The method of forecast output coincidence used to determine if sectors are demand-sided or supply-sided in an input-output framework mixes two effects, the structural effect (choosing between demand and supply side models) and the effect of an exogenous factor (final demand or added-value). The note recalls that another method is possible, the comparison of the stability of technical and allocation coefficients, generalized by the biproportional filter: if for a sector, after biproportional filtering, column coefficients are more stable than row coefficients, then this sector is declared as not supply-sided (but one cannot decide that it is demand-sided anyway), and conversely.