One of the deepest puzzles in economic development define themselves along arbitrary lines on a map. Many people recognize the photograph of a wealthy neighborhood separated by a wall from a crowded informal settlement. This sharp, local contrast reflects an underlying, politically enforced divergence.

This reality is clear at international borders. Consider the city of Nogales, sliced by the fence separating Arizona from Sonora. Both halves share geography, historical roots, and culture. Yet average household income on the U.S. side is three times higher. Residents of Nogales benefit from dependable law and order, access to public services, and accountable democratic institutions. Their southern neighbors lack these conditions. The Korean Peninsula provides a similar natural experiment. Following its split, North and South Korea began with comparable starting points, sharing culture and ecology. Today the South’s high living standards contrast sharply with the North’s poverty, representing a tenfold difference in income.

These dramatic global income disparities, where the gap between the richest and poorest nations can reach 50 to 100 times, demand an explanation rooted in fundamental causes. Three major hypotheses attempt to account for this divergence: the Geography hypothesis, linking prosperity to fixed factors like climate or disease burden; the Culture hypothesis, pointing toward inherited beliefs or ethnic traits; and the Institutions hypothesis, emphasizing the man-made constraints and incentives that shape human interaction.

A key contradiction challenges the Geography and Culture theories: the Reversal of Fortune. If prosperity were tied to fixed, inherent characteristics, regions that were highly prosperous and urbanized in 1500 should still be wealthy. The central valleys of Mexico and Peru, sites of the Aztec and Inca civilizations, should have remained rich. Instead, among former European colonies, the reverse holds. The previously wealthy, densely settled areas are now comparatively poor, while sparsely settled areas like the United States and Canada became the richest. Geography cannot explain the reversal because geography did not change; culture cannot explain it because the same European cultures built different institutions depending on whether local conditions favored settlement or extraction. This inversion suggests that the colonial experience transformed institutions, initiating the profound divergence in prosperity observed in the modern world.

GDP and Urbanization in 1500

Proximate and Fundamental Causes

Neoclassical growth models, following Solow (1956), Cass (1965), and Koopmans (1965), explain differences in income per capita using proximate explanations. These models focus on measurable factors of production: the accumulation of physical capital ($k$), human capital ($h$), and total factor productivity ($A$). While necessary for determining output, these inputs are not the root cause of prosperity. As economic historians Douglass North and Robert Thomas observed: “the factors we have listed (innovation, economies of scale, education, capital accumulation, etc.) are not causes of growth; they are growth.” Factor accumulation and innovation describe the mechanics of income creation, but they cannot explain the deeper, fundamental cause that drives the accumulation process. The first wave of subsequent growth theories (Romer 1986, Lucas 1988) emphasized externalities but remained within the neoclassical tradition, explaining differences primarily through preferences and endowments.

The essential question persists: Why do some countries accumulate $k$, $h$, and $A$ while others do not? Even Aghion and Howitt’s Nobel Prize-winning model of growth through creative destruction (1992, 2025 Nobel) represents an advance over earlier approaches by formalizing how innovation drives sustained growth. But their framework, along with other endogenous growth models, treats the institutional environment as given. Their model explains why creative destruction generates prosperity, not why some societies allow it while others block it. Why do some countries invest heavily in R&D while others do not? Aghion and Howitt attribute this to parameters like research productivity or the profitability of innovation. But these parameters themselves reflect institutional choices about patent protection, competition policy, and whether incumbents can block entrants. As the Nobel Committee noted, creative destruction “creates conflicts that must be managed constructively; otherwise innovation will be blocked.”

The empirical failings of neoclassical theory further expose this deficit. Neoclassical models predict convergence, suggesting poorer nations should catch up, yet reality shows persistent divergence, with the income gap between the richest and poorest nations growing. The Lucas Paradox demonstrates this failure: capital does not flow readily from capital-rich to capital-poor nations where returns should theoretically be highest. High expropriation risk and insecure property rights in poor nations override high theoretical returns, providing little incentive for investment. No amount of capital accumulation can occur when the rules of the game make investment unprofitable.

The institutional hypothesis argues that variations in economic institutions are the major source of cross-country differences in economic growth and prosperity. Institutions, the “rules of the game in a society,” determine the constraints and structure incentives for economic actors. They determine whether innovators capture returns, whether incumbents block new technologies, and whether education systems produce researchers. This hypothesis establishes a hierarchy of causality: Institutions (fundamental cause) → Capital/TFP/Human Capital (proximate causes) → Income.

The Institutional Hypothesis

The institutional hypothesis distinguishes between two fundamental systems that determine a nation’s prosperity or poverty: inclusive and extractive economic institutions.

Inclusive economic institutions provide the essential foundations for sustained prosperity and economic growth. They secure private property rights for a broad cross-section of society, ensuring individuals capture the returns from their investments in capital and new technology, thereby motivating factor accumulation. They foster a level playing field and open markets that enable the entry of new firms and technologies, facilitating creative destruction, where efficient innovations replace obsolete ones and drive productivity growth. This framework relies on a centralized state to maintain law and order, protect these rights, and provide crucial public goods like education and infrastructure. Britain after the Glorious Revolution (1688) exemplifies this arrangement, where patent protections encouraged the innovations that sparked the Industrial Revolution. South Korea’s shift toward inclusive institutions after democratization enabled its rapid transformation.

Extractive economic institutions, by contrast, allow a narrow elite to extract incomes and wealth from the rest of the population. They impose insecure property rights on the majority, destroying the incentive to invest or innovate. High barriers to entry protect incumbent monopolies. Innovation is actively suppressed because creative destruction threatens the existing economic privileges and political power of the elite. Colonial Latin America under the encomienda system extracted labor and resources for Spanish elites while blocking broad-based development.

These institutions endure not from ignorance but from conscious choice by the politically powerful. Elites prefer maintaining their extractive system because they value controlling a large share of a small, stagnant pie over accepting a more distributed share of a larger, growing pie. Those who hold political power cannot credibly promise not to use that power to expropriate assets in the future. Institutional changes that increase total wealth but erode elite control are blocked. The powerful successfully impede socially beneficial institutional changes that threaten their political and economic status.

Establishing Causality

Acemoglu, Johnson, and Robinson’s 2001 paper, “The Colonial Origins of Comparative Development,” set out to establish that institutions cause prosperity. But correlation between good institutions and high incomes proves nothing on its own. Wealthy societies can afford better courts, more professional bureaucracies, stronger property protections, so the causation might run in reverse. Or some third factor might drive both: Montesquieu believed tropical climates bred despotism and laziness simultaneously, which would generate a strong correlation between bad institutions and low income even if institutions had no causal effect whatsoever.

Reversal of Fortune

The challenge is finding something that affected institutions but could not possibly have been caused by current prosperity or by anything else that independently affects income today. Such instruments are rare. You need a variable that mattered historically, that left institutional traces, but that has no plausible direct connection to the modern economy. AJR’s found that European settler mortality rates in the colonial era fit this description almost uniquely well.

Their econometric framework is two-stage least squares, built around a structural equation:

$$\log y_i = \mu + \alpha R_i + X_i’\gamma + \varepsilon_i$$

Here $y_i$ is income per capita, $R_i$ measures institutional quality (protection against expropriation), and $X_i$ includes controls like latitude. The coefficient $\alpha$ captures the causal effect of institutions on income, but estimating it directly by OLS yields biased results because $R_i$ correlates with the error term $\varepsilon_i$. The first stage addresses this by modeling institutional quality as a function of the instrument:

$$R_i = \zeta + \beta \log M_i + X_i’\delta + \nu_i$$

Here $M_i$ is settler mortality. The fitted values $\hat{R}_i$ isolate variation in institutions driven only by colonial mortality, purged of the endogenous components that bias OLS, and the second stage regresses income on these fitted values to recover $\alpha$.

This strategy works only if two conditions hold. The instrument must actually predict institutional quality, the relevance condition, since weak instruments bias 2SLS estimates toward OLS while making inference unreliable. And the instrument must satisfy exogeneity: settler mortality can affect income only through institutions. If mortality influenced income through any other channel (human capital, culture, current disease burdens) the instrument would be invalid and $\alpha$ would remain unidentified. Relevance can be tested. Exogeneity must be defended on substantive grounds.

The historical logic makes settler mortality a compelling instrument. Where malaria and yellow fever killed Europeans at extraordinary rates, colonizers could not realistically plan to stay. They established outposts for extraction rather than settlement, building institutions designed to move resources out rather than protect broad property rights, because they themselves would not be around to benefit from such protections. Where mortality permitted mass European migration, settlers arrived with families and intended to build permanent communities. They imported the institutional frameworks they knew from home: secure property, rule of law, constraints on executive power. They would live under these rules themselves. Self-interest drove institutional choice in both cases, but the time horizons differed completely.

These arrangements persisted because they generated self-reinforcing coalitions. Extractive institutions concentrated power among elites who blocked reform. Inclusive institutions created broad constituencies who defended them.

Settler Mortality and Institutions

The first stage confirms the instrument’s power: settler mortality alone explains 26 percent of the variation in institutional quality today, well above conventional thresholds. Higher colonial mortality means worse institutions centuries later, and the relationship is tight.

Defending the exclusion restriction requires showing that historical mortality has no direct effect on current income. This is easier than it first appears. The diseases that killed European soldiers on the Gold Coast in 1840 pose no threat to Ghanaian workers today. Modern malaria incidence bears little relationship to historical settler mortality: populations have shifted, medical interventions have transformed disease burdens, and the death rates that shaped colonial policy have no persistent biological channel to contemporary productivity. When AJR control for current disease environments, geography, climate, religion, legal origin, and European descendants, the results prove robust. The instrument captures a historical selection process, where colonizers chose to settle versus extract, that left institutional traces without epidemiological ones.

The two-stage least squares estimates deliver a baseline coefficient of $\hat{\alpha} = 0.95$ with a standard error of 0.16. Comparing the 25th and 75th percentiles of institutional quality among former colonies, this implies a 450 percent income difference, against an actual difference of 540 percent. Institutions explain roughly three-quarters of cross-country income variation.

That the IV coefficient exceeds OLS is itself informative. If reverse causality dominated, instrumentation should shrink the coefficient, but instead it grows. This suggests measurement error in institutional variables creates more attenuation bias than reverse causality creates upward bias. Adding latitude changes nothing about the institutional effect, yet latitude itself becomes insignificant. Distance from the equator, long thought to drive prosperity directly, loses all explanatory power once institutions are properly instrumented.

The results prove robust. Excluding the Neo-Europes actually increases the coefficient to 1.29. Continent dummies leave the institutional effect unchanged while the African dummy is insignificant. Whatever makes African countries poorer is institutional inheritance, not continental fixed effects. Restricting to the highest-quality pre-1840 mortality data strengthens the results further. Two limitations deserve acknowledgment: the identification works only for former colonies, so generalizing requires corroborating evidence from natural experiments like the Korean split, and within Africa the first stage weakens enough that the estimates lose significance.

Critics have targeted both data quality and the exclusion restriction. Some argued the mortality estimates combine sources of varying reliability, but restricting to the best data only strengthens the results. Others questioned whether settlers brought human capital or culture that persists independently; controlling for European population share shows no attenuation. The overidentification tests support the framework: European settlements in 1900 as an alternative instrument produces nearly identical estimates, and settler mortality shows no direct effect on income once institutions are instrumented. The identification holds.

But China…?

China seems to be a fitting counterexample. Spectacular growth for four decades under single-party rule seems to contradict everything the institutional hypothesis predicts. If inclusive institutions are necessary for sustained prosperity, how did an authoritarian regime lift 800 million people out of poverty?

The answer requires distinguishing political institutions from economic institutions, and recognizing that China transformed the latter while preserving the former. Mao’s China had extractive institutions on both dimensions: the state controlled production, collectivized agriculture, and allocated resources from Beijing. The results were catastrophic. Deng Xiaoping’s reforms after 1978 changed the economic rules dramatically, and the scale of this transformation is difficult to overstate.

Start with property rights. Under Mao, private ownership was ideologically forbidden. The reforms introduced the household responsibility system, allowing farmers to keep surplus production after meeting quotas. This simple change, letting people benefit from their own effort, unleashed agricultural productivity gains that fed the early growth surge. Township and village enterprises emerged to fill gaps the state sector ignored. The private sector expanded into spaces the state had monopolized. By the 2000s, private firms produced most of China’s output and employed most of its workforce. Property rights remained imperfect, but they were incomparably stronger than before.

The urbanization story is even more striking. In 1978, over 80 percent of Chinese lived in rural areas, most engaged in low-productivity agriculture. Over four decades, 500 million people moved from villages to cities, arguably one of the largest migrations in human history. This was inclusion on a massive scale: hundreds of millions gaining access to labor markets, factory wages, and economic opportunities their parents could never have imagined. The productivity arithmetic is straightforward. Moving a farmer earning subsistence wages into a factory job paying several times more generates enormous gains through structural transformation alone. Within manufacturing, Hsieh and Klenow (2009) show that reducing misallocation across firms, letting productive plants expand while unproductive ones contract, added roughly 2 percent per year to Chinese TFP from 1998 to 2005. China was growing not by pushing the technological frontier but by using existing resources more sensibly.

Competition reinforced these gains. The Special Economic Zones established in the 1980s were expansions of economic access: entrepreneurs who would have been imprisoned a decade earlier could now start firms, hire workers, and keep profits. Shenzhen grew from a fishing village to a manufacturing metropolis because millions of people suddenly had permission to participate. Export orientation imposed further discipline. Firms competing for foreign contracts faced selection based on performance rather than connections. What the SEZs and export sector demonstrated was the same principle the theory predicts: let people in, let them compete, and growth follows. So China grew not because of its extractive political institutions but despite them. The growth followed institutional reforms that allowed economic participation and market competition, precisely what the institutional hypothesis would expect.

The question is whether this can continue. Sustained growth eventually requires creative destruction: new firms displacing old ones, new technologies rendering existing investments obsolete, new elites challenging incumbents. Expanding access is compatible with authoritarian control. Tolerating threats to the powerful is not.

China’s current difficulties illustrate the asymmetry. Many imbalances have been documented for years, acknowledged at the highest levels, and yet remain unresolved. This is not a story of leaders too ignorant to see or too corrupt to act. It is a story of systems that lack reliable means to override the accumulation of vested interests, to compensate losers from reform, to force coordination when every actor has reason to wait. Inclusive political institutions do not solve this problem. They create pressure for inclusive change, sometimes sufficient, sometimes not, that extractive systems lack entirely.

The institutional hypothesis has built a strong empirical case. The econometrics extend beyond settler mortality. AJR’s 2005 study of Atlantic trade shows that where pre-1500 institutions already constrained monarchs, commerce strengthened merchant coalitions who then extracted further institutional reforms. This virtuous cycle explains why England and the Netherlands pulled ahead of Spain and Portugal despite similar Atlantic access. The identification strategies differ, the samples differ, but the results converge.

The framework accounts for everyday observations that neoclassical models cannot. Poor countries do not catch up as Solow predicts; capital does not flow where returns are theoretically highest. Once institutional quality enters the regression, these puzzles dissolve. The proximate factors neoclassical models emphasize, capital, human capital, productivity, are themselves consequences of institutional arrangements: answers to how growth happens, not why it happens in some places and not others.

Yet explaining why nations succeed or fail is easier than explaining how failing nations might change. The historical record shows transitions happen. Britain after 1688, South Korea after democratization. But they appear contingent, driven by crises and coalitions rather than predictable forces. Critical junctures where elites fracture, moments when excluded groups gain coordination capacity, shocks that make reform less costly than resistance: these conditions remain easier to identify retrospectively than to bring about. Neoclassical models explain the mechanics of prosperity once the right institutions exist; they cannot explain how nations escape the political equilibria that block those institutions from forming. That is a question for history, not optimization. The wealth of nations depends on the historically contingent rules of the game. How to change them remains the harder problem.

Picture: Caravaggio, The Cardsharps (1594)