The Fed Board has published a paper titled “How Large is the Output Cost of Disinflation?” by Robert Tetlow. The paper is particularly important because Tetlow is a Senior Adviser in the Monetary Affairs Division, one of the Division at the center of monetary policy at the Fed Board.
To keep in mind
New research confirms that disinflating the US economy is expected to imply a large output cost. This research, which in our experience tends to be very influential among staff members and the Board, supports the idea that the FOMC will remain very hawkish in the near-term. Based on this, in our view the FOMC will lean against any easing in financial conditions even in the case the labor market will slowdown. The risk of doing too little continues to be larger than doing too much.
What the paper does
The paper compares a large set of estimates (and drivers) of the so-called “sacrifice ratio”, that is the cumulative output loss necessary to disinflate the (US) economy by one percentage point. The paper carries out three exercises: (i) reviews the relevant empirical literature (15 papers in total), (ii) compares the estimates of the “sacrifice ratio” across a large set of theoretical models (40 in total), and (iii) carries out a selection of disinflation experiments using the FRB/US model assuming different expectations formation mechanisms.
Results
Regarding the existing empirical literature, Tetlow (2022) makes clear that pinning down a precise measure for the output cost of disinflation is challenging, but as a statistical matter it appears to have risen significantly over time and currently is probably quite high by post-war historical standards. Table 1 of the paper, reported below, shows the estimates of the “sacrifice ratio” from the relevant literature.
Two evidences emerge from Table 1: there is considerable variation across estimates (part of which comes from considering different sample periods), and there is a tendency of the sacrifice ratio to increase over time, consistent with the flattening of estimated Phillips curve.
Regarding the existing theoretical models, Tetlow (2022) compares a large set of general equilibrium models using the Macroeconomic Model Data Base by Volker Wieland. Table 2 of the paper, reported below, summarizes the relevant statistics of the distribution of estimated sacrifice ratios.
As in the survey summarized in Table 1, there is a lot of variation in the sacrifice ratios in Table 2. That said, the sacrifice ratios are mostly clustered around the mode of 7. More importantly, the sacrifice ratios computed from these models are notably higher than the bulk of the historical values shown in Table 1. This happens because, as mentioned above, relying on more recent data tends to result in the estimation of flatter Phillips curves which in turn results in higher sacrifice ratios. (For the record, the sacrifice ratio estimated using the Smets and Wouters (2007) model -arguably the canonical, medium-scale, estimated DSGE model of the U.S. economy- is 5.7).
Regarding the simulations using FRB-US, Tetlow (2022) simulates the sacrifice ratio assuming different expectational adjustment costs (that is, where decisionmakers believe inflation is headed in the long-run) and the information set of the agents (that is, whether decisionmakers have full knowledge of the structure of the economy, including monetary policy, and use that knowledge to formulate their expectations). Without entering into the details of the model, Table 5 of the paper reported below shows the main results of the simulations.
The rows of the table show the different assumed processes for long-term inflation expectations, from “credible” (in which expectations are always anchored at target) to “accelerationist” (in which there is no anchor). On the other hand, the columns of the table show the assumptions about decisionmakers’ full knowledge of the structure of the economy from “MCE” (full knowledge) to “VAR-based” (limited knowledge). The main takeaway from this exercise is that the sacrifice ratio increases moving from the top left of the table towards the bottom right. The main intuition behind this result is a traditional DSGE feature: disinflating the economy becomes more costly if (i) agents’ expectations are not fully anchored, (ii) if decisionmakers do not have full information about the economy, and (iii) monetary policy is not fully credible or well communicated. In the “public model” scenario in which long-term expectations vary gradually over time (but remains anchored) and agents have only a limited information set, the sacrifice ratio is 6-8.
Comment and implications for the Fed staff
Tetlow (2022) is an important paper that, in our view and experience, will be influential among staff members and the Board. No matter how one tackles the issue of the sacrifice ratio (empirically, theoretically or with FRB-US simulations) the answer tends to be the same: the cumulative output loss necessary to disinflate the US economy is quite large. Furthermore, the most reliable estimates (which suggest a sacrifice ratio of about 7) are well above the sacrifice ratio implicit in the latest SEP (and Fed staff forecast inferred from the minutes). Indeed, in their most recent SEP, FOMC participants suggested that the output to be forgone in the course of reducing inflation to 2 percent by 2025 would render a sacrifice ratio of about 2.2 percentage-point years of real growth, that is less than one third of what FRB-US and other models suggest.
In other words, Tetlow (2022) is a reminder that the FOMC (and the Fed staff) are still forecasting an “immaculate disinflation” according to historical standards. For the time being, we expect the FOMC to remain hawkish and raise the FF rate to 5%-ish because the risk of doing too little is way higher than doing too much. But history, economic theory, and models’ simulations teach us that chances are that ex-post either GDP growth will prove lower than assumed in the latest SEP or inflation will prove more persistent.