The determination of the phase shift and the light shift in our optically p
umped Cs beam is reported here. We show that an error on the phase differen
ce value determined by our fit method, without beam reversal, can be induce
d if the light shift is not taken into account. If the experimental data ar
e corrected for the light shift, the results of the fit method are in very
good agreement with the beam reversal results. This allows us to reduce the
related uncertainty in the accuracy budget of our standard. The resulting
overall accuracy of the standard is now estimated to be 6.3 x 10(-15).