When users have ex-ante demands over a common resource and when resource size is not sufficient to cover all the individual demands, there is a need to establish a rationing rule. I test whether the choice of the rationing rule impacts the individual decision to self-insure, i.e., to invest in a secure alternative resource, instead of relying on a free but uncertain common resource. Four rationing rules, empirically relevant for water management, are compared using a laboratory experiment. According to Nash predictions, the investment in self-insurance is the same with the four rules. However, the experimental data show that agents' decisions are impacted by the rule. Coordination on the optimal self-insurance level is higher with the no allocation rule. However, total gains are higher with the constrained-equal awards rule, and their variability is reduced. Rules which are defined as a proportion of posted demands, such as the proportional and constrained-equal losses rules, induce sub-optimal levels of self-insurance.