Quelle valeur ajoutée pour les banques françaises ?

This paper examines critically the current method to measure bank value added in national accounts and propose a new method adjusted from risk premium, which better fits the economic cycle. By applying this new method on an original quarterly database for French banks, we show that on average the banking production is overestimated by 50 %. In this way, the average share of the banking sector in the French GDP between 2003 and 2015 decreases from 2,9 % to 2 %. Furthermore, an empirical analysis based on ARDL models reveals that the new method is cointegrated with the economic cycle whereas the current one is not. In the end, our results put forth the evidence that the new method fits better the guidelines set out by the European Union and the United Nations.