
William Easterly on Benevolent Autocrats and Growth 05/30/2011
What this covers
William Easterly and Russ Roberts work through the paper "Benevolent Autocrats" and the popular claim that poor autocracies outgrow poor democracies. Easterly says autocracy is confounded with commodity and oil dependence and thin financial systems, which independently produce volatile growth. He also calls the observation that most fast growers are autocrats an inversion error, since it confuses the odds that a success is an autocrat with the odds that an autocrat succeeds. Roberts brings in Bueno de Mesquita's selectorate theory and notes that even Adolf Hitler faced political limits. Easterly adds that the benevolent-autocrat story is circular and non-falsifiable: high growth proves benevolence, and collapse becomes a learning period.
The middle of the conversation covers a lot of ground. Easterly compares the 1980s hype over Japan's government-business cooperation with today's enthusiasm for China, and he suggests that state-owned enterprises may inflate China's measured growth, though not enough to explain all of it. He argues that the timing of civil wars shapes growth outcomes. Democracy, he says, works through lobbying and direct contact with officials, not only elections. He rejects Francis Fukuyama's critique, citing the Hayekian knowledge problem. Roberts pushes back on the presumption that local government is more accountable, using Montgomery County's taxi monopoly.
The last stretch turns to what economists can and cannot know. Roberts notes that the growth literature has identified some 145 significant factors. Easterly separates durable principles, such as gains from trade, from their application, and he reads the Washington Consensus as sound in principle but flawed in how it was imposed. He credits Kenneth Arrow with the insight that whole systems can produce results opposed to intentions.
Easterly argues the 'benevolent autocrat' view rests on cognitive biases and statistical noise: autocracies have far more volatile growth, high-growth episodes cannot be credibly credited to leaders, and growth miracles mean-revert, so autocracy is not a reliable growth strategy.
- Autocracies have much wider growth variance, so their top outcomes are a high-tail draw rather than a repeatable recipe
- Most variance is within leaders' tenures, not between leaders, undercutting leader-based credit
- Growth miracles almost always mean-revert, so China's growth should not be extrapolated
This asset isn't compiled yet
You're seeing its claims, ranked. Compile it to build the argument threads, weight them, and check each claim against your library — the full view.
Growth variance is much higher under autocracy: mature democracies grow near 2% per capita year after year, while autocracies range from about -2% to +6%, so most high-growth outcomes come from the high tail of a volatile distribution.
“the variance of growth outcomes under autocracy is much higher than it is under democracy... You could be at minus 2%, you could be at plus 6%”
Autocracies are confounded with low-income features that independently predict volatile growth, such as commodity and oil dependence (where both prices and output swing sharply), low financial development, and undiversified economies, so autocracy is standing in for these features.
“autocracy is just standing in for low-income economies with lots of lots of other features that also predict very variable outcomes.”
The benevolent autocrat theory relies on the simplistic idea that autocrats want high growth and are unconstrained in reaching it, when in reality autocracy is a strategic game among the leader, revolutionaries, demonstrators, elites, and the military.
“the other big flaw in the benevolent autocrat theory is that we have this incredibly simplistic idea that the autocrats have desire high growth and they have are totally unconstrained... really what's going on under autocracy is there's like strategic interaction”
Bruce Bueno de Mesquita's selectorate theory shows that as the group controlling leadership succession grows, its members do not want a single thing, so what the selectorate wants is an emergent, not a unified, process.
“He talks about the selectorate, the people who have control over who's going to be the leader... They don't want a single thing.”
If a leader's growth fluctuates wildly during their term, it is unjustified to attribute the outcomes to that leader; the 'they start stupid and then get smart' story to explain swings is circular reasoning, since the explanation is adjusted to whatever the outcome was.
“sometimes when sort of all else fails, people will fall back onto purely circular reasoning... that was their learning period”
In the 1980s, Japan's growth was touted as proof that government-business cooperation was wise, the same kind of hype now applied to China, until growth stopped around 1990 and reverted to a mediocre rate that has persisted since.
“All the same sort of hysterical things now being said about China were being said about Japan right up to the very last moment in the period around 1990”
Top-down planners give people precise courses of action despite having no evidence and a poor track record, and the task of economics is to show how little people know about what they imagine they can design.
“the top-down people are happy to tell you what that course of action is even though they have no evidence and miserable track record. And task of economics is to demonstrate to men how little they really know”
People attribute great events to great personalities, a 'leadership attribution bias' rooted in our desire for a hero, which makes us believe in benevolent autocrats.
“we just like to attribute great events to great personalities. You know, and that's called leadership attribution bias.”
Economic principles fail without supporting culture and institutions: a culture of individual responsibility and trustworthiness supports the legal rules and enforceable contracts needed to realize gains from trade.
“if you have a culture that promotes kind of individual responsibility and trustworthiness, that kind of supports the institutional rules”
The fact that most high-growth countries are autocracies is a classic inversion error: people confuse the probability that a success is an autocrat with the probability that an autocrat is a success, and these are not the same.
“We don't want to know what is the probability you're an autocrat if you're a success. We want to know what is the probability that if you are an autocrat, you will be a success.”
Kenneth Arrow's point that the workings of an entire system can produce effects very different from, and even opposed to, intentions is probably the most important contribution economic thought has made to understanding social processes.
“The notion that through the workings of an entire system effects may be very different from and even opposed to intentions is surely the most important intellectual contribution that economic thought has made”
Economists broadly agree on principles that work in the long run, such as gains from trade, specialization, and comparative advantage, and a market system that lets people trade and specialize is what delivers them, even though economists cannot reliably predict next year's growth rate.
“We do know a lot of things that work in long run, even though we can't reliably predict that they will give you a 6% growth rate next year.”
Civil wars drive growth outcomes by timing: a civil war early in the period followed by reconstruction predicts high later growth, while a civil war late in the period that destroys a prosperous economy predicts low growth.
“if you had a civil war early in the period we're considering and then you should have destroyed the economy and then there's rapid reconstruction after the war that predicts high growth.”
The benevolent autocrat hypothesis can be made infinitely flexible: any high growth is credited to benevolence and any negative growth to a learning period, so it reduces to circular reasoning, though it could still be true.
“the hypothesis becomes so infinitely flexible that it just has boiled down to pure circular reasoning. But it could be true.”
Easterly rejects Fukuyama's critique: a centralized government pursuing a society-wide outcome lacks the needed knowledge, while Hayek's dispersed local knowledge is held by people who have incentives to apply it in a decentralized system of liberty.
“if we're talking about a centralized government trying to achieve a whole society-wide outcome, then that kind of knowledge of governments what to do simply does not exist.”
Democracy works through far more than elections: citizens who care intensely about an issue get results through direct contact with officials, lobbying, and other channels, so fixating on elections misses how democracy actually responds.
“anything citizens really do care an awful lot about, they are going to find a way in a democratic system to get heard.”
Autocracy is a strategic game among many players, including the leader, revolutionaries, demonstrators, elites, and the military, so outcomes are often unintended, and unlike markets autocracy has no mechanism to ensure unintended outcomes are good.
“the difference between the unintentional autocracy and the unintentional market is that there's no mechanism in autocracy to make sure the unintentional outcome is a good one.”
The original Washington Consensus principles, such as avoiding huge deficits and high inflation, letting markets set prices, the rule of law, and private property, are innocuous, and it failed in application because an elite in Washington imposed them through IMF and World Bank loans without knowing local economies.
“the reason it failed as um as a Washington Consensus is is what was implied in those words... the principles... it's in the application stage, not in not as principles.”
The specific knowledge economists lack is how to achieve a given society-wide outcome through central direction, while what they do have is a set of principles showing that individual liberty leads to prosperity when institutions protect it.
“what we don't have is a kind of specialized decentralized... the kind of specialized centralized knowledge that would allow a some central authority to achieve a given society-wide outcome. But what we do have is a set of principles”
Government has a legitimate role in providing public goods, and democratic accountability works better at the municipal level, but democratic feedback is a much cruder mechanism than the market's finely calibrated feedback, so the whole economy should not be turned over to government.
“democratic feedback mechanisms... is a much cruder mechanism than the market mechanism. ... a very crude bad badly calibrated feedback system in democracy”
Even dictators face political constraints: Hitler backed down when German housewives protested in Berlin during the war, and Stalin, Hitler, and Mao could not do whatever they wanted because they had to manage political capital within the army and populace.
“Even Stalin and Hitler and Mao couldn't do quote whatever they want... there is political capital that those people have to shepherd even when they're autocrats”
The growth literature has identified over 140 (about 145) factors as significant in different growth models, which shows how hard it is for economists to know what causes growth.
“over 140, I think it was 142 or 145 140 something factors have been found significant in different growth models.”
A large non-market sector and state-owned enterprises that book investment at high values can inflate measured growth, producing 'growth of crap', though this cannot explain all of China's miracle.
“a lot of it's it's not hard to fudge the growth because there's still a large non-market part of the economy... so it may just be the growth of crap.”
Development economics was shaped in the 1930s and 1940s by the rapid Soviet five-year-plan growth and the Great Depression, which overemphasized central planning; a similar overreaction is happening now with China and the Western crisis.
“development started off being much more about Soviet central planning than it was about market economics.”
A foreign advisor should not be sent in to fix Zimbabwe; instead, reform should be run by seasoned local politicians with a stake in policy reform who know the relevant institutions and political constraints.
“don't hire us some uh you know, white guy from NYU who knows nothing about Zimbabwe. Hire you know, some seasoned politicians with something at stake in the game of policy reform”
China's rapid growth is best explained as a change of regime: a shift from a destructive totalitarian government after decades of chaos to one that was less destructive, combined with an overseas Chinese diaspora that exploited demand from democratic market economies for world trade.
“you switch to a government that even if you don't give it any credit for doing anything positive... at the least stopped being horribly destructive”
There is a general presumption that local municipal control works better than national control, but local areas may actually have less effective accountability because few citizens pay attention to local elections or outcomes, as in Montgomery County's taxi monopoly.
“Most people don't pay any attention to their local um elections... there's no glory in local politics. And so, we don't spend a lot of time fascinated by the county council”
The benevolent autocrat hypothesis is non-falsifiable at the single-episode level: one high-growth autocrat has many possible explanations, so the only way to assess it is to look at the whole pattern of evidence.
“that one episode has many many different explanations and it's basically impossible to prove on the basis of one episode what caused it or or did not cause it.”
Of 124 countries, 89 were autocracies and only nine of them produced extraordinary growth, so most autocracies are not successes even though most extraordinary growth stories occurred under autocracy.
“How many of those are autocracies? Uh autocracies are uh 89. So, there's 89 autocracies, only nine of them are successful.”
Just as a baseball hitter's early-season .500 average on 20 at-bats regresses, a country's seven-year growth spurt of 10% should not be extrapolated to 100 years, yet people fail to apply this same small-sample logic to growth.
“if you take a very small number of observations, yes, some countries are going to be able to record a remarkably high growth rate average over 7 years”
Brazil grew about 10% a year during its 1967 to 1975 miracle, and Ivory Coast boomed in the 1960s and 1970s with a Shanghai-like skyline in Abidjan, yet both later collapsed or reverted, showing miracles do not persist.
“do you remember the Brazilian miracle that happened from 1967 to about 1975 when Brazil was growing at 10% per year?”
Growth miracles almost never last: most last only about 5 to 7 years, a few such as China's last around 25 years, and every study shows strong mean reversion pulling above-average countries back toward the world average.
“growth miracles never last... Most of them only last 5 years or 7 years... There's very strong what we'd call mean reversion in growth rates”
Hayek's key insight is that growth is driven by innovation whose content no one knows in advance; because central authorities lack this knowledge, the decentralized market, through many independent efforts, is the best system to discover what works.
“growth is about innovation. It's about uh the emergence of what we should what we want when we see it. We don't even know ahead of time what we're going to want.”
The art of economic reform is applying basic economics ('Econ 101') to local conditions, which requires people who understand those local conditions in depth.
“the art of economic reform is really um figuring out a way in which you apply Econ 101 to local conditions”
The economics field's formative focus on the Soviet five-year plans and the Great Depression, when Soviet growth looked rapid, made development economics initially favor Soviet central planning over market economics, showing how small amounts of data distort the field.
“what everyone remembered was the rapid growth of the Soviets under the five-year plans and the Great Depression in the market economy”
Most of the high growth variance under autocracy lies within the tenures of long-serving leaders rather than between leaders; the average growth of different autocratic leaders does not differ much, which contradicts the idea that good leaders explain the high-growth outcomes.
“it turns out that this really high variance of growth under autocracy is actually not between leaders... most of the variance, the high variance of growth is actually variance within the terms of long-serving autocratic leaders.”
We attribute too much success to the top person in both sports and economics: Mike Shanahan won two Super Bowls with Elway, then lost his effectiveness at Washington, showing that the players, not the coach, often drive results.
“He's got two Super Bowl rings. He also had um a quarterback named Elway who was really good. Without Elway, he hasn't won any.”
The World Bank Growth Commission spent $4 million gathering views from about 200 academics and think-tank figures and reached almost no conclusions about achieving high growth, except that sustained rapid growth (about 7% GDP, roughly 5% per capita) over 25 years or more requires strong political leadership.
“The World Bank spent $4 million to sort of get all the collected wisdom of literally like 200 academics”
Zimbabwe's finance minister ended hyperinflation by stopping money printing and adopting the US dollar as the means of exchange, and this worked better because an insider implemented it than it would have if imposed by an outsider.
“introduce the dollar as a means of exchange and get rid of the domestic currency... that did the trick getting rid of inflation. And I think it worked much better because an insider did it”
Since economists struggle to systematically understand what causes growth, a benevolent autocrat would have difficulty implementing growth intentionally.
“since economists struggle to systematically understand what causes growth, it's hard to believe that a benevolent autocrat um would be able to to implement growth intentionally.”
Autocracy is like a risky gamble, like putting all savings on a single roulette bet: it produces both extreme winners and extreme losers, and sharp negative growth declines (below -0.5% per year) are roughly as likely as big successes.
“you basically can think of autocracy as this very risky bet... you could get very very rich or you could get completely penniless”
China's measured growth may overstate true growth in two ways: migration from countryside to city moves previously unmeasured home production into market statistics, and specialization gains from urbanization have nothing to do with governance.
“we have an enormous migration of people from the countryside to the city where previous economic activity was outside the market and now it's going to be measured by the market”
The logic of the invisible hand extends beyond private goods to society and individual liberty: centralized authorities lack the expertise to raise growth, so liberty should let private, social, and political entrepreneurs act, rewarded for positive social returns and penalized for negative ones.
“let's leave things to the spontaneous efforts of not only you know, private market suppliers and businesses and entrepreneurs, but also social entrepreneurs and political entrepreneurs”
Francis Fukuyama's New York Times review claims Hayek was not Hayekian, arguing that governments can also stumble toward good policy through trial and error, just as markets do.
“Francis Fukuyama reviewed the new edition of the Constitution of Liberty and Fukuyama made the claim that Hayek is not Hayekian.”
Market systems that let people trade and specialize, realizing comparative advantage, are the mechanism that produces prosperity in the long run, and individual liberty institutions allow these gains to flourish.
“if you design institutions to allow individual liberty to flourish, then the reason that individual liberty leads pragmatically to prosperity in the long run are all these principles of gains from trade”
The key unresolved question is not whether autocracies have high variance but whether that variance can be credibly attributed to good and bad leaders; the paper argues it cannot.
“Was it Were we really justified in attributing high high and low growth to the good and bad leaders under autocracy?”
Autocracies tend to have incompetent or dishonest statistical departments, and in China there are strong political incentives to exaggerate measured growth, which some analysts argue is happening.
“there's obvious political incentives to exaggerate the measured growth rates... there is a strong camp arguing that Chinese growth is exaggerated”
Even though the paper gives the benevolent autocrat hypothesis a fair hearing and finds no supporting theory or evidence, the cognitive biases it describes predict that readers will keep believing in benevolent autocrats regardless of the argument.
“the prediction of this paper is that this paper which shows you that benevolent autocrats are false also predicts that you will continue to believe they are true.”
Thomas Friedman's praise of one-party autocracy led by enlightened leaders rests on an assumption that autocrats can follow the advice of intellectual elites like himself, which is a fantasy of unconstrained rulers.
“the reason he likes autocrats so much is that they have no political constraints on following the advice of brilliant intellectuals like Tom Friedman”
The Growth Commission's one strong conclusion was that sustained 7% GDP growth (about 5% per capita) over periods of 25 years or more requires strong political leadership, and in its examples all cases except Japan were autocracies, so 'strong political leadership' is a euphemism for benevolent autocracy.
“growth at such a quick pace meaning 7% GDP growth, which which would be about 5% per capita growth... requires strong political leadership. Which is then I sort of uh then I sort of decoded that that in fact if you look at their actual examples, all of them except Japan were autocrats.”
Friedman's first sentence, that one-party autocracy 'has its drawbacks', understates the record, which includes concentration camps, the gulag, and famines that killed around 10 million people.
“Concentration camps, the gulag, those would be too. then the famines that killed 10 million people, yes.”
Hayek's critique of drawing lessons from a single data point, such as the claim that World War II cured the Great Depression, applies to China: one outlier case is unrepresentative and confounded by other factors.
“Wow, one data point and you're jumping for joy. ... There is a tendency to take one data point that could be very unrepresentative.”
Concluding from one data point that a single success is evidence of a general pattern is a common error, as Hayek criticized when people jumped for joy over one data point on whether war cured the Great Depression.
“Wow, one data point and you're jumping for joy”
John Mueller argued that Mexico's PRI winning every election made it a dictatorship only on election day, since for the other 364 days political forces constrained the party to respond to pressures to stay in power.
“the PRI party wins every year every election, so Mexico's a dictatorship, really... but that's really only on election day.”
Even if autocracy raises the chance of high growth, that should not be taken as a reason to favor it, because one must still compare the number of successful autocracies with the far larger number of democracies.
“even because you need to know how many democracies there are and Right. Very hard to keep those two things straight.”
Economics has spent about 200 years since Adam Smith refining a body of knowledge that is still very useful for identifying the principles of prosperity.
“economists have we have spent 200 years since Adam Smith, you know, kind of honing this body of knowledge which still is a very useful body of knowledge about what are the principles of prosperity.”
Local politics lacks glory, so citizens invest their attention in national figures and the presidency, leaving local officials such as county council members free to act without scrutiny.
“there's no glory. There's not much glory in in local politics. And so, we don't spend a lot of time fascinated by the county council”
People prefer top-down solutions partly because human dependence begins in the womb and lasts through childhood, so roughly 15 to 20% of life is spent dependent on others, making bottom-up salvation hard to imagine.
“I think it starts in the womb... it's roughly 15 to 20% is spent dependent on others.”
Montgomery County, Maryland has a taxi monopoly in which drivers must belong to a single company that gives substantial money to the county council head, and most citizens know this yet do not care, which Easterly calls a little taste of autocracy.
“we have a monopoly on taxi services... you have to be a member of this company and that company strangely enough gives a lot of money to the head of the county council.”
Most of the most dramatic high-growth examples over the past 40 to 50 years come from undemocratic states such as Singapore, China, and South Korea, with Japan being a rare democracy that sustained high growth for a while.
“many of the most dramatic examples of high rates... are from undemocratic states that that are autocratic. Right. Correct. Singapore, China, Korea”
The 'Beijing consensus' proposes that authoritarian high growth is a model to follow, challenging the 'Washington consensus' that property rights, rule of law, and open markets yield growth.
“a Beijing consensus replacing the Washington consensus, in which uh authoritarian high growth would be the kind of you know, model to follow”
A benevolent autocrat is any leader of a non-democratic society who gets credit for high economic growth, such as Lee Kuan Yew for Singapore's miracle or Park Chung-hee for South Korea's growth.
“Benevolent autocrat is any leader in a in a non-democratic society that gets the credit for high economic growth”
Friedman's judgment that China's leaders are enlightened appears to rest on their green energy policies (wind and solar) and their broader economic policies, not on any intimate knowledge of them.
“I assume he's drawing the conclusion that they're enlightened from the fact that they are following green policies”
Easterly began the paper willing to give the benevolent autocrat hypothesis a fair shake and finished convinced there is no theory or evidence supporting benevolent autocrats.
“I'm convinced by the end of the paper that there's no there's no theory or evidence to support benevolent autocrats”