This paper analyses the optimal monetary policy of a government facing
an election, whose disinflationary 'abilities' are uncertain, under t
he assumption that reducing inflation is costly because of backward-lo
oking contracts. It is shown that if the government likes to be in pow
er it can choose to 'do nothing' on the inflation front in order to av
oid risking electoral defeat should disinflation prove too costly. The
costs of inflation reduction are worth bearing if initial inflation i
s sufficiently high: therefore it is possible to observe quick disinfl
ations that however stop short of low inflation.