Zhang Weiying: Pundits are not prophets

Prominent PKU economist warns against the conceit of forecasting—and the damage done when policymakers take it on faith.

Zhang Weiying is among the most prominent economists in China. Beyond shaping China’s critical economic reform trajectory from the 1980s to the 1990s, he co-founded the China Centre for Economic Research (CCER), the predecessor to the current National School of Development (NSD), and served as the Dean of Guanghua School of Management, both at Peking University.

The Boya Chair Professor of Economics delivered an interesting speech on the hubris of expertise on 19 August 29, at the opening ceremony for the EMBA Class of 2025 at NSD.

Zhang later published an article based on his speech in Caixin on September 15, which is also available on the NSD’s official WeChat blog.

Zhang’s opening address for NSD’s EMBA Class of 2024 was also published in The East is Read.

张维迎：为什么专家的预测经常错得离谱？

Zhang Weiying: Why Are Experts’ Predictions So Often Wildly Wrong?

In an uncertain world, people yearn to know the future, and many experts are keen on forecasting it. But have you noticed that the vast majority of expert predictions are unreliable—sometimes wildly wrong?

Failed Expert Predictions Are Everywhere

Here are a few typical examples:

A widely shared story online goes like this: a well-known investment bank with a formidable research team published ten big calls at the start of 2023; by year’s end, a review found that none of the ten had come true.

Demography should be relatively easy to forecast: the field is mature and the data are ample. Yet experts’ predictions have often been wildly off the mark. Before 2015, as the second-child policy was debated, some well-known demographers forecast that China’s total fertility rate (TFR) would rise quickly to around 2.0, even above 2.4. In reality, after the policy change, the TFR kept falling, from 1.7 in 2016 to 1.08 in 2024—far below those forecasts. Birth projections were similarly off by around five million: experts anticipated nearly 15 million births in 2024, but the actual figure was about 9.54 million.

Consider the forecasts made by internationally renowned experts. In 1968, Stanford University biologist Paul Ehrlich published the influential book The Population Bomb. He predicted that, because of rapid population growth, even if every emergency measure were undertaken, hundreds of millions would still starve to death in the 1970s, and that “at this late date” a substantial rise in the world death rate was unavoidable. In reality, the opposite occurred: global death rates declined substantially, and so did birth rates, driving the population growth rate downward. Neither developed nor developing countries saw famine on a large scale. In fact, per capita calorie intake rose markedly across all countries, and grain prices fell sharply in the years following the 1973 oil shock.

Energy is the field experts most love to forecast, and the one where almost all influential forecasts misfire. Humanity began using petroleum in 1859. By 1865, the British economist William Jevons, one of the founders of marginal utility theory, was predicting imminent coal exhaustion and that oil could not replace it. Subsequent developments proved this completely untrue. In 1919, the U.S. Bureau of Mines forecast that oil production would peak within two to five years and then decline. President Wilson even convened experts to consider synthetic fuels. However, the anticipated downturn never arrived. Instead, oil output surged continuously.

In 1947, the U.S. State Department’s Petroleum Division declared that “sufficient oil cannot be found in the United States”; that forecast also proved wrong. Sentiment later swung to optimism. In 1970, President Nixon’s Cabinet Task Force on Oil Import Control reported that the United States would remain essentially self-sufficient in oil for at least another decade and the international price of oil would remain low; yet the 1973 OPEC oil shock caught forecasters flat-footed. In 1977, President Jimmy Carter warned of an impending, comprehensive energy crisis; in fact, oil prices began falling after 1978. In 2005, when prices hit about $40 per barrel, Steve Forbes, the founder of Forbes magazine, predicted an imminent drop, but prices continued to rise, peaking above $140 in 2008. At that peak, many authoritative analysts projected a further rise to $200; instead, prices reversed, sliding to roughly $30 by 2016 and even to about $16.55 in April 2020. Even with the war in Ukraine, today’s oil price is only a little over $60 per barrel.

Turning to macroeconomic forecasts: in his 1938 book Full Recovery or Stagnation?, Harvard professor Alvin Hansen, often called “the American Keynes” and a key architect of the Council of Economic Advisors and the U.S. Social Security system, used detailed data to predict that the world economy would sink into permanent stagnation. At the time, even the most optimistic observers saw little prospect of postwar recovery. In reality, the world economy entered nearly three decades of extraordinary expansion after World War II, an upturn unprecedented in both duration and speed.

In 1980, U.S. inflation topped 10 per cent. The 1981 The Handbook of Forecasting featured fourteen major predictions, all projecting that inflation would keep rising for at least a decade. Some experts even suggested that the uptrend could last until 2030. Yet within two years of publication of the handbook, U.S. inflation had fallen to 3.8 per cent and then remained below 5 per cent through 2020, with only a few brief exceptions.

On forecasts of the international landscape, during the 1950s and 1960s, Western economists generally believed that the Soviet Union’s economic growth far outpaced that of the United States and thus the USSR would overtake the United States to become the world’s largest economy. One case especially worth noting is the American economist and Nobel laureate Paul Samuelson, whose Economics was the best-selling economics textbook for half a century. In editions revised in the 1960s, he predicted that the Soviet Union would surpass the United States roughly between 1984 and 1997. In the 1980 edition, seeing that the earlier forecast was unlikely to be met on schedule, he pushed the overtake date to 2002–2012. Yet only a few years after that edition appeared, the Soviet Union collapsed. In later editions, he removed this forecast without explaining why.

From the late 1980s to the early 1990s, at least thirty-five books worldwide predicted that Japan would overtake the United States to become the world’s largest economy. Some even asserted that the twenty-first century would be a Japanese century and that the United States might become a satellite of Japan. One figure to highlight is Clyde Prestowitz, an economist who served as a trade advisor during the Reagan era. In 1988, he wrote Trading Places: How We Allowed Japan to Take the Lead, a book that did not mention China at all. Twenty years later, he reinvented himself as a leading proponent of China taking over global leadership. Whether that judgment is accurate remains to be seen.

Economists’ predictions about reform policies often miss the mark as well. In 1981, during Prime Minister Margaret Thatcher’s tenure, 364 university professors in the United Kingdom signed a formal statement urging the public to reject her proposed reform programme. Yet events showed that the reforms she implemented helped Britain break free from prolonged economic stagnation and re-emerge on the international stage.

When Javier Milei took office as Argentina’s president at the end of 2023, dozens of top economists around the world co-signed an open letter warning that his reform agenda could harm ordinary Argentines. However, more than a year on, and based on the situation so far, Milei’s measures have produced a “manmade miracle” of economic recovery: for the first time in 123 years, Argentina has eliminated its fiscal deficit, and the annual inflation rate has plunged from 211.4% in 2023 to 43.5% by mid-2025. Especially notable, 1.7 million Argentine children have escaped poverty since he took office. In April this year, the economy grew 8% year on year.

In 1977, 800 experts on international affairs forecast the future global configuration of political systems, focusing in particular on what things would look like 25 years later, in 2002: would the number of socialist regimes increase, decrease, or stay the same? More than 25% predicted the 2002 tally would match that of 1977, and another 45% predicted an increase. In reality, in 1977, there were roughly 15 to 20 socialist states worldwide, but by 2002, only five remained. In short, most experts were off the mark.

Arnold Toynbee, one of the twentieth century’s most eminent historians and author of the twelve-volume A Study of History, made a series of sweeping forecasts. In a 1962 book on the population explosion, he predicted that by the end of the twentieth century, a powerful, unconstrained international body would be created to control the production and distribution of food—“the first genuine executive organ of world government that mankind will create for itself.”

As early as 1952, he had forecast the world of 2002 “will have been unified politically through the concentration of irresistible military power in some single set of hands.” He thought the unifying power would not necessarily be the United States, but it was a strong candidate. In 1966, while musing about the possibility of a “Russo-American consortium”, he added that if the Cold War antagonists could not achieve this, China would. In any event, freedom would certainly be extinguished. In a 1970 interview, he went further, saying that the world would be “held together and kept at peace in the year 2000 by an atrociously tyrannical dictatorship”. Yet in the decades that followed, the number of democracies around the world steadily increased.

I have already listed enough examples to show that experts’ forecasts are often unreliable. In 1985, a young psychologist at the University of California, Berkeley, Philip Tetlock, launched an important study. He recruited 284 experts—political scientists, economists, and journalists—and, under conditions of anonymity, asked them for their views on future political and economic trends. Over nearly twenty years, he posed a wide range of questions and collected 27,450 forecasts. A comprehensive analysis yielded a basic conclusion: experts did not outperform random guesses, beating the “dart-throwing chimpanzee” only by a narrow margin.

Tetlock also found that the biggest differences among experts lay in how they thought. Poor forecasters failed to cope with complexity and uncertainty; they tried to compress problems into a single core theoretical theme and stamp out predictions with a template. The small minority who did better were the opposite: they used no template, drew on multiple sources, and tried to integrate them. They were used to self-criticism, continually questioning whether what they believed to be true was in fact true. When they erred, they admitted it and adjusted. Most of these experts were comfortable viewing the world as complex and uncertain. Hence, their forecasting performance was somewhat better.

In 2012, columnist Dan Gardner published Future Babble, a book that examines the success rates of forecasts in fields such as demography, resources, the economy, geopolitics, and science. His core finding is that experts are seldom right; even when they are, it is usually due to luck, like a stopped clock that still shows the correct time twice a day. More troubling, if a clock always runs five minutes fast, it will never show the correct time. Most of the examples cited earlier in this article are drawn from Gardner’s book.

Why Aren’t Experts Prophets?

Why are the forecasts of these erudite experts so unreliable? The simple answer is that the future is uncertain, while forecasting experts tend to assume it is certain. The future contains countless possibilities and is shaped by many factors. Yet experts know only a portion of those factors, not the whole, and are constrained by cognitive limits.

What I want to emphasise is that the future is not only uncertainty, but also indeterminacy. The usual notion of “uncertainty” gives the impression that there is an objective external world whose future has already been set by history and will gradually reveal itself over time, like a fertilised egg becoming an embryo and then a baby. We feel uncertain only because our knowledge is insufficient; as knowledge accumulates, uncertainty diminishes. That may describe most natural phenomena (such as weather), but it is far removed from human society. In human affairs, the future’s shape is not objectively given, nor is it fully determined by history. The kind of future we will live in depends at least in part on our choices—your choices, others’ choices, the choices we make now, and those we make later. Even if you are clear about and steadfast in your own choices, you cannot know or fully influence the choices of others, least of all the reach of human imagination. What a person chooses is tied to imagination, and imagination is not wholly determined by the accumulation of past knowledge.

Thus, the world as a whole is nonlinear. The birth of a new idea, the occurrence of a chance event, the emergence of an ambitious individual, the invention of a new technology, or a sudden virus (such as the COVID-19 pandemic) can all change the course of history. And by their nature, these things are unpredictable. This is what is meant by “a hair’s breadth of error leads to a thousand miles of difference.”

Hence, Newton said, “I can calculate the movement of the stars, but not the madness of men.” [An apocryphal quote said to be connected with Newton’s losses in the 1720 South Sea Bubble. —Yuxuan’s note] The human world is nothing like the natural world. Forecasting experts, however, often think otherwise: they assume the world humans inhabit resembles the Newtonian physical universe, like celestial motion governed by fixed laws—such that, if only the initial conditions and the equations of motion are known, the future can be predicted. This is called “status quo bias.”

A typical case is Robert Malthus. In his 1798 An Essay on the Principle of Population, he predicted that because population grows faster than the food supply, humanity would remain trapped in poverty, and living standards could never truly rise. Yet, as is known to all, with the rise of the Industrial Revolution, humanity escaped what is now called the Malthusian trap. Why did Malthus make that prediction? He based his conjecture entirely on the historical data of his time and failed to anticipate the impact of technological progress on production and on fertility.

Experts often possess extraordinarily deep knowledge in a specific field, but their breadth is limited. Yet even experts with wide-ranging knowledge have only a very limited grasp of the future. Hence, a pattern follows: forecasts are most likely to be correct when they are least needed, and least likely to be correct when they are needed most.

Here, I want to draw particular attention to technological change. Since Malthus, one of the most important features of human society has been the continual emergence of new inventions and innovations. Energy is a striking example. The amount of “oil reserves” on Earth is not “given”; it depends on exploration and extraction technologies. The reason experts keep getting oil-price forecasts wrong is that they cannot foresee inventions and innovations in exploration and extraction. For instance, in the 2000s, when experts were predicting a global energy crisis, they did not anticipate the breakthrough in shale oil technology. The philosopher of science Karl Popper once asserted that it is impossible to predict inventions. The story below vividly illustrates this point:

Some time in the Old Stone Age you and I are discussing the future and I predict that within the next ten years someone will invent the wheel. “Wheel?” you ask. “What is that?” I then describe the wheel to you, finding words, doubtless with difficulty, for the very first time to say what a rim, spokes, a hub and perhaps an axle will be. Then I pause, aghast. “But no one can be going to invent the wheel, for I have just invented it.” In other words, the invention of the wheel cannot be predicted. For a necessary part of predicting an invention is to say what a wheel is; and to say what a wheel is just is to invent it. It is easy to see how this example can be generalised. Any invention, any discovery, which consists essentially in the elaboration of a radically new concept cannot be predicted, for a necessary part of the prediction is the present elaboration of the very concept whose discovery or invention was to take place only in the future. The notion of the prediction of radical conceptual innovation is itself conceptually incoherent

—Erik J. Larson, The Myth of Artificial Intelligence

[When Larson presents this example in his book The Myth of Artificial Intelligence, he was actually quoting Alasdair MacIntyre’s After Virtue, where MacIntyre recounts Karl Popper suggesting it. I have been unable to locate a primary source in which Karl Popper himself uses the exact “wheel” analogy. —Yuxuan’s note]

Looking back over the past several decades, a number of technologies have emerged that have profoundly affected human society. Yet before they appeared, no one foresaw them. In 1940, no one foresaw the transistor; in 1950, the integrated circuit; in 1960, fibre-optic communications; in 1970, the personal computer; in 1980, the World Wide Web; in 1990, the mass adoption of mobile phones; in 2000, the mobile internet; in 2010, the success of artificial intelligence. Therefore, predicting new inventions is extremely difficult, if not impossible.

Why Do Unreliable Forecasts Persist?

Why do such unreliable predictions keep reappearing? The answer can be understood through the economic framework of supply and demand.

On the demand side, people want forecasts. Humans are born averse to uncertainty because uncertainty means a lack of control over the future. The desire to control the world drives the invention of forecasting—the belief that predicting the future enables control over it.

One point is worth noting: fear of uncertainty actually has driven humanity to embrace many things. Some accept a planned economy because its advocates promise to eliminate the uncertainties produced by capitalist “economic crises.” Planning claims to make everything proceed in orderly, predetermined steps, leaving no one to worry, and thus won favour in many countries.

When does a fugitive finally get a good night’s sleep? The very night of capture. At last, the dust has settled. This illustrates that a definite bad outcome can sometimes bring more inner peace than an indefinite good one. A large body of research suggests that as long as individuals retain even a sliver of control over their surroundings, they feel happier.

This also explains why humans engage in superstition. The eminent anthropologist Bronisław Kasper Malinowski, mentor of Fei Xiaotong at the London School of Economics, wanted to know when people turn to divination. He concluded that in times of extreme uncertainty, people are more likely to seek the aid of the gods.

Of course, modern people tend to trust experts rather than fortune-tellers, since experts possess specialised knowledge and are more rational. Yet in practice, as noted earlier, when it comes to predicting the future, the difference between the two is not that great.

On the supply side, since people want forecasts, demand creates supply. Why do some experts love predicting the future? Because making forecasts demonstrates their value and signals superiority. Since the Scientific Revolution, confidence in the authority of knowledge has grown steadily. Experts who issue bold predictions often do so with great certainty, because only by projecting rock-solid confidence can their claims appear credible. If a forecaster seems hesitant, others are quick to doubt the prediction.

Some experts also relish forecasting because accountability is asymmetric. If a prediction proves correct, fame follows; if it fails and the event never occurs, the forecast is soon forgotten, and the expert typically bears little responsibility. Forecasting is like buying a lottery ticket: hit the jackpot and you’re rich; miss, and the loss is trivial. Even when errors are flagged, the expert can claim the prediction itself wasn’t wrong—only the conditions changed. Humans can rationalise almost any behaviour, and on this score, experts are especially adept. For example, if my GDP forecast doesn’t materialise, I can say, “I was talking about potential growth.”

Thus, although expert forecasts are often unreliable, unreliable forecasts keep coming back. There is always a market for forecasters.

As Philip Tetlock has argued, there are “hedgehog” experts and “fox” experts. Hedgehogs cling to one big idea, convinced it captures the essence of all facts; they believe they see the future clearly and harbour no doubt about their judgment, so they project absolute confidence. Foxes are more cautious: they know the world is complex and unpredictable, and that their own knowledge is limited and insufficient to grasp the whole; therefore, they regard forecasting the future as a rash undertaking, and if they must forecast, they do so with great care.

Tetlock found that the relatively more reliable forecasts come from foxes, not hedgehogs. The hedgehogs’ core error is precisely what F. A. Hayek called the “fatal conceit”: we know very little in reality yet pretend to know a great deal—or believe we do—and thus issue bold predictions. This is what Laozi warned against: “not to know and yet think we do know is a disease.”

Experts must remain humble, for the world is exceedingly complex; no matter how much knowledge anyone possesses, each person is, in truth, ignorant. More knowledge does not equal better forecasts. As Nobel laureate David Gross put it, “The most important product of knowledge is ignorance.” As an Arabic proverb says: Those who claim to foresee the future are lying, even if later proved right.

It must also be stressed that many forecasts are remembered not for their accuracy but for their impact on policy. When mistaken forecasts shape policy, the consequences can be disastrous. In fact, many experts issue forecasts precisely to influence policy. A classic example is population policy. As early as around 2000, some argued for relaxing the one-child policy, but they were marginalised. Mainstream experts insisted that once it was relaxed, the population would quickly rebound to an unmanageable level. Because of unreliable expert forecasts, the policy shift came more than a decade later.

Are all forecasts unreliable? Not necessarily. What social scientists can offer is what Hayek called “pattern prediction”—conditional forecasting, or forecasting about human behaviour. Ludwig von Mises and Hayek long ago predicted that a planned economy would inevitably produce widespread poverty and shortages and is thus unsustainable, and events proved them right. Yet it is not possible to predict which countries will adopt planning, nor the exact day a planned economy will collapse. China has pursued reform for more than forty years, yet its neighbouring North Korea still operates a planned economy; the timing of change there cannot be foreseen.

Three Suggestions

Finally, a few suggestions for the students here:

First, do not worship experts. The more emphatic an expert sounds, the less reliable and less trustworthy their forecasts tend to be. As noted earlier, Professor Philip Tetlock used Google click counts as a proxy for the prominence of 284 experts and found that the more famous the expert, the less dependable their forecasting performance. Harvard professor Clayton Christensen, known for his work on innovation, once said: “The only thing we may know for sure when we read experts’ forecasts about how large emerging markets will become is that they are wrong.”

Second, believe in yourself. Humanity long dreamed of flight, yet it did not become conceivable until the late nineteenth century. In December 1899, Scientific American, the leading popular science magazine in the United States, predicted that heavier-than-air flying machines would, at best, be dangerous, deadly amusements—marvellous toys at best with no commercial or military utility. [I have not been able to find primary or secondary sources for this prediction. —Yuxuan’s note.] The Wright brothers rejected that forecast, believing in themselves, they successfully invented the aeroplane in 1903.

From the aeroplane’s invention until the late 1920s, planes were rarely used to carry passengers and mostly transported mail. Their chief competitor was the airship, which far surpassed aeroplanes in altitude, range, and passenger capacity. In 1928, German entrepreneur Hugo Eckener led the Graf Zeppelin on a round-the-world flight and launched regular transatlantic service between Germany, Brazil, and New York, while aeroplanes could fly only a little over 600 kilometres. On that basis, in 1929, the U.S. Department of Commerce’s Aeronautics Branch issued a report declaring that Eckener’s circumnavigation and transoceanic “triangle” route proved airships superior to aeroplanes for transoceanic flight, and that aeroplanes were suited only to short distances. [I have not been able to locate this report. —Yuxuan’s note.] Among airship supporters, this judgment became an article of faith. But American entrepreneur Juan Trippe did not accept it. Convinced that aeroplanes were the best—and the future—of air transport, he persevered, and in 1936 his Pan American Airways finally began transoceanic flights by aeroplane. Today, convenient air travel exists thanks to entrepreneurs like Trippe, who refused to take experts’ forecasts and instead trusted their own convictions.

Third, what will you learn at the National School of Development (NSD)? NSD brings together outstanding scholars and experts from many fields, but I must underscore this: they are highly accomplished in their domains and can teach valuable knowledge. By absorbing what is taught, you can expand your store of knowledge and strengthen your ability to think about the future. But remember: NSD scholars are not prophets, still less fortune-tellers. You should come to NSD with the right attitude—you are here to learn, not to expect professors to predict the future. Professors teach knowledge, and knowledge concerns the past, not the future, though it may help make sense of what lies ahead. Some professors may inspire your wisdom and sharpen your thinking, enabling a clearer view of the future, but what they impart is not a definitive answer about the future.