Time-Varying Pricing May Increase Total Electricity Consumption: Evidence from Costa Rica

We study the implementation of a time-varying pricing (TVP) program by a major electric utility in Costa Rica. Our data come from approximately 7500 households who either remain in the TVP program, join, or leave during the period of the study. Because of particular features of these data, we use recently developed understanding of the two-way fixed effects differences-in-differences estimator along with event-study specifications to interpret our results in terms of bounds on the true estimate. Similar to previous research, we find that the program reduces consumption during peak-hours. However, in contrast with previous research, we find that the program increases total consumption. We explain the differences between our results and the typical finding with a simple economic framework that is based on the setting we study – common to many low-and-middle-income tropical countries – very few households have heating or cooling devices. While previous research used data from rich countries in which the use of heating and cooling devices drives electricity consumption is prevalent, in our setting, since there is no room for technological changes, behavioral changes to reduce consumption during peak hours are not enough to offset the increased consumption during peak hours. Our results serve as a cautionary piece of evidence for policy makers interested in reducing consumption during peak hours – the goal can potentially be achieved with TVP, but the cost is increased total consumption.