The rationale behind the adoption of decreasing targets between 1999 and 2001 had to do
with the nature of the recent inflation in Brazil. It is important to distinguish between an
inflationary process and a temporary inflation rise due to a shock. In the first case, there is a
continuous acceleration in the price level. In the second, there may be only a once-and-for-all
change in the price level, with no further upward pressure. The Brazilian case belongs to the
second category: the currency devaluation that started in mid-January 1999 was a shock that
forced a realignment of relative prices. Before it occurred, Brazil was experiencing price stability:
average CPI inflation was 1.7 percent6
in 1998.
As there were no indications of the presence of an inflationary process in Brazil, a
gradualist disinflation strategy was not recommendable. The CPI inflation rate should return to its
1998 level as soon as the relative prices realignment is finished. Thus, it was not only possible but
also desirable for the government to set a decreasing inflation target path.
An important issue to discuss is the choice of the full inflation rate as reference for the
target, and not some core inflation measure. Perhaps, the best technical procedure would have
been to purge some items from the full index, exempting it from temporary and once-and-for-all
shocks. Nevertheless, adopting a headline index was essential for credibility reasons, at least in
the beginning of IT implementation. Unfortunately, Brazilian society has experienced several
price index manipulations in a not so distant past, and so would be suspicious about any change
related to suppressing items from the target index.
Another related issue is the absence of escape clauses in the institutional arrangement. In
the case the targets are breached, the Central Bank Governor will be required to issue an open
letter addressed to the Minister of Finance explaining the underlying causes, the measures to be
adopted to ensure convergence to the targets and the time period required for these measures to
have an effect.
The combination of the use of headline inflation and the absence of escape clauses
justifies the adoption of the relatively wide 2 percentage point tolerance interval around the
central target, and certainly makes the announced targets much tighter than they may initially
appear.
It is important to emphasize that monetary policy decisions should be taken on the basis
of the widest information set available. Therefore, a mix of models should be under consideration
when looking for an adequate reaction function, and producing inflation forecasts and their
probability distributions. It should also include private sector perceptions about the expected path
of economic variables, extra-model information, leading indicators and any other judgmental
knowledge that helps predict inflation.