Executive Summary

This document presents a professionally formatted transcript and analytical summary of the Rate Check podcast by Deutsche Bank Research, recorded on October 7, 2026. The episode features macro and market insights from hosts Henry Allen and Shreyas Gopal, alongside guest analysis from Senior European Economist Yacine Rouimi and Energy Analyst Michael Hsueh.

The discussion centers on two primary market developments:

• The European Sovereign Debt Market and developments in France: With Yacine, the hosts discuss a historic widening of the French-German (OAT-Bund) 10-year government bond yield spread, the newly presented French budget, and implications for the ECB if market volatility persists.
• The Global Energy Market: With Michael, the hosts discuss the divergent trends between normalizing crude oil flows from the Strait of Hormuz and persistent bottlenecks in refined products, as well as the price outlook for oil under various different scenarios.

Cleaned & Structured Transcript

Introduction

[00:00]

Rate Check, where macro meets markets, a podcast from Deutsche Bank Research.

Henry Allen: Hello and welcome to Rate Check, a podcast from Deutsche Bank Research. I'm Henry Allen, macro strategist, and as ever, I'm joined by my co-host and FX strategist, Shreyas Gopal.

Shreyas Gopal: Hey Henry, great to be back.

Henry Allen: It looks like a lot has been happening in markets in the last couple of weeks since we last recorded. Run us through what's been going on.

Shreyas Gopal: There is almost too much to cover in one go, Henry, but I would highlight two themes of particular note. Interestingly, there is a stark difference between the two.

The first is in the energy space, where we have seen a lot of headlines and news, but that hasn't necessarily translated into significant volatility.

The second theme—which will be the focus of our first half—is the rise in French yields, which has been the most volatile segment of the fixed income markets of late.

On the energy side, there have been various headlines around European countries agreeing to release more fuel reserves. This seems to have reduced the prospect of the US imposing a diesel-export ban. At the same time, the actual flow of oil through the Strait of Hormuz appears to have improved in recent weeks. Despite these developments, energy markets are in a very similar position to our last recording, with the front-month Brent contract hovering around the $100 per barrel mark. I am looking forward to diving into these drivers and forecasts in detail during the second half of this episode with Michael Hsueh, who joins us from Singapore.

Deep Dive: French Yield Volatility and the OAT-Bund Spread

[03:00]

Shreyas Gopal: The second theme relates to French yields, which have emerged as a broader driver of European markets across various asset classes. To provide some context: French specialists and bond market experts have long been aware of the growing spread between 10-year French government bonds (OATs) and their German equivalents—the OAT-Bund spread. This borrowing cost differential had been steadily rising from about 60 basis points six months ago to around 100 basis points by mid-September.

From a broader market perspective, such as in the currency space, this widening had very little impact until last week. That changed last Thursday when we saw sharp moves in the European Government Bond (EGB) space. The 10-year yield spread between France and Germany widened sharply to nearly 160 basis points by the end of the week, reaching levels not seen since 2011–2012. Henry, I think you have a statistic regarding the pace of that move.

Henry Allen: Yes. Looking at Bloomberg data going back to German reunification in 1990 on a daily basis, a weekly spread widening between the 10-year French and German yields of over 30 basis points has never been observed. In terms of the weekly scale of the move, it was historic. Even if, in absolute terms, it only returned to the levels of 2012, the weekly widening pace was unprecedented for decades.

Shreyas Gopal: Another detail that caught the attention of FX and broader market participants was the movement at the front end of the curve. Usually, we expect the two-year point of the curve to be driven primarily by monetary policy expectations and the ECB, without seeing major dislocations between France and Germany. However, last Thursday saw very sharp moves at that short end, which translated into a broader weakening of the euro. Traditional safe-haven currencies, like the Swiss franc, have performed well over the past few sessions.

Some of these moves have partially retraced at the time of recording on Wednesday, amid a busy news cycle regarding the French budget. The figures presented in the budget last Thursday were broadly in line with expectations, suggesting other market dynamics were at play. For instance, our rates colleagues noted that market structure and positioning elements may have amplified these moves.

Today, we will focus on the fundamentals: the numbers, policy options, and implications for the ECB. We are delighted to be joined by Yacine Rouimi, our senior European economist who covers France in close detail. Thank you for taking the time to join us, Yacine.

Yacine Rouimi: Thank you, Shreyas. Thank you, Henry.

French Budget Fundamentals and Fiscal Tightening

[06:00]

Henry Allen: Yacine, France has not run a budget surplus since 1974—over 50 years ago. Could you explain why the deficit is at its current level, how we arrived here, and how much fiscal tightening we can expect in the upcoming budget?

Yacine Rouimi: Those are very valid points. Let us first look at the targets. The government is aiming for a deficit of 5.0% of GDP in 2027, down from 5.4% this year. Looking at these numbers on the surface, one might assume the fiscal effort is minimal. Under the hood, however, there is a substantial amount of fiscal tightening in this budget.

The reason the deficit remains elevated is because several significant, pre-committed spending items are pushing aggressively on the deficit—specifically, debt interest and defense. There are also structural areas that are difficult to curb, such as pensions and healthcare, though the government is attempting to slow their growth. To put this in perspective, the combined increase in debt service and pensions next year represents an additional €24 billion in spending, equivalent to 0.8% of GDP. This is highly significant.

These spending pressures are a bit stronger this year because a large portion of public spending in France is indexed to inflation. This upward pressure on select non-discretionary items is a persistent trend. In fact, cutting the deficit today is more difficult than at any point over the past 15 years.

However, none of this is a surprise to economists. The targets have been known for some time, and there were very few surprises in this budget. To answer your question, Henry: there are pockets of tightening, though they are offset by structural costs. By historical standards, this is a tight budget. It contains €43 billion of consolidation measures on top of the €11 billion of measures already in the pipeline from previous budgets.

Crucially, there is sufficient tightening to remain compliant with European rules. The European Commission assesses "net expenditure," a metric that excludes rising interest costs and the cyclical impact of a weaker economy. These mitigating circumstances will play in France's favor.

ECB Intervention and the Risk of Contagion

[09:00]

Henry Allen: Yacine, the financial stress of the past week has spilled over into other asset classes, including credit and the euro. Do you think the ECB will be able to continue raising rates if this financial stress persists?

Yacine Rouimi: Yes. Generally, the ECB remains highly sensitive to market stress. As President Christine Lagarde noted in her testimony to the European Parliament, the ECB considers rising long-term yields as an organic brake on the economy. Higher yields slow growth and reduce the pass-through of inflation, lowering overall inflation forecasts and reducing the need for an aggressive policy response.

We believe the probability of a highly restrictive ECB stance has decreased. Given the scale of the energy inflation shock, the potential for second-round effects, and the stickiness of core inflation, we still expect the ECB to deliver one final hike to 2.75%, or potentially 3.00%. However, raising rates to 3.25% or 3.50%—as markets priced in until recently—would be excessive in our view.

Henry Allen: What happens if the sovereign debt sell-off becomes disorderly? Under what conditions would the ECB step in to stabilize the markets?

Yacine Rouimi: The rules of engagement are clear. The ECB's toolkit, specifically the Transmission Protection Instrument (TPI), is designed to counter unwarranted and disorderly market dynamics. In this regard, France remains technically eligible for ECB support because the 2027 budget is compliant with its European fiscal obligations.

However, we believe the ECB will hesitate to intervene immediately, as there is a consensus that the current pressure on French sovereign debt is tied directly to domestic fiscal choices. The ECB will want to avoid moral hazard. They will monitor the situation closely, focusing on whether disorderly market movements threaten financial stability or transmission of their monetary policy mandate.

The ECB would become particularly concerned if we saw spillovers into the French financial system or contagion to other Eurozone member states. Fortunately, the risk of systemic spillover is lower now than during the Eurozone sovereign debt crisis. The domestic banking sector is highly capitalized and has limited exposure to French sovereign debt, insulating the financial system and preventing a sovereign doom loop.

If a sell-off triggers broader European contagion and forces the ECB to act, we would expect a temporary, targeted, and fully reversible operation with a clear exit strategy rather than an open-ended commitment. That said, the ECB will act as a backstop. It will not tolerate irrational pricing, liquidity spirals, or one-way flight-to-safety dynamics.

We have a few months before the 2027 budget is formally approved and six months until the presidential election. French fiscal fundamentals remain weak, and political uncertainty is high. However, we have seen encouraging signs of a pro-stability stance from political parties like Rassemblement National. The ECB stands ready to intervene, but the hope is that it will not need to.

Shreyas Gopal: That was fascinating. Thank you very much, Yacine, for joining us. Given how busy the calendar is, we look forward to having you back soon.

Yacine Rouimi: Thank you.

Global Energy Markets: Crude vs. Refined Products

[12:00]

Henry Allen: Another prominent theme in global markets is the oil market. Brent crude remains above $100 per barrel at the time of recording, and the US-Iran conflict continues. This is an opportune time to bring in our oil analyst, Michael Hsueh, joining us from Singapore. Welcome, Michael.

Michael Hsueh: Thank you, Henry. Hello, Shreyas.

Henry Allen: Michael, you recently updated your oil forecasts. Could you run us through your baseline and the key drivers behind your revisions?

Michael Hsueh: Yes, we have revised our oil price forecasts upward. In our base case, we are projecting Brent crude at $95/bbl for Q4. We have also modeled alternative scenarios: a diplomatic breakthrough could bring prices down to $85/bbl in Q4, while a prolonged disruption could drive prices to $105/bbl and subsequently to $130/bbl in 2027. This contrasts with our base case where we expect prices to average $75/bbl in 2027.

Several moving parts are driving this outlook. First, the geopolitical and diplomatic situation is noticeably worse than in June and July, prior to the breakdown of the Memorandum of Understanding (MOU). Second, we must account for the fact that crude oil flows out of the Middle East have improved. By mid-September, ship-to-ship transfer signals indicated improving transport dynamics beyond what was captured in the hard data.

By early October, we gained further clarity confirming that crude oil flows are essentially operating normally. In contrast, refined product flows remain severely disrupted, operating at only one-third of their February levels. We are seeing a distinct divergence between the crude and refined product markets, alongside a decoupling of the diplomatic situation from physical commercial traffic. Nonetheless, conditions look set to remain tight in the near term, with crude in a slightly better position than refined products.

Henry Allen: Your report also outlines a pessimistic scenario where Brent crude could spike to $130 or even $150 per barrel—well above this year's highs. What conditions would trigger that scenario?

Michael Hsueh: We established that $130 to $150 range because it represents the extreme scenario we originally thought might unfold much faster back in March. At that time, we estimated that a three- to four-month supply outage would push prices to those levels.

The primary reason we have not reached those levels is the remarkable demand adjustment we have seen. Some of this is structural demand destruction, but much of it is simply a reduction in refinery runs. Additionally, the rerouting of shipping around the Strait of Hormuz has been highly effective, allowing nearly normal crude flows to continue.

To enter that pessimistic $130–$150/bbl range, we would need to see a prolonged return of severe supply constrictions. With crude flows currently near normal, the diplomatic impasse alone is insufficient to drive prices that high. There has been a clear decoupling of the diplomatic friction from actual commercial traffic flows.

Risk Premiums and Refinery Bottlenecks

[16:00]

Shreyas Gopal: Michael, you published a piece exploring the disconnect between recovering physical oil flows and the persistent risk premium embedded in Brent prices. While you expect this premium to unwind in your baseline, it remains sticky. You pointed to the diplomatic impasse, but also to inventory drawdowns. What is the latest on the inventory front?

Michael Hsueh: There are two main reasons prices have not adjusted downward. First, the diplomatic environment is vastly different from the June MOU. The June agreement represented an effective ceasefire, whereas now, re-escalation remains a constant threat. The market cannot confidently assume that healthy flows will persist uninterrupted for the next three months.

Second, between June and now, we have experienced three additional months of inventory drawdowns. Although crude drawdowns flattened out somewhat in June due to improving flows, we have since had consecutive months of declining crude inventories. Consequently, there is a larger inventory deficit to clear before the market can comfortably return to a surplus.

Shreyas Gopal: Looking at refined products, headlines regarding potential diesel-export bans have occasionally spilled over into the FX space over the last few weeks. The probability of a US ban appears to have diminished following the European and G7 fuel reserves release. How are you factoring these refined product risks into your forecasts? Has this risk receded, or does it remain active?

Michael Hsueh: It appears safe to assume that a US diesel-export ban is off the table, which President Donald Trump has publicly confirmed. This is important because a ban would have sidelined some US refining capacity at a time when the global market could least afford it.

However, we are still dealing with a significant amount of global refining capacity offline across the Middle East, China, and Russia. This structural bottleneck makes refined product prices a much larger issue than crude prices alone.

While crude prices are currently at similar levels to May and June, European gasoil (diesel) prices are 30% higher. This is not purely a crude supply story. As the industry saying goes, "nobody consumes crude oil; they consume gasoline, diesel, naphtha, and jet fuel."

The European reserve release and US regulatory adjustments allowing off-road red-dyed diesel to be used on-road will not have a lasting impact. They do not address the core supply-demand imbalance. If we project current conditions forward—steady crude flows but constrained refined products—crude prices may face downward pressure, but refined product prices are unlikely to see near-term relief.

Henry Allen: We have discussed the downside risks, but let us consider an optimistic scenario. Your report outlines a "diplomatic breakthrough" pathway. If we achieve a breakthrough, could oil prices fall as rapidly as they did in June during the run-up to the MOU?

Michael Hsueh: Yes, I believe they could. While the path to a diplomatic resolution is difficult to map out, a breakthrough would quickly shift the analytical focus back to the physical market. Analysts and market participants are generally predisposed to project significant surpluses in the medium term—surpluses that could easily exceed the deficits we have experienced over the last two quarters.

Therefore, crude oil prices could fall just as rapidly as before. The only caveat is that the baseline crude inventory situation is tighter now than it was in June. While I am not certain prices would return to the exact lows of earlier this year, a diplomatic breakthrough would undoubtedly trigger a swift and material decline in Brent crude.

Henry Allen: Thank you for joining us, Michael. It was great having you on the podcast.

Michael Hsueh: Thanks, Henry.

US Market Outlook and Closing

[21:00]

Shreyas Gopal: That brings us to the end of another episode of Rate Check. Before we close, Henry, what other global macro themes should our listeners monitor over the coming weeks?

Henry Allen: One of the most interesting market developments over the last week was how aggressively markets dialed back expectations of a Federal Reserve rate hike at the late October meeting. While a December hike is still priced in, the probability of an October move was sharply reduced. This shift was driven by downward revisions to recent US PCE inflation data and a weaker-than-expected payrolls report, which also included downward revisions to prior months.

Interestingly, despite the volatility in global bond and energy markets, the S&P 500 managed to close at another record high yesterday for the first time since August. We are seeing many divergent market signals pointing in opposite directions, all building toward a critical round of central bank decisions at the end of the month.

Shreyas Gopal: There is certainly plenty of ground to cover in our next episode. Thank you very much, Henry.

Henry Allen: Thank you, Shreyas.

This podcast has been produced by Deutsche Bank and may contain research as defined in MiFID II. The information discussed is believed to be reliable and has been obtained from public sources believed to be reliable, although Deutsche Bank makes no representation as to its accuracy or completeness. Opinions, estimates, and projections discussed constitute the current judgment of the speaker at the time of recording. They do not necessarily reflect the opinions of Deutsche Bank and are subject to change without notice. For further important information, please visit research.db.com.

This is Rate Check, where macro meets markets, a podcast from Deutsche Bank Research. To ensure you never miss an episode, subscribe now.

Disclaimer

This transcript has been generated using artificial intelligence and may contain minor inaccuracies or omissions. For complete accuracy and context, please refer to the original podcast recording here.

Executive Summary

This episode of Macro Matters on Podzept features David Wilcox, a senior fellow at the Peterson Institute and director of US Economic Research at Bloomberg Economics, discussing the Federal Reserve's approach to data collection and analysis, particularly in light of the Data Sources Task Force. The conversation highlights the immense breadth of data consumed by the Fed, from official statistics to anecdotal evidence and "alt data." Key themes include the evolution of data analysis at the Fed, the challenges and opportunities presented by big data and artificial intelligence, the critical importance of benchmarking alternative data sources, and the need to preserve the integrity and independence of federal statistical agencies. David Wilcox emphasizes that the Fed's engagement with data is not new but an ongoing effort to sharpen its vision of the current economic situation to better inform future monetary policy decisions.

Cleaned & Structured Transcript

1. Introduction to Macro Matters and Guest Introduction

[00:00] [Unidentified Speaker]: Podzept, the podcast from Deutsche Bank Research, with interviews on current economic and financial topics. Listen as economists and analysts from Deutsche Bank present their views.

[00:14] Matthew Luzzetti: Welcome. You are listening to another episode of Macro Matters on Podzept, a series where we discuss some of the best ideas coming out of Deutsche Bank Research. I'm Matt Luzzetti, Chief US Economist, and I'm once again joined by Matt Raskin, US Head of Rates Research. Welcome again, Matt.

[00:30] Matthew Raskin: Great to be back, Matt. Today's episode is the fourth in our series of podcasts with outside experts on the Fed's task forces. Today's will focus on the Data Sources Task Force, which has the objective of, "improving the quality and timeliness of economic signals that inform the Fed's policy judgments." This seems like an important topic for the Fed to consider.

[00:59] Matthew Raskin: For this discussion, we're very happy to be joined by David Wilcox. David is currently a senior fellow at the Peterson Institute and the director of US Economic Research at Bloomberg Economics. Prior to these positions, David served for many years on the staff at the Fed Board, including as Deputy Director and Director of the Division of Research and Statistics, or RNS. That division is responsible for developing data and analysis for use by the Board and the FOMC, for providing leadership on economic and statistical research, and for supplying data for public release. Before the Fed, David served as Assistant Secretary at the Treasury and as a Senior Economist at the Council of Economic Advisors. He really has an abundance of relevant experience and expertise for this discussion today. David, welcome. Thanks very much for joining us.

[01:56] David Wilcox: It's good to be with you, Matt and Matt.

[01:59] Matthew Luzzetti: Thanks, David. Very excited to have you here. You're probably a perfect guest to run through all these key topics from the data task force perspective. Maybe just to lay out how we're thinking about proceeding here. Upfront, we're hoping to give a little background on data analysis and data consumption at the Fed. Maybe a second section, we'd like to focus a little bit more on the role that AI might play, or big data, or alternative data sets could potentially play. This is something that I think has been highlighted by Chair Warsh. And then finally, we'll get into some specifics around this task force, what might come about and over what timeline.

2. Data Consumption and Evolution at the Fed

[02:41] Matthew Luzzetti: So David, maybe just to kick off, can you just give us a sense from your time at the Fed of the breadth of data and other information the Fed would typically consume over an FOMC cycle?

[02:56] David Wilcox: There's an immense amount of data that gets vacuumed up, synthesized, and digested by the Fed staff, both at the Board and at the 12 Reserve Banks. There's a common saying that "the Fed looks at everything" before they make a decision. And I think that's about right. I would qualify that by saying that they look at everything, roughly speaking, weighted by its information content. So important indicators are going to get a lot more attention than noisy ones or irrelevant ones.

[03:35] David Wilcox: You can think of what the high-profile series would be: things like job creation, the payroll employment numbers, the unemployment rate, inflation, of course. Those are, I've just outlined the two legs of the Fed's mandate, and so of course the Fed is going to pay most attention to those indicators that bear directly on the dual mandate. But in terms of the mechanics of it, they're going to look at a huge array of official economic statistics from US statistical agencies: the BEA, BLS, Census Bureau.

[04:19] David Wilcox: They're also going to look at anecdotal evidence. For example, prior to each FOMC meeting, staff at each of the Reserve Banks contact a really astonishing array of people in their respective districts to hear their perspectives on the current economic situation. That means they're talking with and consulting with members of the board of directors at each of the Reserve Banks who themselves come from many different professional walks of life. They're talking to businesses, community organizations, economists, market experts, labor representatives, consumer representatives. And all of these are compiled into a document called the Beige Book, which is published on Wednesday, two weeks before the Open Market Committee meeting.

[05:16] David Wilcox: Increasingly, beyond all that, the Fed has also paid close attention to other indicators of current economic activity, including so-called alt data. And those efforts have been aimed at attempting to take advantage that we live in an era where we're bathed in economic information. A lot of it arises more or less by accident as a side byproduct of the core activity of a business, for example, a payroll processor. But that doesn't mean that it's not potentially fantastically valuable. And so the Fed, like many other analysts in the economy, has become quite sophisticated about processing all of that and trying to distill a meaningful signal about what's going on, most especially what's going on with regard to the dimensions of the economy that matter for monetary policy, namely price stability or inflation and maximum employment.

[06:33] David Wilcox: I should also mention, by the way, that the Fed for a long time has itself been a producer of economic data. It has been in the measurement business itself. The industrial production and capacity utilization operation has been going on at the Fed for over a century, and I have the coffee mug to prove it, by the way. In addition, the financial accounts of the United States, formerly known as the flow of funds accounts, come out of the Federal Reserve. The Fed publishes data on commercial paper and a lot of other important data.

[07:19] Matthew Luzzetti: That really outlines how much information processing Fed officials are doing ahead of each FOMC meeting. I wonder how that might have changed over time. What data do they emphasize? Some of the official public statistics versus private estimates, whether web scrape versus manual collection data, surveys, anecdotes versus hard data? Can you give us a sense of maybe how the emphasis on those has changed over time or at different parts or points in the cycle?

[07:53] David Wilcox: Today there's a lot of data that are available that simply didn't exist 10 or 20 years ago. The idea when I was a buck private on the staff at the Federal Reserve Board, the idea of web scraping was something that couldn't possibly have existed because that was in the pre-Internet era. And I think it's fair and accurate to say that the Fed itself and its staff have attempted to keep up with the times, but more impressively have even attempted to lead the development in these areas.

[08:36] David Wilcox: I'm very proud of the fact that 15 years ago we started in what was then called and I think remains, still goes by the name, "expanded measurement agenda." We were interested in putting in front of the Open Market Committee additional information beyond what was available from the statistical agencies, if that would help sharpen the committee's view about the current economic situation. And so we led the parade in many ways on the development of employment indicators, consumer spending indicators that would help not just be cool because we would know something nobody else would know, but precisely because it would be useful to committee decision-making along the way.

[09:36] David Wilcox: We drilled a lot of dry wells. We attempted experiments that didn't work out well in terms of providing an indicator that seemed like it had adequate reliability to make it be part of the decision-making information set for the Open Market Committee. But at the same time, we came up with some indicators which have been well documented, by the way, on the Board's public website in the form of working papers and such, that we thought were really very helpful to the committee in terms of making its information base more complete and giving it a little sharper vision about what was going on in the economy in real time.

[10:28] Matthew Luzzetti: Can I ask maybe the process by which you would determine if an indicator helped in the process or not? Is it benchmarked against some of the official data? Is it a leading relationship relative to other official data? How exactly would you determine whether a metric was useful? I think it's an important question as we sort through all these alternative data sets that are out there. How do we determine that something would actually be useful for monetary policy making?

[10:56] David Wilcox: That's a good question and it's one where I think much less attention is paid than generally ought to be paid. Let's take an example of the employment indicator that we developed. We were working with ADP, which was and remains, to my understanding, the largest payroll provider, payroll services provider in the country. ADP had approximately 20% of private employment as clients. That's comparable to the sample size of the establishment survey, the so-called CES, which is the survey of employers that gives rise to the job count on the first Friday of most months.

[11:51] David Wilcox: Some of the employers who were clients of ADP would also be reporters to the BLS. And so getting that information through ADP wasn't of much additional value to us. Where the incremental value came was from the fact that there would be ADP clients who were not BLS reporters. And by having access to their employment experience, we could broaden the range of information available to the Open Market Committee. The goal wasn't to know what employment was in a particular month because that's not knowable in real time as a practical matter. The goal was to take the range of uncertainty and narrow it a little bit if we possibly could.

[12:51] David Wilcox: And by augmenting or complementing the official BLS reading, unemployment in the preceding month, with additional information from ADP, we showed through well-tested statistical techniques that, yes, indeed, we were able to narrow the range of uncertainty a little bit about what was going on in that preceding month. You asked what's the process involved in demonstrating statistical value. It's a lot of hard, nitty-gritty demonstration work. It's showing the extent of revisions as time goes by. It's showing that the statistical pipes will work reliably, that if a data provider promises to have their information on a certain schedule, are they actually able in real time to perform on that basis?

[14:06] David Wilcox: That was particularly relevant, for example, in some instances where we were looking at natural disasters. A flood hit the city of Houston, and we had access on a very close to real-time basis to card swipe information, credit cards, debit cards, stored value cards. Remarkably, the provider of that information was able to get us information on a close to real-time basis, even in those dire circumstances. If the line went dark when disaster hit, that information source was nowhere near as valuable to us. But having information about what the spending experience was of retail establishments in the disaster-affected area did a lot to help us discern what the impact of that hurricane and associated flood would be in the upcoming retail trade estimates. So it's not just proof of concept, it's proof in the real world. Can it perform under actual circumstances reliably? Does it narrow the range of uncertainty? Does it give us access to a wider view, a wider angle view of the US economy?

[15:40] Matthew Raskin: That's great context, David. It also makes very clear that the Data Task Force's focus on improving the quality and timeliness of data available to the Fed is not something new. It's a part and parcel of what the Fed staff do in the regular course of business. So I'm curious how that historically has worked. Has there been any sort of centralized push within the Fed system around this, or is it more distributed and organic, and the sections that are responsible for covering consumption or the labor market, in the course of their business, identify and on a routine basis enhance the intake and processing and analysis of data?

[16:31] David Wilcox: What I can speak most knowledgeably about is what the practice was at the Board. That's where I served, after all. And there, responsibility for what we called current analysis, which was monitoring of the situation in the economy today, was distributed across sector specialists. So we had a consumer spending sector specialist, we had a labor market specialist, a prices and wages specialist, a business investment, an international trade, and so forth. And the encouragement was for each of those people to be entrepreneurial, to go out and seek new sources of information that would help illuminate the picture.

[17:23] David Wilcox: Look, I think it's actually useful to—here's a homely metaphor for what's going on. Operating monetary policy is a lot like driving a car with a very poor set of eyeglasses on. That set of eyeglasses, under the best of circumstances, corrects your vision only to, let's say, 20/100. You'd like to have 20/20 vision as you peer into the future. That's not the reality. Our goal was to try to correct the Open Market Committee's vision from 20/100 to, if we possibly could, 20/80. We wanted them to recognize with as much clarity as possible what the current economic situation was, and hopefully by doing that, sharpen our view of the future a little bit. We had no pretense that we were going to solve the more fundamental problem of forecasting accuracy, but we did believe and acted on the belief that if we knew the current situation of the economy a little more accurately, we could deliver that much more accurate forecasts of what would go on.

[18:46] David Wilcox: So the type of mental discipline of trying to go out and develop new information sources, that's been part of the DNA at the organization for a long time. Is there room for improvement? You bet. Absolutely. But it's not a new mental discipline to the staff, nor to policymakers. It's part of the muscle memory of the organization.

[19:20] Matthew Raskin: That's really helpful because I think that if you're not intimately familiar with the Fed and hear some of the implicit criticisms that have led to this and other task forces, you might be inclined to think that Fed staff and policymakers are reliant solely on official statistics and have a fairly static lens and approach to analysis around those. And I think what you've just outlined makes very clear that is not at all the case.

[19:57] David Wilcox: Look, that's very far. I get it and I agree that is a common perception. It's very far from the reality. And I would invite your listeners, go visit the Fed's working paper series and see the papers that have been published over the years reporting on the Fed's staff's efforts to use and actually implement the expanded measurement opportunities that were available to us, including, as I've mentioned, employment data coming from ADP, card swipe data coming from a company called Fiserv. And those are probably the two main flagship efforts, but that's far from the only efforts that have been underway.

[20:54] David Wilcox: I want to underscore that another part of the muscle memory of the organization is to understand deeply and interrogate the inner workings of the official statistics. Let me give you one famous example of this. It was Board researchers in the mid-1990s who developed some of the earliest evidence that the Consumer Price Index was overstating the true rate of increase in the cost of living. That prompted Alan Greenspan in the mid-1990s to testify to Congress to that effect. Congress heard that and was sufficiently agitated about that, that they established a Boskin Commission. That Boskin Commission came to the conclusion that, yes, indeed, the Consumer Price Index at that time was overstating the true rate of increase in the cost of living by a material amount, and that there were practical implementable steps that could be taken to remediate that situation. And to their great credit, BLS staff proceeded to do a lot of the groundbreaking research that was required to take on board the remedial steps. And that was a huge development and it came initially out of Board staff research.

[22:27] David Wilcox: Another famous example that really gave rise to Alan Greenspan's reputation as the "maestro," as is noted in the title to Bob Woodward's book, that circumstance came around when Greenspan became convinced that productivity was increasing faster than the official data were showing. And so Alan Greenspan worked with two of my colleagues at the time, Larry Sliffman and Carol Corrado. And they did the hard work to convince themselves that that was actually the case. And it was on the basis of that nitty-gritty, deep-in-the-weeds research that Greenspan was able to persuade his colleagues that it was prudent to allow the unemployment rate to run lower through the late 1990s than conventional wisdom had it. And they could do that without generating inflationary pressure. And so once again, that deep familiarity, that intimate knowledge of how the economic data are put together, in this case, importantly, driven by Chairman Greenspan himself, really generated enormous economic benefits for the country as a whole.

[24:05] Matthew Luzzetti: Maybe one recent example of that as well, maybe not as impactful as those that you mentioned. I remember Board staff research earlier this year looking at the software and accessories component of the inflation basket. We are going to get revisions to that later this month in terms of PCE. And I think it's building on some of that research, maybe stating that component in particular was overstated.

[24:33] Matthew Luzzetti: Another topic that comes up a lot, especially in recent years, has been the extent of revisions that we've seen in key economic indicators, payrolls being, I think, the number one indicator that comes up with this. Is your sense that official statistics such as payrolls have become less reliable over time?

[24:47] David Wilcox: I don't think so. I've actually taken a very deep dive in looking at precisely that question. I published a paper on the Peterson Institute's website where I go through in considerable detail what the revision history of the payroll employment numbers have been. And to be honest, there's a bit of a puzzle there. We've all seen that response rates have come down. And so it stands to reason that you might think that revisions would have gotten larger. But if you scale for the size of the labor force, if you scale for the fact that there are more workers in the workforce today than there were 10 or 15 years ago, there's no evidence that the size of revisions has gone up. What's remarkable is that by the time you get to the third estimate of employment, response rates remain really high. They're still in the 90% level. And so by the time we get to that third estimate, we still have a very good reading on what employment was. And it turns out we don't have a materially worse estimate even in the first reading, and that frankly is a bit of a surprise.

[26:18] David Wilcox: There are two cautionary flags I would point to. One is that it's true, response rates are down. Households and businesses have become more reluctant to participate in surveys, and that poses a serious threat. We've already seen in the United Kingdom where they had to suspend publication of their workforce survey, which is the equivalent of the survey that gives rise to our unemployment rate. And that was pretty much a national catastrophe for the United Kingdom. The second red flag is that we've seen some worrisome developments at the statistical agencies, particularly at the Census Bureau, that despite their high dedication traditionally to ethical standards to nonpartisan statistical publication, there have been some developments lately that have given us cause for concern. We can go into those in more detail, but what I want to leave your listeners with unequivocally is a sense of urgency that those of us who pay attention to economic data need to watch these developments really carefully, need to be on high alert for the possible subversion of the integrity of economic data. The fact that the US economic statistical system has long been regarded as the gold standard around the world for providing quality information about the condition of the US economy, that's a national asset. And we need to work hard to preserve the value of that asset.

[28:13] Matthew Raskin: Is there anything in particular there that you're mindful of or looking to for evidence of subversion of this data? What should our listeners be focused on there?

[28:25] David Wilcox: There are a couple of things. Let me reframe your question just slightly, which is how will we know that something's gone wrong? I think a couple of ways. One is, first and foremost, we'll probably see a wave of resignations from the agency wherever the malpractice has occurred. These people are dedicated public servants. They're committed to doing the job in the traditional manner, and I have every confidence that if there's an attempt at manipulation for political purposes, the civil service, the core engineers of the statistical system, won't stand for it and they will leave. They will let us know and their whistleblowing will be absolutely essential for helping put in place corrective actions. We haven't seen that yet.

[29:30] David Wilcox: Second, and partly, frankly, as a consequence of some of the enhanced data techniques that we've been talking about, there's a pretty wide community of outside analysts who have a fine microscope on the information that's being put out by the BLS, the BEA, and the Census Bureau. And so we'll have good capability, not perfect capability, to detect if there are any marked departures from existing techniques. This actually arose in a rather obscure way, which turned out not to involve any statistical meddling, but I think it demonstrated the surveillance capability of the outside community. And that occurrence involved a little tiny component of the PCE price index, which is the inflation measure that the Fed tries to control to 2%. There was a methodological change there that BEA implemented without announcing to the outside world. And it caused inflation specialists to get pretty agitated because some of the people who had the best track records for forecasting accuracy had a big forecast error that month. What the heck had happened? BEA didn't announce it in their statistical release. And so they called the BEA office and the staff person on the other end of the line said, "Oh yeah, we made a methodological change." That should have been communicated ahead of time. It wasn't. And in my view, that was a communication, to adopt World Cup vocabulary, that error was an own goal of quite a serious type. But what came out of that was a demonstration of the proof of concept that the outside analyst community has pretty good surveillance capability on what's going on inside. You can check for consistency. You can test whether the numbers are adding up.

[31:58] David Wilcox: Because we're now doing tons of web scraping and data accumulation from other sources, we're able to develop independent estimates. We won't have perfect ability to determine whether some meddling has occurred, but it's a lot better than nothing. And I think that small experience with the legal services index demonstrates the capability quite well.

[32:34] Matthew Luzzetti: David, I was actually going to raise exactly that point specifically for legal services because I remember very well over the past year when you had the PCE pulling from the PPI rather than the CPI. But now we are also getting a revision to that methodology more formally at the end of this month, which I think is also probably a positive outcome there.

[33:00] Matthew Luzzetti: Another data topic that gets a lot of attention, and I am certainly focused on, in particular how it feeds into monetary policy decisions, is this idea around residual seasonality, the potential for even seasonally adjusted data to still show some seasonal fluctuations. For example, it looks like things like inflation, core inflation, is higher over the first few months of the year than it is over the middle few months of the year. And it probably creates a lot of complications for monetary policymakers that are trying to catch turning points in the data by focusing on shorter-run trends. So is this something that you are focused on? Do you think the Fed is fully aware of it and reflecting it into their own monetary policy decisions?

[33:39] David Wilcox: Yes, absolutely. Seasonality of all sorts is a distraction. It's a complication. It's a pesky annoyance. And in the best of all possible circumstances, traditional seasonal adjustment takes care of that. But if you've been around the business as long as either of you have or as I have, we've all lived through circumstances where routine seasonal adjustment doesn't get rid of all of the calendar-related effects that infest raw economic data series. And so, yes, absolutely. Residual seasonality is sort of like that pesky gnat that annoyingly flies around your head and you wish you could sort of bat it away. But I think it actually also goes, and here I want to put in a plug for economic measurement as a discipline. It goes to show that creating the most useful economic series is really tough. It's really challenging. This is not folks just merely being counting. This is conceptually difficult at the highest levels. It involves an intimate marriage with economic theory and data science and practical considerations like, are the promised data sources going to be delivered on time?

[35:19] David Wilcox: So seasonal adjustment is a bit of a field for unsung heroes. But boy, they're heroes in my book because they're dealing with tough and challenging problems. And you'd think that you've seen it all, and then some new calendar circumstance arises that flummoxes even a very well-designed seasonal adjustment routine. So yes, absolutely. It's one more on the list of challenges to making sense of the economic situation in real time.

3. Big Data, AI, and the Future of Economic Measurement

[36:08] Matthew Raskin: So David, maybe we can briefly turn to a discussion of big data and artificial intelligence and how that might factor into any efforts to improve the quality or timeliness of data sources for the Fed. So you're the Director of Economic Research at Bloomberg. Bloomberg has created a number of alternative data series using big data approaches. I wonder if you could just talk a little bit about that experience and what, if anything, you've learned from it.

[36:44] David Wilcox: I would say big data is the next frontier. But as I've suggested, it's actually a frontier we've been pushing on for the past decade, 15 years. Artificial intelligence is going to help us push forward on that frontier more aggressively and make faster progress. But it's already clear, it's eminently clear, this is going to remain really challenging. What you can do with big data can be pretty remarkable. You can get much more granularity, as my Bloomberg colleagues are demonstrating. They're scraping huge amounts of price data in real time and generating a lot more detail on pricing, consumer pricing, for example, than is available through the official economic statistics. And sometimes that can be really helpful to economic interpretation. For example, Anna Wong told me yesterday, Anna's the person who was heavily involved in precisely this effort, told me that that granularity is really helpful in certain cases in interpreting the impact of tariffs on particular categories of goods, because some items will be subject to tariffs, others won't. And it's very useful to be able to tell the difference between the two.

[38:26] David Wilcox: It's not the magic solution to the quest to understand the current situation of the economy with perfect clarity. Why not? Let me give you a couple of reasons. The main one is that big data sets are accidental creations. They arise from the business activity of Luzzetti Enterprises or Raskin Inc. Raskin Inc. was not established for the purpose of economic measurement. It was established for the purpose of making widgets. And so you might have great data on widget production, widget supply, widget materials upstream from your production facility. All of that's tremendously useful to you as a business owner engaged in the widget industry. It is not representative of the US national economic experience. And so while I'd love to get my hands on your proprietary data set on the manufacturer of widgets, I'm not going to mistake that for itself being a representative portrait of the US national experience.

[39:58] David Wilcox: Here's one idea I don't think I've mentioned before, which is the following. The Fed sets a national monetary policy. It doesn't get to set a separate monetary policy for the manufacturing industry versus the agricultural industry versus the construction industry. It doesn't get a separate monetary policy for the West Coast, separate from the Southeast, separate from the Northeast. It sets one national monetary policy. And so having a representative portrait of the nation is incredibly important. And what we find over and over again is, generally speaking, there's only one source that tells us what constitutes a national economic portrait, a representative portrait of the entire experience of the country. And that, generally speaking, comes from an economic census or other data product that's created by the official agencies.

[41:18] David Wilcox: Do I want to get my hands on vacuuming up information from every possible source? You bet. But the first thing I'm going to do is try to weight those private sources using federal economic indicators to make sure that I'm getting a balanced national portrait. And the experience of the past 10 or 15 years with big data construction is littered with examples where a failure to do that kind of balancing, which is oftentimes referred to as stratification, leads to misleading conclusions. So that's one example of how at one and the same time, big data collection, sophisticated AI-backed processing, incredibly valuable. And it's clearly an element of the future. It's also not the magic solution. And we all need to be attentive to creating data that actually are fit for purpose.

[42:31] Matthew Luzzetti: It's a really important point about benchmarking to official statistics, even with metrics that might be quite representative. So ADP themselves benchmarks their own data to the QCEW to ensure that they are actually capturing the national trends. So even a metric which in many ways is probably far more representative than other metrics, they're using government data to benchmark to ensure that they're actually capturing the trends, which to me is incredibly important.

[42:59] David Wilcox: If I can just expand on that for one second, Matt. I've been gone from the Fed for over 7 years, going on 8 years. So what's true today may be quite different from what I experienced when I left the Fed at the end of 2019. But even there, there would be times when people would come into my office at the Federal Reserve and they would say, "Boy, we're getting a very different answer than ADP is." What was remarkable about that statement was we were working with the very same microdata that ADP itself was, and we were deriving quite different conclusions about the current state of the labor market based on the very same information base. So I mentioned that by way of saying this is tough, hard work that demands creative insight, demands brilliance and experience. It's not a matter of just adding up an Excel spreadsheet and looking at the sum total at the bottom of the column.

[44:22] Matthew Luzzetti: Yes, and I think all of the topics that we mentioned as well about residual seasonality, seasonal adjustment, all of those topics become, in my experience, far more complicated when you're referencing these alternative data sets, private sector data sets. And that could be a source of some of the tensions or differences that you find.

[44:37] Matthew Luzzetti: Maybe just a final question on thinking through big data and AI, AI specifically. Where would you think it provides the most promise? Is it in collecting new data? Is it in analyzing existing data? Or is it in something else?

[45:00] David Wilcox: I kind of hesitate to limit the focus. I think it's all of the above. With AI, we will have the capacity to process a much broader spectrum of information. With big data, we'll be able to get much more granularity like Anna Wong was talking about, and we'll be able to understand the inner mechanisms, what's driving the current pace of inflation, what factors are important, what factors are not. All of that's going to be important, so I kind of hesitate to rule anything out.

[45:37] David Wilcox: I do want to introduce one other thought that's absolutely essential here to thinking about what's realistic, what's achievable, what's feasible. The job of monetary policy is to try to deliver price stability and maximum employment. But as your listeners will have heard, monetary policy operates with a lag. Monetary policy can do almost precisely nothing about employment conditions today, inflation today. When the Committee makes a decision as they did last week at the September meeting, that's going to have its maximum impact four quarters, six quarters, eight quarters ahead. And so the relevant dimension of uncertainty is not really what's going on today. The relevant dimension of uncertainty is what's the world going to look like one to two years from now?

[46:51] David Wilcox: Is it worth sharpening our vision about today's circumstance in order to get a better bead on where the economy will be four to six to eight quarters ahead? You bet it is. Don't misunderstand me. I'm not saying that that isn't the case. However, I do think it's incredibly important to be realistic about what's attainable here. The real dimension of uncertainty is that future dimension, not today.

4. The Data Sources Task Force: Hopes and Impact

[47:27] Matthew Raskin: Very interesting. So David, maybe just to wrap up, we'll focus in very directly on the workings of this task force. And I'll kind of bundle two questions around that. I guess the first is, given what we know about the leads of this task force, so Raj Chetty, Doug McMillan, and Kevin Murphy, what are some logical conclusions, findings, recommendations that you think they might reach? And relatedly, overall, what is your prediction for the impact of this task force on how the Fed collects and analyzes data that's relevant to the conduct of monetary policy?

[48:13] David Wilcox: So the answer to your first question is I really don't know. I have no idea what the specific perspectives of the three members, Raj Chetty, Doug McMillan, and Kevin Murphy, what they will bring to the table. That's part of why I frankly find this pretty intriguing. It's a little like a mystery story where I don't know what the final chapter is going to be. I think it's particularly intriguing in this case. They all bring quite different perspectives, among other things. Doug McMillan obviously brings the perspective of a CEO of a major corporation who's used to having very close to real-time certainty about the current condition of the business enterprise that he's running. Raj Chetty, Kevin Murphy, all three of them are intellects of the highest order. So my suspense or uncertainty has nothing to do with that. To my knowledge, none of the three of them have been involved in economic policymaking from a government perspective. And so I don't know how they will be able to map their expertise and their experience into the problem confronted by the Fed.

[49:46] David Wilcox: In terms of what impact they might have, let me contort that a little bit to addressing the directions I hope they will push in. I hope they'll give even greater momentum to the idea that the Fed needs to understand the dynamics of the economy as accurately as possible to fulfill its monetary policy mission. I hope I've given a sense that that is baked deep into the DNA of the organization, but there never should be a missed opportunity to reinforce the urgency of pressing ahead on that margin.

[50:36] David Wilcox: I hope that the task force will think carefully about what the proper assignment of governmental roles is here. I am deeply convicted that it's the Fed's job to conduct monetary policy. It's the BLS, BEA, and Census Bureau's job to measure the economy. And there is tremendously important, in my view, to keep those two functions separate. I think it would be quite unfortunate, for example, for the Fed to be put in charge of designing the inflation yardstick that will be used to assess the Fed's own performance. There's a tremendous amount of virtue in the fact that the measurement responsibility is lodged in a separate government agency from the monetary policy function. That doesn't mean that the Fed shouldn't be at the scientific frontier in terms of understanding the data, in terms of trying to augment the official measures, make its vision a little bit sharper. But I think there are some really serious hazards that could result from blurring responsibility across those two areas.

[51:58] David Wilcox: And I hope the task force will underscore how utterly important it is that we maintain a vibrant federal statistical system. The task of measuring the economy, as I've mentioned, belongs first and foremost to those statistical agencies. That becomes a classic public good. And for those who don't speak this vocabulary on a daily basis, what's a public good? It's something where if I benefit from it, my using that good does in no way diminish your ability to benefit from it. So when the BLS measures the labor market more accurately, that can benefit everybody around the globe. It can benefit Deutsche Bank and its customers. It can benefit the Federal Reserve. It can benefit foreign countries as they try to understand the US economy with a little greater sharpness of vision. It can benefit policymakers at the state and local level as they try to design policies that are appropriate for local economic circumstances. This is a positive-sum game. If we measure the economy better, the benefits flow broadly and they're not diminished by one actor using it.

[53:49] David Wilcox: So that's a pretty broad array of hopes that I have. I am foursquare behind the importance of prosecuting with as much vigor as possible the job of measuring the economy well. I do think there are some governance-related issues that need to be attended to to make sure that the Fed stays in its lane, very much consistent with the kind of message that Chairman Warsh has been saying, that it sticks to the conduct of monetary policy and doesn't fundamentally get involved in the publication or design of official economic statistics.

5. Conclusion

[54:33] Matthew Luzzetti: Well, thanks very much, David. I think that's probably a great place to leave off. It's been an incredibly insightful conversation for us, I think for our listeners, taking us through how the Fed over time has collected and assessed data. Some of the benefits and risks maybe around big data and AI, some of the considerations around revisions and residual seasonality, and then some of the hopes that we might see from this task force. So thank you for taking your time to speak with us today.

[55:01] Matthew Raskin: I'll just echo that. I learned a lot from this discussion, David. I really appreciate you joining us and actually leave just feeling really optimistic based on your expertise and thoughtfulness around the job that the Federal Reserve does in this and around all of its responsibilities. So I really appreciate your time and again, think it was a great discussion.

[55:25] David Wilcox: I really appreciate the opportunity to join you. And if this conversation results in a listener or two or five deciding, "You know what, economic measurement sounds like a pretty darn exciting career, I think I'm going to go into that," that would be all the better.

[55:43] Matthew Luzzetti: That'd be a big success, no doubt. Thank you, David. If you'd like more information on anything discussed today, please reach out to your Deutsche Bank sales representative. This is Matt Luzzetti and Matt Raskin. You've been listening to Macro Matters on Podzept.

[55:57] [Unidentified Speaker]: Podzept, the podcast from Deutsche Bank Research. This podcast has been produced by Deutsche Bank and may contain research as defined in Mifid II. The information discussed is believed to be reliable and has been obtained from public sources believed to be reliable, although Deutsche Bank makes no representation as to its accuracy or completeness. Opinions, estimates, and projections discussed constitute the current judgment of the speaker at the time of recording. They do not necessarily reflect the opinions of Deutsche Bank and are subject to change without notice. For further important information, please visit research.db.com.

Key Takeaways

1. Comprehensive Data Approach: The Fed consumes an immense breadth of data, including official statistics (BEA, BLS, Census Bureau), anecdotal evidence (Beige Book), and alternative data. This comprehensive approach is integral to informing FOMC policy judgments.
2. Evolution of Data Analysis: The Fed's engagement with data is dynamic, evolving from pre-Internet methodologies to actively leading "expanded measurement agendas" incorporating new data sources. This involves sophisticated statistical work to assess the reliability and value of new indicators.
3. Benchmarking is Crucial for Alt Data: While "big data" and "alt data" offer increased granularity and timeliness, they are often accidental creations not inherently representative of the national economy. Benchmarking these private sources against federal economic indicators is essential to avoid misleading conclusions.
4. AI's Role in Data: Artificial intelligence is expected to accelerate progress in data processing and analysis, providing greater capacity for broader information spectrums, enhanced granularity, and deeper understanding of economic mechanisms.
5. Integrity of Official Statistics: The integrity of US federal statistical agencies is paramount. Vigilance against potential subversion of economic data is crucial, as is maintaining the independence of data measurement from monetary policy functions.
6. Challenges in Economic Measurement: Issues like residual seasonality and data revisions highlight the inherent complexities of economic measurement. Experts dedicated to seasonal adjustment and robust data collection are essential "unsung heroes."
7. Monetary Policy Lag: The primary uncertainty for monetary policymakers lies in forecasting the economy one to two years ahead, given the lags in monetary policy effectiveness. Sharpening the vision of the current economic situation helps improve these longer-term forecasts.
8. Data Sources Task Force Expectations: The task force is anticipated to reinforce the urgency of accurate economic measurement and to emphasize the importance of distinct governmental roles: the Fed conducting monetary policy, and statistical agencies measuring the economy as a public good.

Disclaimer

This transcript has been generated using artificial intelligence and may contain minor inaccuracies or omissions. For complete accuracy and context, please refer to the original podcast recording here.