Value of information in portfolio selection, with a Taiwan stock market application illustration

•Presents a methodology for computing the value of perfect expected return information in portfolio selection.•The information is then a key reference point for judging how much to spend on research.•In a developing economy such as Taiwan, the value of information is very high.•For the required computations, this paper demonstrates a procedure that compares favorably to Cplex.

Despite many proposed alternatives, the predominant model in portfolio selection is still mean–variance. However, the main weakness of the mean–variance model is in the specification of the expected returns of the individual securities involved. If this process is not accurate, the allocations of capital to the different securities will in almost all certainty be incorrect. If, however, this process can be made accurate, then correct allocations can be made, and the additional expected return following from this is the value of information. This paper thus proposes a methodology to calculate the value of information. A related idea of a level of disappointment is also shown. How value of information calculations can be important in helping a mutual fund settle on how much to set aside for research is discussed in reference to a Taiwan Stock Exchange illustrative application in which the value of information appears to be substantial. Heavy use is made of parametric quadratic programming to keep computation times down for the methodology.