9
nU
j
– mU
i
> C/p
U
i
(nU
j
/U
i
– m) > C/p
(7)
(8)
Figure 1 shows graphically how the different parameters affect policy change. Change
occurs if C/p is smaller than the threshold given by the surface. It appears clearly that the
likelihood of change, which is related to the size of the volume under the surface, increases
as m becomes smaller, n becomes bigger, and U
j
becomes bigger relatively to U
i
. In other
words, as m decreases, n increases and U
j
/U
i
increases, change is possible for increasingly
high values of C and low values of p. Note that change is possible even if m (the
effectiveness of the existing policy) is high, but only insofar as n (the effectiveness of the
alternative policy) is also high and U
j
is somewhat bigger than U
i
. Note also that if n is
small, change is possible but unlikely, given that either U
j
must be much bigger than U
i
or
m must be very low (or both). This means that the new policy must seem at least
moderately effective in order to be chosen. A poor m or a big U
j
/U
i
can compensate a
weak n, but only to some extent. Intuitively, this makes sense: ineffective policies (for
example, policies that have performed poorly elsewhere) are unlikely to be adopted.
< Figure 1 about here >
Applying Figure 1 to the example of the liberalisation of telecommunications, it can be
seen that public monopolies (policy i) are more likely to be abandoned in favour of
competitive markets (policy j) if the effectiveness of the former (m ) declines, if the
effectiveness of the latter (n) is quite high, and if the latter becomes more attractive in
relation to the former (U
j
>U
i
). The interplay between effectiveness and payoffs must be
stressed. Firstly, a change in relative utilities in favour of telecoms liberalisation does not
automatically lead to policy change. The new policy must have some effectiveness. A very
weak effectiveness can to some extent be compensated by very big payoffs, but our model
Convention