Foretelling what makes people pay: predicting the results of field experiments on TV fee enforcement

One of the current challenges in field experimentation is creating an efficient design including individual treatments. Ideally, a pilot should be run in advance, but when a pilot is not feasible, any information about the effectiveness of potential treatments' to researchers is highly valuable. We run a laboratory experiment in which we forecast results of two large-scale field experiments focused on TV license fee collection to evaluate the extent to which it is possible to predict field experiment results using a non-expert subject pool. Our main result is that forecasters were relatively conservative regarding the absolute effectiveness of the treatments, but in most cases they correctly predicted the relative effectiveness. Our results suggest that, despite the artificiality of laboratory environments, forecasts generated there may provide valuable estimates of the effectiveness of treatments.