Our main finding from this investigation is that the trend in
ation rate has a powerful
adverse effect on economic performance, but only when it rises above 3 percent. For example,
an increase of trend in
ation from 3 to 10 percent reduces the average annual level of GDP
in the median simulation by 6.8 percent. It has qualitatively similar effects on the other
macroeconomic indicators that we examine.
Our paper demonstrates how an ACE approach can deal with questions not readily
addressed by conventional equilibrium analyses, and also how to tackle some common objections
to ACE methodology. One such objection is that simulations can be sensitive to
numerical values assigned to parameters. We handle this objection, first, by calibrating our
parameter values to U.S. data and, second, by performing a sensitivity analysis, which shows
that our main finding is qualitatively robust to changes in parameter values.
Another common objection to ACE is that it is subject to the "Lucas critique." In our
context, when trend in
ation rises, one would expect individual behavior to adjust to this
change in the environment, but a model with exogenous behavioral rules does not allow as
much adaptation as a rational - expectations analysis would. To deal with this critique, we
first use our sensitivity analysis to identify which behavioral parameters are most important
for our results. Specifically, we find that the markup applied by a firm when setting its retail
price is particularly critical - the effect of in
ation is greatly magnified by having a small
markup. We then modify our model to allow markups to adapt to the environment through
the selection mechanism implicit in the firm exit process. We find that the economy-wide
average surviving markup ends up being higher when in
ation is higher, but we also find
that our main result remains intact when we take this endogenous behavior of markups into
consideration.