Modeling the Determination of Rate of Return in Transaction Contracts Using Artificial Neural Networks

Islamic banking includes profit and loss sharing (PLS) and transaction contracts. Transaction contracts have fixed rates of return, which in turn form a base for allocating the financial resources to PLS contracts. In Iran, the rates of return in transaction contracts are determined by Central Bank. In this research, we compute and estimate the rates of return in transaction contracts using the Multi-Layer Perceptron (MLP) Artificial Neural Networks and Radial Basis Function (RBF). This research is an extension and improvement of Dadgar and Firoozan (2012) work. Data used for algorithms is the real data gathered from manufacturing workplaces having more than 10 employees. Our results show that two networks are of good accuracy to estimate the coefficients of shadow cost function, and most of them are approximately equal in two networks.
Compared to econometric method, the proposed model has no sampling limitation. This method accounts for all of 14000 manufacturing units in 2007, and consequently the computational errors are much less than those of econometric calculations. According to the results, the estimated rate of return for transaction contracts is 15%. This rate in comparison with the prevailing rate, i.e. 12%, reflects a 20% deviation in determining rate of return, which causes undeniable costs on the economy and allocation of limited resources.