Black markets are estimated to represent a fifth of global economic activity, but their response to policy is poorly understood because participants systematically hide their actions. It is widely hypothesized that relaxing trade bans in illegal goods allows legal supplies to competitively displace illegal supplies, but a richer economic theory provides more ambiguous predictions. This paper evaluate the first major global legalization experiment in an internationally banned market, where a monitoring system established before the experiment enables us to observe the behavior of illegal suppliers before and after. International trade of ivory was banned in 1989, with global elephant poaching data collected by field researchers since 2003. A one-time legal sale of ivory stocks in 2008 was designed as an experiment, but its global impact has not been evaluated. It find that international announcement of the legal ivory sale corresponds with an abrupt ~66% increase in illegal ivory production across two continents, and a possible ten-fold increase in its trend. An estimated ~71% increase in ivory smuggling out of Africa corroborates this finding, while corresponding patterns are absent from natural mortality and alternative explanatory variables. These data suggest the widely documented recent increase in elephant poaching likely originated with the legal sale. More generally, these results suggest that changes to producer costs and/or consumer demand induced by legal sales can have larger effects than displacement of illegal production in some global black markets, implying that partial legalization of banned goods does not necessarily reduce black market activity.