This is the second part of a mini-series on FRB-US. Part I is here. In part I we have explained the “consistent” FRB-US database, as well as the “unconstrained” forecast. In this note, we show the model forecast using what, in our view and estimates, is a more reasonable Phillips curve. After that, we simulate a large financial shock and show what monetary policy could do in that event. As already anticipated yesterday, under all simulations it is virtually impossible to get 100bps cut of the FF rate in 2023 (at most, we get a 50-60bps cut with the FF rate around 4.8% in Q4:2023).
The technical issue
The “consistent” Phillips curve forces inflation to converge to target. As explained in part I, FRB-US contains a Phillips curve specification that results in an inflation process converging towards target by the end of the medium-term under different scenarios. The two equations governing the dynamics of core PCE price inflation in FRB-US can be seen here. Without entering into the details (in case, #AskTilda), the key is that core PCE prices are essentially driven by: (i) the so-called PTR (a proxy for pi*) which is an error correction mechanism in deviation from the 2 percent target, and (ii) the price markup (gap) which is measured as the weighted average of detrended labor share. After a careful investigation and a certain number of simulations, we concluded that they both converge towards target almost under any circumstance, forcing core PCE inflation with them. The point is: how realistic is that? Is it possible to embed a more realistic Phillips curve in FRB-US?
A more realistic forecast
A more realistic Phillips curve results in a higher path of core inflation. In order to understand what happens with a more realistic Phillips curve, we have modified FRB-US. Specifically, we have allowed a specification that mimics Detmeister et al. (2014), which in turns mimics the partial equilibrium Phillips curve model of the Fed staff framework. Under this specification, inflation is still “anchored” (it is still forced sooner or later to converge to target) but the persistency of the process is (much) higher than in the “consistent” FRB-US dataset. The results of this simulation are shown in Figure 1.
Figure 1. FRB-US December 2022 SEP “consistent” (blue) and alternative Phillips curve (orange)
Note: the figure shows the forecast of the publicly available FRB-US database consistent with the December 2022 SEP (blue line) and our simulation (the orange line) assuming a Phillips curve specification as in Detmeister et al. (20214). Real GDP growth is shown as YoY. “Core inflation rate” refers to core PCE price inflation (YoY) at quarterly frequency. The Federal Fund rate and the unemployment rate are expressed in percent.
As shown in Figure 1, if the model is allowed to embed what seems to be a more realistic Phillips curve, FRB-US delivers a forecast which is far from a soft-landing / immaculate disinflation. Indeed, the model pushes the FF rate to a peak of 5.98% in Q3:2023. The higher terminal rate and the fact that the model is “unconstrained” generates a protracted recession and an increasing unemployment rate (which reaches 5.4% in 2025). Nevertheless, core PCE price inflation (bottom right panel in Figure 1) is persistent and remains above target at the end of the medium-term. In a nutshell, it seems to us that this is precisely the scenario that markets were slowly pricing in before SVB.
What happens with a large financial shock?
The Fed can cut rates in 2023 but it would take a lot. Given what happened in the turbulent weeks behind us, we evaluate how the Fed could react in case of a large financial shock. In FRB-US, financial shocks are generally simulated either by shocking the real expected rate of return on equity or BBB risk premium. Here is what we have found:
1. A one-time 100bps shock to BBB risk premium results in a forecast which is not materially different than the one in Figure 1. The path of GDP and the FF rate can be seen here. The reason is that growth does slow down and the unemployment rate is marginally (3-4 tenths) higher, but not nearly enough to disinflate the US economy. This evidence, once again, confirms the suspect that it is hard to imagine a disinflation without a recession. Maybe in the end it will happen; but it would be a surprise to the model(s).
2. Under all simulations we have tried, if the Taylor rule remains inertial (that is, if one assumes that the FOMC will react to a slowdown or a recession with the same speed as it is reacting to inflation), then there is no way of getting the Fed to cut rates in 2023. For instance, assuming a 300bps shock to BBB risk premium generates a recession, but in the model the Fed keeps the FF rate elevated in 2023 to make sure it disinflates before cutting it in 2024 (the path of the FF rate can be seen here).
3. Because in history the Fed has cut the FF rate faster than it has increased it, we have run some simulations assuming a non-inertial Taylor rule with a higher coefficient on the output gap (so the monetary authority reacts more and faster to a downturn). The path of the FF rate and GDP growth can be seen here (the reader wants to look at the orange lines only in this chart). The takeaway is that under this scenario, the Fed does cut the FF rate in 2023 but less than what markets seems to expect right now (in the simulation the FF rate peaks at 5.3% and ends 2023 at 4.8%). In this scenario, the Fed is successful in disfinflating the US economy but it takes an 8-quarter long recession.
Implications for the FOMC and conclusion
There are 3 takeaways from our exercise. First, working with FRB-US confirmed that the Fed staff and the FOMC are, once again, at risk of underestimating the persistency of the inflation process (for a discussion about why, in our view and experience, the Fed staff underestimated inflation in the Tealbook judgmental forecast, please read this note from last September). Second, under the assumption that there is no trade-off between price stability and financial stability, the Fed could/should raise the FF rate up to 6%; and even in this event the model is unsure the Fed could successfully disinflate the US economy. Finally, it is hard to see an aggressive Fed when it comes to cut rates in this environment, unless one thinks we are about to experience a deep recession (which is a non-linear event). All told, it remains unclear which side the Fed will prioritize, if any. The upcoming FOMC should bring some clarity. What is pretty clear, however, is that despite SVB and the financial stability concerns, the inflation problem has not magically gone away and the Fed cannot give up now. Otherwise, we will pay a higher price tomorrow.