In 1956, Robert Solow published a twelve-page mathematical paper that would win him the Nobel Prize in Economics thirty-one years later. The paper built a model of how economies grow: output depends on capital, labor, and a third factor that Solow could measure but not explain. That third factor — what he called total factor productivity, or TFP — turned out to account for roughly half to two-thirds of all economic growth in the United States. Solow had created the most important model of economic growth in the history of the discipline, and at its center was a variable labeled "technology" that the model simply assumed existed and grew at a fixed rate. He had named the residual, estimated its size, and left its explanation to others. It would take another thirty years before Paul Romer built a theory of where TFP came from.
The puzzle Solow was trying to solve had been building for more than a century. Before roughly 1800, per-capita income across the world had been essentially flat for millennia — the vast majority of humanity lived at the edge of subsistence, and generation after generation was born into roughly the same material conditions as the last. Then, beginning in England and spreading to Western Europe and North America, something changed. Per-capita incomes began compounding. Not just growing — compounding. A growth rate of 2% per year doubles income in 35 years and multiplies it tenfold in 115. By the time Solow was writing, the gap between the richest and poorest countries was already enormous, and it was widening. Why had growth started? Why did it continue? And why, a century after industrialization began, were most countries still poor?
These questions are not merely academic. They determine whether hundreds of millions of people remain in poverty, whether the climate crisis can be addressed without catastrophic sacrifice, and whether the next century looks like the extraordinary enrichment of the last two hundred years or a reversion to the historical norm of stagnation.
"Ideas are the instructions that allow us to rearrange matter to create more value. Unlike physical goods, ideas are non-rival: one person using an idea does not prevent another from using it." — Paul Romer, Journal of Political Economy (1990)
| Growth Theory | Key Theorist | Explanation of Growth | Policy Implication |
|---|---|---|---|
| Neoclassical (Solow) | Robert Solow, 1956 | Capital accumulation + exogenous technology | Savings rate, capital investment |
| Endogenous growth | Paul Romer, 1990 | Ideas and innovation created within economy | R&D subsidies, education investment |
| Institutional | Acemoglu, Johnson, Robinson, 2001 | Property rights and rule of law | Governance reform, institution building |
| Human capital | Theodore Schultz, Gary Becker | Skills and health embedded in workers | Education, health systems |
| Big push | Rosenstein-Rodan; Jeffrey Sachs | Coordinated investment crosses threshold | Foreign aid, coordinated public investment |
| Geographic | Jared Diamond, Jeffrey Sachs | Climate, disease burden, natural resources | Infrastructure, disease control |
Key Definitions
Total factor productivity (TFP): The portion of output growth not explained by capital or labor inputs. Represents technology, knowledge, organizational efficiency, and innovation. Also called the "Solow residual."
Endogenous growth theory: Models in which technological progress and innovation are explained within the model itself, rather than assumed to arrive at a fixed exogenous rate. Associated with Paul Romer and Robert Lucas.
Convergence hypothesis: The Solow model's prediction that poor countries should grow faster than rich ones because capital is scarcer and its marginal returns are higher. Confirmed in "conditional" form (countries converge when institutions and policies are similar) but not unconditionally.
Inclusive vs extractive institutions: Daron Acemoglu and James Robinson's distinction between institutions that disperse economic and political power to allow broad participation (inclusive) and those that concentrate power to extract resources for a narrow elite (extractive).
Human capital: The skills, knowledge, and health embodied in people, treated as a form of capital that can be invested in through education and healthcare. Theodore Schultz and Gary Becker developed the modern theory.
Technology diffusion: The process by which innovations spread from the frontier countries where they originate to follower countries that adopt and adapt them. Explains much of the East Asian growth miracle.
The Solow Model and Its Famous Gap
Robert Solow's model begins with a simple production function: output is determined by capital, labor, and technology. Capital exhibits diminishing returns — the hundredth machine added to a factory adds less to output than the first. This implies that capital accumulation alone cannot sustain growth indefinitely. As capital accumulates, its marginal return falls, and eventually new investment exactly replaces depreciated capital, leaving the capital stock constant. This is the "steady state" — a point at which, absent technological change, per-capita income stabilizes.
The model generated a clean prediction: poorer countries, with less capital per worker, should have higher returns to investment and should therefore grow faster. Capital should flow from rich to poor countries. Poor countries should catch up to rich ones. This became known as the convergence hypothesis.
Empirical tests of convergence produced an important nuance. Unconditional convergence — all poor countries growing faster than rich ones — did not hold. Countries like sub-Saharan African nations did not converge to Western income levels despite having far lower capital per worker. But "conditional convergence" — convergence among countries that shared similar institutional quality, educational attainment, and policy frameworks — was well supported. The OECD economies converged substantially during the postwar period. Japan, South Korea, and Taiwan converged rapidly to advanced-economy income levels. Poor countries with poor institutions did not converge, regardless of capital scarcity.
The Solow residual — the share of growth unexplained by capital and labor — was the model's most important and embarrassing result. When economists decomposed U.S. growth since the late 19th century into contributions from capital accumulation, labor force growth, and the residual, the residual dominated. The model that was supposed to explain growth explained only a minority of it. The rest came from something called "technology" that the model treated as exogenous — falling like manna from heaven at a fixed rate. Solow's model was essential. It was not sufficient.
Endogenous Growth: Where Technology Comes From
Paul Romer, then at Stanford, published his solution in 1990: a model in which technological progress is endogenous — created within the economy as a result of decisions by firms and individuals. The key insight drew on a fundamental property of ideas. Physical goods are rival: if you use a piece of equipment, I cannot simultaneously use it. Ideas are non-rival: once the formula for a pharmaceutical or the design of a more efficient engine exists, it can be used by any number of people without being depleted. This property means that ideas can have increasing returns: a large economy or a densely networked one can generate more ideas and benefit more from them than a small or isolated one.
Romer's model explained several features of growth that Solow's could not. It explained why growth rates had not declined as economies accumulated more knowledge — because knowledge, unlike capital, does not depreciate in the same way and generates its own further growth through spillovers. It explained why large economies did not converge automatically to small ones — because returns to scale in idea production mean that larger economies can sustain higher innovation rates. And it provided a policy framework: investments in human capital (education, science, research) that increase the pool of people generating and using ideas have spillovers beyond their private returns.
Robert Lucas, in complementary work, emphasized human capital externalities: an educated population generates productivity benefits for everyone around it, not just for the individual who acquired education. This makes investments in education a public good with returns above what private markets alone will generate.
Romer won the Nobel Prize in Economics in 2018. The committee cited specifically his work on endogenous technological change and its implications for growth policy.
Why Institutions Are the Substrate
Endogenous growth theory explained where TFP came from — knowledge, innovation, human capital. But it did not explain why some countries generated much more of it than others, or why the same technologies that powered growth in East Asia had not done so in sub-Saharan Africa or much of Latin America. That question directed economists toward institutions.
Douglass North, in "Institutions, Institutional Change and Economic Performance" (1990), argued that formal and informal rules — property rights, contract enforcement, political stability, social norms of trust — are the substrate within which economic activity occurs. Without secure property rights, entrepreneurs cannot capture the returns to their investments. Without contract enforcement, the gains from specialization and trade are limited. Without rule of law, political connections substitute for productive efficiency as the path to economic success. North won the Nobel Prize in 1993, largely for this framework.
Acemoglu, Johnson, and Robinson turned North's framework into an empirical research program in a landmark 2001 paper in the American Economic Review. They proposed a clever instrument for institutional quality: settler mortality in former European colonies. Where disease environments made permanent European settlement dangerous (much of tropical Africa and Asia), colonial powers established extractive institutions — designed to transfer resources to the metropole rather than to build productive economies. Where Europeans could settle safely (temperate North America, Australia, New Zealand), they built European-style inclusive institutions. Settler mortality in the 17th and 18th centuries, mediated entirely through its effect on institutional quality, predicts current income levels with remarkable strength. Geography and culture have little residual explanatory power once institutions are controlled for.
The political dimension of this argument was developed fully in "Why Nations Fail" (2012). Acemoglu and Robinson argued that the persistence of poverty is fundamentally a political problem, not a technical one. Extractive institutions persist not because governing elites are ignorant of the growth-promoting alternatives but because they are aware: inclusive institutions would also disperse political and economic power, threatening the position of those who benefit from extraction. The critical junctures of history — moments when institutional change was possible — were navigated differently by societies with different power configurations, producing path-dependent divergences that persist for centuries.
The East Asian Anomaly and Technology Diffusion
If institutions are the substrate of growth and innovation the driver, the East Asian economic miracles require explanation. South Korea in 1960 had per-capita income roughly comparable to Ghana. By 2000 it was a high-income economy with globally competitive manufacturing and technology sectors. Taiwan, Singapore, and Hong Kong followed similar trajectories in compressed timeframes. How?
These cases did not primarily involve frontier innovation — they involved technology diffusion, the adoption and adaptation of technologies already developed in advanced economies. The developmental state model — governments that actively identified strategic industries, directed credit, protected infant industries, and invested heavily in education — played a central role. Dani Rodrik has argued that industrial policy in successful developers worked not by picking winners in the abstract but by facilitating learning and technological upgrading in sectors where follower countries could compete.
High savings rates (40%+ of GDP in some periods), massive investment in secondary and tertiary education, disciplined macroeconomic management, and aggressive export orientation all contributed. The combination of strong state capacity, embedded autonomy (bureaucracies insulated from narrow interest group capture), and market mechanisms distinguished successful Asian developmental states from failed imitators elsewhere.
The broader point is that technology diffusion — not innovation — is the realistic growth path for middle-income countries. This requires openness to foreign direct investment and technology transfer, educational systems that produce workers who can learn and adapt imported technologies, and institutional environments that provide adequate returns to the enterprises undertaking the adaptation.
Human Capital: Education and Health as Investment
Theodore Schultz's insight, developed in the early 1960s and extended by Gary Becker, was that spending on education and health is not consumption — it is investment in productive capacity that yields returns over time. The modern human capital framework treats years of schooling, cognitive ability, and health as inputs to economic production on a par with physical capital.
The evidence on returns to education is extensive. Jacob Mincer's earnings equations, estimated across dozens of countries and time periods, consistently find that each additional year of schooling raises earnings by 8-12% on average, with higher returns in low-income countries where educated labor is scarcer. Eric Hanushek's meta-analyses have shown that the quality of education — not just years of attendance — is the critical variable, with large effects of test-score performance on economic growth rates.
Claudia Goldin and Lawrence Katz's "The Race Between Education and Technology" (2008) traced the relationship between educational attainment and wage inequality in the United States across the 20th century. Their core thesis: the college wage premium fluctuated with the balance of supply and demand for skilled labor. When educational attainment grew rapidly, the premium was modest and growth was broadly shared. When the technology-driven demand for skilled workers accelerated while educational attainment stalled in the 1980s, the premium widened and inequality grew. Economic growth and its distribution are jointly determined by the relationship between technological change and human capital investment.
The Foreign Aid Debate
Few questions in development economics generate more controversy than whether foreign aid promotes growth. The debate has been clarified — if not fully resolved — by the distinction between macro-level aid effectiveness and the effectiveness of specific interventions.
At the macro level, Sachs's "Big Push" argument draws on S-curve models of growth: countries below a threshold of productivity may be trapped in a low equilibrium where marginal returns to investment are insufficient to escape, and large-scale aid can shift them to the high-growth equilibrium. Easterly's response, drawing on decades of failed aid experience, was that no such threshold had been empirically demonstrated and that aid created dependency, distorted local markets, and allowed bad governments to avoid necessary reforms.
The RCT revolution — randomized controlled trials of specific development interventions — has shifted the conversation toward what works rather than whether aid works. Banerjee and Duflo's "Poor Economics" synthesized evidence from field experiments across many countries. Direct cash transfers, it turned out, did not reduce work effort or generate dependency — they increased consumption, facilitated small productive investments, and improved household welfare in measurable ways. Microfinance produced smaller effects than early enthusiasm had suggested. Health interventions, particularly insecticide-treated bed nets, showed very high returns when distributed freely rather than at market prices. Deworming studies produced dramatic initial results that proved more difficult to replicate on re-analysis, generating one of development economics' most contentious methodological disputes.
Banerjee, Duflo, and Kremer received the 2019 Nobel Prize for this body of work. The prize recognized not just specific findings but the methodological contribution: importing the randomized controlled trial from medicine into development economics, creating a standard of causal evidence that had been largely absent from the field.
Climate Change as a Growth Constraint
The relationship between economic growth and climate change operates in both directions. Growth, historically powered by fossil fuels, generates the emissions that drive climate change. And climate change itself is increasingly understood as a constraint on growth, particularly in poor tropical countries.
Solomon Hsiang, Marshall Burke, and Edward Miguel's 2015 paper in Nature estimated the global economic impact of temperature using a panel of 166 countries over fifty years. Their finding: the relationship between temperature and economic output is strongly nonlinear, with productivity peaking at a moderate temperature and falling sharply at higher temperatures. Agricultural output declines with heat stress. Industrial productivity falls as workers become less effective in extreme heat. Cognitive performance decreases. Applied to projected warming scenarios, they estimated that unchecked climate change could reduce global per-capita income by 23% by 2100 relative to a no-warming baseline, with the largest losses concentrated in countries that are already hot and poor — amplifying existing inequality.
This finding connects the growth economics literature to the structural transformation literature: as economies develop, they move from agriculture (highly temperature-sensitive) to manufacturing to services (less sensitive). Countries that have not yet made this transition are more vulnerable. Climate policy, from this perspective, is also development policy: the same countries that have the most to lose from warming have the least capacity to adapt, reinforcing the case for internationally funded adaptation alongside emissions reduction.
What We Know and What Remains Open
Growth economics has come a long way since Solow's 1956 residual. We now understand that technology and knowledge, not capital accumulation, drive long-run growth; that innovation is endogenous and can be promoted by investments in human capital and research; that institutional quality — property rights, rule of law, inclusive political economy — is the substrate without which other growth inputs cannot realize their potential; that technology diffusion can enable rapid growth in middle-income countries with the right policy environment; and that specific development interventions can reliably improve welfare even in the absence of overall institutional transformation.
What remains genuinely contested includes the relative weight of geography versus institutions, the conditions under which specific forms of industrial policy work, the mechanisms by which educational quality translates into growth, and the long-run effects of climate change under different trajectories. These are frontier questions in the Solow sense: real uncertainty, legitimate competing hypotheses, and ongoing empirical work that will gradually narrow the range of defensible answers. The growth economics research program is not complete, but its accumulated knowledge has already informed policies that have contributed to the dramatic reductions in poverty seen over the past half-century — reductions that would have seemed utopian when Solow was writing.
References
- Solow, R. M. (1956). A contribution to the theory of economic growth. Quarterly Journal of Economics, 70(1), 65–94. https://doi.org/10.2307/1884513
- Romer, P. M. (1990). Endogenous technological change. Journal of Political Economy, 98(5, Part 2), S71–S102. https://doi.org/10.1086/261725
- North, D. C. (1990). Institutions, Institutional Change and Economic Performance. Cambridge University Press.
- Acemoglu, D., Johnson, S., & Robinson, J. A. (2001). The colonial origins of comparative development. American Economic Review, 91(5), 1369–1401. https://doi.org/10.1257/aer.91.5.1369
- Acemoglu, D., & Robinson, J. A. (2012). Why Nations Fail: The Origins of Power, Prosperity, and Poverty. Crown.
- Banerjee, A., & Duflo, E. (2011). Poor Economics: A Radical Rethinking of the Way to Fight Global Poverty. PublicAffairs.
- Goldin, C., & Katz, L. F. (2008). The Race Between Education and Technology. Harvard University Press.
- Burke, M., Hsiang, S. M., & Miguel, E. (2015). Global non-linear effect of temperature on economic production. Nature, 527(7577), 235–239. https://doi.org/10.1038/nature15725
- Hanushek, E. A., & Woessmann, L. (2012). Do better schools lead to more growth? Cognitive skills, economic outcomes, and causation. Journal of Economic Growth, 17(4), 267–321. https://doi.org/10.1007/s10887-012-9081-x
See also: What Is Capitalism | Why Inequality Grows | What Made the Industrial Revolution Happen
Frequently Asked Questions
What causes economic growth in the long run?
Long-run economic growth is driven by three broad factors that economists have identified across different theoretical frameworks. The first is capital accumulation: when households save and firms invest, the stock of productive capital — machines, buildings, infrastructure — grows, and with it the economy's capacity to produce. But capital alone is subject to diminishing returns: each additional unit of capital adds less to output than the previous one, so capital accumulation eventually slows and stops. The second factor is labor — more workers producing more output — but labor force growth is limited by demographics and is not a source of sustained per-capita growth. The third and most important factor is technology, or what economists call total factor productivity (TFP): improvements in the efficiency with which capital and labor are combined to produce output. Technological progress, knowledge, organizational improvements, and innovation can in principle grow without limit, making them the ultimate driver of sustained long-run growth. Robert Solow's Nobel-winning model demonstrated that without technological change, growth must eventually reach a steady state. Paul Romer extended the model to explain where technological change comes from — arguing that ideas, unlike physical goods, are non-rival (many people can use the same idea at once) and have increasing returns, making innovation self-reinforcing. More recent work by Daron Acemoglu, Simon Johnson, and James Robinson has emphasized that technology and capital can only do their work within the right institutional environment: secure property rights, enforced contracts, and rule of law are preconditions for growth, not automatic byproducts of it.
What is the Solow model and why does it have a 'residual'?
The Solow growth model, published in 1956, is the foundation of modern growth economics. It describes how an economy's output depends on its capital stock, its labor force, and a residual term representing technology or total factor productivity (TFP). The model's core prediction is that capital exhibits diminishing returns: as you add more capital to a fixed amount of labor, each unit adds less to output than the last. This means that capital accumulation alone cannot sustain long-run growth. The economy reaches a 'steady state' where the new capital created each year just replaces what has depreciated, and further capital accumulation stops. Without technological progress, per-capita income stagnates. The 'Solow residual' — the portion of output growth that cannot be attributed to measurable inputs of capital and labor — turned out to be the dominant explanation of growth. When economists measured how much of American economic growth since the late 19th century was explained by capital and labor accumulation, and how much was left over as TFP, the residual accounted for roughly half to two-thirds of all growth. Solow had built the most important model of growth, and it pointed to a crucial gap: the model's most important variable, TFP, was unexplained. It was not until Paul Romer's endogenous growth theory in the late 1980s that economists had a serious framework for understanding where TFP growth comes from — and the answer (ideas, knowledge spillovers, human capital) had major implications for policy.
Why do some countries grow fast and others stay poor?
The divergence in economic outcomes across countries is one of the central puzzles of development economics, and economists have proposed several complementary explanations. The Solow model's convergence hypothesis predicted that poor countries should grow faster than rich ones because capital is scarce and its marginal returns are higher. Empirically, this prediction is only conditionally true: countries converge in income when they share similar institutions, policies, and educational levels, but unconditional convergence — poor countries automatically catching up — does not hold. Countries with extractive political and economic institutions, weak property rights, or unstable governance do not attract investment or encourage innovation, so capital and technology do not flow to them at the rates needed for catch-up growth. Acemoglu, Johnson, and Robinson's research showed that colonial history systematically affected institutional quality, which in turn explains large shares of current income differences. Countries where European colonizers settled permanently (and therefore built European-style institutions) are much richer today than those where colonizers established extractive systems to transfer resources to the metropole. East Asian success stories — South Korea, Taiwan, Singapore — suggest that rapid technology transfer and diffusion, combined with high savings rates, educational investment, and competent state capacity, can compress into decades the growth that took Europe centuries. But these models have proven difficult to replicate in other contexts, suggesting that geography, initial conditions, and specific institutional configurations all matter in ways that general models struggle to capture.
What is the role of institutions in economic growth?
Douglass North defined institutions as the formal and informal rules that structure human interaction — property rights, contract law, political rules, social norms. His foundational insight, developed in 'Institutions, Institutional Change and Economic Performance' (1990), was that economic growth depends fundamentally on institutions that reduce transaction costs, protect property rights, and make credible commitments possible. Without secure property rights, entrepreneurs have limited incentive to invest because the returns to investment can be expropriated. Without contract enforcement, the gains from specialization and exchange cannot be captured. Without rule of law, political connections determine economic outcomes rather than productive efficiency. Daron Acemoglu and James Robinson extended this work by distinguishing 'inclusive' institutions — which disperse political and economic power and create incentives for broad participation in economic activity — from 'extractive' institutions, which concentrate power to extract resources from the many for the benefit of the few. Their 2012 book 'Why Nations Fail' argued that the persistence of poverty is primarily a political problem: extractive institutions persist because they benefit those with political power, who resist the institutional changes that would promote growth but dilute their advantage. Their cross-country evidence showed that institutional quality, as measured by protection against expropriation, predicts current income levels better than geography, culture, or colonial origin alone. The policy implication is sobering: institutions are deeply path-dependent and resistant to outside imposition, making the engineering of growth through external advice or aid more difficult than purely economic models suggest.
What does the geography vs institutions debate tell us?
Jeffrey Sachs and Daron Acemoglu represent the two poles of a genuine and important debate about the deep causes of income differences across countries. Sachs argues that geographic and ecological factors — particularly the burden of malaria and other tropical diseases, soil quality, and distance from navigable water — directly constrain growth by reducing agricultural productivity, increasing mortality, and reducing the human capital accumulation that education and child survival make possible. On this view, poor countries face real biophysical constraints that require direct intervention — health systems, agricultural technology, infrastructure — before institutional improvements can take hold. Acemoglu and colleagues counter that the correlation between geography and income largely disappears when you control for institutional quality, and that geography affects income primarily through its effects on institutions. Their key insight is that European colonizers built different institutions depending on disease environments: where mortality was high (tropical Africa and Asia), they built extractive institutions without settling permanently; where mortality was low (temperate North America, Australia, New Zealand), they settled and built inclusive institutions. The current income differences between former colonies largely reflect this institutional divergence, not geography directly. Both positions have empirical support, and the current consensus leans toward institutions as the dominant factor while recognizing that geographic constraints can themselves impede institutional development. The debate matters for policy: if geography is a direct constraint, health and infrastructure investments have high payoffs independent of institutional reform; if institutions dominate, the policy agenda is fundamentally political.
Does foreign aid help poor countries grow?
The debate about foreign aid and economic growth is one of the most contested in development economics, and the evidence is genuinely mixed. Jeffrey Sachs, in 'The End of Poverty' (2005), argued that poor countries can be trapped in a low-equilibrium poverty trap — too poor to save enough to invest in the capital needed to grow — and that large-scale aid can provide the 'Big Push' needed to move them onto a self-sustaining growth trajectory. William Easterly, in 'The White Man's Burden' (2006), responded that decades of aid totaling over a trillion dollars had failed to generate sustained growth in Africa and elsewhere, that aid distorts local markets and creates dependency, and that the development establishment systematically ignores its own failures. Both macro-level arguments are difficult to test rigorously because the causal identification problems are severe. The most persuasive recent evidence comes from randomized controlled trials (RCTs) of specific interventions. Abhijit Banerjee, Esther Duflo, and Michael Kremer won the 2019 Nobel Prize in Economics for pioneering this approach. Their work and that of other RCT researchers found that specific, well-targeted interventions can produce durable improvements: direct cash transfers increase consumption and investment without reducing work effort; certain health interventions (insecticide-treated bed nets, for example) have large returns; savings facilitation programs help households accumulate productive assets. The shift from asking 'does aid work?' to asking 'which specific interventions work, in which contexts, with what delivery mechanisms?' has made development economics considerably more useful, even if it has also revealed the limits of grand theories.
What are the most important drivers of growth today?
Contemporary growth economics has converged on a more pluralistic view than earlier theories suggested. Technology and innovation remain the fundamental source of sustained long-run growth, but the mechanisms through which innovation occurs and diffuses have received more attention. For advanced economies at the technological frontier, growth requires continued investment in basic research, education, and the institutional environment for private innovation. For middle-income countries, technology diffusion — adopting and adapting technologies developed at the frontier — can sustain rapid growth, as it did in East Asia. For low-income countries, the institutional prerequisites for growth remain paramount: security, rule of law, and basic public goods provision. Human capital — education, health, and cognitive development — is consistently among the most robust predictors of long-run growth across methodological approaches. Claudia Goldin and Lawrence Katz's work showed that the expansion of education in the 20th century United States drove broad-based growth and reduced inequality, while a slowdown in educational attainment has contributed to rising wage inequality. Climate change has emerged as a significant constraint on growth prospects, particularly for tropical countries: Burke, Hsiang, and Miguel's 2015 Nature paper found strong nonlinear relationships between temperature and economic output, with warming significantly reducing productivity in agriculture and manufacturing. Inclusive institutions, functioning state capacity, and avoiding conflict remain foundational. The research suggests that while there is no single recipe for growth, the failure modes — extractive institutions, poor governance, underprovision of public goods, isolation from global knowledge flows — are relatively well understood.