A MULTIPLE NONLINEAR REGRESSION MODEL FOR ESTIMATING INPUT-OUTPUT VALUES WITH PARTIAL INFORMATION

The paper aims to find an effective method for estimating input-output (I-O) values in the target year when the original I-O table is incomplete. By examining the input and output relationship hidden in the RAS method, we find a multiple nonlinear regression model embedded in this method to achieve this well. Specifically, each element value in the target I-O table is closely related to its corresponding element in the original I-O table and the input and output growth rates of each sector. The performance of the proposed model is validated by extensive experiments based on simulated and real I-O datasets.