the cycle would repeat itself resulting in permanent oscillations even in the
absence of exogenous innovations. Oscillations are not a consequence of mathe-
matical properties of underlying dynamic equilibrium equations. The path that
the economy follows is chosen in equilibrium and a priori the economy need not
follow an oscillatory path.
Economic theory either perceives downturns in economic activity as shifts
in the potential level of output or as temporary deviations from the underlying
level of potential level of output. This paper falls into the latter category and
illustrates that endogenous revisions of expectations cause output to fluctuate
around the trend. Specifically, the paper shows that overly optimistic expecta-
tions lead to high consumption and to high output and pessimistic expectations
cause consumption to be low and output to fall below the potential level of
output. Moreover, the paper illustrates that a boom caused by consumer con-
fidence and consumer spending can endogenously end with a drastic revision of
expectations and a cut in the level of consumption. The paper, thus, rationalizes
consumption shocks thought to be at the root of the 1990/91, see Blanchard [4],
and 2001 recessions. Moreover, the paper formalizes and gives precise meaning
and clear interpretation to intangible notions such as: overheating, output gap,
overborrowing, or irrational spending sprees etc. Consequently, the paper shows
that consumer behavior can depart from optimal first best spending plans and
that the deviations can become costly ex post and can lead to endogenous revi-
sions of expectation. Finally, the paper provides a justification for the usage of
countercyclical policy measures allowing for smoothing out deviations from the
trend.
The paper resembles most closely the recent contribution by Brock and
Hommes [7] who study the dynamics of a cobweb model with costly forma-
tion of expectations, showing that agents may find it optimal to switch between
different expectation formation technologies if it is costly to collect data. This
paper, contrary to Brock and Hommes, allows agents to choose the optimal
expectation formation technology in the rational fashion based on the level of
utility that it gives, whereas in Brock and Hommes [7] agents choose expectation
formation technologies based on their past performance. Moreover, in this set-
ting we allow agents to switch expectation formation technologies at any point
in time and all agents do indeed use the optimal one. In Brock and Hommes[7],
on the other hand, some agents are artificially constrained to use suboptimal
expectation formation technologies.
Ramey and Evans [18] also present a model that is similar to the one in
this paper. Their model allows to trade between accuracy an income, a feature
exploited in this paper, However, the dynamics in their model are driven by
changing environment. In this paper the changes of environment are endoge-
nous, not triggered by exogenous movements in states of the world, and are
solely due to changes in the behavior of economic agents.
Blanchard [6] studies a model in which agents learn about the economy.
He finds that learning need not lead to a full revelation of the fundamentals
of the economy and that in the limit agents may not know the correct model
of the economy. This paper presents a complementary result. It is shown