Gauti Eggertsson and coauthors (Alessandro Lin, Josef Platzer, and Luca Riva) have published a column on VoxEu titled “Inflation and the Fed’s new framework in the time of Covid-19”. Their contribution is related to the on-going debate on the “immaculate disinflation” (that is, FOMC projecting inflation slowing down without the need of a recession). Please, find below our critical review.
To keep in mind
The new dots plot shows a quick return of inflation towards target in 2023-2024, despite the high reading in 2022 without an imminent recession (an “immaculate disinflation”). While apparently contradictory, the SEP projection is in line with the prediction of most DSGE models. However, in order to believe in the SEP forecast, one has to put a lot of faith on a crucial assumption: the medium-term behavior of inflation being driven by stable long-term expectations. The issue is that the Fed is operating on thin ice because this crucial assumption has never been tested empirically (excluding the post-2009 period when however inflation was running below target).
Monetary policy will remain hawkish until the data will tell us that inflation is no longer an issue.
What the papers do
The authors address two main questions. First, is the new Fed framework (Average Inflation Targeting, AIT) preferable to other Taylor rules in terms of inflation and output stabilization? Second, what is the cost of over-stimulating with respect to AIT?
In order to answer these questions, the authors calibrate a Dynamic Stochastic General Equilibrium (DSGE) New Keynesian model at the zero lower bound and compare the behavior of some variables of interest under AIT and some alternative Taylor rules (a standard Taylor Rule truncated at the zero lower bound, a nominal GDP target rule, a history-dependent nominal GDP target and a symmetric dual-objective targeting rule). The authors simulate the behavior of output and inflation under AIT and an alternative “over stimulative” scenario (in which the Fed keeps interest rates at zero for four quarters beyond what the standard AIT rule would prescribe, under any possible future state of the world).
Results
Regarding the first question, Eggertsson et al. (2022) document that AIT substantially mitigated the recession that otherwise would have been deeper under standard inflation targeting (i.e. the traditional Taylor rule truncated at the zero lower bound). A key conclusion from the authors’ exercise is that a lift-off of the Federal funds rate from zero in the first half of 2022 is consistent with the AIT framework outlined by the Fed, provided the averaging is over a period of around three to five years and that the Fed is giving equal weight to output deviations relative to average inflation. They also find that two alternative make-up strategies –a history-dependent nominal GDP target and a symmetric dual-objective targeting rule– would have reduced the output contraction by even more because in contrast to AIT, these policy strategies explicitly take into account past shortfalls in both inflation and output.
Switching to the second question, the authors results are more controversial. In this exercise, the authors evaluate the costs of over-stimulating by comparing the behavior of output and inflation under AIT and under an alternative rule that keeps interest rates at zero for four quarters beyond what the standard AIT rule would prescribe. For convenience, we report the results of this exercise here below.
According to the authors’ DSGE model, the increase in inflation under “overstimulate” is relatively small and the inflation trajectory is similar under the two scenarios. On the other hand, the increase in output due to the extra stimulus is non-negligible and the output trajectory is materially different. The authors conclude by saying that “assessing the desirability of a more dovish policy therefore depends on how one views the trade-off between a stronger recovery and elevated inflation.”
Note: Eggertsson et al. (2022) results are closely related to the evidence of the NY Fed DSGE model. the NY Fed has released an update forecast of this model yesterday. The model projects that core PCE inflation will stand at 2.8 percent in 2022, at 2.2 percent in 2023, and 2 percent in 2024 and 2025. As previously discussed (for a review of monetary policy in the NY Fed DSGE model please see our post here), the NY Fed model identifies in cost-push shocks the main driver of the current inflationary episode.
Comment and implications for the FOMC
The exercise carried out by the authors (as well as the NY Fed DSGE model) is, in our view, one example of the potential risks of using DSGE models in the current environment. As mentioned, the results of the first question investigated (whether AIT is preferable to other Taylor rules in terms of inflation and output stabilization) are intuitive and (under the authors assumptions) uncontroversial. On the other hand, the results of the second question are much more controversial and should be taken cautiously. According to the authors, in the current environment the cost of an over stimulative monetary policy ends up being small. However, there are two important assumptions behind the authors’ results: first, the output gap is calculated using the deviation of real GDP from pre-Covid linear trend. Because potential is unaffected by the recent crisis, in the model the output gap remains negative in 2022 and 2023 (the blue line in the figure above). For this reason, monetary policy can be pushed beyond accommodative even if inflation is above target. Second, and most importantly, inflation gradually returns to its long-run mean because expectations remain anchored. This is why the trajectory of inflation under the two scenarios is similar: it cannot be materially different.
For this reason, we reiterate that relying on DSGE models in this moment requires a good amount of faith: how much does one believe in the theory, given than it has never been tested empirically (except for the post 2009 period when the shock was on the downside and disinflation was attenuated by downwardly nominally rigid prices and wages)?. The newly released SEP projections embed this feature: inflation is projected at 4.1 percent in 2022 but it returns quickly towards target in 2024 without the need of engineering a recession (a feature that has been labeled “the immaculate disinflation”). The trajectory of the SEP projection reflects not only the nature of the shocks that hit the US economy (quickly dissipating cost-push shocks) but crucially the augmented Phillips curve framework with anchored expectations (please note that, in our view, the Fed Board staff forecast is build using the same approach). So far this cycle, the observed behavior of long-term measures of inflation expectations has been reassuring: most measures have remained broadly stable (“anchored”) despite the runup of actual inflation. But the reality is that relying too much (or exclusively) on these indicators would probably neglect the complexity of the issues and the risk of expectations being a lagging indicator.
Back in December we wrote that because the Fed is forced to operate on thin ice (that is, relying on the idea that long-term expectations remain the main driver of inflation), “focusing on a combination of actual observed inflation and long-term inflation expectation is probably a wise approach for the Fed”. Whether disinflating the US economy will require a recession is unknown at the moment: the FOMC does not know, economists are unsure, and the theory is ambiguous (that is, it depends on the model and whether one thinks that long-term expectations still play a crucial role). This seems to be the new Fed approach: the FOMC has finally realized that relying too much on the theory is too risky when actual inflation is so high. Therefore, the FOMC has come up with a strategy to reduce the tail risk event: monetary policy will remain hawkish until the data will tell us that inflation is no longer an issue.