This paper presents a theoretical model that explains two real-world observations on defined benefit pension funds: that is, (i) most pension plans are overfunded, but not by much, and (ii) commonly used pension benefit formulas base their calculation on years of service and/or (final) wage. This study comes up with the "cooperation hypothesis" to provide an economic rationale for these two observations. The argument is that overfunding is a way that the firm can commit itself to efficiency-enhancing cooperation with workers. Also, for this arrangement to be viable, any incentive for defection has to be stripped of workers. This can be done by basing the pension benefit formula on the worker's effort. Furthermore, this paper finds out that the standard theory claiming extreme patterns of pension funding to be the optimal strategy can only remain valid in the present study when the workers' effort is exogenously given to the model.