EC9A1, Problem Set 2 G. Trigilia, February 2016 Exercise 1 In the example from the lecture notes about Costly State Verification (CSV), find the optimal contract. Solution: From the lecture notes, we know that the optimal contract is debt, characterized by a face value D and verification in bankruptcy states. From the break even condition for the financier we get (in thousands of $): Z D [x − 40]f (x)dx + (1 − F (D))D 200 = 80 because income is uniformly distributed between 80 and 420. Further, from the formula for the CDF of a uniform: 1 − F (D) = (420 − D)/340 and from the PDF: f (x) = (340)−1 . Plugging inside and solving the integral we get: 200 = (680)−1 [760D − D2 ] The quadratic equation has a unique feasible solution (the negative root), that is: D∗ = 288.4. Exercise 2 A corporate raider considers raising K > 0 funds to finance a takeover bid. He knows that either the bid is successful, in which case he will get RS back from his investment (with probability p), or it is unsuccessful and he will get RF ∈ (0, RS ). Investment is subject to costly state verification, and the audit cost is µ > 0. 1. First, show that the pledgeable income is maximized, and the audit cost minimized, if the lender gets the full income in case (i) a failure is reported; and (ii) no audit takes place; 2. Second, show that if the face value of debt D < RS , a random audit policy dominates deterministic audit. 1 Solution: (1) Denote the probability of audit in case of reported failure as yF , and the repayment by SF . For a given D ∈ (K, RS ), the break even condition for the financier reads: pD + (1 − p)[SF − yF µ] = K The pledgeable income equals the LHS of the break even condition above. It is increasing in SF and decreasing in yF . The incentive compatibility constraint reads: RS − D ≥ (1 − yF )[RS − SF ] and notice that it is must be binding. To see why, argue by contradiction. Suppose it is not binding. First, this could depend on the fact that we are verifying too often (yF is too high). Now, because this has an adverse effect on the pledgeable income, we could decrease yF and achieve higher pledgeable income levels without violating incentive compatibility. This is in contradiction with our current goal of maximizing the pledgeable income, hence it could not be. The second possibility is that we are choosing a too low SF . Again, a higher SF would increase the pledgeable income satisfying incentive compatibility, hence reaching a contradiction again. As a result, the incentive constraint must be binding. A final observation is the following: suppose that SF < RF , and the associated audit probability is yF . Then, there exists a pair (yF0 , SF0 ) such that (i) SF0 > SF and yF0 < yF , and (ii) it makes the incentive constraint binding as well. From the break even condition for the financier, it becomes obvious that the prime contract increases the pledgeable income relative to the original contract we started with. (2) As a result of part (1), it must be that SF = RF , and the optimal audit probability yF is the one that makes the incentive constraint binding: D − RF RS − RF which is a well defined probability because RS > D > RF > 0. Furthermore, yF < 1 implies that random audit is optimal. Could you argue intuitively why this must be the case? RS − D = (1 − yF )[RS − RF ] ⇐⇒ yF = Exercise 3 For ease of notation, in this exercise assume that yS and yF denote the probabilities of NO liquidation, conditional on the reported state of the project. As before, assume that D < RS . 2 The incentive constraint for the successful entrepreneur not to claim failure reads: RS − D + yS RC ≥ RS + yF RC z; ⇐⇒ (yS − yF )RC ≥ D The optimal contract solves the following problem: S max p R − D + yS RC + (1 − p)yF RC s.t. {D,yS ,yF } (yS − yF )RC ≥ D, and p D + (1 − yS )L + (1 − p)(1 − yF )L ≥ I A few observation help us in solving the problem: 1. The latter constraint (zero profit condition) must be binding. To see why, note that if it does not bind one can lower D and this (i) still satisfies the incentive constraint; (ii) increases the objective function – hence contradicting the optimality of a contract with non-binding zero profit condition; 2. At the optimal contract we have: yS = 1. To see why, argue by contradiction. Start with 1 > yS . We know that D > I > 0 because the project is risky. Therefore, by incentive compatibility it must be that yS > yF . Now, consider an alternative contract with yS0 = yS + , for some small > 0. To keep the zero profit condition binding, we have to set also D0 = D + L. The incentive constraint holds because RC > L, and the objective function increases by p[RC − L] > 0, contradicting the optimality of the original contract; 3. Finally, the incentive constraint must be binding. Again we argue by contradiction. First, notice that we must have yF < 1. Suppose though that the incentive constraint is slack, and consider an alternative contract such that yF0 = yF + . To make the zero profits binding, we need to set D0 = D + L(1 − p)/p. It is again easy to check that the objective function increases by (1 − p)(RC − L) > 0, reaching the desired contradiction. As a result of these observations we can use the substitution method and solve for the optimal contract. I leave the last step to you. 3