The increasing availability of supermarket scanner data covering expenditures and prices
on a wide range of products has created new opportunities for national statistical institutes,
including the possibility of publishing more reliable indicators of monthly price changes. We
discuss and evaluate the properties of different multilateral index numbers for measuring
high frequency price changes, drawing on household scanner data. We find that use of the
Caves-Christensen-Diewert-Inklaar (CCDI) index, updated using the mean splice, is to
preferred for both theoretical and empirical reasons.