The project aims to build a generic game-theoretic model of a government`s credible commitment to a costly long-term policy. In some cases, such commitments have proven credible (e.g., successive Norwegian governments have adhered to the fiscal policy rule for the Government of Norway Pension Fund Global). However, in other cases, such credibility seems to be lacking (e.g., ambitious long-term climate change policy). We will develop a novel game-theoretic model to elucidate under which conditions a government can maintain a costly long-term policy, given that it depends on continued electoral support and that the outcome is also influenced by stochastic processes ("moves by nature"). Select implications of this game-theoretic model will be tested through carefully designed laboratory experiments.