Interview with Jason Furman (Harvard University, chair of the Council of Economic Advisers under President Obama)
Jason Furman is the Aetna Professor of the Practice of Economic Policy jointly at Harvard Kennedy School (HKS) and the Department of Economics at Harvard University. Previously, Jason was the Chair of the Council of Economic Advisers under President Obama.
Interview conducted 11th December, 2024.
Q. How do you characterise “Bidenomics”? What do you think its objectives were and how do you think it performed?
To some degree, “Bidenomics” was a label for what the administration was able to do, which tended to be things related to manufacturing, climate change, microchips, but not the things related to some of his initial ideas on the care economy – child tax credits and the like. It was a label that made a virtue out of necessity. To some degree, it was just a catch-all label for taking credit for anything good about the economy.
How do I assess it? In some ways, the macro economy was good; in other ways, it was bad. I can unpack that for hours if you want, but broadly speaking: productivity growth has been very good; labour force growth has been very, very good; as a result, there’s been strong real economic growth. There’s also been a huge amount of inflation and persistent high interest rates, and real wages are well below their previous trend. I think a lot of that has many, many causes, of which policy is just one contributor.
In particular, “Bidenomics” certainly seems to be about manufacturing. But to date, there’s not much evidence of a manufacturing resurgence. Employment has been on the same down slope, production has been flat, and productivity has been terrible. The one thing that might be meaningful has been a huge increase in structures investment – maybe that foretells some lagged increase in manufacturing in the future, but for now, the manufacturing story is a bit too soon to tell.
Q. How has the U.S. economy performed relative to other countries over the past four years? What explains that performance?
First, we have had narrowing wage differentials in the United States. That started around 2014. It started, by the way, with an unemployment rate still around 6%. So it’s hard for me to put that down to “Bidenomics.” It’s also hard for me to put that down to a hot economy when it wasn’t a super-hot economy for much of the time.
It looks almost like a continuous process of wage compression over the last decade.
Part of that, at the bottom end, is the increasing minimum wage (which a lot of states have done, even if the federal government hasn’t). Some of it is that some of the forces that were increasing inequality have come to an end. There’s not much more decline in unionisation, for instance, because unions have mostly declined to very little. It’s hard to keep going at the same pace of decline we saw before.
A lot of the factors that explained higher inequality were factors that gave you higher inequality, but not permanently rising inequality. So maybe we’ve now asymptoted to where we were supposed to be and even moved off it a little bit. I haven’t been paying as much attention to comparative wage-inequality trends, but in terms of overall growth, the United States has more favourable demography, has more immigrants, and immigrants work at a higher rate. It’s also had more productivity growth. Basically, all the different structural ingredients have been more favourable in the United States than they have been elsewhere.
Q. Given your experience in the response to the 2008/09 recession, was the Biden administration right to risk over-shooting its fiscal stimulus (the ARP, American Rescue Plan) to avoid the mistake of under-shooting last time?
I have no doubt, first of all, that the American Rescue Plan was too big. There are basically two things that I think caused that. One is “fighting the last war,” as you said: “too small a response in the financial crisis – let’s not make that mistake again.” The second is an inertial process in the policy system, where policymakers form a view and don’t rapidly update it as data changes.
Many of the elements of the American Rescue Plan – like the size of state and local assistance – were fixed by House Democrats in May of 2020, back when they thought the unemployment rate was going to rise to 20%. Instead, the unemployment rate fell to 6%. Rather than saying, “Oh, that idea we had nine months ago, we need something smaller now,” it became “That’s what you’re fighting for, and when you can finally get it, you’re happy you got it.” By the way, that was the exact opposite in 2008 and 2009; the economy then was worsening very rapidly, so if you formed your view early it was going to be outdated and too small if you didn’t update it quickly.
I would assume the American Rescue Plan both added to economic growth and added to inflation. I don’t think there’s any perfect decomposition, but I’m pretty sceptical that it added much to economic growth. The economy in the first quarter of 2021- basically before it went into effect and the money got spent – grew at over a 5% annual rate. Households at the beginning of 2021 were in a vastly better position financially than they had ever been – more money in bank accounts, more excess saving, more room on credit cards, lower delinquencies and defaults, etc. So I think as soon as supply returned, there was a huge capability for demand to be there to meet that supply.
If you look at what output did, it got basically to nearly the pre-COVID projection by the end of 2021, all the way there by 2022. I just don’t see how there would have been insufficient demand to accomplish that even without the ARP. Moreover, I don’t think supply was sufficiently elastic such that, if you produce a lot of extra demand, it shows up in supply. We saw all sorts of supply-chain issues. I tend to think that we would have had a pretty similar real GDP recovery – most of the extra demand went into prices. But I can’t entirely prove that, obviously.
Q. How do you account for that relative economic strength going into 2021? Was it the CARES Act and the immediate pandemic response under the late Trump administration?
There was another $900 billion that was passed at the end of 2020. Basically, in 2020, households spent much less than normal, and they got income that was much, much higher than normal. So they already had trillions of dollars of excess savings. You just needed the economy to reopen.
I don’t think it was all supply shock – had [the economy] been a closing and then reopening but people didn’t have money, you wouldn’t have had demand come roaring back once things reopened. But people had more than enough money.
States and localities got about $500 billion, I think, from the American Rescue Plan. They were in a better-than-normal fiscal situation when that plan passed, so it was pretty clear they also weren’t going to be a drag on growth. That was a big aggravating factor in the slow recovery from the financial crisis – state and local governments were a major drag then. Here, it was clear they weren’t going to be.
Q. Do you think the stimulus money, even if it had been smaller, could have been spent differently on different priorities or through different channels?
I think there were two models. One is to do something smaller that does the genuine emergency stuff – funding for vaccination, the health infrastructure, etc. That’s basically all that was needed. Maybe some extension of unemployment insurance, not because there weren’t jobs – there were tons of jobs available – but because maybe it was OK to slow people’s move back to work a bit.
The second model would be: “Let’s take advantage. It’s an emergency, so let’s do some structural things at the same time. We’ll do $1.9 trillion, but spend it out over four years, and it’s a down payment on various things.”
We did that second model in 2009. We had a lot of money for green energy as part of the stimulus, for electronic medical records and so on. Arguably, that was a mistake, and by the way, some of us did not like it at that time. Some of us were very opposed to high-speed rail because we needed as much money in 2009 as possible. But part of the attitude was: “Don’t let a good crisis go to waste.”
Now, in this case, in some sense their judgment was vindicated in that they were able to come back and do quite a lot on things like climate change and infrastructure. So the fact that they didn’t fit it into the first bill, they did get to it later on – but in some sense, they risked it not happening, because who knew whether another bill would pass and how tight the margin was?
Q. Relative to Europe, the US has seen much more job switching and business creation since the pandemic. Does that explain the US’ different productivity performance? Why?
On business creation, I have no idea what’s going on and I don’t know what these businesses are. I don’t think anyone else who’s talking about them does, either. But tell me if you’ve found someone who does.
On employment: yes, the employment rate went down much more in the United States than in other countries during the pandemic. Some of that is just optics: somebody in the UK wasn’t working but was still getting a paycheck; someone in the U.S. wasn’t working but was getting money from the government. Functionally, it’s the same, but it’s recorded differently in the statistics. However, by the end of 2022 or 2023, the U.S. employment rate was still down, whereas it was up in most other places relative to pre-pandemic. So this goes beyond optics.
I’m very haunted by the possibility that the flip side of that is the higher U.S. productivity growth. There are two mechanisms for that. One is a reallocation, where in the US people move to better things, and in other places, they got stuck in whatever they were in before. The second is labour hoarding. Elsewhere, people were labour-hoarding and that doesn’t speak much to your long-run technological potential. In the short run, that shows up as lower productivity growth.
I’m not positive either interpretation is correct. Part of why I’m not positive is that the U.S. historically has had a higher elasticity of employment to the business cycle compared to other countries. Our unemployment rate also went up more than other countries during the financial crisis, for example. So there may be something in the nature of our labour-market institutions.
But yes, I’m haunted that in the US we need to take more seriously what can be done to have faster employment recoveries; and in Europe, people need to take more seriously that some of the low productivity they’re concerned about might be the flip side of strong employment rates.
Q. If labour productivity in the US disguises lower employment (and vice versa in Europe), does that mean we would expect the US and Europe to converge again over time?
The divergence has been going on for quite a while. If you look at the stock market, it’s easy to explain the divergence: a couple of huge tech companies. But the stock market is pricing a present-value concept, not actual realised productivity. Certainly, the Draghi report puts much of the difference, over a longer window, down to the technology sector.
I haven’t seen a sectoral comparison for the last couple of years, but one interesting thing is that it’s definitely services where we’re getting productivity in the U.S. Manufacturing productivity, I believe, has been pretty flat. The composition of productivity growth in the U.S. is shifting. That also speaks to industrial policy as manufacturing policy – it’s really about a small fraction of workers, and it’s not necessarily what’s been paying off recently. It’s definitely not the thing that’s been paying off in recent years.
Q. In the UK debate, there is significant concern Baumol’s “cost disease”: as we move into a service-dominated economy, we have to accept lower productivity growth. Why is there a difference between what’s happening in the U.S. and in the UK?
Over a longer horizon, services productivity has been a lot lower, but it’s just recently that services productivity picked up while manufacturing has fallen. So the composition of productivity growth we’re seeing has changed substantially.
The idea that you can do more with fewer hours – especially when you have technology – means so many technological innovations now are in the service sector, not manufacturing. I’m not sure how Baumol’s disease will hold up if AI can replace me as a lecturer. Then all we’d have left is, say, classical music performance, where it still requires four humans and 30 minutes to play a Beethoven quartet. Everything else, the AI might be better at. Perhaps it would even play Beethoven better, but people wouldn’t prefer it.
Q. And if you had to hypothesise about business creation, what would you say?
My question is: if I already know productivity growth and employment growth, does that give me any additional information? If it’s an imperfect proxy, if I didn’t know anything about those, I’d cling to this business-creation data as the one piece of data I had. But I already do know employment and productivity growth. Should I feel differently about two economies that have identical employment and productivity growth but one has higher small-business creation and the other has lower? Maybe it’s worse if you have a bunch of small businesses instead of super-exciting big businesses. I just don’t know whether it matters in a positive or negative way, once I already have the other information.
The data series people are looking at doesn’t go back far enough. I’ve asked whether anyone has research or a basis for interpreting it, and no one’s given me an answer. Maybe somebody does and just hasn’t shared it with me, but I’m not saying there isn’t an answer—only that I haven’t seen one.
I do think that the intellectual ecosystem in the last few years, at least in some spaces, did a better job surfacing positive data – because there are more wonky, numerate people who were cheerleading Biden than there were trying to find contradictory numbers. So I think the U.S. comparative GDP performance got amplified more than the U.S. comparative employment performance did for that reason, and less thought went into it. I think the small-business data got elevated less by the nerdy macroeconomic analysis that looks at central banks and more by people enthusiastic about the last few years—rightly or wrongly enthusiastic.
Q. Moving to industrial policy in the second half of the administration – IRA, CHIPS etc – what do you think the positives? Where would you critique Biden’s industrial policy?
I’d separate CHIPS and the IRA.
On CHIPS, I don’t think it’s that much of a positive innovation. If you want more of something, you give subsidies. They ran it in a pretty businesslike way. When it passed, my view was it wasn’t self-executing – we’d have to see how it was implemented. I think so far, so good, with a huge asterisk around Intel. They brought in a CEO from Blackstone, I believe, so they approached it in a deal-making style. Normally, the private sector is better at making deals than the government. But if the goal is to get microchips that are 20% worse in quality or price, the private sector won’t do that, so you need the government. Basically, that’s a national security argument, not an infant-industry argument.
If you talk to businesses, a lot of them preferred the investment tax credit to grants because there was less red tape. So maybe the government attached too many conditions; but for the most part, I think the “everything bagel” approach to childcare was more in press releases than reality. Most of that wasn’t binding. My guess is that by 2035, we’ll be making more advanced microchips in the U.S. than we would have otherwise. Does it pass a cost-benefit test? Probably, but we’ll have to see.
One thing the CHIPS Act didn’t have was a fully integrated approach – it basically gave subsidies. It would have been nice if it had also addressed immigration and permitting more basic research. It did authorise some research funding, but that wasn’t fully appropriated. Possibly government procurement or guaranteed-purchase approaches could have been considered too. So I can imagine five tools for CHIPS; they mostly did one, and didn’t do the other four.
On climate, the approach was more holistic: you have regulations that tell you to make more efficient cars, you have subsidies to buy those cars, money to build charging stations, subsidies for greener electricity, and subsidies for manufacturing the greener equipment. If anything, the climate approach might have been too holistic. There’s a big difference between installing solar panels in the U.S., which I wholeheartedly support, and making solar panels in the U.S., which, if it means more expensive, lower-quality products, I’m not on board with. And I’m not sure we could get them only 20% more expensive – China is just very good at scale.
Then you have charging stations. That was done through a state program rather than a federal program, so almost nothing has happened yet. I think eventually that money will be spent, and we’ll get more charging stations, but we’re not there yet. They were a bit slow to push NACS, the North American Charging Standard, which might help address some of the chicken-and-egg problems.
Some pieces didn’t fire on all cylinders. Some pieces weren’t necessary. On an up-or-down vote, though, I’d vote yes on the whole package.
Q. Were there other microeconomic policies pursued by the administration that you’d highlight that might be undervalued as part of the Biden administration’s toolkit?
Well, there’s one big one: infrastructure. That’s not particularly novel – it’s just the government spending money on infrastructure. Everyone has wanted that. In nominal terms, there was a big increase; but in real terms, after inflation, infrastructure spending is actually down relative to Trump and Obama. That highlights how it’s not just how much money you spend, but how you spend it. If you try to spend it too quickly, you run up against an inelastic supply curve and end up with higher prices rather than more quantity.
That’s a cautionary tale for infrastructure. It’s also a hopeful sign for medium-sized economies with fiscal constraints that maybe you can do a lot without necessarily spending more money – if you can spend existing money better, by reducing obstacles.
On smaller micro stuff, I think the administration’s approach to AI has been generally good. They’ve encouraged voluntary transparency – that builds public trust in AI without unduly inhibiting rapid change. If Congress had passed a bad law, maybe the administration would have signed it – I’m not sure – but Congress never did. Some proposals had language that was overly focused on DEI (diversity, equity, inclusion), without considering that humans are quite biased and maybe it’s easier to de-bias AI than to de-bias humans. But none of that was really binding. If you read the Wall Street Journal editorial page, you might hear about something posted on a website implying “hire these types of people or you can’t proceed.” But I don’t think it has slowed results in practice.
Another huge micro piece is antitrust. The administration moved in the right direction – greater toughness but overshot the mark by a decent amount. That’s particularly true of some of the remedies proposed in the Google case, likely also the Amazon case. We’ll see what the courts say, but in the Google case, especially, the proposed remedy on AI would inhibit their ability to be a real competitor in the AI space. Also, there’s probably been too much chilling of merger activity overall.
Q. You’ve mentioned a couple of examples – conditions around grant funding or DEI requirements – where you think there’s a gap between the rhetoric and the reality. Is it that the administration has tools in regulations that they’re choosing not to use, or are they trying to implement the rhetoric but finding it doesn’t work because people can easily get around it?
In both broadband and electric vehicle chargers, there’s a lot of money available and very little has happened so far. You look at the websites and they say you should hire these types of people, so folks say, “Well, that’s why no one’s doing anything,” but I suspect that’s not really the binding constraint. In the CHIPS program, I think 98% of the analysis was on “How many chips are you going to make, how fast, and where?” and only 2% was about “What’s your childcare program?” The childcare piece got a lot of press, but it wasn’t binding because everyone already had something close to it. In broadband and EV chargers, it’s mostly the states deciding who gets the money.
To be clear, I think diversity and inclusion are very worthwhile goals. But if you have a mission – say, building EV chargers – you probably don’t want to risk that mission failing because you placed excessive emphasis on an entirely separate set of goals. You can’t accomplish everything at once.
Q. These bills had objectives beyond economic growth – national security or climate. How might you redesign CHIPS or the IRA so they still deliver on climate or security goals but boost economic growth further?
For CHIPS, I’d push more high-skilled immigration. For the IRA, a carbon tax of maybe $80 a ton would have achieved the same emission reductions. That would have been more efficient, and it also would have generated money that you could use for other things. And I’d put less emphasis on ‘Buy American’ requirements, which I think yield negative emissions reductions and negative economic growth – lose-lose.
Overall, a decent fraction of the people implementing these programs understood they were doing cost-benefit thinking, trying to ensure benefits exceed costs. Some people spoke of solving all problems at once, but there were also a fair number of clear-eyed people.
Q. What lessons would you take from the U.S. experience over the last few years for the UK?
I don’t know if there’s anything different from what I’d have said a couple of years ago: try to figure out how far you can go without spending huge sums of money. Brexit is a net negative, but you can mitigate some of that because you have more freedom on the regulatory side now. I don’t know how much that’s been utilised.
I believe the U.S. is among the worst, and France among the best, in terms of infrastructure costs. I’m not sure where the UK ranks, but it might be good to relentlessly focus on that. Immigration is always a fraught issue, but perhaps you can at least target high-skilled immigration.
As for targeted programs, I’m not sure what I’d do on microchips if I were the UK – maybe free ride on everyone else. Even the huge sum the U.S. spent is just “a thing in the bucket” compared to global microchip investment.
There’s not something dramatically new here. You can take anything you like in the U.S. and adapt it, but I’m not sure the data from these three weird years suddenly changes big lessons.
Q. So fix the UK’s foundations rather than copying the US?
I think so. It’s only three years of data in an extremely weird time. It would be surprising if you learned a new lesson that caused you to totally change your mind.
In some ways, it just makes me more worried about certain things – like the fact that the US can generate a lot of GDP but not commensurate employment. I’m now even more sceptical about the “hot economy” thesis: that a hot economy is what narrows inequality and raises real wage growth. A lot of policymakers were sure that when you heat the economy, wages rise faster than prices. I surveyed macroeconomists about this in 2018, and they were sceptical. It’s theoretically ambiguous, and I think the last few years bear that out. But I don’t see the UK being in a position to heat its economy like that anyway.
ENDS