Pre-announced policies often generate anticipation effects that may end up in unin- tended consequences. But little is known about the extent to which the actual implementation of the policy can offset these effects. Previous research have shown that the announcement of an illegal crop substitution program made coca cultivation increase substantially in Colombia, but the net effect of the policy has not been estimated. We use detailed data on both coca cultivation and substitution payments at 1km x 1km grid squares to estimate the net effect of the policy. Our data also allows us to study the geographical spillovers of the program to non-targeted neighboring areas. Using a difference-in-differences empirical strategy, we nd that program recipients reduced illegal crops by 94% with respect to the pre-program mean. Surprisingly, the reduction in neighboring (non-targeted) grid areas is of similar magnitude. However, these reductions are not enough to compensate for the large increase in coca growing that took place between the announcement and the implementation of the policy, and thus the net effect is negative. This suggest both that the early announcement was a mistake that led to a substantial one-time cost, but the ongoing substitution efforts will have the intended effects if continued.
