfolios.(2)The ongoing DEA evaluation method for multi-period portfolio which simply applies risk-return framework separately on every period of the multi-period portfolio has overlooked the wealth connections between one period and another.(3)Most of the traditional DEA models assume production possibility set with technological convexity.For some portfolio evaluation like ones considering high-order moment,however,this assumption cannot be justified,and thus the convergence of DEA in theoretical sense cannot be guaranteed.(4)Since extant studies just keep an eye on efficiency and rankings,the driving factors of portfolio efficiency are far less understood.All of the limitations mentioned above call urgently for proper DEA evaluation method for multi-period portfolios which attaches great significance in guiding investment practices for both individual investors and fund managers.This is the motivation of our paper.We offer here feasible alternatives for evaluating multi-period portfolios.Specifically,under the framework of DEA theory and the theory of multi-period portfolio optimization model,we have conducted the multi-period DEA model with diversification,multi-period linear DEA model,and multi-period FDH model,fitting to portfolios with different characteristics of data respectively.We have also made simulation analysis and empirical analysis to validate our models.In general,we arrive at the following findings:First,we conduct multi-period DEA model with diversification considerations.After specifying the input-output process of multi-period portfolio,we define the efficiency of multi-period portfolios with different orientations basing on the distance of the portfolio to be evaluated and the real frontier.Next,the stochastic connection conditions for multi-period portfolio are deduced by clarifying the changing process of wealth.And then,we replace these conditions with their first moment and secondary moment constraints.Following this process,we can always get equivalent and certain constraints to that stochastic connection conditions and further conduct two-period certain model with diversification if the return s on assets satisfy continuous distribution conditions.The convergence of that model can always be theoretically justified.Second,we conduct multi-period linear DEA model.Different from how we conducted multi-period DEA model with diversification consideration where we directly replace the stochastic connection conditions with their first moment and secondary moment,we adopt another practice.That is,we linearize the first moment and secondary moment conditions,and get our multi-period linear DEA model.When the returns on assets satisfy continuous distribution conditions,we build a more feasible two-period DEA model for portfolio evaluation.The convergence of this model can also be guaranteed.Third,we conduct multi-period FDH model.Under the framework of mean-variance and high moment,we,basing on the changing process of wealth,deduce the required connection conditions for the portfolio to be evaluated in the multi-period evaluation model.Then,the first moment,secondary moment,and higher moment constraints are introduced to take the place of the original stochastic connection conditions.Besides,further discussions are made on the consistence of the efficiencies acquired from FDH and the real efficiencies,which meanwhile reflected the convergence of multi-period FDH model.Finally,we use the above-mentioned models to make dynamic analyses on the performance of open-ended funds in China,models chosen by the data characteristics of different types of funds,comparisons made on the consis tence and discrepancy of evaluating results,and analyses done on the rationality and practicability of our models.In addition,diving factors are examined in our paper.We choose the managers' abilities of security selection and timing,and fund size as explanatory variables to see on which degree they exert their influences on f und efficiency.Robust tests are made by using sample from different time period.