摘要：We exploit scanner data to track payment choice for grocery purchases for a large panel of households over three years. We particularly focus on the role of expenditure size in determining payment choice. While the use of a long panel for these purposes is novel, the introduction of household fixed effects has little effect on our estimates.
