R.Mehra and E.C.Prescott,The equity premium 159 time to be poorer substitutes than consumptions at widely separated dates. Perhaps introducing some features that make certain types of intertemporal trades among agents infeasible will resolve the puzzle.In the absence of such markets,there can be variability in individual consumptions,yet little variabili- ty in aggregate consumption.The fact that certain types of contracts may be non-enforceable is one reason for the non-existence of markets that would otherwise arise to share risk.Similarly,entering into contracts with as yet unborn generations is not feasible.6 Such non-Arrow-Debreu competitive equilibrium models may rationalize the large equity risk premium that has characterized the behavior of the U.S.economy over the last ninety years.To test such theories it would probably be necessary to have consumption data by income or age groups. Appendix The procedure for determining the admissible region depicted in fig.4 is.as follows.For a given set of parameters u,8 and eqs.(10)-(14)define an algorithm for computing the values of Re,Rf and Re-Rf for any (a,B)pair belonging to the set x={(a,B):0<as10,0<B<1,and the existence condition of section 3 is satisfied. Letting R=h(a,B)and Re-Rf=h2(a,B),h:XR2,the range of h is the region depicted in fig.4.The function h was evaluated for all points of a fine grid in X to determine the admissible region. The experiments to determine the sensitivity of the results to the period length have model time periods n=2,1,1/2,1/4,1/8,1/16,1/64 and 1/128 years.The values of the other parameters are u=0.018/n,8=0.036/vn and =0.5.With these numbers the mean and standard deviation of annual growth rates are 0.018 and 0.036 respectively as in the sample period.This follows because =0.5 implies independence of growth rates over periods. The change in the admissible region were hundredths of percent as n varied. The experiments to test the sensitivity of the results to u consider u=0.014, 0.016,0.018,0.020 and 0.022,=0.43 and 8=0.036.As for the period length, the growth rate's effects upon the admissible region are hundredths of percent. The experiments to determine the sensitivity of results to 8 set=0.43,u= 0.018and8=0.21,0.26,0.31,0.36,0.41,0.46and0.51.The equity premium varied approximately with the square of 8 in this range. 6See Wallace(1980)for an exposition on the use of the overlapping generations model and the importance of legal constraints in explaining rate of return anomalies