"... In PAGE 11: ... 3.3 Hypotheses Equilibrium predictions based on subgame perfection for the two bargaining treatments are summarized in Table1 . In this table the bargaining outcome is stated as the share M1 (as investor) obtains in equilibrium.... In PAGE 11: ... The expected efficiency losses are due to the fact that the predicted investment levels fall short of the first best investment level of 50. The predictions in Table1 lead to hypotheses regarding investment behavior, bargaining behavior and efficiency. With respect to investment behavior we formulate hypotheses based on... ..."

"... In PAGE 15: ... 3.3 Hypotheses Equilibrium predictions based on subgame perfection for the two treatments are summarized in Table4 . In this table the bargaining outcome is stated as the share the buyer (as investor) obtains in equilibrium.... In PAGE 15: ... In this table the bargaining outcome is stated as the share the buyer (as investor) obtains in equilibrium. The predictions in Table4 lead to hypotheses regarding investment behavior and bargaining behavior. With respect to investment behavior we formulate hypotheses based on comparative statics predictions and hypotheses based on point predictions.... In PAGE 17: ... Depending on the outside bid in two adjacent periods, we then predict that the investment level will increase, decrease or remain the same (cf. Table4 ). These predictions can be confronted with the actually observed changes.... In PAGE 28: ... As subgame perfection predicts agreement to be reached immediately (cf. Table4 ), a theoretical foundation for such a distinction is lacking.28 It does appear, however, that the length of the bargaining game under the OO-game is significantly affected by the value of the outside bid itself (as was the case for conflict of opinion reported in Table 10).... ..."

"... In PAGE 29: ...elf-insure. As a result, the ratio of ct=gt (i.e., the proxy for shifts in the distribution of wealth between the private and public sectors) uctuates widely over the business cycle, as documented earlier, and the uctuations in government expenditures are particularly costly (since the average of gt is nearly 7 times smaller than the average of ct, and hence the curvature of the CRRA payofi function, with the same parameter, is more pronounced and yields larger utility changes for the uctuations in gt). Table4 compares the long-run moments of macroeconomic aggregates in the MPE and the social optimum to ofier more insight on the determinants of the large welfare costs of imperfect risk sharing in the model. The mean values of the variables are almost identical in the two economies, so the welfare costs do not arise because the social planner has access to resources not available to the agents in the competitive equilibrium.... In PAGE 30: ... Table4 shows, the standard deviation of public expenditures (private consumption) is 13.5 (2.... ..."

"... In PAGE 22: ... An M2 rule speci es a probability distribution over at most 2m possible actions by conservative and adaptive rules and assigns a best response to it. While the number of pure-action best responses to mixed actions may exceed 2m, one can construct a stage game similar to Table6 to restrict them to 2m as well. Any higher level mixed rule can be also restricted to have beliefs with nite support.... ..."

"... In PAGE 27: ...(Insert Table2 about here.) We note that the effects of peer influence are in some ways similar to what is observed in terms of the addiction effect.... ..."

"... In PAGE 7: ... Parameters are chosen based on three considerations: rst, the Stackelberg equilibrium and Nash equilibrium under ACP are su ciently far away from each other; second, the normal form games with discrete strategy space has a unique Nash equilibrium under ACP3; third, most of the payo s are positive in both normal form games. [ Table1 about here.] Table 1 tabulates the parameters and equilibrium values for both mechanisms.... In PAGE 7: ... [Table 1 about here.] Table1 tabulates the parameters and equilibrium values for both mechanisms. In the second column we let 1 = 16:1 and 2 = 20:1.... ..."

"... In PAGE 23: ...For all schooling outcomes, children living in intact families have substantially and significantly higher mean outcomes than any other family structure. In Table5 , we estimate two models of family structure using the stable blended- family sample.11 Model (A) is a parsimonious model where family structure is measured as years outside of an intact family (nonintact); the specification is appealing in part because we only have 135 observations.... In PAGE 23: ... For each outcome, the first two columns are OLS specifications and the next two columns are fixed effects specifications. In the first OLS specification, Model (A), the normalized 11 Only three of the four schooling outcomes are presented in Table5 because only 4 individuals in the... ..."

"... In PAGE 23: ...10 Let ri max,i a,b, be the largest r for which firm i is willing to invest. For the positive and negative cases these are listed in Table2 . Note that in each case, rb max ! ra max , so for r between the two values there is a unique equilibrium.... ..."