CFM Talks To: Joanne Hsu

Dr. Joanne W. Hsu (pronounced “shoo”) is the Director of the Surveys of Consumers and a Research Associate Professor at the Institute for Social Research at the University of Michigan. She earned her PhD in economics at the University of Michigan and her AB in economics and international relations at Brown University. Her research is primarily in the fields of household finance, labor economics, and survey methods, with a current focus on financial sophistication and cognition, and consumer experiences with debt. She previously served as a principal economist in the Division of Research and Statistics at the Federal Reserve Board of Governors, where her work included the Survey of Consumer of Finances and the consumption forecast, as well as a visiting professor at the Department of Economics, Howard University.

CFM: The Michigan survey garnered a lot of news coverage as the aggregate headline consumer sentiment indices fell to a near all-time low in April earlier this year. While having rebounded recently, what in the underlying data drove the significant drop? 

JH: Many folks might not be aware, but the survey interview provides many opportunities for respondents to freely comment on open-ended questions, which provides critical context for our headline measures. By April, about two-thirds of consumers spontaneously mentioned tariffs – these were completely unprompted, unsolicited references [see figure 1 in this report]. Note that it’s very unusual for a majority of consumers to agree on anything! Digging into the interviews made it abundantly clear that the top factors on consumers’ minds were, and continue to be, cost of living and high prices. This has been the case since the tail end of the pandemic and is manifesting this year with widespread concerns among consumers that trade policy might pass through to them in the form of higher prices.

It looks to me that consumers correctly anticipated just how severe the tariff announcements would be, as evident in the initial April 2 announcement of the “reciprocal” tariffs. This is probably why sentiment started to decline well before April. As tariff developments quieted down in June and July, consumer views improved.

CFM: Given a recent snappy rebound, after the even snappier drop, how do you respond to critique that consumer sentiment is, or has become, perhaps too susceptible to short term events or news cycles?

JH: This year’s sentiment trends appear to be influenced heavily by tariff policy developments. I don’t think anyone really anticipated just how volatile trade policy would end up being this year; it’s been unclear if any particular announcement would last for the short or long run. It is, however, certainly true that the pandemic coincided with a sharp acceleration in how people get and consume their news, particularly with the proliferation of podcasts and the evolution of social media during this period. Before, say around the turn of the last century, the average person might read the news once a day in the morning paper or watch the evening news. Now you cannot shut it off – it’s available and constantly updating, available all the time on smartphones, etc. It would be more surprising if news and information developments didn’t influence consumers’ views of the economy.

For example, in the post-pandemic period, a large number of people indicated having heard more negative news about inflation – even higher than in the early 80’s – this despite inflation having been much higher back then (see figure 1 in the insert below). We published a special report on this topic last year and are in the process of collecting new data about the sources of economic news followed and trusted by consumers.

When we look at the post-pandemic period, sentiment hit its all-time historic low in June 2022. Consumers kept spending even though they felt terrible about the economy. This spending was made possible because consumers were supported by strong and reliable incomes stemming from historically tight labor markets, as well as pandemic-era excess savings. At that time, they had the resources and buffers to continue robust spending even if they did not feel optimistic about the trajectory of the economy.

Fast forward to today – now we see negative views about not just inflation, but also the labor market and their own personal incomes as well. That is one of the major differences between now and three years ago: negative views on multiple fronts – not just inflation – have dragged headline sentiment lower. If consumers believe that labor markets are going to weaken and that their own incomes are no longer reliable, then the fundamental support for continued, robust spending just isn’t there anymore. That to me is something of real concern and is important for market watchers to pay attention to.

CFM: That leads neatly into one of my follow-up questions about a mismatch between hard and soft data. This strong belief that labor market conditions will worsen is not fully supported by the hard data (although it is deteriorating). The soft and hard data is not really singing from the same hymnbook. Do you have any working theory for this current mismatch?

JH: I would say that hard data has always sung from yesterday’s hymn book; consumer sentiment data has always sung from tomorrow’s hymn book. Hard data is by definition always backward looking, whereas we ask consumers to project into the future – expectations that shape the decisions they make today, which affect their future as well as the future of the economy. Previously, consumers broadly agreed that labor markets had been fairly strong, but now, I think the sour mood of consumers is exacerbated by a belief that unemployment can only go up at this point. The share of consumers expecting unemployment to rise in the year again is currently at levels last seen during the Great Recession.

Again, these unemployment expectations link back to consumers’ worry about trade policy. Not only do consumers believe that tariffs could lead to an increase in prices, but also that tariffs may make it very difficult for businesses to continue hiring. In short, consumers expect high tariffs to pass through to multiple dimensions of the economy.

There is a duality, which is both lower employment and own income expectations – consumers are just not as confident about their prospects as they were a few years ago.

CFM: How concerned are you about the quality of national economic data going forward?

JH: Reliable, high-quality national statistics about the economy are critical to policymakers, businesses, academics, market participants, and the public as a whole. The data collection environment has indeed gotten more challenging over the course of the last couple of decades, but there is a tremendous amount of scientific research and knowledge about how to produce high quality economic data and reliable statistics. However, doing so requires more resources, not less; it requires leadership that is plugged into the latest research and innovation in this space; it requires a continued commitment to accuracy and rigor, free from political considerations. One concrete step the federal government can take now to advance these commitments is to reinstate the Federal Economic Statistics Advisory Committee (FESAC), whose purpose was to advise “the Directors of the Department of Commerce’s statistical agencies, the Bureau of Economic Analysis (BEA) and the U.S. Census Bureau, and the Commissioner of the Department of Labor’s Bureau of Labor Statistics (BLS), on statistical methodology and other technical matters related to the collection, tabulation, and analysis of federal economic statistics.” This committee was terminated in February because “The Secretary of Commerce has determined that the purposes for which the Federal Economic Statistics Advisory Committee (FESAC) was established have been fulfilled.”¹ Reconvening FESAC, and other advisory committees, would help ensure the continued reliability of national economic statistics.

There really is no substitute for the comprehensive, high-quality statistics provided by the BLS and other statistical agencies; even large-scale private data producers rely on benchmarks drawn from national economic data.

CFM: U.S. citizens are more than ever invested in the stock market. And the US equity market post-COVID, barring a few blips, has been on quite a tear. Again, this seems contradictory to the exceptionally sour mood, given that sentiment has historically reacted positively to changes in the US stock market. What do recent data suggest about consumer sentiment across different wealth groups?

JH: Let me send you a graph as a backdrop (see figure 3 below, courtesy of Dr. Hsu). It shows trends in sentiment for the group with large stock holdings compared to those that are not participating in the stock market. Historically, the wealthier you are, the more favorable your sentiment is. This isn’t any sort of mystery. But in 2022, as inflation in the US reached its peak, sentiment for all wealth groups converged and with each reaching historic lows.

In the two or so years since then, particularly in the second half of 2024, we saw a huge improvement in sentiment for the top tercile of stock holdings, while the sentiment of people with no stock holdings remained basically unchanged. The ‘normal’ wealth gap was hence restored, and in fact grew to historically large levels.

Incidentally, I think this is one of the things that helps to explain why consumer spending was so strong coming out of the pandemic. The groups that generate most of aggregate spending – high income and high wealth consumers – were specifically the groups whose sentiment recovered quite quickly in 2022-2024. Thus, the aggregate trends were in fact consistent with the survey data.

Turning back to recent market trends: I don’t think most consumers are closely watching their portfolios – most are merely participating passively through their 401(k)’s. However, even among the ones who do – likely the most invested – we observed a steep plunge in sentiment in 2025. This is despite being precisely the consumers who should have seen their portfolios remain strong amid all this policy uncertainty. The dominant trend, even among high wealth consumers, continues to reflect worries about the impact of trade policy on the greater economy, despite their portfolios likely having shown solid gains. If the mood of the wealthiest, highest spenders sours, it would bode poorly for aggregate consumer spending.

CFM: There is another discrepancy I find curious, this between various proxies of inflation expectations. The median inflation expectations from the Michigan survey for prices changes one year from now peaked at +6.6%, while one market-based proxy – inflation swaps, barely moved. Why?

JH: It shouldn’t come as any surprise that they don’t match up. Market-based inflation expectations reflect views from a totally different set of people than those of consumers. They’re different actors following different types of data, incorporating different types of information.

You of course know there’s a lot of chatter about how financial markets are banking on the ‘TACO’ narrative. This is not something that has really surfaced in the consumer survey. Consumers are expecting continued instability and uncertainty around trade policy – I think they are taking the administration at their word that tariffs will be higher. I think market participants and consumers generally have fundamentally different understandings of how the current administration is doing policy.

“Consumers are expecting continued instability and uncertainty around trade policy – I think they are taking the administration at their word that tariffs will be higher.”

CFM: Could you share a few words on the discrepancy we have observed between the Michigan Survey and other surveys’ consumer inflation expectations, especially the Fed’s, where the gap was most notable.

JH: Differences between the Michigan survey measurements and other surveys are largely a consequence of their different methodological approaches. The various options that survey designers consider all have their own pros and cons, and decisions may reflect fundamentally different philosophies — one option is not necessarily better than the other, but they often will yield different estimates.

One key methodological choice is related to sample design. The Michigan survey employs a rotating panel design such that, each month, the majority of our respondents are brand-new interviewees. Thereafter, we will only re-interview a respondent up to two more times, 6 months apart. In contrast, many surveys re-interview the same people over and over, often on a high-frequency basis (as an example, the New York Fed re-interviews respondents repeatedly for up to 12 months). This distinction is very important because, when it comes to re-interviewing people specifically about economic sentiment and inflation expectations, their views tend to get better with each re-interview. We’ve known this for decades and it is well-documented in the literature. Survey methodologists call this the panel conditioning effect. When you interview the same people over and over about their economic attitudes, you’re likely to get more favorable readings over time that may not necessarily reflect an improvement in underlying views of the economy. This isn’t the case for all survey topics, but it’s specifically an issue for economic sentiment questions.

Secondly, the specific design of survey questions involves trade-offs, including between simplicity and conceptual precision. When I was in grad school, a professor told me, half-jokingly, “Psychologists are great at measuring concepts they can’t define, while economists are great at defining concepts they can’t measure.” The different ways one can elicit inflation expectations illustrate this distinction. From its founding in 1946, the Michigan survey has prioritized simplicity in order to gather usable responses from the broadest swath of the population, and to capture the types of beliefs that they may consciously or subconsciously employ in their economic decision-making. The Michigan approach to inflation expectations thus asks for a single number: “By about what percent do you expect prices to go up/down on average, during the next 12 months?” This is easy for people to understand, and most people will provide a number; very few people skip a question like this. More complex question designs will tend to yield more missing responses.

One alternative approach is to ask respondents to fill out a full probability distribution. The upside of this design choice is that a probability distribution might be closer conceptually to what one might plug into a formal economic model, but complex interview questions typically come with their own downsides.

The New York Fed’s headline inflation expectations questions employ the probability distribution approach: what’s the percent chance that the rate of inflation will be 0 to 2%, 2 to 4%, 4 to 8%, 8 to 12%, and 12% or higher, and symmetrically on the deflationary side, the percent chance that the rate of deflation will be, 0 to 2%, etc., all the way down to 12% or higher deflation. In total, a respondent is asked to supply probability weights to ten different bins, and those weights must add up to 100.

The fact that the bins of the probability distribution are presented symmetrically from 12% or more inflation down to 12% or more deflation, means that visually, 0 looks like the average, “middle”, or “neutral” response, when in fact historically speaking 0, let alone deflation, is quite rare. This matters because survey respondents in many contexts often naturally gravitate to the middle of a visual scale. Here, these tendencies could potentially attenuate measurements of inflation expectations.

In yet another approach, the Bank of Italy’s survey tells their respondents the most recent inflation rates measured in Italy and the Eurozone. With this design choice, respondents will gravitate to those current numbers, creating less dispersion in responses, but doing so likely makes it more difficult to detect shifts in underlying expectations.

I do want to be clear that there is no objectively right or wrong answer on this front; these differences in methodologies simply reflect different philosophical approaches and measure different underlying concepts. As such, there is value in collecting data from all of these approaches. One should not expect them to hit the same number, nor should you expect them to move in parallel at all times either. That’s why policymakers don’t look at just one data source, they look at the whole suite of available data. They look at market-based proxies, they’re looking at us, at the Conference Board, at the New York Fed, etc

CFM: Another fairly new and potentially interesting source of macroeconomic data is ‘prediction’ markets. Do you assign any serious value to these as a complementary source of data?

JH: I must caveat that I don’t have a ton of knowledge about prediction markets but let me just speak philosophically. I think prediction markets are conceptually similar to market-based measures – participants are putting down money based on where they think inflation, rates, etc. are going to be in the future.

I think the extent to which one might find that valuable depends on who you think is participating in these markets, and whether you think those participants will meaningfully expand your understanding of how different parts of the population have different expectations for the future. If the investors who buy and sell the assets used for existing market-based measures closely resemble participants in prediction markets, then both measures are likely to provide the same kind of signal. If the participants are very different, then “prediction” markets could very well provide additional information.

CFM: Do you liaise with policymakers or the Fed in terms of interpreting and applying the findings of the survey? And if so, what are you hearing from policymakers specifically when it comes to consumer sentiment that they find particularly worrying or concerning at the time?

JH: I worked at the Federal Reserve Board of Governors for over ten years before I took my current position at the University of Michigan, and I also serve on the Academic Advisory Council at the Chicago Fed, as well as the American Economic Association’s Committee on Economic Statistics. I continue to have regular conversations with former colleagues throughout the Federal Reserve System and elsewhere in the federal government. Recent discussions have related to the influence of tariff policy on consumer behavior and inflation expectations. One of the neat things about the Surveys of Consumers is that it goes beyond general consumer sentiment or inflation expectations. We ask questions about many different sectors and dimensions of the economy that are important to policymakers. We have questions on housing markets, vehicle buying conditions and financing choices, labor markets, and many more; the Fed has groups of researchers that focus on these specific areas.

CFM: The underlying panel data of the survey is, as you just alluded to, extremely rich, with plenty of very granular questioning. Is there anything in this underlying data that stood out of late?

JH: One thing that has stood out to me, and which caught quite a bit of social media attention, has been partisan trends in consumer sentiment. What I think a lot of people don’t realize is that partisan differences in the data are nothing new. We’ve been observing partisan differences in sentiment and expectations since at least the Reagan administration.

Moreover, there’s a wealth of research that highlights the same partisan differences in actual consumer behavior: Democrats tend to spend more under a Democratic president than under a Republican president, and vice versa for Republicans. Similarly, Democrats tend to file more patents, become entrepreneurs at higher rates under a Democratic president, and again, vice versa for Republicans. Recent Fed research has even shown similar patterns among professional macroeconomic forecasts and tax evasion!

All that said, aside from presidential transition periods, sentiment of Republicans, independents, and Democrats tend to move together, even if there are level differences between the three groups. Independents tend to be in the middle and generally match the national trends.

So perhaps I shouldn’t actually have been surprised about what we saw with consumer expectations by political affiliation this year, given the long-standing historical precedent of parallel movement. Between February and April, consumers of all three political affiliations exhibited a decline. Even Republicans, who are likely to be broadly supportive of the White House’s policy agenda, reported that the outlook for the economy had worsened over those months. Republicans, too, share the concerns of others that large tariff hikes may well worsen inflation, or otherwise lead to some pain, at least in the short run; the increasing worry about tariffs can be seen across the entire political spectrum. This was visible in the data despite the fact that, as typically is the case under a Republican president, Republicans generally hold much more favorable views than Democrats, with independents in the middle.

CFM: And that extends to inflation expectations – I do recall that there was a significant spread between the Democrats and the Republicans. With Republicans even expecting deflation.

JH: Yes, that’s right. As tariff developments accelerated, the gap just ballooned where Democrats expected that a worst-case scenario with very high inflation was on the horizon, whereas Republicans expected prices to fall imminently. It’s not that they’re living in different worlds, but they do have dramatically different views on the potential consequences of economic policies. As time went on, Democrats realized the world does not appear to be collapsing, and Republicans moderated their views as well with an increase in inflation expectations. Since then, we have seen both ends of the political spectrum converge a bit. I think everyone’s still trying to figure it out with respect to tariffs, but people are moderating on both sides.

CFM: I would like to conclude with a final question about the survey itself. What do you think are some of the major challenges and opportunities for the survey to remain relevant?

JH: I’m excited for the future of the survey. Given we are in the field every week of the year, we have the unique opportunity to add new questions to contextualize what we’re seeing as and when the need arises, and we’re able to collaborate with colleagues from around the University of Michigan to do so. For example, we are collecting new information about where people get their economic news and what sources of information they believe are reliable. In light of questions arising about Fed independence, we’re also collecting information on whether trust in institutions like the Federal Reserve has changed. Additionally, a major area of focus for us is understanding the heterogeneity of views across the US population, which helps us identify areas of vulnerability among consumers. Taken together, I think these will help us understand how expectations are formed and how they pass through to consumer behavior and the rest of the economy.

On a similar note, I think one of the big challenges – and this is not unique to our survey – is maintaining public trust in the survey. I’ve been doing a lot of public engagement to make sure people understand what we’re doing – not only potential respondents, who generously share their time by completing our surveys, but of course our data users as well. I enjoy hearing from our data users and speaking to folks like you who consume our data in various ways.

I’m very lucky that the Surveys of Consumers lives here at the University of Michigan. Our Institute for Social Research is where survey science was born; the department of survey methodology here remains second to none worldwide. I feel truly fortunate that we’re shoulder-to-shoulder with pre-eminent social science and data collection experts, which means we can continue to innovate and publish data that is accurate and timely for policymakers and other data users, and continue to shed light on the outlook for the consumer.

Joanne spoke with André Breedt, VP in the research team based in Paris.

¹ Drawn from the FESAC website here.

DISCLAIMER
The text is an edited transcript of an interview with Dr. Joanne Hsu in July 2025 and have been edited for length and clarity. The views and opinions expressed in this interview are those of Dr. Hsu and may not necessarily reflect the official policy or position of either CFM or any of its affiliates. The information provided herein is general information only and does not constitute investment or other advice. Any statements regarding market events, future events or other similar statements constitute only subjective views, are based upon expectations or beliefs, involve inherent risks and uncertainties, and should therefore not be relied on. Future evidence and actual results could differ materially from those set forth, contemplated by or underlying these statements. In light of these risks and uncertainties, there can be no assurance that these statements are or will prove to be accurate or complete in any way.