We experimentally test a model of public good bargaining due to Bowen et al. (Am Econ Rev 104:2941–2974, 2014) and compare two institutions governing bargaining over public good allocations. The setup involves two parties negotiating the distribution of a fixed endowment between a public good and each party’s individual account. Parties attach either high or low weight to the public good and the difference in these weights reflects the degree of polarization. Under discretionary bargaining rules, the status quo default allocation to the group account (in the event of disagreement) is zero while under the mandatory bargaining rule it is equal to the level last agreed upon. The mandatory rule thus creates a dynamic relationship between current decisions and future payoffs, and our experiment tests the theoretical prediction that the efficient level of public good is provided under the mandatory rule while the level of public good funding is at a sub-optimal level under the discretionary rule. Consistent with the theory, we find that proposers (particularly those attaching high weight to the public good) propose significantly greater allocations to the public good under mandatory rules than under discretionary rules and this result is strengthened with an increase in polarization. Still, public good allocations under mandatory rules fall short of steady state predictions, primarily due to fairness concerns that prevent proposers from exercising full proposer power.