What this covers

Suresh Naidu, an economist at Columbia University, speaks with Sean Carroll about how real labor markets diverge sharply from the frictionless supply-and-demand model taught in introductory economics. The conversation centers on monopsony—a firm's discretion to set wages below the competitive level because workers face frictions and cannot instantly flee to better jobs—and its role in sustaining inequality. Naidu argues that empirical research from the past 20 years, built on natural experiments rather than convenient correlations, has overturned classical predictions: minimum wage increases destroy far fewer jobs than Econ 101 predicts, matched worker-firm data reveal that 15–20% of wage variation comes from employer choice rather than worker productivity, and when firms cut wages by 10%, only about 30% of workers leave over a year rather than the 100% the frictionless model demands.

The discussion ranges across how economic rent functions differently from rent-seeking, why labor union structures in the United States differ fundamentally from European sector-wide models, and how inequality itself—independent of material living standards—damages democratic health through zero-sum competition for status and political power. Naidu argues that a 40-year rise in inequality requires institutional remedies, particularly a revived labor movement, rather than faith in self-correcting markets. The conversation also touches on broader methodological shifts in economics toward causal identification, the ideological divisions within the discipline, and Gibrat's Law as an explanation for wealth concentration under capitalism. Naidu places these arguments in historical context, tracing the intellectual lineage of monopsony back to Joan Robinson and 1940s labor economists, and questions whether the frictionless market model remains a useful starting point for understanding labor, land, or credit markets.

Sharpest takeaway

Naidu argues that real labor markets are pervaded by frictions and employer wage-setting power (monopsony), so the frictionless supply-and-demand model misleads on policies like the minimum wage, and that rising inequality is a structural problem requiring institutional remedies—especially a revived labor movement—rather than faith in self-correcting markets.

  • Empirical identification shows minimum wage increases destroy few jobs, contradicting the competitive model
  • Workers don't instantly leave when wages are cut, revealing employer discretion (monopsony)
  • Inequality has risen for 40 years and historically only fell via catastrophe, so policy and labor institutions are the peaceful alternative

The argument · threads12 threads · 48 claims
0.66

Unions and collective institutions historically curbed monopsony; their decline has removed the main check on employer wage-setting power.

6 pointscentrality 5/5
  • Clark Kerr argued in 1950 that monopsony, while relevant for laissez-faire labor markets, would become irrelevant because the future labor market would be administered and unionized; he may have been right that unionization curbs monopsony but got the timeframe completely wrong, since unionization later collapsed.

    “he might have got the effect of unionization on monopsony right, but he got the timeframe of unionization totally wrong”

  • Over roughly 57 years, the institutions that took wage-setting power out of firms' hands—chiefly unions—have declined due to a mix of trade hammering union manufacturing jobs, right-to-work laws hastened by globalization, and relentless government hostility to unions since the 1970s, eroding the main institutional check on employer monopsony power.

    “a lot of the forces that have countervailed monopsony have just been in slow decline. So like union... So a lot of the things that took the wage-setting power out of the hands of firms have just over 57 years have just declined”

  • Beyond tax policy, something like the labor movement must come back for political reasons: in the 1950s, when 35% of the American workforce was unionized, it aggregated ordinary people's voices politically, trained working-class people to run for office, and broadly compressed the income distribution—functions a wealth tax alone cannot perform.

    “something like the labor movement has to come back both for political reasons, because it was a way for so many regular people to have their voices aggregated, to be expressed politically”

  • American, British, and Australian unions are organized firm-by-firm (enterprise basis), which makes organizing an uphill battle, whereas in Europe unions are sector-wide and cover nearly everyone; this structural difference makes it hard to imagine US unions returning in their old form, though something that curbs monopsony power needs to return.

    “In Europe, unions are sector-wide and they got everyone's in them or everyone's at least covered by them. And in the US, it's been this firm by firm kind of organizing model”

  • Unions are a better solution than government regulation for curbing monopsony because the government can impose a minimum wage but cannot know the mix of wages and benefits workers want or how much of that mix is consistent with a firm's business model, whereas collective bargaining can customize agreement terms to the particular workplace and workers.

    “the government can't know is exactly A, what's the mix of wages and benefits that workers want; B, how much of that mix of whatever workers want are consistent with the firm's business model”

  • Organized labor's power dropped sharply in the early 1980s, driven not only by the 1981 recession but by a tipping of employer norms in the 1970s: employers who previously refrained from using their legal powers against strikers (out of norms) suddenly began firing strikers and hiring permanent replacements, with the Reagan administration acting as a green light.

    “you have all of a sudden this tip in norms among employers, where before they were kind of playing softball with unions”

0.64

Labor markets exhibit substantial frictions enabling employers to set wages below competitive levels.

5 pointscentrality 5/5
  • Monopsony is the labor-market analog of monopoly: instead of taking the market wage as given, a firm has discretion to set the wage because changing it a little doesn't lose all workers immediately, so employers can choose a lower wage and accept somewhat higher turnover.

    “The monopsony is basically the flip side of that for the labor market, where you think about wages.”

  • When an employer cuts wages by 10%, standard theory predicts 100% of workers leave instantly for better jobs, but the data show roughly 30% leave over a year—proving labor markets have substantial frictions and that employers therefore have discretion to set wages below the fully competitive level (monopsony).

    “when I cut my wage by 10% as an employer, what fraction of my workers disappears in the next, say, one to three years? And standard economic theory would say 100%.”

  • Matched worker-firm data show that 15% to 20% of wage variation is explained by which firm a worker is at rather than who they are, contradicting the frictionless 'law of one price' for labor that says wages should depend only on a worker's human capital and be equalized across firms.

    “something between 15% and 20% of the wage variation that we see is explained by the firm that you're at, not just who you are”

  • Leaked Sam's Club HR slides show that a store needing a wage different from the national wage policy must apply to Bentonville headquarters for a special exemption, demonstrating that firms set wages by national policy rather than simply taking the local competitive market wage—evidence of wage-setting discretion.

    “if a Sam's Club needs to set a wage that's really different from the national wage policy, they need to apply to Bentonville, Arkansas for a special exemption”

  • A national retailer gave raises that depended discontinuously on prior wage (a penny above versus below a threshold produced very different raises), and few workers quit when they got a smaller raise—showing low wage-responsiveness, while the very existence of such kinked, bunched wage rules indicates real room for human decision-making rather than a smooth market-clearing process.

    “if you had, if you were making like one penny over some threshold you were, got a very different raise than if you were making one penny below”

0.59

Wage stagnation despite low unemployment points to monopsony and institutional decline, not automation or trade alone.

2 pointscentrality 5/5
  • Trade with China has had a huge, real effect on American jobs—a major factor in the labor market—and operated substantially by hammering unionized manufacturing jobs, whereas the evidence that automation has leveled American employment is, by contrast, premature.

    “I'm not going to deny the trade with China has had a huge effect on American jobs and that's like a big, that is a big factor. Automation I think it's a little bit premature”

  • Despite record-low unemployment and a hot labor market, wages are barely rising—a pattern that goes against Econ 101 and also against automation and trade stories (which should show up in unemployment too), pointing instead toward a different category of explanation such as monopsony combined with the decline of countervailing institutions like unions.

    “we're currently at record low unemployment, like kind of insanely low unemployment, hot, hot labor market. And yet we're seeing that it's really difficult to pick up increases in wages”

0.47

Minimum wage increases reduce separations and turnover more than employment, offsetting job losses through lower quit rates.

3 pointscentrality 4/5
  • When the minimum wage forces employers to pay a higher wage, they get lower turnover as a byproduct; one paper found that new hiring falls but separations fall by the same amount, so overall employment is roughly unchanged—the offsetting reduction in quits explains why few jobs are destroyed, though it does cost firms profits and is partly passed to consumers.

    “There's a paper that looked at this in the US that found that new hiring falls, but separations fall the exact same amount. And so that's why it's kind of a net loss.”

  • Decades of identification-based studies have roughly settled (about 80-95%) that the minimum wage does not destroy many jobs—if it kills any, the effect is very small, on the order of a 10% increase costing no more than 0.1% of jobs—directly contradicting the Econ 101 prediction that raising labor costs must reduce employment.

    “There's a paper that looked at this in the US that found that new hiring falls, but separations fall the exact same amount. And so that's why it's kind of a net loss.”

  • Roughly 15 years of identification-based studies using lotteries, distance to universities, and similar instruments converge on a return to an additional year of schooling of about 9% to 12% in additional earnings, especially in the high-school-to-college range, after controlling for ability and family background.

    “We roughly get a ball park number of between 9% and 12%”

0.47

The frictionless competitive model does not fit reality but shapes how economists theorize and measure economic activity.

3 pointscentrality 4/5
  • To get any role for central-bank monetary policy in a model, you need sticky prices where not all firms adjust prices instantaneously; New Keynesian models achieve this via tricks like monopolistic competition and staggered price setting, which have themselves become a baseline because it is very hard to model monetary policy without such frictions.

    “If you want a world in which there's any kind of role for monetary policy, for example, from the central bank, you need a world in which prices are sort of sticky.”

  • The frictionless competitive model is not purely theoretical—a surprising amount of 'folk theorizing' based on it gets embedded into how core economic measures like national income are constructed in government statistics, which matters for how we actually measure the world.

    “there's a surprising amount of folk theorizing that goes into even core economic measurement”

  • The perfectly competitive frictionless market is analogous to the ideal gas—powerful and tractable as a starting point—but for the labor market, land market, and credit market it does more violence to reality than is acceptable, and there is not yet a good substitute benchmark; quantitatively accurate models now adopt departures from the frictionless world as baseline.

    “the perfectly competitive frictionless world is analogous to the ideal gas”

0.44

Neoliberalism accepted government intervention for competition and basic income but assumed markets fix problems; its approach is exhausted.

17 pointscentrality 2/5
  • Parts of economics describe big impersonal forces—patterns of global trade, macroeconomic forces—that constrain and shape everyday life much like gravity or atmospheric pressure, and understanding them reveals the tools and constraints for manipulating them (e.g., what a minimum wage or central bank interest rate can do).

    “much in the same way that gravity or ecology or atmospheric pressure is a real thing in our lives and we don't necessarily understand it all, but what physics and the rigorous study of these things do, is let you understand these forces”

  • Rent in economics is technically a payment above and beyond the next best alternative (opportunity cost)—so a professor willing to do the job for one-tenth the salary earns the rest as rents—and rent-seeking can be productive (entrepreneurs filling unmet needs chase rents) or pejorative (raising barriers to entry to grab a larger share of a fixed pie rather than growing it).

    “rent technically in economics is just defined as like a payment for something above and beyond its next best... It's payment that I would guess, and there's a next best alternative”

  • A practical alternative to UBI is expanding the Earned Income Tax Credit so non-workers receive a positive amount rather than zero, while working more yields even more—providing a basic income floor while preserving incentives to work, potentially the best of both worlds.

    “make it so that you don't get zero if you're not working, you're just getting some positive amount, but then if you work more, you get even more via the [EITC]”

  • A deep deficiency of economics is that it always reasons in terms of smooth optimization—shifting a little at the margin between choices—which fails badly for systems with big discontinuities, cliffs, and gaps (like climate tipping points), where marginal thinking is inappropriate and a different decision calculus is needed.

    “we're always thinking in terms of smooth optimization problems... but if you're dealing with these systems that have these big [cliffs]... then that kind of thinking at the margin, that economists are kind of trained to do, doesn't work super well”

  • Economics is ideologically divided in a way most fields are not, partly because there are actual conservatives in economics where other fields lack them; one can predict an academic's campaign contributions reasonably well from the language of their academic articles, since divergent real values exist within the same discipline.

    “it's ideologically riven because there's actual conservatives in economics in a way that there aren't in lots of other fields”

  • Long-horizon discounting calculations, such as effective altruists' estimates of AI risk many generations out, involve multiplying very large numbers by very small ones (or dividing by tiny numbers), so you can get almost any answer you want; the tools for thinking about such trade-offs are ill-suited because the magnitudes are entirely up in the air.

    “you can make small quantities blow up by dividing by like a really, really small number”

  • Economists are much more involved in policy-making than almost any other social science, with some working on issues like climate change, though Naidu does not think the field has more of a responsibility to do so than physicists do.

    “you'll find that economists are much more involved in policy-making than almost any other social science”

  • Recent research by Piketty, Saez, and Zucman developed distributional national accounts, which track GDP by percentile rather than just in aggregate, so one can look at median rather than mean GDP—a statistic that captures how evenly growth is distributed.

    “the idea of distributional national accounts, which is that instead of just keeping track of the aggregate GDP of a country, let's keep track of it by percentile”

  • The CORE open-source economics textbook deliberately presents introductory economics differently from every other (largely identical) textbook by emphasizing that economics is not just supply, demand, and the frictionless model—analogous, per Simon DeDeo, to what the Feynman lectures were to physics teaching—giving students a library of gadgets for thinking about the world.

    “what we're really trying to do is trying to say that economics is not just supply and demand. It's not just the frictionless model.”

  • An economic intuition often overlooked in climate trade-offs is that future generations will likely be richer due to technological change and better equipped to deal with climate change, so for any given level of climate change cross-generational trade-offs might reasonably account for that inequality—unless we destroy the planet, which exposes the limits of marginal thinking.

    “future generations are also going to be richer than us, just because of technological change”

  • Naidu is skeptical of a basic income partly because a lot of the gains would be undone by monopoly pricing (sellers raising prices to capture the extra cash), and because work—having to interact productively with people unlike you—is a real social glue; a UBI that doesn't raise the floor enough to let people opt out of work isn't transformative but just a little extra cash.

    “there's a way in which we'll, A, see a lot of the gains undone by monopoly pricing, and secondly that there's a real sense in which work and the fact of having to go and interact productively with people that are not like you is just like a real great glue”

  • In an economy where the biggest resources are ideas—high-fixed-cost, zero-marginal-cost goods—putting them in the shackles of private ownership through patents and monopoly rights may not give society the best chance of success, suggesting private property should be treated as a more elastic concept for such goods.

    “an economy where the biggest resources are ideas that putting those in the shackles of private ownership and in patents and in monopoly rights, it's kind of not necessarily doing ourselves the best”

  • Because assets have stochastic multiplicative returns—they grow at a roughly random rate—Gibrat's law implies an extremely skewed power-law distribution of wealth; taxing wealth doesn't eliminate this but damps the growth parameter, keeping the distribution skewed but thinning the tail.

    “this idea of Gibran's law for wealth, that basically because assets have these stochastic multiplicative returns so they just kind of grow at a random rate, that that can get you an extremely skewed [power-law] distribution of wealth”

  • Neoliberalism, clearly stated in a 1950 Milton Friedman article, is distinct from libertarianism: it accepts an active government whose function is not to distort the market but to provide antitrust enforcement and some basic income provision, while broadly trusting markets as the baseline mechanism and viewing wide-ranging regulatory intervention as unnecessary because the market would fix problems.

    “a very clear statement of it is in a 1950 article in Milton Friedman, where he actually just literally explains what the doctrine of neoliberalism is, and it's not libertarianism”

  • The only way inequality of this type has historically fallen is through social catastrophes—internal revolution, external war, or similar—as documented in Walter Scheidel's The Great Leveler; the bet of progressive policy is that modern society has enough tools of government and reason to reduce inequality peacefully through policy without going through war.

    “the only way in which we've seen inequality of this type fall in the past is with social catastrophes”

  • The right way to evaluate capitalism is not against the pre-capitalist past (which is not the correct counterfactual) but against other possible future ways of organizing the economy that, given current technology, might produce more human flourishing, less inequality, and more environmental sustainability; pretending the current system is the best possible does us no favors.

    “Looking to the past doesn't generate the right counterfactual; what you want to think about is another possible future.”

  • Neoliberalism is past its sell-by date and the origins of many current social problems can be laid at the feet of policy reforms conducted in its name during the 1970s, 80s, and 90s and even through the Bush and Obama administrations—reflecting a real ideological commonality across those administrations rather than an empty label.

    “when I just think of the origins of so many of the social problems that we've got now, I just feel... Like it just feels like the history is borne out that a lot of those can be laid at the feet of policy reforms that were conducted in the name of neoliberalism”

0.31

Since the late 1990s, economics shifted toward causal identification through natural experiments rather than convenient correlations.

2 pointscentrality 3/5
  • Since the late 1990s, economics shifted from mining datasets for convenient correlations to a causal-identification discipline that seeks natural experiments with comparable treatment and control groups, allowing ideological articles of faith about how markets must work to be empirically tested.

    “the rise of this causal revolution in economics really sort of... Also changed how we use data, so that we really became attentive, that you couldn't just mine a data set and look for whatever correlations that you wanted”

  • Throughout the 20th century, economics was dominated by brilliant mathematical theorists (Arrow, Samuelson), and doing 'good economics' meant building elegant mathematical models; those who weren't top theorists did empirical work, creating a status hierarchy with model-builders at the top.

    “the whole 20th century of economics was dominated by basically brilliant mathematical theorists”

0.31

Inequality is intrinsically harmful because status, power, and representation are zero-sum, damaging democratic society.

2 pointscentrality 3/5
  • Even if everyone is materially better off, inequality is intrinsically harmful because important goods are zero-sum—relative status, political power, cultural representation—so when income inequality correlates with these, some people are made genuinely worse off (very low status), which is bad for a healthy democratic society and itself justifies reducing inequality.

    “there's lots of dimensions of things that are important, that are in fact zero sum... like relative status in society... Because relative, it's got to be zero sum”

  • Nobody can deny that inequality has increased over the last 40 years in the US, UK, and almost everywhere, though the increase is much more muted in continental European countries and larger in the more anglo/neoliberal countries; it has also risen in fast-growing developing countries like India.

    “nobody can deny that inequality has increased over the last 40 years”

0.31

Capitalism's institutional virtue is rewarding useful production to strangers rather than conquest or domination, a historical break.

2 pointscentrality 3/5
  • Capitalism's success in raising material standards of living represents a genuine historical break—Cosma Shalizi's line that 'the singularity already happened and it was the industrial revolution'—because letting people buy and sell in arm's-length transactions is a remarkably effective way to coordinate human wants over long distances.

    “Cosma Shalizi, another SFI professor, has a great line which is like the singularity already happened. And it was the industrial revolution.”

  • Capitalism's distinctive virtue is that the way to get prestige, status, and material resources is by making things other people want to buy—an institutional break from human history, where status came from killing people or conquering territory—so it makes the basis of social reward providing useful things to anonymous strangers, a more win-win arrangement.

    “the way that you get prestige and status and material resources in capitalism is by making things that other people want to buy”

0.29

Information asymmetries about wages give employers bargaining advantage; transparency shifts power toward workers.

1 pointcentrality 3/5
  • The social norm against asking each other about salaries gives employers an informational advantage; a Berkeley study randomized showing some public employees a link to coworkers' pay and found those discovering they earned below peers were more likely to leave, showing information transparency shifts bargaining power toward workers.

    “they actually randomized showing some workers a link to the pay of their co-workers and others not and looked to see the job search and quit behavior”

0.29

Monopsony power and labor-market frictions are not recent; institutionalist economists recognized wage-setting discretion in the 1940s-50s.

1 pointcentrality 3/5
  • The idea that unregulated labor markets look like monopsony is old, not a recent invention: 1940s-50s institutionalist labor economists who set wages on the National War Labor Board recognized firms had a 'band' of wage-setting discretion, and Joan Robinson developed monopsonistic competition in the 1930s through constructive arguments with Marxists.

    “in the 1940s and '50s labor economists... very much had the idea that the labor market when it was not regulated looked like monopsony”

0.17

Markets require boundaries; some restrictions like zoning serve communities and coalitions despite being economically distortionary.

2 pointscentrality 2/5
  • Not all restrictions on markets are bad even though they are distortions and deviations from an efficient benchmark; some, like certain zoning restrictions, may be justified by the communities they support, the political coalitions they enable, or other non-economic values—against the confident neoliberal instinct to simply remove all such distortions.

    “there are some restrictions on markets that are in fact good, even though they are distorting and even though they are deviations from some sort of a more efficient benchmark, because of either the kinds of communities that they support, because of the political coalitions they allow”

  • The 13th Amendment is an example of the law ruling out a voluntary, mutually-improving market transaction: you are not allowed to indenture or sell yourself even if willing, a restriction that was conscious at the time given Mexican peons in debt bondage in the New Mexico territory.

    “my favorite example of a bit of the Constitution that rules out. Some of this is actually the 13th Amendment, which says that I'm literally not allowed to indenture myself”