Jason Furman
What this covers
Jason Furman diagnoses US economic underperformance through the lens of policy failure rather than structural inevitability. The conversation covers his case that four correctable constraints—rising concentration, restrictive land use, weak labor market fluidity, and too-low immigration—account for much of the slowdown in productivity growth. Rather than retreating to protect manufacturing or adopting human-rights framings for technology regulation, Furman argues the policy lens should stay trained on productivity itself. He walks through evidence on monopoly power and investment, the distinctive economics of acquisition versus organic growth, and why immigration of talented people represents the single highest-leverage policy lever available.
The conversation ranges widely across where these constraints bite. Furman examines spatial mismatches created by restrictive zoning in high-productivity cities, the poor labor-market outcomes revealed by China shock literature, and mechanisms by which concentration dampens the investment incentives firms face. On technology specifically, he argues Human Rights vs Competition Framing matters for design—that antitrust and merger scrutiny address competitive concerns more cleanly than privacy-as-rights approaches, while also contending that not all tech tools (facial recognition, data portability) warrant blanket prohibition. The discussion also covers intellectual puzzles: why productivity growth was unusually high in certain periods rather than why it fell, whether internet benefits have been net positive when compared against displaced activities, and the constraints on sustainable debt even at low interest rates. Throughout, Furman acknowledges tradeoffs and contested ground rather than declaring final judgment.
Furman argues that US economic underperformance stems primarily from solvable policy failures — rising market concentration, restrictive land use, weak labor fluidity, and too-low immigration — and that boosting productivity, not protecting manufacturing or restricting tech via human-rights framings, is the right lens for most economic policy.
- Increased industry concentration and monopoly power reduce investment incentives and contribute to the productivity slowdown
- More immigration of talented people is the single highest-leverage productivity policy
- Tech competition concerns are best addressed through antitrust and merger scrutiny rather than treating privacy as a competition problem
Increased market concentration has reduced business investment incentives and correlates with where investment shortfalls actually appear.
- Part of the US investment shortfall is benign (the intangible economy needs less business investment than heavy industry did), but part reflects policy failure: increased concentration in certain industries has created greater monopoly power, which reduces firms' incentives and need to invest.
“the increase in concentration in certain industries has led to greater monopoly power and has reduced some of the incentives and need to make investments in those industries”
- Macro measures of market power are limited and known to be problematic, but they still have predictive power — they correlate with profitability, with markups at the aggregate level, and match data features like the rising return to capital relative to the safe Treasury rate; they mislead in some sectors like retail but hold up on balance.
“it does seem associated with things like profitability, with things like markups at an aggregate level”
- Research by Thomas Philippon and Jan Eberly shows that predictions based on increasing industry concentration closely match where investment shortfalls actually appear, supporting a causal link between market power and underinvestment.
“the predictions you’d get by looking at increasing concentration are pretty similar to the predictions you’d get in terms of where you see the investment shortfalls”
- In the health sector, market power demonstrably leads to higher prices; for drugs this creates a genuine tradeoff between innovation and prices, but for hospital mergers there are very few efficiencies — the main one being the ability to pay nurses less.
“I’m quite confident that in the health sector, market power leads to higher prices.”
Immigration of talented innovators is the single highest priority for boosting productivity and economic growth.
- The single highest-priority policy to boost productivity is more immigration of talented people who add to innovation, bring ideas, and start businesses; secondary measures are much-increased federal R&D (which has fallen as a share of GDP since the 1960s), business tax reform with investment incentives like expensing and an expanded R&D credit, and investment in American higher education.
“number one, two, and three would be more immigration, more immigration of talented people that will add to innovation here in the United States”
- Micro evidence, including Jennifer Hunt's research, shows that not only do immigrants patent, but native-born Americans end up patenting more when in the presence of those immigrants — strong enough micro data that we don't need macro decade-level inferences to establish immigration's innovation benefits.
“Jennifer Hunt found that the Americans end up patenting more when they’re in the presence of those immigrants”
- While Furman would always prefer more immigration to less and believes the US could economically support 700 million to a billion people, he argues Bryan Caplan's open-borders case does not take seriously enough the transition issues and costs — the speed at which political, cultural, and economic changes occur is genuinely relevant.
“Bryan Caplan’s book, which I loved, didn’t take completely seriously some of the transition issues and some of the transition costs”
Acquisitions warrant antitrust scrutiny more than organic growth because they presume less efficiency than organic expansion.
- Organic growth (like Walmart opening better stores one by one) carries a presumption of efficiency that should be left alone, but growth by acquisition does not: Facebook holds three major social media companies because it bought two, and almost every Google component except the search algorithm was purchased — making them more like mergers to monopoly that warrant scrutiny, whereas Amazon (mostly organic) is lowest on the concern list.
“Facebook has three major social media companies all in one because it bought two of those three. Almost every component of Google was, except the search algorithm, purchased at some point along the way.”
- EU policy toward big tech should not be lumped together: antitrust enforcement (e.g., blocking the Siemens-Alstom merger that infuriated both France and Germany) is technocratic, independent, and blind to nationality, whereas the digital tax proposals come from politicians and fit public-choice predictions — the politicians are right that they're missing taxes but wrong to use sector-specific levies jury-rigged to target American companies.
“They’re wrong to have sector-specific levies and proposals, in some case, jury-rigged just to get at American companies.”
- Anticompetitive-conduct rules can easily be applied only to the largest players, but rules framed as human rights cannot — just as you wouldn't exempt small businesses from rules against killing their workers, a privacy right framed as a human right should apply regardless of firm size; GDPR's framing as a human right rather than a competition matter makes it cumbersome and easier for big firms to navigate than small ones, though the human-rights approach still has merit.
“That’s easy to do if the code of conduct is around anticompetitive behavior. That’s much harder to do if the code of conduct is about something that’s considered a human right.”
Poor US labor market fluidity—restrictive land use, occupational licensing, healthcare job lock—amplified recession damage and weakened matching models.
- Both the 2001 and 2007-09 recessions produced prolonged long-term unemployment, involuntary part-time work, and declining participation because they hit when labor market fluidity was already low — fewer job-to-job moves and less business job creation/destruction; the underlying causes include restrictive land use, occupational licensing, and healthcare-related job lock, and fixing fluidity would make matching models work better in the next shock.
“if you have a less fluid labor market, then a shock will have all of these bad consequences”
- There is a sizable spatial mismatch within the US: the places where people can most productively contribute — the Bay Area, Boston, Washington — are unnecessarily expensive due to artificial constraints on building that drive people out and keep people from coming in, so YIMBY policy would meaningfully raise productivity.
“places like the Bay Area, the Boston area, the Washington area — are ridiculously and unnecessarily expensive to live in because of artificial constraints on building that drive people out and keep people from coming in”
- The China shock literature is a major contribution to labor economics but less important for thinking about trade; it revealed weaknesses in US labor markets — poor job matching, low mobility after shocks, inadequate retraining and wage insurance — and together with observations about declining prime-age male labor force participation, it changed Furman's view of labor market dynamism far more than his view of trade.
“the China shock literature is a fantastic contribution to labor economics and is less important in thinking about trade”
Productivity growth is realistically constrained to around 1-1.5 percent, requiring focus on what drives growth rather than blame for historical slowdowns.
- Extrapolating the last 50 years of average productivity growth yields a GDP growth rate around 2 percent; the realistic ceiling for productivity growth is about 1.5 percent if lucky, though it has recently been closer to 1 percent — Furman positions himself as a realist who expects reversion to historical trend rather than the pessimist's stagnation or unrealistic acceleration.
“broadly speaking, we’re in a world of 1.5 percent productivity growth if we’re lucky. Unfortunately, it’s been more like 1 [percent] lately”
- While less infrastructure investment and R&D plausibly contributed to the productivity slowdown, when you actually quantify that type of public investment it explains only a couple tenths of a percent; the better framing may be to ask not why productivity slowed but why it was unusually fast in special periods like the post-WWII 1950s-60s and the brief 1995-2005 window.
“when you go and quantify that type of public investment, it’s a couple tenths of a percent of the answer”
- Most unmeasured internet benefits (Google, online travel, online shopping, Wikipedia) entered the productivity base 5-15 years ago, so they raised the level of well-being but add little to measured growth now; further, net benefits must be computed against displaced activities (YouTube is only slightly better than TV), yielding a decent impact on the level but only a small impact on the growth of well-being over the last decade.
“They might add to the level of productivity, but they were in the level. They were in the base a while ago now.”
Surveillance tools, fintech competition, data portability for large firms, and facial recognition require process-based oversight rather than outright bans.
- Furman opposes broadly banning facial surveillance, arguing it is an extension of existing legitimate tools — police can already look at someone on the street and post wanted posters — and that since all law enforcement violates some principle (locking people up, taking rights away), the right approach is to surround surveillance with processes that respect balances of rights rather than to discard a tool that could keep people safer.
“You’re allowed to have a police officer on the street looking at somebody. You’re allowed to post a wanted poster for somebody.”
- Mandatory data portability should be turned on for large companies, where lack of portability creates a barrier to entry that entrenches them, but turned off (or not applied) for small companies — even though this raises a genuine concern that requiring portability into incumbent standards could lock in current ways of doing things and stifle innovation.
“with the large companies, the lack of portability creates a barrier to entry. It entrenches them and reduces competition. So I’d turn the dial one way for them, and for smaller companies, I’d turn the dial the other way”
- The fact that payments are not simple and free between any two people anywhere in the world is the lowest-hanging fruit in fintech for billions of people, and it is something we should be fully able to solve in the near future — best pursued through more competition, including big tech entrants, provided rules are set up front.
“the fact that payments aren’t simple and free for anyone anywhere in the world... to me seems like the lowest-hanging fruit for billions of people”
The US can sustain primary deficits given low interest rates but should explore borrowing limits rather than accept a 400 percent debt trajectory.
- Borrowing more is not a free lunch even at low rates: there is ultimately a resource constraint at the planetary level, and while the US can relax its own constraint by financing consumption from abroad, that consumption (and any foreign-financed investment) must be repaid and will come out of future US well-being.
“if what we’re doing with the rest of the world is financing consumption and financing investment, that will come out of our well-being in the future. So no, I don’t think we have a free lunch there.”
- Because interest rates are below the growth rate, the US can sustainably run a primary deficit; but the current primary deficit of about 4 percent of GDP implies debt would asymptote to roughly 400 percent of GDP — possibly survivable, but Furman would rather explore how much more borrowing is feasible slowly rather than commit to that trajectory.
“our primary deficit is set to be about 4 percent of GDP. That means our debt under that plan would asymptote to about 400 percent of GDP”
- Over the last 140 years real rates have been lower than current levels only about 25 percent of the time and higher 75 percent, and rates are lower in most advanced economies, suggesting roughly symmetric near-term risk; but at the extremes the risk is asymmetric because rates cannot fall 10 points while they could rise 10 points — a manageable tail risk one can hedge by borrowing long-term to lock in low rates.
“Rates can’t go down by 10 percentage points. Rates could go up by 10 percentage points, so I think there is some tail risk there.”
Manufacturing deserves no special policy priority relative to services; current policy over-emphasizes goods production.
- Manufacturing itself should not be an important objective of US policy — it is one type of good job among others, and policy should not privilege where physical things are made over where services are made; if anything, current policy over-emphasizes manufacturing and under-emphasizes services.
“I don’t think that manufacturing itself should be an important objective of US policy. It’s one type of job.”
- The basic textbook proposition — that trade has winners and losers, where winners are diffuse with smaller gains and losers are concentrated with large losses — is exactly what the China shock literature confirmed.
“the winners were diffuse and tended to have smaller gains, and the losers were concentrated and tended to have large losses. That seems to be exactly what the China shock literature has found.”
- National security arguments about supply chains should be taken seriously but treated with great wariness because many are tendentious and unpersuasive; in semiconductors specifically, the diversified world supply and ability to scale up domestically meant the concern was protectionism dressed as national security.
“there, it was protectionism under the guise of national security”
Macro indicators like inflation have deteriorated in meaningfulness due to quality measurement challenges and low volatility.
- The macro indicator that has most deteriorated in meaningfulness is inflation: we don't measure it well — increasingly because more of the economy is in hard-to-measure-quality areas, healthcare being the most notorious — and it is low enough that it's hard to get excited about it.
“the one that probably has deteriorated in meaningfulness is the measure of inflation”
- The 1980s' brief period of unusually high interest rates did lasting damage to economic forecasting because people treated that anomaly as normal; much of the subsequent decline in rates is simply a return to the pre-1980s trend, and analyses of long-term rate declines must also account for changes in the risk profile of bonds, which may now warrant rates below the pure safe rate.
“so much damage was done to economic forecasting by the experience of the 1980s”
Savings rates are not a binding constraint given low capital costs; Social Security expansion addresses retirement better than private savings.
- Furman is not very worried about the savings rate because the cost of capital is currently very low, so savings is not the binding constraint on business investment; for retirement security he would expand Social Security rather than try to raise private savings, and if increased savings were desired, reducing the budget deficit is one route.
“the first place I would go for retirement security would be to expand Social Security rather than to raise savings”
Social media's zero price is illusory; consumers pay through higher advertised product prices, privacy loss, and foregone innovation.
- The 'zero price' of social media is illusory because Google and Facebook are advertising companies that dominate the online advertising empire; consumers pay through higher prices on every advertised product (just as steel costs are embedded in steel-containing goods), through giving up privacy and data, and through unseen lost innovation from products that never existed because entry was too hard.
“I am paying more for any product that has to pay more in advertising fees to Google and Facebook, which dominate the online empire.”
Economics pedagogy faces severe challenges in retention and application, requiring more rigorous evidence-based course design than currently exists.
- The scariest finding in the economics pedagogy literature is that, a year after an economics class, students who took it perform on questions like 'what's the opportunity cost' about the same as those who didn't — making the highest and hardest-to-verify teaching objective ensuring students recall and apply core ideas 15-20 years later when facing real issues.
“compared performance on a test a year after an economics class, people who had taken the class to people who hadn’t... And the two groups performed quite similarly.”
- Even at elite universities, course design is done with far less randomized, rigorous evidence than economists demand of public-policy questions; Furman redesigned Harvard's 600-student intro economics course using mostly judgment rather than evidence, and is painfully open to the likelihood that better methods exist right next to him that he can't identify because the evidence isn't there — so claiming higher ed runs at 30 percent capacity is only useful if accompanied by knowing how to reach 50 or 100.
“we did all of that with, frankly, less randomized trial, rigorous evidence than I would do on most of the public policy questions that I work on because it’s simply not there”
Policy-economist career paths are idiosyncratic and harder to replicate as think tanks become more tribal and partisan.
- Furman's policy-economist career path is highly idiosyncratic and probably wouldn't work for even one in ten people who tried it, and it may be slightly harder to replicate now because think tanks have become more tribal and partisan, leaving less space for independent thinking that isn't useful in the short term.
“what I’ve done has been pretty idiosyncratic. I don’t know that it would work out for even one out of ten other people that tried it.”
- Furman is skeptical of place-based policies: algorithmic, formula-driven ones (extra Medicaid or longer unemployment benefits triggered by a region's deeper recession) can make sense, but regional development efforts to create clusters or industries have a history of failure, and the political system implements them worse than any other policy area because Congress is least interested in expert input on which locations get resources.
“The regional development... I think that’s littered with a history of failures.”
Blockchain has no meaningful advantage over existing systems and cannot match payments industry transaction throughput.
- Furman is strongly bearish on blockchain: very little of its proposed use seems like something that couldn't be done more cheaply and efficiently with prior systems, and blockchain cannot come close to the transaction throughput per second that Visa and MasterCard already handle; he also notes that the technology's enthusiasts all have a stake in its success.
“very little of it seems like something that couldn’t be done more cheaply and efficiently with the types of systems we had in the past”
Economists should co-author papers holding different estimates to surface disagreement sources and make assumptions transparent to readers.
- Furman advocates pairing economists who hold different point estimates to co-author papers — as he attempted with Robert Barro on the 2017 tax law — so that readers get either a reliably understood truth or a transparent crosswalk of the differing assumptions driving disagreement; in their case they converged on the causal estimate but still drew very different normative conclusions.
“force people who have two different sets of numbers to write a paper together, so we can have something that we can reliably understand is the truth on the issue”
Brexit imposes a couple to five percent GDP loss over the long term through microeconomic inefficiencies rather than an immediate shock.
- Brexit's economic cost is primarily a medium-to-long-term loss from microeconomic inefficiencies — worse supply chains and worse import/export composition — totaling a couple percent of GDP (possibly two, possibly five) accruing slowly over the long run, rather than a large up-front hit.
“I put Brexit primarily in a medium- to long-term cost of the set of microeconomic efficiencies associated with somewhat worse supply chains”
New York City provides exceptional childhood autonomy through accessible transit and diverse experiences.
- New York City is an amazing place to grow up because children can have great autonomy — riding the subway and bus and accessing many types of food and experiences.
“Children can have a lot of autonomy in New York. They can ride the subway.”
Mixed banking and commerce holds no clear benefit and presents precautionary concerns absent specific problems to solve.
- Furman sees no large benefit to allowing banking and commerce to mix in the US and, on the precautionary principle, no particular problem that allowing greater mixture would solve — though he holds this without conviction and would be open to changing his mind.
“on the precautionary principle, I don’t see any particular problem we’d be solving by allowing a greater mixture than we do”
Post-2008 US unemployment recovery matched historical pace despite weaker GDP growth, consistently beating forecasts.
- Across roughly two dozen OECD countries over 50 years, unemployment rates can rise very quickly but the fastest any country brought it down was 0.7 percentage points per year; the post-2008 US recovery in the unemployment rate was about the same or slightly faster than the 1980s and was the one variable that consistently beat forecasts, so it was not abnormally slow even though GDP growth was weaker.
“the fastest that any OECD country achieved was 0.7 of a percentage point per year”
Healthcare quality depends far more on behavior, genetics, and social environment than medical care, yet healthcare offers more marginal improvement leverage.
- Behavior, genetics, and social environment all matter far more than healthcare for differences in mortality and health outcomes, but because it is much easier to know how to change healthcare than those other factors, there may be more marginal gains achievable in healthcare even though it is not the number-one factor in health.
“behavior, genetics, social environment — I think all matter a lot more than healthcare”