The Effect of Public Sector on the Financial Sector:An NN Approach in a View of Complexity

摘要

In the existing literature there are many works which use NN for forecasting the financial market. Generally, these are centred on technical analysis and/or endogenous variations. A different approach is based on the use of "fundamentals". In this latter approach, the effect of fiscal variables between the fundamentals is disregarded. We considered both technical-endogenous and fundamental variables at the same time for the forecasting. Particularly, we used the payment due by taxpayers (estimating the daily's burden) as explicative for differences in the level of Comit30 index. The sensibility of financial behaviour to the tax burden seemed to be characterised by a '"threshold" effect; that is to say a complex - non-linear behaviour. Final forecasting reached is quite good. It is possible to demonstrate that the correct consideration of exogenous shocks, together with endogenous behaviour of the financial market, could be the key passage for a correct use of neural networks in complex systems. More, the importance of a proper non-linear consideration of the interrelation between fiscal and financial sectors was highlighted.