Keeping up with the FOMC
Following the Employment Situation Report for November released last Friday, I changed my outlook for Federal Reserve Policy from what I had predicted to date - a change in 2024 from a Fed focused on fighting inflation to a Fed concerned about rising unemployment. The employment data reported was so strong that inflation continued to be the dominant policy driver into 2024.
When the FOMC today left policy rates unchanged, it was not a surprise to anyone - I had recommended raising rates, and the yield curve gave that a 1.6% chance. Having a good understanding of probabilities, I, too, expected the Fed to remain on hold.
What I found unsettling was that Chair Powell outlined a weaker labor market, and continued proposing economic growth for 2024 at only half the long term trend rate, where my forecasts remain growth at a tick above the 2.5% long term average. As a consequence, the Fed trade-offs between Maximum Employment and Price Stability (PCE Core inflation) were in favor of taking a forward-looking stance on both the very positive trend in inflation, and considering the effects of economic slowdown on employment.
There comes a time when we look at models and realize there is something that is not capturing what goes on in the world that we are trying to replicate. My models are very simple compared to those of the Federal Reserve, but if one gets the direction of change and the turning points right, it is "good enough" for my purposes. I missed an apparent policy turning point that I had abandoned only five days earlier!
I had planned to spend a day or two over Christmas to review the labor market models of mine, as I suspect they could be improved. I have a list of question marks, of which some regard the massaging of the data and some regard the time window used for training the models. My fear is always that I start seeing how to build the larger and more complex model, and I reject that path, as the only certainty is that complexity will always make the model opaque, require eternal maintenance, and errors can survive in code for years before you get to the trigger level of a variable that makes the model crash or producing more rubbish than usual.
I quickly looked through the model inputs and formulas to see if there were any evident errors or reliance on data series that could account for the insensitivity - or the opposite - that brought my results to follow a trajectory 75 bps above the FOMC projections of Fed Funds. The first suspect was the inflation time series, but there it was - PCE Core and 2024 forecast of 2.4%, exactly as the FOMC SEP predicts. Given that I have the Fed classified as 100% inflation focused (it is a quantitative assessment, a formula), I was more than curious about the source of my mirage.
The fact that the FOMC has become relatively forward leaning and dovish has its roots in their obsession with the Soft Landing and, implicitly, their idea that Inflation is Transitory in this cycle. The market is slowly coming to a realization that the Fed might pull off the former, and be right about the latter.
So, I looked at my model(s) and focused on the employment data and GDP projections. Essentially, Real GDP growth is a broad measure that in the Fed models for the right level of the Fed Funds rate is a very broad proxy for their Maximum Employment mandate. I am always on the Fed press conference calls (sinceDotCom crash) and listen carefully to language, retraining to understand each of the FOMC Chairs. Powell is an open book, transparent and a Gentleman. We, the audience, complicate his message. If you try to do the work preceding his speeches, people tend to bow in appreciation and recognition. If you do not, people tend to respond as the cognitively distracted do - with criticism and rancor. I always criticized the Fed for talking about the narrow unemployment measure and always using that for their reports. Given the variations in labor force participation, the measure is all over the place, and very generous to policy makers when participation is low. Today's remarks recognized the significant rise in participation. If you listen carefully to Powell, he will always insist that the Committee looks at broad data, and not single parameters. I looked at my model and substituted the narrow unemployment series with the participation adjusted series, and then we were aligned with respect to the turning point.
The decision to replace one data series I think is misleading at best with my favorite measure of Unemployment in the US and in any country resolved the issue. The bonus was unexpected, but dangerously self-gratifying. The recommendations for the Fed Funds rate also eliminated the alarms regarding recession potential. The US 10-year Treasury Yield minus the Fed Funds rate warns of recession ahead in a 4 - 8 quarter time frame. This fourth quarter of 2023 is the fifth consecutive quarter when the indicator signals Recession Ahead, as it dips to 0.2% or below. With the model parameter change, the recommendation eliminates the recession warning - i.e. the Fed Funds forecast is commensurate with a rational FOMC decision.
The corrected model now proposes that the Fed SHOULD cut Fed Funds to 3.75% - 4.00% range in Q1 2023. The model particularities remain - they are recommendations, not forecasts.
The rate cut motivations and scope lie in the nature of participation adjusted unemployment. The long term average broad unemployment rate (the Fed will likely look to U6) is 7.0%. The average for Q4 2023 is 9.7%. We project this to improve to 7.7% during 2024 as Labor Force Participation rises from 62.7% to 63.6%, coming inside the Long Term normal range of Participation for the first time since Q1 2020 (COVID outbreak). Whereas the politically correct narrow measure signals an economy at Full Employment, Participation Adjusted Unemployment redimensions unemployment to be a much larger issue than the current level of inflation. This creates a Q1 2024 Policy Shift to focus on the Maximum Employment mandate.
As an unintended bonus of the truth-seeking rush over here, I realized that the debate over the Stars - the equilibrium rates of Fed Funds (r*) and unemployment (u*) - had me change my proposals, as I looked at the formulas and the statistical estimates. The Target Fed Funds rate also comes into question when we open the Pandora's Box.
We come back to some of the fundamental beliefs of economists that lie behind policy targets. We tend to believe as a community that the world is changing, hopefully for the better. We see improvements resulting from human collective efforts. The metrics change.
Price Stability
The Fed's 2.0% inflation target is based on the median inflation over the last 30 years in the US, based on the Core Personal Consumption Expenditures Price Index. If you substitute the inflation measure - for example using headline PCE-PI or Core CPI - you get a different target. If you change the range of the estimate, e.g. from 30 to 60 years - you get a different target. If you change the estimation method from Median to e.g. Harmonic Median - you get a different target.
Inflation has been an infrequent issue over time. Hence, if we limit the estimate of the target to the 1994 - 2023 time frame, we miss all periods with inflation at levels recently experienced. Model metrics will not properly classify and project inflation phenomena as we recently experienced. In theory, we could say that inflation is a long wawe phenomenon and if you need to train your models to understand prices, you are best served with the longest time range available. However, the counterpoint is that information and communications technology has evolved dramatically since 1871 (start of timeseries at our end) and now we can deal with price stability with other time frames. I keep the one generation - 30 years - time range for target calculation. The big question is if you build your forecast on the data universe or a time restricted data set. To date I have been for using every data point available. I am rethinking that, but it is an exercise in philosopy. What is the purpose of forecasting? Do you want to align with the Fed or do you want to make "best estimates"? There is no answer, but don't fight the Fed and don't fight the ticker.
I would recommend that the market and the Fed both sees Price Stability as achieved when PCE-PI inflation is in the 0.9% - 2.6% range, and that the Target Rate of Inflation is revised to 1.7%. The implication is that the Fed will see the PCE-PI inflation for March at the end of April and on May 1 announce that inflation is in the long term normal range, and hence "under control". The Policy Question lingering is why the FOMC would entertain cutting rates before May 1. It would not seem to be motivated by the Price Stability mandate.
Recommended by LinkedIn
Unemployment
Narrow unemployment fails to take into account changes in Labor Force Participation as discussed above, and so we need to consider which time series measure to use as our reference, which statistical concept to use for estimates, and which time period is relevant for our estimates.
Narrow unemployment stands at 3.9% as of November 2023. When announced last Friday I considered that to be below the 4.0% Full Employment level. Today I propose that US Full Employment occurs at 3.4% or below on the narrow measure, and so justifying Chair Powell's remarks. On average, the US economy was at Full Employment in the period Q4 2022 through Q3 2023. If the Fed cuts rates as we propose (3.25% - 3.50% at end 2024), the economy will return to Full Employment in Q4 2024 (elections quarter). Recent studies of the Equilibrium Unemployment rate of the US has ranged in the 4% - 5% range. I estimate this at 5.2%, down from 5.7% estimated a week ago, and paired with the corresponding Fed Funds rate discussed below.
Broad unemployment stands at 9.7% this quarter, well above the 7.0% long term Equilibrium Unemployment rate (u* or NAIRU), but within the long term normal range of broad unemployment. Full employment on this measure occurs at 3.6%, close to the measure for narrow unemployment. We are quite distant from Full Employment, but moving towards the equilibrium rate.
The point is that policy based on narrow unemployment would tend to indicate tighter money as the policy stance, while broad unemployment indicates the opposite. The Fed is clearly indicating that Price Stability will be achieved by March 2024, and so the refocus to (broad) unemployment is now their official monetary policy stance, regardless of the politically correct communication of unemployment based on the more favorable narrow measure.
Fed Funds
The debate about the long term equilibrium rate of Fed Funds has raged since the Great Recession and it was recognized that after reaching Peak Real Fed Funds in the period Q1 2007 - Q1 2009 at 3.0%, it has fallen to about 0.5% today. The previous Bottom Real Fed Funds was seen in the period Q3 1977 through Q3 1979. It is a cyclical phenomenon that appears to last two generations (sixty years) to complete.
In this context one can distinguish between a number of reference rates.
The Fed tends to reference the long term equilibrium Fed Funds rate at their inflation target plus the current real equilibrium rate. So, 2.0% plus 0.5% = 2.5%.
Alternatively, one could accept the proposals above and so 1.7% plus 0.5% = 2.2%.
Alternatively, one could accept the proposition that in the long run, the average real equilibrium rate is 1.8%. This would mean that the nominal rate is 3.5%, where is where we suggest the Fed will have to be in Q3 2024. At that time. broad unemployment will be at its lowest of the year at 7.7%, just 10% above u*.
Where we then propose a 3.25% Fed Funds rate at the end of 2024, and the Fed dot plots indicate an average of 4.6%, it is our view that the Fed might underestimate the chance of arriving at an economic equilibrium during the course of the coming year, and so the stated fear of cutting too late is already our base case.
#fed #fomc #inflation #unemploymment #growth #equilibrium