points of Fi are restricted to ti = 0 and ti = T.
We proceed in two steps: first we show that the mass points at zero are uniquely
determined, and second we prove that (10) and (11) constitute an equilibrium.
Step 1 : From (10), it follows that bi < 1 implies Fi (ti) < 1 for all ti < T :
whenever i chooses a mixed strategy, there is strictly positive probability that ti = T.
In particular, we have Fi (—c/2 + T) < 1, which implies that Fjl (—c/2 + T) = 1.
This is due to the fact that there is a strictly positive probability that i waits until
T, and, as in the case of Fi above, Fjl must be continuous on (0,T). However, as
Fi is constant in (—c/2 + T, T) and c^ < c, j^ strictly prefers tjL = —c/2 + T to all
tjL > —c/2 + T, and therefore Fjl (—c/2 + T) < 1 contradicts the nonexistence of
interior mass points.
With (11), min (bi, bjL) = 0, and Fjl (—c/2 + T) = 1, we get
bi = 0 and bjL = — 1 — (1 — Pl) exp
Pl L
(-1∙∙ T)∣
(12)
bjL is strictly decreasing in T with limτ∣δ∕2 bjL = 1 and limτ↑c∕2-ci∏pH bjL = 0. Hence,
c/2 < T < c/2 — clnpH is a necessary condition for the existence of a mixed strategy
equilibrium.24
Step 2 : It remains to show that (10), (11) and (12) indeed constitute an equilib-
rium. Consider first individual i and suppose that j follows Fjl and Fjh , respectively.
For any ti ∈ (0, —c/2 + T], i’s expected payoff is
which is equal to V — (1 — pL) exp (—1/2 + T/c) c. If i concedes in T, he gets
- Pl
pti
Jo
1
x-
- pL 1 -i + τ-x
-C 2 ' ≡
Pl c
— I, τ-ti
dx — (1 — pL) c- 2+ c
(c + ti)
r~F+ττ 1 — pL 1 i + τ- t c /c ∖
— Pl X------' 2 + ð dx — (1 — Pl)(t+ T)
Jo Pl c 22 J
24To be precise, if T = c/2 — clnpχ, we get bjL = 0 and bi > 0 is not uniquely determined.
Hence, there exists a continuum of mixed strategy equilibria where i’s payoff is υ — c, as in the pure
strategy equilibrium. We omit this case in order to simplify the exposition.
29