Using micro price data, we empirically evaluate what price-setting moments are
informative for monetary non-neutrality. Kurtosis of price changes has none, or
even a negative association, contrary to the notion in the literature. Kurtosis over
frequency of price changes is informative about monetary non-neutrality but only
because the frequency has a strong negative association. Neither pricing moment
is a sucient statistic, explaining little variation in monetary non-neutrality. We
show that menu cost models can match empirical price responses, in particular
the negative association between kurtosis and monetary non-neutrality. Menu cost
models predict a positive relationship as posited in the literature only if random
menu costs are the source of excess kurtosis and raise the \Calvo-ness" of the model.