Unfortunately, we are not yet finished because we do not have closed
form formulas for the expectations in this expression. To provide
sufficient conditions, we need to bound these quantities. In the re-
mainder of the proof, we construct a number k- where
This makes p 1 <p 2 k- a sufficient condition for (A.2). Of course, this
assumes that p 2 ≥0, and so we impose this as an additional con-
straint to be satisfied. In practice, we find that for the ranges of πL,
πH, λ 1 , and λ 2 allowed by the model, p 2 is always positive. However,
we do not attempt a proof of this.
The first step in bounding the expression
is to bound E(q). To do this, note that
Consider the function g(q) defined on [q- ,q–]. The idea is to bound
this function above and below over this interval by straight lines,
parallel to the line passing through the endpoints of g(q), namely (q
,
g(q
)) and (q–,g(q–)). In other words, suppose that we bound g(q)
above by g–(q)=a+bq. The slope of this line is
and awill be such thatGiven thatEq(g(q)−q)=0,
we must haveEq(g–(q)−q)≥ 0orE( )E( ) ,abqqq
a
b+−≥≤
−01inf ( ( )).
qqq[,]
abqgq
∈
+− = 0bgq gq
qqqq q q
qq=−
−=−− +
−<() () ( ) (() ())
,1
2
1∆∆E( ) E( ) E (E( ))
E(( ( )) (()))
E(()).qq qq
qq qq
gqttqtt
qt t t t
qt
t==
=++−
=++ 11
1
2
1
2
∆∆E( )qqq++^12 E ( ( )q∆∆( ))kq
q qq
<++
E( )E( () ())
.∆∆
2456 BARBERIS, SHLEIFER, VISHNY