(Bloomberg) -- Words are powerful. For the US economy, no-one’s more so than those of Fed Chair Jerome Powell.

Bloomberg Economics’ Fed sentiment index — powered by a natural language processing algorithm based on more than 60,000 Fed headlines — shows that in December, Powell delivered a major pivot. By hinting at a swifter shift toward rate cuts, he gave markets a boost and helped the economy dodge a downturn.

So far, so good — but there’s a catch. Four months on, with demand powering ahead and inflation stuck above target, Powell has been forced into a reversal. That started on the margins of the International Monetary Fund’s April meetings, when he said “it is appropriate to allow restrictive policy further time to work” —  pushing the prospect of rate cuts further into the distance.

That was a move in the right direction. But our index shows it only undid a fraction of the stimulus unleashed by his December pivot. That means there’s more to do to bring inflation back under control. We think that means more hawkish words to come, endorsing the recent upward move in yields – perhaps as soon as this week’s Fed meeting. 

Theories of Resilience

A year ago, Bloomberg Economics’ view — and the consensus in the market - was that the price for bringing inflation under control would be a recession, likely starting before the end of 2023. That’s not how things have played out.

Growth in the second half of 2023 was brisk. Even as first quarter GDP missed expectations, a 3.1% expansion in sales to US firms and households  showed the economy continuing to hum along at the start of 2024.

“Bottom line” tweeted Jason Furman, the former head of the White House Council of Economic Advisors, the GDP data “confirms that the real side of the economy remains very healthy but the nominal side is too hot.”

There are three possible explanations for why crystal balls turned cloudy. The first, advanced by proponents of Modern Monetary Theory, is that higher interest rates are boosting income for consumers. If that’s correct, Fed hikes are a growth driver rather than a drag, and the answer to high inflation is rate cuts.  

That’s a provocative idea — and one that has captured the interest of the markets. But it’s hard to back up with either theory or data — the numbers show net interest income has been a drag on spending power.

A second possibility is that US growth potential — and so also the level of interest rates needed to cool inflation — has gone up.  “It could be that we have a higher neutral funds rate” said Cleveland Fed President Loretta Mester, speaking in March. If that’s the case, then the 525 basis points of hikes the Fed has delivered since March 2022 aren’t sufficient to tame inflation. More are required.

In theory, that’s possible. In the last three years, millions of immigrants have arrived in the US — promising a boost to the size of the labor force. President Joe Biden’s industrial policy aims at a made-in-the-USA manufacturing renaissance. New technologies like artificial intelligence hold out the prospect of a step change in productivity.

In practice, that idea too finds little support in the data. It takes time for migrants to integrate into the labor force. Investment growth is below trend. An AI-driven productivity surge remains more science fiction than fact.

That leaves the third and — in our view — most plausible explanation: Powell’s December pivot.

At the press conference following the December meeting of the Federal Open Markets Committee, Powell struck a markedly dovish tone. Taking the markets by surprise, he acknowledged the Committee had discussed conditions for rate cuts and wouldn’t have to wait till inflation was at 2% in order to move.

The Fed hadn’t shifted policy — the federal funds rate hadn’t moved — but Powell’s language had changed, and that sent an important signal.

Terminal clients: Global Rate Cut Bet Unwind Still Weighing In Dollar’s Favor 

Ellen Meade — a professor at Duke and former Fed economist — was an early pioneer of using advances in data science to parse monetary policy signals. “Our analysis shows that natural language processing can strip away false impressions and uncover hidden truths about complex communications such as those of the Federal Reserve” she wrote in a 2015 research paper, co-authored with Miguel Acosta.

Inspired by Meade’s early insight, Bloomberg Economics has built a Fed sentiment index. It’s based on a natural language processing algorithm trained to read news headlines on Fed speeches and press conferences, and score them on a scale from ultra-hawkish to super-dovish. Following Powell’s press conference, the index lurched dovish – showing the Fed had moved markedly closer to delivering a first cut.

Dovish Dodge

For the markets – and the economy – Powell’s words matter. Anticipating an earlier than expected rate cut, the benchmark two-year Treasury yield fell from 4.7% the day before Powell’s press conference to a low of 4.1% in mid-January. The impact of those lower borrowing costs, and a renewed rally in the equity markets, rippled through the economy – giving a fresh impulse to growth.

How big was the impact? Building on a model developed by Michael Bauer and Eric Swanson – respectively economists at the San Francisco Fed and University of California, Irvine – we find it was the biggest monetary policy shock of the current cycle. The stimulus impulse was bigger even than that of Powell’s dovish words following the collapse of Silicon Valley Bank in March 2023.

What would have happened if Powell hadn’t delivered that December surprise? It’s impossible to say with any certainty. Our view, though, is that the US was headed for a recession.

Data at the time showed the three-month moving average of unemployment rising from a low of 3.5% in early 2023 to 3.8% in October — edging toward the 0.5ppt increase that typically signals the start of a downturn. The Fed’s Beige Book — a collection of anecdotes about the state of the economy — confirmed the bleak picture. Payroll numbers looked robust, but census data from the Bureau of Labor Statistics suggest the final revisions released in 2025 will be lower. 

Powell’s pivot occurred at the perfect moment, and hit with sufficient force to prevent a downward spiral. Unfortunately, there’s now a price to be paid. Giving a fresh impulse to growth does the same to inflation. We estimate Powell’s pivot likely added about 0.5ppt to inflation over the year.

That, in our view, is why inflation prints so far this year have come in above consensus expectations. Indeed, the upside surprises have come in no small part from the financial sector – the part of the economy that responds most rapidly to Fed signals. It’s also why we’re forecasting core inflation will end 2024 above 3% - up from 2.8% in March and drifting even further above the Fed’s 2% target.

Getting Burns-ed

For Fed chairs with an eye to the verdict of history, the example to avoid is Arthur Burns — vilified for his failure to tame runaway inflation in the 1970s. The one to aspire to is Paul Volcker, who inherited double digit inflation from Burns and did what it takes to get it back under control — even at the expense of a recession.

For Powell, the benefit of the December pivot is that it kept the US economy on track for a soft landing. The risk, as he stepped on the growth pedal and reignited the inflation impulse, is that his reputation gets Burns-ed.

Perhaps that’s why Powell has already started a reverse pivot. Speaking on that April panel on the margins of the IMF meeting, he acknowledged that the “recent data have clearly not given us greater confidence” on disinflation. The Fed can keep rates steady for “as long as needed” to bring price changes back to target, he said.

Our Fed index picked up the change in tone, edging hawkishly higher and signaling that a first rate cut had retreated further into the distance. At the same time, it showed that Powell’s April shower had only doused a fraction of the dovish impulse unleashed by his December pivot.

That raises the question, will the Fed have to deliver more hawkish surprises in order to tighten financial market conditions, and bring disinflation back on track? In our view, the answer is yes. And that process could start as soon as this Wednesday’s press conference.

That would repeat the pattern from early 2023, when a dovish pivot following the collapse of Silicon Valley Bank was followed by a hawkish reversal after the shock faded. FOMC meetings in June and July, and the annual symposium at Jackson Hole in August are additional opportunities to deliver a hawkish surprise.

Markets have already shifted expectations for Fed cuts this year from 160 basis points in forecasts at the end of 2023 to 35 basis points in late April. That’s plausible – favorable statistical effects mean inflation is set to drift down into the summer. A July cut is still in play.

Even so, mid-year will likely mark a low point for core inflation before it starts to drift up again into year end. Anticipating that, the Fed may well duck the July meeting. At that point, with inflation rising on an annual basis and the Presidential election looming, the window for cuts may be closed till the end of the year.

The last few years have been humbling for forecasters. In 2021, few saw the post-pandemic surge in inflation coming. In 2022, even with prices spiraling higher, few expected the Fed to raise rates above 5%. In early 2023, the consensus was that the price of bringing inflation back to earth would be a recession.

It was, in our view, Powell’s December pivot that enabled the US to pull up ahead of a hard landing. In the months ahead, a reverse pivot could yet mean that the landing – when it does finally occur – is harder and bumpier than many in the markets expect.

Methodology

The Fedspeak Index is an NLP index that aims to quantify the Fed’s communication sentiment. We employ a machine-learning model, fine-tuned and trained on thousands of human-labeled annotations of Bloomberg news headlines on Fed officials’ communications, set to mimic our interpretation of the text.

The sample covers more than 6,200 unique speaking engagements and statements made by FOMC members since 2009, totaling more than 60,000 Bloomberg News headlines. After filtering the headlines for relevancy, the NLP algorithm was applied to about 47,000 relevant headlines. The algorithm is 71% accurate at scoring headlines between hawkish, dovish and neutral, and 96% of scores are within one-point of the true reading.

The monetary-surprise model is an extension of the Bauer and Swanson (2022) structural vector autoregressive (SVAR) model to include both the risk and the credit sub-index of the Chicago Fed Financial Condition index and the unemployment rate, in addition to the CPI, industrial production, and the two-year Treasury yield. We use monetary-policy surprises in Bauer and Swanson (2022) as a proxy for the monetary-policy shock. The SVAR is estimated using data from 1990-2019 with Bayesian methods.

--With assistance from Stephanie Davidson.

©2024 Bloomberg L.P.