Some remarks on the Pariser-Parr-Pople method