Bank output calculation in the case of France: what do new methods tell about the financial intermediation services in the aftermath of the crisis?

Since the onset of financial crisis, the System of National Accounts method for measuring the value added in the banking sector has become subject to criticism. Some authors argue that the value added by banks should be the residual net interest income after subtracting the required term and risk premiums on loans and deposits. For the first time, we apply this method to evaluate bank output for France for the period 2003 to 2012. First, we show that on average, using the traditional method, bank output is overestimated by between 31% and 74%. This overestimation is especially pronounced in times of financial stress. Second, we establish that the proposed new method is robust to the choice of various reference rates. Third, we find negative FISIM (Financial Intermediation Services Indirectly Measured) on deposits from 2009. Finally, we check the existence of a single or multiple structural unknown breaks in the long run relationship between retail interest rates and driving market reference rates. We find existence of break dates that are coincident with negative FISIM on deposits. We explain this result by a change in banking behavior that may result from the new banking regulation on liquidity and from banks' adaptation to "near" zero interest rate policy