Author: Kyla Scanlon
Translation: Deep Tide TechFlow
Deep Tide Guide: Why do macroeconomic indicators appear overwhelmingly positive while the general public widely feels extremely pessimistic? Chicago Fed President Austan Goolsbee engages in an in-depth conversation with renowned financial scholar Kyla Scanlon. Goolsbee admits there is an unprecedented disconnect between traditional consumer confidence used to predict spending and actual consumption, and the Fed firmly chose to trust hard data when setting interest rates. However, in reality, the cumulative price increases of conspicuous living costs (gasoline, food), the stagnant labor market of "low hiring and low firing," and the potential local overheating risk triggered by a $1.5 trillion AI infrastructure boom are widening the gap between macro data and public sentiment.
Featured Image: Chicago Fed President Austan Goolsbee discusses the conflict between macroeconomic perception and hard data with financial scholar Kyla Scanlon.
One year ago, I had the privilege of interviewing Chicago Fed President Austan Goolsbee. At the end of that conversation, he astutely warned that we might be about to plunge into an intense conflict between hard Data and public sentiment Vibes. A year later, today, I sat down with him again in Chicago to explore how this tug-of-war has evolved and the true perception ordinary people have of their own economic situation.
This conversation took place during the "Why Consumers and Economists See the Economy So Differently" symposium co-hosted by the Chicago Fed and the University of Chicago NORC. During the dialogue, President Goolsbee dissected in detail why consumer confidence indices and actual consumption expenditures have become severely disjointed, the prioritization of Fed decision-making, the true temperature of the labor market, and whether the AI infrastructure frenzy is causing economic overheating.
The Clash Between Hard Data and Public Sentiment: Why Does the Fed Firmly Believe in Hard Data?
Kyla: When we spoke last year, you predicted a fierce battle between data and public sentiment. At what stage is this battle now?
Austan Goolsbee: This family feud is still ongoing; it has never stopped. It started with arguments at the Thanksgiving dinner table, extended through Christmas, and has now entered a new year. But I think, within the Fed, hard data has essentially won, because over the past year or even year and a half, our interest rate policy has been entirely based on real macro numbers. However, the public's good sentiment has never returned: economic indicators have remained robust and respectable, while the public's sentiment remains very pessimistic. That's the reality we are in.
Kyla: You once said, "The data tracking dog only sniffs for real data," but the disconnect between public sentiment and data is obvious now. When you look at consumer spending, do you think it matches public confidence? Has actual consumer behavior surprised you?
Austan Goolsbee: Compared to the frustration people verbally express, actual consumer spending behavior has indeed surprised me somewhat. The top research team at the Chicago Fed has been closely tracking the mapping between the two. In fact, the correlation between sentiment and spending didn't just break recently; it has been gradually weakening over many years, and recently, the correlation has almost hit rock bottom.
Public sentiment is crucial on many social dimensions. But if you've been relying on sentiment indices as a core leading indicator for predicting consumption spending (the Fed historically did), the current disconnect forces you to reduce your dependence on it. You'll never hear me say to throw away any data completely, but consumer confidence has clearly moved down in my personal decision-making priority list.
Chart: The composite index constructed by the Chicago Fed shows the correlation between consumer confidence and actual consumption spending has fallen to a historic low.
Kyla: So, in your decision priority list, what indicators are at the top?
Austan Goolsbee: The first is still the core ratios of the labor market. I prefer to look at ratio-based indicators rather than the absolute monthly non-farm payrolls figure, as the latter is influenced by multiple question marks like immigration scale and statistical methodology. Therefore, the unemployment rate, job openings rate, layoff rate, and hiring rate are my core focus.
Secondly, on the consumption side, I closely watch the actual hard data for retail sales. Using solid hard indicators for macro forecasting provides the highest precision guidance. Although economic forecasts always come with volatility, the marginal incremental value from layering sentiment data onto hard indicators at present is very weak.
Kyla: Many people worry that negative sentiment will eventually become a self-fulfilling prophecy, meaning the economy will ultimately be dragged down by bad sentiment. What do you think?
Austan Goolsbee: Consumer sentiment is extremely dramatic; its ups and downs are far more volatile than smooth real variables. Historically, a sharp drop in consumer confidence has often been a reliable leading signal of an impending recession, which is why the market becomes extremely paranoid and panicked when seeing low confidence indices.
However, in these past few years where sentiment indices have persistently sounded alarms, the widely predicted recession has never truly materialized. So, I believe the outcome of this debate has tilted towards hard data. If sentiment indices cannot effectively predict current or future actual economic trends, perhaps they are better suited as references for measuring consumers' medium- to long-term inflation expectations, a direction the Fed is actively researching.
Why Does the Public Feel So Bad? The "Salient Pain" of Living Costs
Kyla: There are many theories about the causes of low public sentiment. Former White House Council of Economic Advisers Chairman Jared Bernstein points to cumulative price levels as the culprit; others blame social media algorithms for amplification; and some attribute it to crushing rigid expenditures on housing, childcare, and eldercare. What's your assessment?
Austan Goolsbee: I believe there is a significant Salience psychological mechanism at play here.
Gasoline prices are an extremely conspicuous public price. If you drive every day and see roadside diesel signs showing $6 per gallon, you instinctively feel the overall economy is terrible. Supermarket food prices also have strong public salience. In a high inflation environment, even if the rate of inflation slows, the high absolute price levels translate into overwhelmingly negative evaluations in sentiment surveys.
Additionally, partisan political polarization is contaminating all economic perceptions. Sentiment surveys today have, to some extent, become substitutes for political polls. On election day, even when faced with purely factual, objective questions (e.g., whether the unemployment rate has risen or fallen over the past year), respondents' answers are often entirely determined by their own political stance and the incumbent president's party.
Finally, we did experience a phase of shrinking real wage income. When real purchasing power declined, public dissatisfaction was completely justified. Even though nominal wages later caught up, and even with tax cuts, as long as ordinary people witnessed daily bill increases outpacing their paychecks for a period, this psychological trauma and negative feedback persist.
Kyla: The public generally feels there hasn't been a moment to breathe since the pandemic, with one wave after another. Social media pushes price hikes in all areas of life to everyone's eyes with unprecedented density, coupled with employment anxiety brought by AI, leaving people feeling lost.
Austan Goolsbee: People's discontent is justified. Having to shell out $70, $80, or even $90 every time you fill up your gas tank—anyone in that environment would grow weary.
This reminds me of my 30 years teaching at the University of Chicago. At the end of the semester, MBA students rate professors on categories like organization, clarity, and engagement. If students genuinely like a professor, even if he was unprepared and flustered in the first class, all scores will be high. Conversely, if students dislike a serious, rigorous professor, they'll give poor ratings across all dimensions.
Current public sentiment indicators sometimes reflect this all-negative rating mechanism: when consumers feel immense life pressure, they are not only extremely pessimistic about the current economy but also become more aggressive in their future inflation expectations. Does this reflect real supply shock expectations, or are people giving the current economy a poor rating? The two are often difficult to separate simply.
Chart: Data from the New York Fed shows about a 1 percentage point difference in real consumption growth between different income groups, indicating a slowly evolving K-shaped divergence.
The Stagnant Labor Market of Low Hiring and Low Layoffs, and Concerns Over AI Infrastructure Overheating
Kyla: Regarding the labor market, a year ago you summarized it as "low hiring, low layoffs." Does this pattern still hold?
Austan Goolsbee: Yes, this unusual coexistence of low hiring rates and low layoff rates continues. It is indeed somewhat unique. I thought it was entirely caused by macro uncertainty, with businesses choosing to wait and see; but if this state persists for several years, it's no longer just temporary uncertainty.
In the hard data, the current hiring rate is as low as during an economic recession, while the layoff rate is as low as during a super-boom, both in an unusually stable state. This indicates the entire job market is not deteriorating but is in a highly sticky, stable equilibrium.
Actually, when I engage deeply with industrialists at the grassroots level in Illinois, many business owners instead emphasize that the labor market is trending towards tightness. Especially for skilled technical workers, intense competition from the massive AI data center construction boom across industries has made hiring significantly more difficult, forcing companies to pay wages far exceeding budgets from six months ago.
Kyla: Is the massive capital expenditure on AI data centers causing local economic overheating?
Austan Goolsbee: Regarding AI infrastructure, I hold two cautious perspectives:
First, the more grandiose the vision for the future, the stronger the incentive for companies today to invest heavily in building data centers, converting high valuations into real capital expenditure. This easily drives up short-term demand, risking economic overheating and making it harder for central banks to control inflation.
Second, the core issue is whether this frenzy will spread beyond its intended orbit to the entire society. If the aggressive expansion of AI data centers comes at the expense of crowding out investment in other industries, then overall societal demand is not overheating; it's merely a reallocation of relative capital weights. But if the AI frenzy fully spills over, driving up labor, electricity, and raw material costs across the entire supply chain and pushing up overall inflation, then the central bank must perform its classic aggregate demand management duties.
No one would pour $1.5 trillion in capital expenditure without intending to give away the final product for free. Therefore, either future AI software and computing service pricing will be extremely expensive, or the entire industry's valuation system will inevitably face deep, fundamental reassessment. If capital expenditure fails to deliver explosive real productivity leaps, the profound pain we witnessed during the 2000 internet bubble burst may replay in a new form.
Core Insights and Deep Observations from the Report
Combining this conversation and exchanges with scholars at the Chicago Fed symposium, the following are several crucial core insights:
1. Sentiment is no longer an effective leading indicator for spending: For decades, the Fed closely watched sentiment indices, as pessimism often accompanied consumption tightening; but now this chain has severely dulled. The Fed has significantly reduced the predictive weight of sentiment indicators in its policy models, relying more on hard indicators like retail sales and employment flow ratios.
2. Absolute price Levels are the source of pain: The University of Michigan survey shows that in August, 53% of U.S. households cited high living costs as the main reason for their worsening financial situation (compared to just 7% in 2019). Although current CPI growth has moderated to 3.7%, overall price levels have cumulatively surged nearly 30% since early 2020. As economist Brian Fabes aptly summarized: "Ordinary people live in absolute price levels, not in inflation rate changes."
3. One person, one vote versus one dollar, one vote: University of Michigan director Joanne Hsu points out that sentiment surveys count by head (one person, one vote), while GDP, stock market, and retail data are weighted by money (one dollar, one vote). Since 2023, real retail spending by high-net-worth households with annual income over $125,000 grew by 7.6%, while spending growth by low-income households earning under $40,000 has completely stagnated at zero. The strong consumption of wealthy groups masks the real plight of lower-income groups.
4. Labor market enters a low-flow state: The hiring rate (3.3%) is comparable to the average during past recessions (3.5%), while the layoff rate (1.0%) is lower than the historical expansion average (1.3%). Companies are reluctant to lay off experienced key personnel but are extremely strict in hiring new employees, leading to a significant freeze in new social job opportunities.
5. The central bank won't pay for sentiment but vows to defend the 2% inflation bottom line: President Goolsbee reiterated that the Fed's duty is not public relations reassurance but using monetary policy tools to ensure price stability returns. Only by firmly bringing the long-term inflation rate back to the 2% target path can public anxiety over living costs and deep-seated insecurity be fundamentally resolved.








