Productivity & Growth
Why Some Economies Get Richer and Others Do Not
Introduction
In 1820, the average person in Britain was already better off than the average person in China, India, or Mexico - roughly three to four times richer, by the Maddison Project's historical estimates, though nothing like the chasm that came later. Two centuries later, average income per person in high-income countries is about twenty-five times that in low-income countries, and the gap between the very richest and poorest countries exceeds a hundred to one. Most of that difference was not luck or natural resources. It was productivity. A worker in modern Germany produces in a single hour roughly what a worker in many developing countries produces in a full day, often using the same body, the same eight hours, and the same basic willingness to work hard. The difference is everything around that worker: tools, training, supply chains, electricity, transportation, legal systems, and the accumulated knowledge baked into every step of the production process.
For many economists, productivity is the most consequential economic variable, and it is one of the least discussed. It is the slow, quiet engine behind rising wages, longer life expectancy, more leisure, better medicine, and the gap between countries that have escaped poverty and countries that have not. Over a single year, productivity growth of 1% versus 2% looks trivial. Over fifty years, 1% a year raises output per hour by nearly two-thirds, while 2% a year multiplies it by about 2.7. Understanding how productivity actually works, why it slowed down across rich economies after 2005, and why it might or might not accelerate again is essential for understanding almost every long-run economic question on the table today.
What Productivity Is
Productivity is output per unit of input. The most common version is labor productivity: how much economic value a worker produces per hour. If a baker produces 100 loaves of bread in an eight-hour shift, her hourly productivity is 12.5 loaves. If she upgrades to a better oven and produces 200 loaves in the same shift, her productivity has doubled. She did not work harder. The world around her got better at converting her hours into bread.
That distinction matters enormously. People often assume rich countries are rich because their citizens work harder. The opposite is closer to the truth. Workers in many low-income countries put in longer hours under more difficult conditions than workers in wealthy ones, and they produce far less per hour. The gap is not in effort. It is in the machinery, infrastructure, training, organizational know-how, and accumulated institutional knowledge surrounding each hour of work. An hour of work in South Korea produced roughly thirty times as much in 2023 as in 1960, by Penn World Table estimates. Korean workers did not become thirty times more diligent. The system around them became dramatically more capable.
Economists usually distinguish three kinds of productivity. Labor productivity is output per worker-hour. Capital productivity is output per unit of physical capital, like factories or machines. Total factor productivity is what is left over after you account for both labor and capital, the residual that captures everything else: better organization, better technology, better processes, better institutions. For many economists, that residual is the most interesting variable in the field, because it is where most long-run growth comes from, and it is the hardest to measure or explain directly.
Total Factor Productivity: The Mystery Variable
In 1957, economist Robert Solow published a short paper that reshaped the field. He looked at American economic growth between 1909 and 1949 and asked a simple question: how much of the rise in output per hour of work could be explained by giving each hour more capital to work with? The answer was startling. Output per hour roughly doubled over those four decades, and Solow could trace only about an eighth of the increase to more capital. The other seven-eighths came from neither more capital nor more labor. Something else was happening. Solow called it "technical change," a shorthand for any shift in how inputs become output, and the field came to call it the Solow residual, or total factor productivity.
Total factor productivity, often shortened to TFP, is best understood as a measure of how efficiently an economy converts its inputs into outputs. When a country invents a better assembly line, adopts a more effective management practice, or develops a logistics network that delivers goods faster, TFP rises. When regulations get smarter, supply chains get tighter, or workers learn to use new tools more skillfully, TFP rises. The variable bundles together everything that is not labor or capital, which is to say it bundles together most of what we mean when we talk about innovation, knowledge, and institutional quality.
The discomfort in the field is that TFP is partly a "measure of our ignorance," in the economist Moses Abramovitz's 1956 phrase. Economists can calculate it, but they cannot decompose it cleanly into specific causes. Was the gain due to a new technology? A better-trained workforce? A reform that improved competition? More efficient capital markets? All of these blur into a single number. That ambiguity is why arguments about productivity policy get so heated. People with very different worldviews can all point to TFP movements and tell consistent-sounding stories. The data alone usually cannot adjudicate between them.
Why Some Countries Get Rich (and Stay That Way)
Why is South Korea, which had a lower income per person than Ghana in 1960, now about as rich per person as Britain, France, or Italy, once local prices are taken into account? Why did Argentina, which was among the ten richest countries in the world before the First World War, fall into the middle of the pack? Why does Switzerland, small and with few natural resources, keep producing some of the world's highest output per worker decade after decade? These questions have answers, but the answers are contested, and the contest matters because different answers point to different policy choices.
The institutional view is associated most prominently with "Why Nations Fail," the book by Daron Acemoglu and James Robinson. It argues that rich countries are rich primarily because they have developed inclusive institutions: legal systems that protect property rights broadly, governments accountable to citizens, schools that educate widely rather than narrowly, and markets that allow new firms to challenge old ones. Poor countries, on this account, are usually trapped in extractive institutions that protect a narrow elite and discourage the broad-based investment that drives long-run productivity. The division of Korea is the field's favorite illustration: a shared culture and history, a North that started out no poorer than the South, and by the 2020s a South with more than twenty times the North's income per person. Institutions, on this reading, are decisive.
Other careful researchers emphasize different ingredients. Joel Mokyr argues that what separated the rich world after 1750 was a "culture of growth": a shift in attitudes that made it socially respectable to apply systematic knowledge to practical problems and to challenge inherited authority on matters of fact. Jeffrey Sachs and other proponents of geography point out that landlocked countries, tropical disease environments, and limited access to navigable rivers correlate strongly with persistent poverty. Some economic historians stress the lucky timing of the Industrial Revolution and the subsequent advantages of being early. China's rapid growth since 1978, under political institutions that Acemoglu and Robinson classify as extractive, is a leading empirical challenge to their framework. They reply that growth under extractive institutions can last for a time but will run out of steam. Chinese development economists argue that a strong developmental state can substitute for the inclusive institutions the theory requires, at least for several decades of catch-up growth; whether that model sustains growth at the frontier remains open. Most current researchers see all these factors as interacting rather than competing. Institutions matter, but they grow inside cultures and geographies, and they take generations to build or destroy. There is no clean recipe.
The Solow Model in Plain Language
The Solow growth model, named after the same Robert Solow, is the workhorse framework taught in nearly every economics program. Stripped of its math, the model says something simple and counterintuitive. If a country only adds more capital (more factories, more machines, more buildings), its growth will eventually slow down and stop, no matter how high its savings rate. The reason is that each new factory adds less to total output than the one before. The first factory in a small economy is transformative. The hundredth factory in a saturated economy adds modest gains. Capital alone hits diminishing returns.
What keeps long-run growth going, in Solow's framework, is technological progress. New ideas, new methods, new tools allow each unit of capital and each hour of labor to produce more than they could before. Without ongoing technological change, an economy converges to a steady state where investment just covers depreciation and growth in income per worker stops. With ongoing technological change, the steady state itself keeps shifting upward, and incomes can rise indefinitely. This insight made the field take innovation seriously as the central long-run question, even though Solow's original model treated technology as something exogenous, falling like rain from outside the economic system.
Decades later, Paul Romer and others built what is called endogenous growth theory, which tries to explain where technological progress actually comes from. In Romer's framing, ideas are different from physical goods. Two people cannot eat the same loaf of bread, but they can both use the same blueprint or the same algorithm. Ideas are nonrival, which means they can be reused without being depleted. That property, Romer argued, is why economies that invest in research, education, and the production of new knowledge grow faster than those that simply accumulate more of the same factories. Growth is, at the deepest level, the discovery and diffusion of useful ideas. The hard policy question is how to organize an economy so that idea production keeps happening, given that ideas, once created, are easy to copy.
The Productivity Slowdown Since 2005
Something strange happened in rich economies after 2005. Productivity growth, which had averaged 2% or more per year through much of the postwar era, dropped to around 1% in most of them. The slowdown shows up in the United States, the eurozone, the United Kingdom, Japan, and most other developed economies, so it is not a measurement quirk in any single country. In the US nonfarm business sector, growth fell from 2.3% a year over 1947-2004 to 1.5% over 2005-2019; since 2019 it has recovered to about 2%, a pickup that Europe and Japan have not shared. The slowdown has been one of the dominant macroeconomic puzzles of the last two decades, and it is one of the few topics where economists across very different camps roughly agree on the data while disagreeing intensely about the cause.
Robert Gordon, in his book "The Rise and Fall of American Growth," argues that the great productivity-driving inventions are largely behind us. Electricity, the internal combustion engine, indoor plumbing, antibiotics, and modern communications each transformed the economy. In his view, the smartphone and the internet, for all their visibility, have not matched them. On Gordon's reading, the postwar productivity boom was a one-time payoff from a unique cluster of foundational technologies, and a return to slower growth is the historical norm, not an aberration. Tyler Cowen, in "The Great Stagnation," tells a related story: the easy gains from earlier breakthroughs have been picked, and the next wave of innovation is harder to translate into broad productivity.
Other careful economists disagree. Some argue that recent innovations are real but mismeasured, especially in services, health care, and software, which official statistics have always struggled to capture; studies by Chad Syverson and by Federal Reserve economists, however, found little evidence that mismeasurement explains the slowdown. Some argue that ideas are getting harder to find: research effort keeps rising while the productivity growth it buys shrinks. Others point to rising market concentration and falling dynamism (the rate at which new firms challenge incumbents) across many rich economies. Some point to aging: one study of US states found that a growing share of people over 60 slowed growth in output per worker. The most likely answer is that several of these are true at once. The slowdown is not one problem with one cause. It is several adjacent problems whose effects compound, which is also why it is so resistant to single-shot policy fixes.
The AI Productivity Question
Artificial intelligence is the first technology in two decades that economists across very different camps take seriously as a possible productivity accelerator. The optimistic case, made by researchers like Erik Brynjolfsson and his collaborators, draws on early empirical studies of generative AI at work. In controlled studies from 2023, AI tools raised customer-support agents' output per hour by 14% (34% for novices) and cut the time taken on writing and coding tasks by 40% to 56%, with the biggest gains usually going to less experienced workers. A 2025 trial found the opposite for experienced open-source developers working in codebases they knew well: AI tools slowed them down by 19%. If the gains generalize and diffuse across the broad economy, productivity growth could meaningfully accelerate over the next decade.
The skeptical case is also coherent. Daron Acemoglu has argued that the effect of AI on total factor productivity over the next decade will likely be modest, perhaps 0.5% to 0.7% in total rather than per year. His reasons: only a fraction of tasks will be affected in that time, and early studies measured easy-to-learn tasks where AI helps most. Robert Gordon and others have noted that previous technologies people called transformative (the internet, mobile, cloud computing) showed up much less in productivity statistics than enthusiasts predicted. What is sometimes called the "productivity paradox" of information technology, where computers showed up everywhere except in productivity numbers, may apply again.
History suggests a middle path. General-purpose technologies (electricity is the favorite example) often take decades to show up clearly in productivity data because firms have to redesign processes and rebuild factories around the new tool. Electricity lifted factory productivity only in the 1920s, four decades after the first central power station, partly because factories at first simply replaced steam engines with electric motors in the same building layouts. The big gains came when factories were redesigned around the flexibility electricity allowed. AI is plausibly on a similar arc. Real productivity effects, if they materialize, will probably arrive over years rather than quarters and will be uneven across sectors. The evidence so far is suggestive but not conclusive: in May 2026, economists at the San Francisco Fed judged that US productivity data "do not yet provide definitive evidence" that the economy has entered a period of high productivity growth.
The Catch-Up Game
Poor countries can grow faster than rich countries for a long time. The reason is mechanical: catching up is easier than pioneering. A developing country can adopt technologies, management practices, and institutional forms that already exist somewhere else. It does not have to invent the assembly line, the container ship, or the modern factory floor. It just has to copy and adapt. This is sometimes called the "advantage of backwardness," a counterintuitive phrase coined by economic historian Alexander Gerschenkron. Countries that arrive late to industrialization sometimes leapfrog earlier stages entirely, going straight to mobile phones without building landline networks, or straight to electronic payments without building thick branch banking.
The post-1950 history of East Asia is the cleanest example. Japan, then South Korea, Taiwan, Singapore, and finally China each posted multi-decade stretches of 7% to 10% annual growth, far faster than any rich country sustained over comparable periods. Growth at those rates doubles incomes every seven to ten years, which is why these economies transformed so dramatically within a single working life. The mechanism is straightforward: import frontier technology, train workers to use it, channel high savings into investment, integrate into global trade, and let productivity catch up to the global frontier. Once a country reaches the frontier, growth necessarily slows, because there is no longer a stock of better practices waiting to be adopted.
The catch-up game is not automatic. For every South Korea, there are several countries that started at similar income levels and grew much more slowly or stagnated. The "middle-income trap," widely discussed in development economics, describes the pattern of countries that grow rapidly to roughly $10,000 to $15,000 in income per person (in 2005 dollars, adjusted for local prices) and then stall, unable to make the institutional and educational transitions required to keep climbing. Why some countries make the transition and others do not is exactly the institutional and cultural question from earlier sections. Catch-up creates the opportunity. Whether a country uses the opportunity depends on the slow-moving variables that determine long-run productivity growth.
Does Growth Have Limits?
Modern productivity growth has lifted billions out of severe poverty. The share of humanity living in extreme poverty fell from about half in 1950 to about a tenth today, while the world's population more than tripled. Growth has also produced rising carbon emissions, depleted ecosystems, polluted air and water, and accumulated waste streams the planet cannot easily absorb. Whether continued growth is compatible with ecological stability is one of the most consequential and contested questions of our time, and much of the answer depends on what kind of growth.
The optimistic case, sometimes called "decoupling," argues that productivity improvements can let economies grow while using fewer physical resources and emitting less pollution. There is real evidence for partial decoupling. Several rich countries have grown their economies since 1990 while reducing total carbon emissions, mostly through a mix of cleaner electricity, more efficient buildings, and the shift from manufacturing to services. The skeptical case, made by ecological economists, argues that this kind of decoupling is too slow and too partial to address the underlying ecological pressure, and that the share of growth coming from truly dematerialized activity is smaller than the headline numbers suggest.
There is also a question about whether continued GDP growth is even desirable in the rich world. Some researchers argue that beyond a certain threshold, additional income produces diminishing returns in terms of human well-being, and that effort spent chasing higher growth would be better directed at distribution, ecological repair, and shorter working hours. Others reply that large studies find well-being keeps rising with income, leveling off only for the least happy, and that growth is essential for sustaining the public services, debt servicing, and pension commitments that aging societies have already built into their institutional fabric. Both arguments have empirical support. Reasonable people of good faith disagree, and the disagreement plays out in different ways in different countries depending on demographic, fiscal, and ecological starting points.
Practical Applications
Productivity is not just a national-statistics topic. The same logic applies inside firms, teams, and individual careers. A worker whose hour produces more value, through better tools, better processes, better skills, gets paid more over time. A firm that improves the productivity of its workers can pay them more without raising prices. A country whose firms keep doing this gets richer. The mechanics scale across all three levels, even if the policy questions look different at each.
For individual workers, the practical implication is that long-run earnings power depends much more on the productivity of the economy around you, and of your own work, than on how many hours you put in. Industries that have steadily integrated better tools, software, and processes have produced rising real wages over decades, even when the work itself feels unchanged. Wages have also risen in industries where labor productivity has barely moved, often services where the worker's time is the product, because those industries must compete for workers with everyone else. Their services therefore grow steadily more expensive. This is called Baumol's cost disease, after economist William Baumol. It explains why teachers, musicians, and barbers earn more in real terms today than in 1900 even though their underlying work has not changed much: their pay is dragged upward by productivity gains in the broader economy.
For voters and citizens, productivity growth is the slow variable that determines what is fiscally possible decades into the future. Pension systems, health-care commitments, public investment, and debt sustainability all rest implicitly on assumptions about future productivity growth. When growth comes in slower than projected, the math everywhere else gets harder. This is why arguments about housing supply, immigration policy, education quality, research funding, and regulatory burden are not just culture-war proxies. They are arguments about the long-run productivity trajectory, even when the participants do not frame them that way. Productivity is rarely the headline. It is almost always part of the explanation.
Productivity is the quietest of the big economic variables. It does not make the news the way unemployment or inflation does, but it is the slow current beneath both. A 2% productivity growth rate compounded over a forty-year working life more than doubles what each hour produces. A 1% rate, sustained as long, raises it by only about half. The arguments about institutions, technology, ideas, and limits all feed into a single question that takes generations to answer: what kind of economy are we building, and how much more can each of its hours actually produce?
