Is the Fed data-driven or data-ridden?
The Federal Reserve constantly reminds us that its decisions are “data-driven” or “data-dependent.” But what does that even mean? And what are the dangers when data loom too large?
To understand the role of data at the Fed, it’s essential to start at the top. Here is Fed Chair Powell opening the press conference last week:
Today, the FOMC raised its policy interest rate by 1/4 percentage point. Since early last year, we have raised interest rates by a total of 5 percentage points in order to attain a stance of monetary policy that is sufficiently restrictive to return inflation to 2 percent over time. We are also continuing to reduce our securities holdings. Looking ahead, we will take a data-dependent approach in determining the extent to which additional policy firming may be appropriate.
And then later in the Q&A:
JEANNA SMIALEK: I wonder if you could tell us whether we should read the statement today as a suggestion that the Committee is prepared to pause interest rate increases in June.
CHAIR POWELL: A decision on a pause was not made today. You will have noticed that, in the statement from March, we had a sentence that said the Committee anticipates that some additional policy firming may be appropriate … So that's a meaningful change that we're no longer saying that we anticipate. And so we'll be driven by incoming data meeting by meeting.
It sounds good, but it’s not well-defined. And misunderstandings about the concept, especially outside the Fed, lead to the data taking many for a ride.
Tomorrow, we receive consumer price inflation for April, and the monthly increase is expected to remain frustratingly high. With the Fed telling us data are the star, it’s unsurprising that markets and Fed watchers wait breathlessly for every data release to predict the Fed’s next move. Here’s one of many examples:
US inflation data in the coming week will provide clues about whether the Federal Reserve can pause its series of interest-rate hikes at next month’s meeting.
The core consumer price index, which excludes food and energy, probably increased by 5.5% in April from a year ago after a 5.6% increase a month earlier. The annual advance would suggest just a slight moderation in the pace of underlying price pressures, and that inflation is proving persistent.
Let’s be clear; the Fed “can” do anything it wants. It can cut rates in June if it wants or raise them by 50 basis points, which it effectively did by hiking into the banking crisis. I agree that the CPI and other data will inform the Fed’s decision. But too often, it feels like the Fed hides behind the “data-driven” label.
Data do not speak for themselves.
The data do not tell us what they mean. We must always draw inferences. We must decide which data deserve attention, which models to feed them into, and how they should help inform decisions. Data are not and must not be driving. Hyping a handful of data points, especially preliminary ones, is the path to bad decisions.
Judgment is always in charge at the Fed. I worked for a decade at the Fed on the staff’s forecast. It’s the one getting attention now for saying a recession is likely in the second half of this year. How did a recession make it into the forecast? The senior staff made the call and put it in. No one data point, frankly, no constellation of data points, tells us affirmatively that a recession is the most likely outcome. In the end, it is about judgment. And not everyone agrees. Chair Powell told us he looked at the same data and came to a different conclusion.
JEANNA SMIALEK: And I also wonder if the Fed staff has in any way revised their forecast for a mild recession from the March minutes. And, if so what a recession, like what their envisioning would look and feel like when it comes to, for example, the unemployment rate.
CHAIR POWELL: So the -- the staff's forecast is - - so let me say -- start by saying that that's not my own most likely case, which is really that the economy will continue to grow at a modest rate this year … So, broadly, the [staff’s] forecast was for a mild recession, and by that I would characterize as one in which the rising unemployment is smaller than is has been typical in modern era recessions
Data are crucial, but Chair Powell and the rest of the FOMC, especially the voting members, drive. The Fed is judgment-driven. Disagreements are healthy:
CHAIR POWELL: Staff produces its own forecast, and it's independent of the forecasts of the participants, which include the governors and the Reserve Bank presidents, of course. And we think this is a healthy thing, that the staff is writing down what they really think. … So, it's actually good that the staff and individual participants can have different perspectives.
One could argue that the range of perspectives is too small at the Fed and the push for consensus too strong, but his remarks are heartening. It is no small feat to go against the judgment of the Chair, and that says a lot about what the staff sees now.
Some data are largely ignored.
Which data get attention and which are ignored is crucial. Again, it’s a judgment. In the past few years, profit margins and strategic behavior by firms are examples. Profit margins are high, and purchase volumes are declining. People are becoming more price sensitive. Demand is declining; it’s inflation that isn’t. Here’s one reason:
Companies from automakers to hoteliers keep on sacrificing sales volume – sometimes by design, sometimes by necessity – in favor of higher prices, a dynamic that will test the Federal Reserve’s efforts to rein in inflation …
Case in point: Kimberly-Clark Corp., the Dallas-based maker of Huggies diapers and Kleenex tissues, raised prices by 10% across all its categories last quarter. Sales volumes did fall by 5%, but its gross margin rose to 33% from about 30% a year ago.
But why isn’t demand coming down even more? One reason could be that income inequality is sizeable, and it will take more than higher rates and 10% inflation to break the spending at the top. And that the top 20% of households account for about 40% of aggregate consumer spending.
There is also evidence that the auto industry is intentionally limiting supply to boost profits, even as supply chains are largely cleared up and labor shortages addressed, but automakers are constraining production:
Ford Motor Co. said this month it is aiming to maintain robust sticker prices, even if that means rolling fewer cars off its assembly line …
The pricing tactics reflect lessons of earlier phases of the pandemic: When Ford and its rivals struggled with a chip shortage, they saw running with a thinner backlog of cars had profit upside.
Automakers’ decisions feed into the Fed’s super-core inflation—PCE services excluding shelter—which includes vehicle insurance, repairs, rentals, leases, and other vehicle services. These categories currently contribute a non-trivial amount to super-core inflation. You don’t hear the Fed talk about these data points on price setting. Profit margins get the textbook treatment instead:
CHRIS RUGABER. [Profit margins] have expanded, did expand sharply during this inflationary period. And while there are some signs that they are starting to decline, many economists note they haven't fallen as much as might be expected, given that we're seeing at least some pullback among consumer spending. So speaking of causes of inflation, do you see expanded profit margins as a driver of higher prices?
CHAIR POWELL. So higher profits and higher margins are what happens when you have an imbalance between supply and demand, too much demand, not enough supply. And we've been in a situation in many parts of the economy where supply has been fixed or not flexible enough. And so, you know, the way the market clears is through higher prices. So to get, I think, as goods pipelines have gotten, you know, back to normal so that we don't have the long waits and the shortages and that kind of thing, I think you will see inflation come down. And you'll see -- you'll see corporate margins coming down as a result of return of full competition where there's enough supply to meet demand. And then it's -- then you're really back to full competition. That's -- that would be the dynamic I would expect.
Many data points, including the ones above, don’t point to “full competition.” Chair Powell does not have to call it greed to call attention to them. We live in a shareholder economy more than in the 1970s, and some industries with fewer players. That’s reality. And we must examine the data and models that help us understand reality.
No, the Fed cannot levy extraordinary profits. No, the Fed cannot require producers to produce more to meet demand. But it can use its bully pulpit to draw attention to the behavior. Putting all the data on the table leads to good policy. Just because demand is the only lever the Fed can pull does not mean it cannot talk about other levers.
Again, the data cannot speak for themselves, so give them a voice.
What happens when data drive?
The Fed constantly says the data drive can lead others to believe it, and that’s problematic. Outsiders may put too much weight on the data and get confused about how the Fed interprets them. Here are two examples where it can go astray.
When markets are told that data drive the Fed, then it’s a natural conclusion for data to drive the markets. And then the loop goes back to the Fed, which does not want to surprise markets. And that is how the Fed can end up being data ridden.
We have had multiple meetings in this cycle where events happened within the blackout period before the FOMC meeting, most notably the failure of Silicon Valley Bank and the opening of a new lending facility, and the Fed did not have (or did not create) an opportunity to communicate its thinking ahead of the vote.
Events in the world are some of the most important data points and some of the hardest to ‘guess’ what the Fed is thinking.
Seasonal factors and revisions.
It’s what you thought you knew but didn’t that breaks your heart. That’s what data revisions are. And they are common, and whether due to changes in underlying data or the seasonal adjustment, they are frustrating. Sometimes they ‘change the story.’ Jumping up and down over a preliminary number is fraught and done all the time.
Seasonal adjustment, which takes out the regular calendar patterns, is important and is a source of the revisions. These procedures usually stay in the background, but it's not pretty when they move out of the darkness. I won’t dive into the details, and I have the utmost respect for the statisticians making these adjustments. But it’s been hard, given the massive swings in spending in recent years. And then, a new source for the price data was introduced in some critical components like new vehicles. The seasonals, revised recently, were calculated across the spliced-together series. And the other complication is that different data sources update their seasonals at different times. The payroll survey does it every month; the CPI does it in February, and PCE does it in November. These things matter.
Data are absolutely essential to policymaking at the Fed. Even so, it’s the people, specifically the people on the FOMC, who decide how to interpret the data and whether to use them. It’s not as simple as reading the latest data release and comparing it to consensus. Looking at the data is a worthy exercise, but it’s not the verdict. Remember that tomorrow, when the CPI rolls in.
Don’t let the data take you for a ride.
If you enjoyed today’s post, please consider subscribing. Financial support is also greatly appreciated. Thank you!
After reading Christopher Leonard’s excellent The Meat Racket , I can see why food prices can be controlled by a small group of wealthy meat corporations. I wonder if getting the Anti trust department involved might be worth including in the debate. Also the media could play a better role but generally doesn’t bite the hand that feeds it Pampers and Huggies have over 50 percent of market share which leaves the consumer in deep doo doo!
Excellent post. I'm just a retired academic of very little brain, but I learned long ago in trying to improve student learning that data are meaningless until purpose turns them into information.