On the design of lottery games

We describe a model of participation in lottery games designed to address the optimisation of tax revenue in state-sponsored lotteries. The model treats participants dynamically and examines a long-run equilibrium. A novel high frequency approximation is used to turn the problem into a static, state-contingent deterministic programming problem. We demonstrate that the solution of this problem has qualitatively plausible properties and then calibrate the model against the United Kingdom National Lottery (UKNL). The results suggest that the current design of the UKNL may not be maximising tax revenue.