Neurofuzzy Portfolio Selection Policies in A Small National Assets Model of Japan (Special Issue of the Faculty of Information Management; Studies Presented to Prof. Susumu Tabuchi on his Retirement and In Commemoration of the Establishment of Department of Business Information and Department of Finance)
幸雄 伊藤 · Institutional Repositories DataBase (IRDB) · 2006
This paper proposes the portfolio selection in National assets of Japan subject to national balance model using neurofuzzy approach. National assets includes the wide range assets from natural resources to financial assets so that asset selection has to be differentiated by their properties and attributes. Having divided by the individual group, the most influential variable has to be found by selecting the weights on hidden layer by the neural network between inputs and outputs. Three expected risks between the actual values of grouping assets, and the mean, desired and estimated values will be investigated and compared by minimizing their risks subject to the national balance model including some asset variables. The constraint of the balance model is constructed and estimated from data for the national flows of funds table. It is stated there exists the difference and characteristics between national asset selection and the corporation by using an illustrative application to national balance model of Japan.