A Forecasting Model for Fund Demand of the Securities Finance Industry

• Main Author: 孫榕麋
• Other Authors: 范志強
• Format: Others
• Language: zh-TW
• Published: 1996
• Online Access: http://ndltd.ncl.edu.tw/handle/32041059156867247503

碩士 === 國防管理學院 === 資源管理研究所 === 84 === 　　According to the Regulations Governing Administration of Securities Finance Industry, the securities finance industry functions as a specialized finance system. It''s main business is to supply loans of fund and securities to investors and brokerage houses for margin transactions purpose. As past experience has shown, heavy trading for an active market will normally bring about tens of billions of fund demand, whereas light trading for an dull market will typically suppress margin transactions and thus only billions of fund demand to occur. Therefore, there if a highly varying character in the fund demand of stock market. In order to sufficiently supply fund for the ever changing market demand, it is an urgent task for the securities finance industry to make an accurate and efficient forecast of its fund demand especially under the current high competition in the industry. 　　In the thesis, three methodologies, ARIMA, state space and vector autoregression, are employed to build forecasting models for fund demand of the securities finance industry. The ARIMA model, though being widely used in various fields because of its simplicity since its development by Box and Jenkins in 1976, unfortunately focuses only in analysis of an univariate time series and fails to take into account the interrelationship between the target variables and other variables. The state space model, in addition to its ability to incorporate the effects of other variables and cover the above deficiency in ARIMA model, can be characterized by it flexibility in using the expectation of a variable as a state variable. However, the vector autoregressive model, developed by Sims in 1980 with the aim of focusing on forecasting ability without emphasizing the analysis of casual relationship between variables, allow the implementation of empirical analysis to be carried out by taking both the same number of variables and length of lags for each variable in each equation. 　　The empirical analysis employs daily data from October 1, 1994 to March 31, 1995. To evaluate the forecasting ability of each model, five statistics are used as comparing criterias for accuracy. To evaluate the cost saving ability of each model, the concept of average cost which is proposed by Charles M. Ermer in 1991 is also introduced as a comparing criteria for cost saving performance. The empirical analysis concludes that the state space model is the best model in forecasting the fund demand of the securities finance industry.