In a survey on methods of payment conducted by the Bank of Canada in 2009, 6% of respondents reported in three-day shopping diaries using two new payment innovations: contactless credit and stored-value cards. However, early adopters of new technologies are not always representative of the wider population. We utilize propensity score matching methods to decompose the impact of these payment innovations on cash usage into selection and causal effects. Our results show that the use of contactless credit and stored-value cards leads to a reduction in expenditure share for cash of 14% and 12% on average, compared with difference in-means estimates of 19% and 20%, respectively. Sensitivity analysis using Rosenbaum bounds quantifies the robustness of our estimated treatment effects to unobserved bias. Accounting for selection could be important for central bank monitoring and forecasting of demand for cash using micro-level data.