substantial cut in bonus. For simplicity, we assume that at all earnings lev-
els other than at the thresholds the incentives for better performance are
positive and constant. (In practice, we suspect that they are steep near a
threshold and more tempered at either extreme.)
The self-interested executive manages earnings to maximize his personal
payoff. In each period, he receives a payoff f(Rt, Rt− 1 ), where Rtis the re-
ported earnings performance in period t. If the manager meets or surpasses
the benchmark, he receives a bonus v(Rt, Rt− 1 ).^16 Thus, we posit the fol-
lowing form for f:
f(Rt, Rt− 1 )=βRt+v(Rt, Rt− 1 ),where
v(Rt, Rt− 1 )=γ if Rt≥Rt− 1
=0 otherwise.The executive’s direct rewards for the current period performance (at a rate
β) are captured by the first term. The previous period’s reported earnings
EARNINGS MANAGEMENT 641
Figure 18.1. Reported earnings in period 2 as a function of those in period 1, hold-
ing L 1 and L 2 fixed.
(^16) If effort boosts earnings, incentives should be strongest where earnings outcomes, given
optimal effort, are most likely. Strong incentives cannot be provided across all outcomes be-
cause executives cannot be paid negatively for poor outcomes. Thus, strong incentives will not
be provided for very favorable outcomes either lest executives be overpaid on average. The op-
timal reward schedule will be steeply responsive near the benchmark since such earnings out-
comes are most likely.