After the Dem Sweep: Talking with G. Elliott Morris

My polling guru on what just happened.

G. Elliott Morris’s Substack Strength in Numbers has been my go-to site for political data analysis since he fired it up, and I’ve been leaning on another site he’s involved with, FiftyPlusOne. So I knew I’d want to talk to him after Tuesday’s election, and they were even more interesting than I expected. Transcript follows.

. . .

TRANSCRIPT:
Paul Krugman in Conversation with G. Elliott Morris

(recorded 11/06/25)

Paul Krugman: Hi everyone. Paul Krugman again. I am having another conversation with G. Elliott Morris, whose “Strength in Numbers” Substack and fiftyplusone.news have become my go-tos for polling elections. We had a few results this week and I want to talk about them and then talk a bit more about the broader debate. Hi, Elliott.

G. Elliott Morris: Hi, Paul. Thanks for having me back here. I think this is a three-peat now.

Krugman: It’s a three-peat, elections and polls keep happening.

So we had a pretty startling Tuesday. Maybe before we get into more analytical stuff, why don’t you just give me some reactions. What do you think just happened?

Morris: I wrote a couple of data driven takeaways—well not a couple, seven of them—on the Substack which I can reference. Then I’ll also mention a couple things from the new data analysis. For the readers, we’re recording this on Thursday. So there’s been about a day to catch up on sleep and digest the findings.

The big thing is unsurprising, but it really bears repeating. This is an electoral repudiation of Donald Trump and an electoral verdict on his unpopularity. I have appeared on your interview show a couple of times now to say essentially that the polls show he’s the most unpopular president ever—save himself in his first term. Trump’s policy agenda is also one of the most unpopular policy agendas in American presidential history, at least since we have surveys, since 1936. In that context, it’s rather unsurprising that Democrats did so well in Tuesday’s elections. They swept all of the statewide races in Virginia, all of the New Jersey races, they picked up two utility offices in Georgia, a red state—or a purple state if you’re very optimistic there as a Democrat—which is surprising. In Pennsylvania, they hold like three partisan justices and they win a lower court race as well as statewide.

In that context, it’s rather unsurprising, but it does affirm what we’ve been seeing in other data, which as good Bayesians, is always important to us. It also gives hard data to members of Congress that might want to fight Trump on things like his tariffs, or immigration, which is very unpopular, either from the right or from the left. It gives them something to point to that’s not just survey data, which is increasingly poo-pooed in Congress, as we might say.

So that’s my big takeaway. There are some smaller things which I’ll mention now. The first is it’s not just that Trump lost, it’s that he lost with voters that he supposedly had a realignment with in 2024. This is Latinos and Generation Z voters in particular. He loses with voters who say the economy is very important to them, which is the single constituency that likely propelled him to victory in 2024 in the first place. So in my article I’m putting out on Friday, I’m going to characterize this as: Trump’s losing his winning coalition; because I think that’s really what’s going on here.

The voters that put him in the White House because they wanted lower prices have said, “he’s not holding up his end of the bargain.” He’s not lowering prices. The supposed Republican ideological gains among Latinos in Generation Z who have tended to lean to the left and who, by the way, still voted for Kamala Harris despite lower margins than they did previously has evidently, evaporated. That’s really worth digesting as well.

Krugman: At one level of dispute we’ve had Trump himself insisting week after week that the polls are fake and that he’s extremely popular. And it’s basically you can argue until you’re blue in the face that “polling, it’s not a perfect science, but it’s meaningful.”

But there’s nothing quite like actual elections to settle that dispute. This sort of says that the polling saying that he’s unpopular and his policies are unpopular is right. But it was even a bigger Democratic sweep than expected. I was taking some heart from your polling averages, and some of the polling you’ve been sponsoring. But I was still particularly nervous a little bit about New Jersey. What came in was: the polls were wrong. The Democrats won bigger than the polls suggested they would. Talk a little bit about why you think that happened?

Morris: Interesting context here, we’ve spent the last three presidential election cycles saying, “the polls are broken,” or what have you—I haven’t said that, but other people surely have—because they’ve underestimated Republicans. Over those three cycles, presidential years 2016, 2020, 2024, the average bias at the state level was about 2.7 to 3 points, depending on how you count it, against Republicans. The error that we saw Tuesday night was about the same in the opposite direction, an underestimation of Democrats. So there’s a sort of flip flopping here of the error in the polls for different years and different levels of elections. Which is always my argument that we should use these surveys responsibly and acknowledge their uncertainty. This is a good reminder that polling error does not always go in the same direction. So that’s the first thing I’d say.

But the error itself, about two points in the Virginia governor’s race according to our average, which was closer than other averages, and I think 6 points, maybe 7 points in the New Jersey governor’s race is slightly larger than average, but not terribly so for an off-year governor’s race, the average polling error for an off-year governor’s race is about 4.5 percentage points. So we did better than average in Virginia, worse than average in New Jersey, average those together: it’s not terribly surprising that we saw a candidate beat their surveys. But the mechanism here is really interesting and will have consequences for surveys in 2026. It looks like what is happening is that pollsters who adjust their surveys so that they match the electorate of the 2024 presidential electorate, something we call “weighting by past or recalled vote,” they underestimated Democrats more than the other pollsters did.

Let me unpack. So today, polls have a response rate less than 1 percentage point, in some cases less than 1/10th of a percentage point. So the people you get are really weird, it’s terrible. The people you get to answering surveys are really, really weird. The way you account for that is to increase the weight on respondents who you think are underrepresented. Based on some benchmarks, we get really good demographic benchmarks from the census. That’s pretty easy. That problem has been solved for the better half of a century at this point, but we don’t have good benchmarks for the partisan composition of the electorate. We don’t really know how many Republicans or Democrats should be showing up in an off year election until the election happens, so pollsters have to make guesses about this. Some pollsters have said that their polls should be representative of the Kamala Harris and Donald Trump distribution of voters in the off-year election. Those people underestimated Democrats because, of course, the electorate was much friendlier to Democrats. You had a lot of Republicans not show up, and you had a lot of people change their votes from Trump and from Trump to Mikie Sherrill in New Jersey.

So the top line conclusion here for me is that pollsters should not be adjusting their samples to be representative by past vote if they’re benchmark is past presidential elections, and they’re not in a presidential election. When they do that, they will probably underestimate Democrats.

Krugman: Possibly a naive question: the past vote is based on what people say, right? I wonder whether—particularly given that there appears to have been a large swing—how many people who voted for Trump are now either not being honest or misremembering strategically because they’re feeling pretty bad about that vote?

Morris: This is kind of like how I strategically misremember my chore responsibility some time. Listen, there is a possible effect long term in polls for people to underreport voting for losers. It’s not a very large effect. The polls that are done online, they ask people who you voted for right after the election, and they can use that data over time so they adjust for this. But other pollsters don’t do that, they’re asking them like in the surveys that I helped with, that we were asking them who they voted for today. So I don’t have data to rule that out as a possibility. It’s possible people see that Donald Trump is really unpopular and they don’t want to say they voted for him last time. I’m not going to rule that out.

Krugman: Last point here. I think you’ve addressed this. There’s always the question of turnout versus persuasion, and it looks like both, but I have the impression as an amateur here that there was an unusually large amount of persuasion in the sense that people who were Trump voters last year turned out to be Democrat Mikie Sherrill voters this year. Is that right?

Morris: Yes. Also, for some context, there was an unusually large swing in favor of the Democrats in Virginia. The Democratic candidate Abigail Spanberger did about 15 or 16 percentage points better than Glenn Youngkin did in 2021. So it’s very hard to get that large of a swing just based on differential turnout, based on Republicans not showing up to vote.

That should set our expectation in the persuasion direction. We don’t have the voter file in Virginia yet, or in most of New Jersey, but we do have the voter file in some counties in New Jersey. Nate Cohn at The Times looked at this, this morning. I think he finds that the partisan composition of the electorate in the places in New Jersey that have reported who voted were about 3 percentage points more Democratic than the people who voted in 2024. The difference between Mikie Sherrill’s margin and Kamala Harris’s margin is closer to 8 percentage points. So you get 5 points of persuasion potentially here of Trump-Sheryl voters, if that’s the type of persuasion Democrats are looking at in 2026 or 2028, then they have a pretty good shot of winning the Senate, probably. That’s a pretty good margin. Just to put that in context.

Krugman: Wow. My caricature of this, and I’m sure that it’s probably just a fraction, but particularly that there were a lot of Hispanic voters who basically voted against inflation in 2024 and kind of realized that Trump is not helping on that front. Also he’s going after people who look like them and they switched over. That’s a really big deal.

Morris: My operating assumption for these voters has been that they voted against Democrats because of their performance on the economy during Biden’s election. I know you and I have talked about the effect of inflation. Again, until Tuesday, we really only had the polls to back up this assertion. The polls were showing a large shift away from Trump among Latinos. We saw the same thing on Tuesday. Union City in New Jersey is an 82% Hispanic city. Mikie Sherrill got a 70 percentage point margin there. With more turnout, more people voting, than voted in the 2024 presidential election. It was only D plus 17. We were looking at a 50 percentage point swing among Hispanics, that’s too large really to be explained by noise or turnout, and in this case we actually have more people voting. So that’s a pretty strong signal that the voters are really dissatisfied with Trump, as the exit polls also suggest, because they don’t think he’s held up his end of the bargain on economics.

Krugman: I wonder, and I don’t think the polls will tell you this, I’m just asking for your opinion. How much the whole deportations, ICE agents, grabbing-people-on-the-street thing may have weighed in on that vote. I’ll share with you an anecdote in a second.

Morris: So it’s a great question. The exit poll asked people what their number one issue was. About 50% of people said “the economy,” if you include taxes, it’s closer to 60. 11% of people said immigration. The most important issue question might suffer from a category error here. The most important issue is mostly going to be economics all the time. So we might be missing people who think that immigration is important or it influenced their vote in other ways, even if it’s not their “most important issue.” But if we’re thinking about the model that voters are using, we have to think about all the information that they’re taking in. Over the last six months, so much of this information has been images of Trump kidnapping people off the streets, including citizens—or I should say ICE kidnapping them and putting them in vans, and coverage of the polls showing that this policy is unpopular. So I have to think that that has an impact on people’s votes, on their general impression of the Trump presidency and that that has downstream consequences. Especially if, again, voters think that Trump is under delivering on the economy. What was your anecdote?

Krugman: Latino voters—I know who are an utterly unrepresentative sample—they tend to be highly educated, high income, they’re people for whom American life has been really good. They carry passport cards around with them all the time because they have darker skin and talking to family members in Spanish they think puts them at risk. They’re not sure that even the passport card is going to be good enough. But the idea that, if you’re Latino—no matter what else—that might be a marker that you’re at risk, that’s looming quite large in their consciousness. I think that has to be a factor out there.

Morris: I mean, yes, psychologically, if you’re the type of voter who voted for Donald Trump because he said he was going to put the economy back in 2018, and you really don’t pay attention to politics at all otherwise, and now all of a sudden, you’re inundated with stories of, in many cases, your family members being taken away and the promises on the economy aren’t coming to fruition, I don’t think you’re over your skis in mapping these things.

Krugman: Any thoughts on Mamdani? I actually think that that’s probably the least important political story out there, because New York—not a critique or anything, but New York—it’s just not like America. But any thoughts on that?

Morris: I’ll say my takeaway from Mamdani’s primary campaign back in June was that this affordability message was very resonant with voters. Of course, there’s also that Andrew Cuomo has lots of problems as a person, as a candidate. So we don’t want to use the electoral result here too much.

But, my takeaways: We essentially saw the same campaign being run in New York City as we saw in Virginia and New Jersey. If you abstract away the particularities of being a mayor of New York City, you’re not going to do free buses in Virginia, for example. But that is a campaign and just focused on affordability and costs and this is what voters say is the most important thing to them. This time Democrats are able to make that argument without the ball and chain of the 2022-2023 inflation crisis dragging them down as it did in 2024.

So, I think Mamdani’s victory is another point in the case that Democrats running on affordability will outperform expectations. We get the same evidence from the other elections. The Georgia election is an election about affordability. The municipal utility district there statewide, the people they elected get to set electricity prices. And there you have candidates running 20 percentage point margin statewide, an almost 30 percentage point swing from previous elections in the Democratic direction. And that, I think, is another really strong signal in the affordability case here for Democrats.

Krugman: Interesting. I thought that you’ve been making that point, I guess others too, that aside from Mamdani being left and Spanberger and Cheryl branding themselves as “centrists”, they were all running basically, “I’m going to do something to make your life more affordable.”

Morris: There’s been some discussion over Mamdani’s alleged “pivot” to the center during his general election campaign. I just don’t think we can make that argument when the guy’s still literally calling himself a “democratic socialist”. I don’t think that that argument goes very far with the voters who aren’t paying attention to his policing policy, whether or not he’s going to keep the commissioner in charge, most people don’t know that stuff. They know what he calls himself. The general orientation of his is about prices. That’s not a progressive or a moderate issue. That’s just something people care about.

Krugman: So you’ve been involved in this mostly internet waged battle—a little bit in other media—all among commentators rather than actual political players, but I think it still matters, where there was a lot of: “the Democrats totally have to change, they screwed up, they’re too progressive, they’re out of touch with the American people.” All of which looks a little bit silly Wednesday morning, but you have not so much a critique of that assessment of the Democrats as a critique of that whole way of thinking about politics. I want to talk about that because that’s something I think about a lot, too. Tell us about the “strategist’s fallacy”.

Morris: So I call this the strategist fallacy. The fallacy is among strategists in politics who map their model of how they make decisions as voters onto the median voter. In this case, the fallacy is of using issue positions to decide who to vote for. So your strategists will look at politics, will look at candidates and say, “controlling for their party identification, their fundraising ability, their incumbency status, these candidates do better or worse because of the issue positions that they take,” or that voters will vote for them because of certain issue positions.

That might be how a strategist would operate, because they know about the issue positions of various candidates, but we know from most political science research that voters have very poor understanding of what candidates actually stand for at the issue-position level. They also have a very poor understanding of what these ideological labels: moderate, progressive, really even mean. The most famous article on this in political science is a 1964 Phil Converse essay, in which he finds basically 10% of the public can distinguish ideological labels. This is higher today, but there’s a great 2017 book that basically redoes his analysis and looks at some new evidence, by Nathan Kalmoe and Donald Kinder at University of Michigan. They find essentially the same thing. About 80% of the American public just cannot distinguish liberal and conservative issue positions. So if we’re trying to think about why candidates would lose, we’re thinking about why voters would make their decisions; it’s a fallacy for us to say that they were making their positions based off of candidates’ issue positions that they don’t know about.

Krugman: Kind of a detour, but if we look at politicians, if we look at roll call votes in Congress, then an issues and ideological positioning framework does work; that you can map politicians into an ideological space and you can map particular pieces of legislation into that ideological space. The candidate’s position does predict their votes within Congress very well, that there really is Voteview, DW nominate (a common measure of ideological distance in Congress). Nobody has any idea what I’m talking about, but there is this stuff. This does kind of work as a member of Congress thing. But most people are not like that, most people have lives and kids and jobs…

Morris: So the other aspect of this is, how much are people actually consuming this information about where politicians stand on individual issues? This is like a “news diets” thing. I’ve said on your show before that about 50% of the American public doesn’t consume political news in anything more than a monthly frequency, they tell us. I think that’s even an overestimate, probably. I think the people that you’re getting in surveys are the really news-oriented people, so the numbers are probably less.

The fact is, that the vast majority of the American public is not consuming the type of information that you would need to know, first off, what issue positions politicians hold and second, what the ideological labeling, the orientation of those, what those issue positions are; so it becomes increasingly hard the more you look at this and the more you build out. This is the work that I’m sort of doing all the time: the more you build out individual models of how people are going to vote, it’s increasingly hard to justify the belief that you can change who people vote for by taking different issue positions.

Krugman: At some level, there’s a free rider problem. How important is it to your individual life to follow the news? Also I think that the even semi-analytical thinking that is involved in doing political analysis is not something that most people do as that’s not what their life is about.

Morris: I think what you just said is really important. Maybe I can sharpen this criticism a little bit too. The New York Times, they wrote this big editorial essay about the value of ideological moderation, which I disagree with. They ascribe the success of Democratic candidates in 2024 to exclusively their issue positions. They don’t mention inflation as a cause of Kamala Harris’s defeat in 2020 a single time. There’s no mention of inflation in subsequent articles from The Times on this—from Ross Douthat this week, the social conservative writer—and in much of the strategist space, which is among people who are also friends with the New York Times editorial billboard, let’s put it that way. In the strategist space, there’s an abstracting away from the type of factors that they cannot control. They can’t say “Kamala Harris lost because of inflation, there’s nothing we could have done about that,” or else people won’t give the money, right? They have to sell their services based on a theory of how elections work. The cleanest theory to draw in 2024 might be: “moderates did a little bit better.” So they go out and they sell that theory to donors, to party activists. It’s a nice, really neat story. They can make graphs of it and put a bow on it and sell it really well, and it’s easier to rationalize for themselves as well.

But clean, easy solutions are not necessarily correct. What’s H. L. Mencken quote on this? “There’s plenty of solutions in life: simple, neat, and wrong.” I think that applies here.

Krugman: I hadn’t really thought about the fact of saying, “look it was basically inflation, so why do you need to pay a consultant to devise your strategy?”

Morris: The 2025 elections are the same thing. It’s just inflation, the economic environment and the message about that.

Krugman: Pointing out what Burn-Murdock at the Financial Times said, which was that 2024 was a “graveyard of incumbents,” basically incumbent parties lost everywhere. That’s an inconvenient fact. What good are the consultants as an incumbent in ‘24 if you were going to lose?

Morris: The other thing I think about is—I have some friends in the strategists space, so I don’t want to be too impolite here but—you do have to think about the incentives of your line of work in the industry. I just think in this case, the incentive is towards coming up with a very simple narrative that that you can sell someone, that can from an altruistic point of view give you something to do, give you a way to solve the problem you want to solve. That’s a good instinct. But I’m in the space of constructing really accurate models of why people make the decisions they do. And in most of these models, you just come up with very little premium on moderation, a really high degree of uncertainty, and the types of factors that drive election outcomes, it’s just not issue positions, just empirically.

Krugman: The story that says, “supply chain disruptions post-Covid led to inflation, which led to incumbents losing everywhere,” that’s a simple story. But it’s not an actionable story. It’s not a story that generates any jobs for consultants. So, yeah it actually ties back a lot with our ‘24 versus ’25. It’s not necessarily that, “Democrats did something different in ‘25.” It’s just that, they’re not the incumbents, and people are still upset about the cost of living and there are other things going on there, we think.

Morris: If I’m one of these proponents of the supposed working class realignment towards the Republican Party over the long term—an enduring majority for them–perhaps among working class voters, I’m having a pretty hard time after the 2025 elections. Because the signal we get is that this wasn’t a realignment. It was a temporary, anti-Democratic (with a capital D) partisan vote stemming from anti-incumbency bias, the type of thing that Republicans are going to have to face as they are now the incumbent presiding over an economy that does not work for working people.

Krugman: I have to say, this is a different issue. That there’s a simple fix is one thing, but the idea that selling the idea that there was a permanent realignment, that Tuesday was really bad for that.

It’s funny, when I saw you writing about the strategists fallacy, I was thinking about the old line about the pundit’s fallacy, which is kind of a subset of that, which is that you have a pet issue or a pet position and not only do you push for that, which is legitimate, but that you argue that that is exactly the position that politicians need to take to win elections. The slight irony there is that the concept was invented back in 2010 by Matt Yglesias, who’s been a big part of these roiling debates about moderation and Democrats and could arguably be accused of doing that himself, that he doesn’t like progressive positions within the Democratic Party and he’s saying, “so what they need to do is ditch the progressives to win elections.”

Morris: I mean, let’s think about the strategic advice from that point of view. The strategic advice from the moderation crowd was that, “you should run a campaign in 2025 against trans women playing in sports, against ‘abolish ICE’ and ‘defund the police’,” right? With the idea that voters cared about those positions. We did not see anyone running those campaigns, and yet we saw a major Democratic over-performance across the board. It just seems very obvious when you look at the polling data that you don’t really need to take the heterodox positions on the things people don’t care about. What you need to do is send a positive signal that you’re going to work on their behalf and you care about them, and that’s what the Democrats did, regardless of their ideological position.

Krugman: One of the things that I was wondering about a little bit, I saw some really odd commentary, to the effect that Spanberger’s 15 point victory in Virginia was somehow less significant than Youngkin’s 1.5 point victory four years earlier, because Spanberger didn’t open up in a new space, whereas Youngkin opened up the space of campaigning against woke—roughly speaking—which was kind of odd, but also had me wondering, is there any evidence that any of these other things—that wokeness or anti wokeness, that this has had any major electoral effects one way or the other?

Morris: I guess I can buy that argument. If you think about politics on the traditional left-right scale, where Democrats need to run more on the right side to win more moderates, let’s say, but what Trump does in 2016 and what he does in 2024, is run on a different ideological spectrum, not a left-right spectrum, but an anti-system spectrum and “we care about you more, we’re going to give you a good economy” spectrum as well. Just for these purposes, we’ll call this on an up-down scale.

So I think what Democrats do is that in 2024, they appeal to the voters who might be a little bit more right leaning, but are extremely more oriented towards the anti-system pro-working class, “deliver for me on the economy” portion of this ideological space, and that’s where they will continue to run. The leverage here that they’re getting is not moving to the right, it’s moving up on the chart. I find that conclusion pretty wild. Democrats have unlocked a lot of leverage by running candidates who can appeal on an affordability message. That’s the thing that people say they care about in American politics, not social conflict.

One other thing here, the voters who cast ballots in Virginia in 2021 are some of the most engaged voters in the country. It’s the exact type of people you might expect to be moved by the anti-woke message. That itself is a spin of what happened in the 2021 campaign, which was basically a negative referendum on McAuliffe, who said that “parents shouldn’t be making decisions in schools.” That’s just like an education kitchen table issue. That should also be our part of the conversation about why Youngkin won, not just “woke”.

Krugman: I just have this general impression, and it comes back to the original people’s positioning, my guess is that the median voter—if such a thing exists—probably spends maybe 30 minutes a year thinking about trans athletes or any of these hot button social issues, except in the sense that they have some general sense of whether the candidates are on their side, it’s probably just no big deal.

Morris: The data here from Virginia, I’ll mention two things. First, the survey data from before the election showed that the Democratic candidate, Abigail Spanberger, was the one who had the advantage on the very issues that people say Democrats do poorly on because they’re “too woke,” in the case of whether or not trans students should participate in women’s or female sports, in this case, Abigail Spanberger had like a 16 percentage point lead over her opponent. So it’s no mystery why she won in the first place if she’s leading on those issues. But again, what would that “anti-woke” Democratic Party strategic advice have said about Abigail? Spanberger said herself, she was “going to lose because of all these ads that the Republican candidate was running against her on trans issues,” in fact, that advice was sort of false. The conclusions from that are wrong. She wins in part because of the advantage that she had on this issue, which had some salience, but not a whole lot.

The second data point I was going to mention was the intensity of the ads from the Republican Party against Abigail Spanberger. It was entirely about her, about the Democratic Party’s affiliation with this sort of “woke crowd” arguing for “trans women to participate in female sports in schools,” which no one’s arguing for, and the ads just completely fell apart. She loses by 15 points because of a bad economic message in an incumbent president from her party, probably.

Krugman: What struck me is how heavily a lot of Republicans are leaning into what they perceive as “social anti-woke” stuff. I don’t know whether it’s that they have their own bad consultants or just that’s who they are.

I’ve been, for obvious reasons, looking up SNAP food stamp data. At the top of the data page for SNAP, which is actually maintained by the Department of Agriculture, is a banner blaming Democrats for the shutdown and saying that instead, “Democrats want to pay for sex change operations for students,” which aside from being false and aside from being totally inappropriate on our government data page, I can’t imagine who they think is even going to look up supplemental nutrition enrollment data and is going to be moved by that kind of banner.

Morris: That is a complete bastardization of the government’s role in providing data to people. In terms of the data, I just keep coming back to this massive change in the voting behavior of people who say the economy is important to them. Trump won that group by 63 percentage points in 2024 and Democrats just won them by 30 points. There’s about a 90 percentage point swing here.

Krugman: Oh my God.

Morris: Just look at that. We don’t need to really do any more narrative about this election, right? Just look at that number. That’s the whole story here. The same is true for 2024, people are just reading way too much into their pet issues or what data might give them something to sell as a campaign consultant.

Krugman: Maybe just to conclude by talking about that, because that reversal on the economy, if there’s something that really puzzles me it is just how sharp that was. I was surprised in 2024 that we had a bout of inflation, but the extent to which people appear to believe Trump’s promises that he could bring prices down kind of shocked me. Then the sheer violence of the snapback on that, we’re—whatever it is—ten months in, and the Trump economy has certainly not been as promised, but it’s not hyperinflation and it’s not a depression. Do you have a theory about why people got so disillusioned so fast on the economy issue specifically?

Morris: I think the framing might be a little off. I think they were always disillusioned. Bringing some of my priors here too—I’m a young person—lots of people my age are having a lot of trouble with economic mobility, social mobility as well. Most people my age don’t think they’ll ever own a house, for example. The idea of the American Dream is not present even for the elite educated in my group. So, that itself I think sets us up for an anti-incumbent sentiment. This is what I’ve been writing about the economy.

People are looking at this as a Dem. vs Rep. issue and to some extent, they can do that. The Gallup polling did show Republicans with a big advantage on the economy, I think a 14 percentage point lead on which party adults trusted to handle the economy. That has evaporated. It’s 4 or 5 points in the Democratic direction now. But that masks just a long term anti-incumbent sentiment among people who say the economy is getting worse. That percentage, 53% of the public, has said on average that the economy is getting worse since 2000, since the pandemic. You can adjust for a little bit of movement around the elections when partisans change how they’re feeling. In 2021, there was a bump among Democrats, and in 2024, there was a bump among Republicans in the percent of people who said that the economy was doing better. But within two months it had trended back toward that long term, pessimistic direction and pessimism is just bad for the incumbent. It’s not bad for the Democrats. It’s not bad for the Republicans. It’s just bad for whoever is in charge. I guess if I was Stephen Miller—the people in the Trump campaign right now—that’s the number I’d be looking at to know what 2026 and 2028 is going to look like: “what percent of the people say the economy is getting worse?” If that number is more than 25-30 points, you’re basically going to lose. There’s no way around that.

Krugman: Yeah. Like I said, it’s not a great economy. It’s not the hot economy that Trump boasts about, but the extent of the pessimism is really quite remarkable. I say that as somebody who’s obviously not at all happy with Trump, but I spent a lot of time arguing about the vibecession and trying to understand why people were pretty negative on what I thought was objectively a pretty decent economy in the later Biden years.

But the vibecession persists now, and to a degree that the consumer sentiment numbers look like the depths of the financial crisis. The economy has its problems, but it doesn’t look like that.

Morris: Right. I hadn’t even compared it to 2009. In 2022, the Inflation crisis itself had lower consumer sentiment numbers than the financial crisis, if I remember correctly. Trump’s number on handling inflation now is basically 30 points underwater. That’s worse than Biden’s numbers were during the depths of the 2022 inflation crisis. Yes, the indicators aren’t as bad as they were then. But you know what we’ve had, Paul, is a president, basically every day, boasting about his tariff policy and making single handed decisions about the trajectory of the global economy. You can’t blame people for a bit of a vibecession there, despite the objective economic data.

Krugman: That’s right. Consumer expectations of inflation following his tariff announcements. I’m anti-tariffs and I think they’re bad and I think they’re inflationary. But the consumer expectations deteriorated far more than crunching the economic numbers say they should have. But I guess that’s the point, he’s highlighting it.

The last issue. So we’re going to have, we think, midterm elections next year. There has been the issue, which is that Trump is wildly unpopular. His policies are wildly unpopular. Democrats have an edge in the generic ballot polls, but it’s fairly modest. Do we believe that? I think you said that the results in Virginia, if you use them to extrapolate, would be at like a 8 or 9 point Democratic advantage, whereas the polls are saying more like three. What do you think is going to happen in the midterms?

Morris: I’ll make an early prediction, if you’re going to force my hand, Paul. So there’s a couple of things working in the Democrats corner here. The first is that there’s a long term trajectory in American politics of the party in the White House not only losing in the midterms, but also losing ground in public opinion over the course of that term. We actually expect the Democrats are plus three right now in the generic ballot. The long term trajectory suggests that they, as the party out of power, would gain about five points in the polls between today, and next November. And that puts them at about plus eight, plus nine. That is maybe coincidentally or maybe not the same projection you would make about how the House popular vote would break down next year, based on the swing that Abigail Spanberger saw from the last Virginia governor’s race. It was 15 points, we should expect a slightly smaller swing from the last House election nationwide. So the Virginia swing does tend to be a bit of a reliable bellwether in our nationalized, polarized, political environment.

I guess I should caveat for your smart readers, we only have five years of electoral data there, so it’s not an ironclad prediction, but the fact that these numbers are the same and that they’re pointing in the same direction should be a source of optimism for Democrats. I’m happy to make this projection because I think it’s relatively secure. Democrats are very likely to take the House, conditional on one thing, which is the Voting Rights Act Section two staying in place. If the Supreme Court got section two of the Voting Rights Act, then southern states can very easily gerrymander Democrats out of their urban seats in states like Tennessee and Alabama. But let’s not get to that [future] point too much. I’m relatively confident Democrats will take the House. If the swing against Republicans is replicated at the state level, then Democrats would also have a potentially good shot at taking back the Senate as well. I mean, I would put that at maybe 28-29% chance at the highest, but that’s a lot higher than it was a week ago, and it’s a lot higher than zero.

Krugman: Wow. So it could be really interesting. Trump may have a difficult couple of years before he runs for his third term. Anyway, the crazy environment we’re living in.

So, revelatory week, the world looks quite different from the way it did Monday.

Morris: Well, it does, but the signal that we’ve got really affirms what I think we believe in and what you spend a lot of time—you and I both spent a lot of time—writing about, which is: the president’s unpopular, he’s lost ground on the things that people care about most. So while the Democratic brand is unpopular, when people are forced to make comparisons between what the Democrats are offering and the Republicans are offering, most of the time they’re picking the Democrats, and when their economic security is the number one thing and Trump is threatening that with 100% tariffs on China because he wants to, voters aren’t going to vote for him. So, I think in some sense the people in the camp of the polls over the last six months have actually been proven right by the election result.

Krugman: Okay. That’s a good place to end, relatively, and if you have my political orientation—an upbeat place to end. Thanks so much for talking with me.

Morris: Well, thanks for having me on again Paul. This was a good conversation.