This paper estimates the consumer surplus from using alternative payment methods. We use evidence from Uber rides in Mexico, where riders have the option to use cash or cards to pay for rides. We design and conduct three large-scale field experiments, which involved approximately 400,000 riders. We also build a structural model which, disciplined by our new experimental data, allows us to estimate the loss of private benefits for riders when a ban on cash payments is implemented. We find that Uber riders who use cash as means of payment either sometimes or exclusively suffer an average loss of approximately 40–50% of their total trip expenditures paid in cash before the ban. The magnitude of these estimates reflects the intensity with which cash is used in the application, the shape of the demand curve for Uber rides, and the imperfect substitutability across means of payments. Welfare losses fall mostly on the least-advantaged households, who rely more heavily on the cash payment option.

This article documents a new set of stylized facts on how pricing moments depend on product age and emphasizes how this heterogeneity is crucial for the amplification of nominal shocks to the real economy. Exploiting information from a unique panel containing billions of transactions in the US consumer goods sector, we show that our empirical findings are consistent with a narrative in which firms face demand uncertainty and learn through prices. Such a mechanism of active learning from prices can strongly influence an economy’s aggregate price level and can thus be important for assessing the degree of monetary non-neutrality. To quantify this, we build a general equilibrium menu cost model with active learning and exogenous entry that features heterogeneity in pricing moments over the life cycle of products. Under this setup, firms engage in active learning to deal with uncertainty on their demand curves. Firms choose prices not only to maximize static profits but also to create signals to obtain valuable information on their demand. In the calibrated version of our model, the cumulative real effects of a nominal shock are approximately three times as large compared to a standard price-setting model. The main intuition behind this result is that active learning weakens the selection effect. Price changes are mainly determined by forces of active learning and, hence, become more orthogonal to aggregate shocks, which reduces the aggregate price flexibility of the economy.