Summary
By early 2023 the US federal funds rate was the highest it had been since the global financial crisis, and was expected to continue to rise. With inflation materially above target, monetary policymakers in the US and in other major economies were facing a challenge unlike any they had experienced in decades. Against this backdrop, our client was keen to understand the prospects for US interest rates. Key questions included:
- “What is the FOMC likely to do when it next meets?”
- “Has the federal funds rate reached a peak yet?”
- “When will the federal funds rate come down again?”
They asked us to develop a tool that would help them to address these questions in close-to-real time.
Key facts
Client
The investment team of a major (re)insurance company
Brief
Develop a dashboard with a live data feed that would enable the client to make judgements about the outlook for the US policy rate of interest in close to real time
Timeline
1 March – 30 April 2023
Focus area
Macroeconomic strategy
Our approach
Fathom applies the same rigorous, proven process to every piece of work; this process is the DNA that makes up our world-leading research.
Our recommendation was that we adopt a probabilistic approach, rather than attempt to provide point estimates of, for example, the most likely month in which the federal funds rate would first be cut. This was accepted by the client after a series of preliminary calls to agree the scope of the work. Our initial step was to determine the time horizons when market pricing and macro data were most informative about FOMC decisions. We were unable to find any macro data that could improve on market pricing in predicting the outcome of an FOMC meeting set to take place the following day. (This result was not surprising.) However, at horizons from three months up to one year ahead, models that used macro data alone were more informative than models that used market pricing alone to predict changes in the policy rate, while models that used both sources of information worked best of all.
Next, we proceeded to estimate three types of probabilistic model.
- The first type tries to predict only the outcome of the next FOMC meeting, by assigning probabilities to each of five possible outcomes, ranging from a cut of 50 basis points or more, through to a hike of 50 basis points or more
- The second type estimates the likelihood that the federal funds rate has peaked when the most recent move was a hike, or troughed when the most recent move was a cut
- The final type estimates the likelihood that the federal funds rate is cut within 3, 6, 12 and 24 months when the most recent move was a hike, or increased within 3, 6, 12 and 24 months when the most recent move was a cut
Delivery
We met the client to discuss all the probabilistic models we had developed, and to gather their feedback on how each performed. We then designed a user-friendly dashboard in Excel, to draw in the latest macro and financial market data (from either Bloomberg or Datastream, according to the client’s wishes) and display a full set of results from all three model types in the form of estimated probabilities. The dashboard presents heatmaps to help the client understand which of the macro or financial market data are driving each of the outcomes. It also allows the client to conduct ‘What if?’ analysis, moving some of the inputs away from their current values and observing the impact on the estimated probabilities.
Once the client had spent a week or two experimenting with the prototype dashboard, enhancements were made to develop a final product, including adding charts to show how each estimated probability evolves over time as new macro data are released and as relevant financial market prices move up and down daily. The probabilities often change direction, shifting significantly as new information becomes available. In the best traditions of John Maynard Keynes, when the facts change our models change their mind.
Key calls
By the autumn of 2023, our suite of probabilistic models was pointing to the growing prospect of a Fed ‘pivot’. The odds that rates had peaked was estimated to be 90% or more, with a slightly greater-than-evens chance of a cut within the next six months, so by the March 2024 meeting. This was not the message coming from the Fed at the time: the September 2023 summary of economic projections, for example, was consistent with just one 25-basis-point cut in the federal funds rate through the whole of 2024, with the median participant feeling that a target range of 5.00% to 5.25% would be appropriate by the end of that year. Investors were doubtful, with market pricing in late September consistent with between two and three 25-basis-point cuts through 2024, with the first scheduled for the July meeting. In that respect, our models were at one extreme, with FOMC members at the other and investors somewhere in the middle.
Our models take no direct account of what FOMC members say, though of course they are guided indirectly by such commentary to the extent that it is reflected in market pricing. Our models do pay particular attention to the macroeconomic data, and by late September weak business surveys combined with slowing payrolls and a policy stance that had become ‘tight’ were all giving a green light to rate cuts. The pivot, hinted at by our models for some months, finally came in December, when a revised summary of economic projections pointed to three rate cuts through 2024. Investors duly recalculated, and by the end of 2023, market pricing pointed to a greater-than-evens chance of a cut by March 2024 — something our models had been indicating for some time.
Get in touch
For more information about our services or to discuss your needs and how we can help, please submit the form below: