Retail payment innovations and cash usage: Accounting for attrition using refreshment samples

We exploit the panel dimension of the Canadian Financial Monitor (CFM) data to estimate the impact of retail payment innovations on cash usage. We estimate a semiparametric panel data model that accounts for unobserved heterogeneity and allows for general forms of non-random attrition. We use annual data from the CFM on the methods of payment and cash usage for the period 2010 - 12. Estimates based on crosssectional methods find a large impact of retail payment on cash usage (around 10 percent). However, after correcting for attrition, we find that contactless credit cards and multiple stored-value cards (reloadable) have no significant impact on cash usage, while single-purpose stored-value cards reduce the usage of cash by 2 percent in terms of volume. These results point to the uneven pace of the diffusion of payment innovations, especially contactless credit.