What this covers
Paul Romer and Tyler Cowen discuss the foundations of sustained economic growth, development policy, and what makes institutions actually work. Romer argues that economists have misread the problem: growth models must account for fishing-out effects as ideas become harder to find, yet a larger integrated global population can still sustain growth if we increase the share of people in research. More fundamentally, Romer contends that development and policy success depend less on transplanting good laws and more on understanding the durable motivations and norms that exist "inside the head vs outside the head" — the distinction between material discoveries like transistors and penicillin versus the harder-to-measure human drives that shape how societies actually function. The conversation circles around institutional mechanisms like charter cities as tools to gradually socialize populations into new cultural norms.
The discussion branches across education, management, urban design, and the persistence of culture. Romer traces how culture as persistent norms — imprinted by founding populations — explains why Maryland's legal guarantee of religious freedom failed while Pennsylvania's endured, and why the United States and New Zealand diverged politically despite shared immigrant sources. He argues culture operates through subtle forces like shame, guilt, and identity that economists typically ignore. On education, he identifies asymmetric information problems, the failure of English orthography, and how schools amplify rather than dampen inequality — then points to Singapore's model of targeting struggling readers with the best teachers. In urban design, he credits Gouverneur Morris's rectangular grid as foundational, and argues successful cities require both design plus decentralized forces. Throughout, he emphasizes that management and organizational reform should start from empirical observation of what actually worked — like W. Edwards Deming's principles at Toyota — rather than from theory alone.
Romer argues that sustained economic growth and successful development depend less on importing laws or technology and more on understanding the durable, hard-to-change cultural norms and human motivations 'inside the head' — and that institutional mechanisms like charter cities and founding populations are how those norms can actually be shifted.
- Growth models must incorporate diminishing returns to ideas (fishing-out) but a larger integrated population can still sustain growth
- Culture and norms persist from founding populations and explain why importing laws alone fails
- Charter cities work as a mechanism to socialize migrating populations into new norms gradually
Founding populations' beliefs and norms imprint durable political institutions that later migrants adopt through socialization.
- Founding populations imprint durable political norms: New Zealand and the US diverged in political language because their founders left England at different times (US amid concern for religious freedom, NZ when economic inequality dominated), and even as large numbers of migrants from the same sources later arrived in both, they were socialized into the existing norms, producing persistent differences in politics and decisions.
“What’s interesting is the persistence of those founding effects.”
- When a population's norms are inefficient and holding it back, an effective mechanism for changing the whole distribution of norms is to let a nonrepresentative subgroup found a new place; as people then migrate in at a moderate rate from the old population, they get socialized into the new place's predominant culture — changing norms more feasibly than trying to change them in place.
“this idea of letting a nonrepresentative subgroup go off and be the founding population in a new place. Then as people go in at a moderate rate from the old population to the new one, they can get socialized into the predominant culture in the new place.”
- Pedestrian behavior illustrates norm change: you couldn't get New Yorkers to adopt Zurich's pedestrian norms by passing or enforcing laws (you'd get a backlash), but if you move a few people to Zurich they start behaving like Zurich residents, and over time you could move everybody from New York to Zurich and shift the whole population's norms — a politically uncharged example of the founding-population process.
“if you do move a few people to Zurich, they’ll start behaving like people in Zurich, and then over time, you could move everybody from New York to a place like Zurich.”
- There is a migration trilemma in which liberal democracies can have only two of three desirable features: local democratic accountability, equal treatment under the law, and the ability to absorb very large numbers of migrants (potentially exceeding the existing population). Existing jurisdictions pick the first two and forbid large-scale migration, because if a small existing population admits far larger numbers under equal treatment and democracy, the new arrivals will collectively vote in a system other than the one the original residents value.
“I think we should talk about a trilemma for migration, which is three things, and we can only have two out of the three.”
- The best resolution to the migration trilemma is a Hong Kong-like arrangement that sacrifices the local-democratic-accountability leg temporarily: commit to large-scale migration with equal treatment under the law, governed by an offshore but accountable authority (as Hong Kong's governor answered to British democracy), and make the transition to local democratic control only once maximum population is reached and migration ends; violating equal treatment instead is too damaging to both migrants and existing society.
“We’re committed to large-scale migration for a period of time. We’re going to give everybody equal treatment under the law, but we’ll make the transition to local democratic accountability when we’ve basically reached our maximum population and the migration process is over.”
- A charter city cannot succeed by importing good laws without the matching culture: Maryland legally guaranteed religious freedom but it failed because its founding population (a Catholic elite over Protestant servants) wasn't committed to it, whereas Pennsylvania's freedom of religion endured because William Penn recruited people who actually believed in it — proving the document didn't matter, the founding population's beliefs did.
“What mattered was that there were a bunch of people in the founding population who were committed to this idea of a separation of church and state and religious freedom. And that’s what made it durable in Pennsylvania in a way it wasn’t in Maryland.”
- There is currently no answer to large-scale migration crises: no outside force can invade and quickly resolve a civil-war collapse, no existing political system can absorb 20 million arrivals without real risk to its political equilibrium, and there's no answer either to the chronic problem of hundreds of millions who want to leave places of low opportunity and high violence.
“there arenoexisting political systems that could absorb 20 million new arrivals without real risk to whatever their current political equilibrium is.”
Norms and motivations—shame, identity, anger, guilt—are complex forces economists neglect but crucial for understanding outcomes.
- Growth and social outcomes are hard to forecast not because people don't average out, but because a social-interaction dimension introduces correlation rather than independent aggregation, so even large groups behave unpredictably — and economists lack a rich picture of what really motivates people beyond 'more is better.'
“there’s a social-interaction dimension to people that means, instead of just aggregating independent entities, you’re getting correlation. It’s actually a pretty general phenomenon that social outcomes are hard to predict.”
- Economics treats motivation as 'more stuff is better,' but subtle forces — norms about right and wrong, identity, shame, anger, guilt — are powerful emotions that don't average out when people interact, and are important for understanding group outcomes; economists need 'culture' as the label for these effects.
“There are these very powerful emotions that don’t average out necessarily when people interact, and that are important for understanding group outcomes.”
- Dividing the world into 'outside the head' (transistors, penicillin, formulas to discover) and 'inside the head' (motivations), the usual economic model assumes complex conscious reasoning but very simple preferences; in fact human motivations are at least as complex as our conscious reasoning, and a lot of the action in development lies there.
“The usual economic model of inside the head is complex conscious reasoning but very simple preferences or emotions or motivations. In fact, the motivations are atleastas complex as our ability to reason in a conscious sense.”
- Economics has behavioral models of individuals and of markets, but lacks an analysis of teams where agents with complicated, correlated motivations engage in teamwork; developing behavioral or broader motivational analysis of teams is the promising direction for management research.
“if you think about the theory of teams, we don’t have an analysis where we have these agents with more complicated motivations, and you think about teamwork.”
- Questions about what is happening to the distribution of income — labor share versus profits — are very important current concerns that traditional growth models simply haven't addressed.
“What’s happening to labor share versus profits? Those are topics that traditional growth models just haven’t addressed.”
Growth can be sustained by increasing the fraction of research workers and improving R&D policy efficiency.
- Even with diminishing returns to ideas, growth can be sustained because a world heading toward 10-11 billion people allows increasing the fraction of people in research, offsetting fishing-out problems; and a bigger globally integrated economy plus better research policy can raise efficiency, since we are not yet at the optimal level of R&D.
“we can offset some of these fishing-out problems just by having more people working on it.”
- To fit the broad sweep of history — long-run growth alongside a huge increase in researchers — models must include something offsetting the rise in researchers, either short-run adjustment costs that make it hard to increase the rate of idea growth, or a fishing-out effect where the more you've discovered the harder remaining ideas are to find.
“Chad Jones has really been leading the push, saying that to understand the broad sweep of history, you’ve got to have something which is offsetting the substantial increase in the number of people who are going into the R&D-type business”
- Characterizing economies requires more than the catch-up-versus-frontier (technology level) dimension; a separate institutional-evolution dimension is needed, since an economy can be institutionally well-developed but technologically backward, or technologically successful with very weak institutions — and a country like China is really several different economies (Shanghai versus rural China).
“there’d also be a separate dimension, which is something like the institutional evolution of the economy. So you could have a very institutionally well-developed economy that’s still very low in terms of its technological success.”
- The best test of a theory is using it to think about problems, and Romer found he needed a different theoretical apparatus to think about a challenge like Tanzania's than a challenge like Germany's — supporting the need for multiple dimensions beyond the technology-level distinction.
“I needed a different theoretical apparatus to think about, say, the challenge in Tanzania than the challenge in Germany.”
- Growing up in the mountain West and then moving to the East Coast may have made Romer (and parallel figure Bill Nordhaus, from New Mexico) more willing to take on unconventional problems outside the accepted consensus, because the cultural difference between regions creates a sense of being an outsider that can filter into one's work.
“I’ve wondered if there was something about the mountain West, then moving to the East Coast that made us both willing to take on things that weren’t the accepted consensus”
Schools amplify inequality at entry; targeted interventions like phonics instruction can substantially reduce educational gaps.
- Children enter school with inequality in preparation and makeup, and current school systems substantially amplify that inequality rather than damping it; targeted attention — such as phonics-based instruction for students with dyslexia early on — could keep struggling readers caught up with peers and substantially reduce inequality in educational and eventually labor-force outcomes.
“make sure our school systems damp that inequality instead of doing what they currently do, which is to substantially amplify that inequality.”
- For children who really struggle with reading, phonics-like instruction is clearly the approach that works, not whole-word or other broken approaches.
“the phonics-like instruction is clearly the thing that works for those kids, not a whole word or these other broken approaches”
- Singapore's school system frequently tests students in math and reading through about fourth or fifth grade and gives those falling behind the best teachers and more instruction, investing extensively in students who might otherwise fall behind, thereby achieving much more equality in educational outcomes.
“And if you’re falling behind, you get the best teachers. You get more classroom instruction. They really invest extensively in the ones who might otherwise fall behind”
- There is a formative period in very early childhood when the care, affection, and emotional warmth from the mother is a big influence on the rest of one's life, providing emotional capacities that come from growing up in a supportive environment.
“there’s this formative period, when you’re a very young child, where the care and affection and warmth, the emotional warmth, with the mother is a big influence on the rest of your life.”
The primary innovation policy is investing directly in graduate students rather than professors.
- The number-one policy to promote innovation is to invest in people, because well-trained humans are the inputs into the discovery process; specifically, redirecting graduate-school funding away from professors as principal investigators and toward promising students directly would give students freedom and control over what they study and where they work.
“My number-one recommendation is to invest in people. Humans that are well trained are the inputs into this discovery process.”
- If students controlled the tuition resources universities compete for, and were oriented toward the far future and big opportunities, universities would become more responsive to the changing landscape — functioning like vouchers for graduate school that force competition between educational institutions.
“If those students controlled tuition resources that universities were competing for, the universities would be more responsive to the changing landscape.”
English orthography's poor sound-spelling correspondence harms reading acquisition and increases inequality.
- Spelling reform is hard partly because class is signaled through pronunciation and spelling, but having two different orthographies has no necessary connection to different pronunciations — in principle people can share a dialect and pronunciation while using different spelling systems, avoiding the class-inequality fragmentation seen in pidginized Englishes.
“having two different orthographies hasnonecessary connection with different pronunciations. In principle, you can have a shared dialect and pronunciation even though the parties had different spelling systems.”
- Among modern languages, English has the worst orthography — the worst mapping between spelling and sounds — which makes it take longer to learn to read and, crucially, increases the variance: more people never get over the hurdle of learning to read at all, a tragedy as English becomes the universal second language.
“English has theworstorthography, theworstmapping between spelling and sounds of any of the existing languages.”
- Spelling reform has always failed in the Anglosphere and is hard to mandate directly, but China — which has the most people learning English — has the incentive and could introduce a rationalized spelling running in parallel with traditional spelling, since translation between the two would be a trivial word-for-word problem solvable by a plug-in.
“If somebody had a big incentive to actually create a rationalized spelling that could operate in parallel with the traditional one, it would be something like China.”
- A default dialect for spoken English — Globish, a slightly simplified global version with different uses of articles — may emerge among second-language learners, who understand each other better than they understand native speakers because native speech carries distracting features.
“There might emerge a default dialect for spoken English that dominates amongst people who learn it as a second language. They understand each other much better than when they talk to somebody like you or me”
Transparent processes, knowledge transmission through reading, and empirical evidence on team management drive effective organizations.
- To make progress on management questions, one should not start from assumed axioms but proceed empirically — go look at evidence of where reforms actually worked in organizations (like Toyota applying Deming's principles), then build a theory afterward about why things like trust matter.
“I wouldn’t sit down and make some assumptions, like, “Here are the axioms of management.” What I’d do is go out and look at the evidence.”
- Too much work in progress impedes the flow-through of a system, like congestion on a street; limiting work in progress (as Kanban does with visual cards) avoids the costs of task switching and distraction and improves coordination.
“too much work in progress can really impede the flow-through of a system... It’s kind of like congestion on a street. You limit the amount of work in progress to avoid the effects of task switching and distractions.”
- Transparency is very helpful in teamwork because when people don't know what others are doing, they have a bias toward the negative and assume something bad is happening; making everyone's work visible (as Kanban does) is therefore a good way to build trust.
“when people don’t know what others are doing, they tend to have a bias towards the negative. They think something bad is happening. Transparency is actually a good way to build trust.”
- Charles Sanders Peirce had a far more sophisticated sense of how science proceeds than the positivist 'machine' picture people describe, and his work on semiotics — the distinction between the sign and the concept it represents — is a profoundly important distinction rather than academic angels-on-a-pinhead worrying.
“had a much more sophisticated sense about how science proceeds than the positivist sort of machine that people describe.”
- Because teams like IT or research units face almost unlimited requests but can process only a subset, having a transparent and rational process for selecting what to work on produces much better outcomes than a covert process influenced by which friends ask for favors.
“If that’s transparent enough to other people and there’s some rational process for making those decisions, you get much better outcomes than if it’s covert and influenced”
- The joy and foundation of scholarship — and of modern life — is that through reading we can access another person's thoughts, even a hundred years later, and sample the most relevant or powerful of them; this is the foundation for the transmission of knowledge and the whole growth process.
“through reading, we can get access to the thoughts of another person, and then you can sample from the thoughts that are most relevant to you”
Universities became less responsive to changing needs post-WWII, and asymmetric information prevents education from functioning as competitive markets.
- The education business has among the biggest asymmetric-information problems, because a young person paying to be educated is highly dependent on the educator's decisions about what to study, so it cannot be treated as a perfect-information competitive market.
“the education business is one of the ones that has thebiggestproblems with asymmetric information.”
- US universities were highly innovative in the first half of the 20th century, inventing whole new schools like chemical and electrical engineering, but have been far less responsive in the post-WWII period; the only big change, computer science, was driven by the defense department rather than universities responding to demand.
“Universities have not been as innovative or responsive in the post–World War II period. The only big change has been computer science, and that was really driven by the defense department in the US government rather than the universities responding to demand.”
Technology effects like electrification and the internet take a decade or two to measurably boost growth.
- The general-purpose technology is a useful abstraction that takes a continuum of possible innovations and defines a discrete category for those with long exploration paths that induce much complementary innovation (like electrification), as opposed to one-off innovations like the paper clip — though whether we can recognize a GPT as it unfolds, versus only in hindsight, is uncertain.
“the general-purpose technology has been a useful way to distinguish a type of innovation that you could think of as being aligned along a continuum.”
- The internet may not have visibly boosted measured growth yet because, like electricity historically, its effects on output may take 10-20 years to appear and may include benefits that current measurement misses.
“It’s also possible that it will have an effect on output that we won’t see for 10, 20 years. This is the kind ofstory that Paul David tells about electricity.”
- Five-year moving averages are too short a window to extract the low-frequency growth component because growth fluctuates substantially decade by decade, largely driven by the macroeconomic stimulus-versus-recession position.
“I don’t think five years is enough to really pull out the low frequency or growth component. Things vary so much.”
Foundational design like Manhattan's street grid enables infrastructure retrofit and provides lasting public value.
- Gouverneur Morris's rectangular street grid for Manhattan was incredibly valuable because it asserted a two-dimensional grid of public space that public authorities could later use for vehicles, pipes, wires, bike lanes, or buses; such a foundation is almost impossible to retrofit if you don't start with it, since there will be no more Haussmann to bulldoze new streets — so Morris's foundational contribution outweighs both Moses and the design-versus-Jacobs debate.
“all it was was an assertion that here’s this two-dimensional grid of public space so that the public authorities will be able to make decisions in the future about how you use it”
- A successful city requires both some top-down design and decentralized market-like forces; you can't have a successful city without some level of design, and Robert Moses's failure was not design itself but being too enamored of the car and getting the design wrong for the infrastructure everyone relied on.
“You need in a city both some design, which you can think of as kind of top down, and then some market-like or decentralized forces. You can’t have a successful city withoutsomelevel of design.”