China Derangement Syndrome

Facing reality in an age of disillusionment

Note: This is the updated/revised version of the original post. An abridged version has been published on Aporia Magazine.

Among developed nations, few topics unite public opinion quite like hostility toward China. In the United States, dislike of China cuts cleanly across demographic groups—political affiliations, age brackets and levels of education alike. Even Asian Americans, including those of Chinese descent, tend to view the country unfavorably. It is one of the few bipartisan reflexes left in American life.

This hostility takes many forms. In the media, China is repeatedly condemned for its undemocratic government and vilified whenever it takes any action on the world stage that could be construed as assertive. On social media, it’s filtered through memes about collapsing buildings or doctored videos of empty cities. And in online intellectual circles—especially on the dissident right—there’s a recurring impulse to explain Chinese success or failure in moral terms: if China stumbles, it’s proof that the people are corrupt; if it advances, it must be cheating. The underlying message is consistent: the Chinese cannot be trusted.

One of the most commonly cited pieces of evidence for this supposed national character defect comes from an experiment that made headlines in 2019. The study, led by behavioral economist Alain Cohn, was published in Science under the title ‘Civic honesty around the globe’.

The authors conducted a large experiment in which they dropped over 17,000 wallets in 355 different cities across 40 countries. Some of the wallets contained cash, while others didn’t. Each wallet contained business cards with a name and email address using realistic but fictitious names. Every wallet was assigned its own unique email address. That way, if someone tried to return a wallet, the researchers could identify exactly which one it was (and whether it contained money or not). Emails were tracked for 100 days after the wallets were dropped. In almost every country, wallets with money were more likely to be returned than wallets without money (which runs against the self-interest axiom of modern economics). Even more striking, though, were the vast differences in civic honesty between countries—as shown below.

One country stands out more than any other: China. Despite having a relatively high average IQ (100.2 according to the latest dataset) and a decent standard of living, it performed abysmally. The share of wallets reported as found was among the lowest in the world. To many readers—especially those already inclined to believe in Chinese moral failings—the conclusion seemed obvious: the Chinese were simply dishonest as a people. Among the online right, where skepticism of social science is usually high, the study circulated as confirmation that Chinese civic life was built on deceit.

The Chinese themselves were, understandably, less than pleased about these results. However, China’s abysmal performance in this experiment was almost certainly due to methodological bias, rather than any genuine defect of national character.

To begin with, one of the authors of the study, David Tannenbaum, admitted that they had to exclude Japan (one of the dissident right’s favorite countries) because it has very different cultural norms for dealing with lost property:

We originally planned to include Japan but after some initial pilot testing we realized that the country was unsuitable for methodological reasons. Japan has a lot of small “police booths” where people can return lost objects. During our pilot tests, we found that Japanese citizens would not contact the owner but instead drop them off at a nearby police booth. This feature made it virtually impossible for us to assign individual wallets to particular drop-off locations.

So the authors deemed Japan to be culturally distinct enough that their methodology might be biased against Japan, but for some reason decided it was okay to include China, a cultural and ethnic cousin. One might speculate what exactly the rationale was for this.

Some years after this lost wallet study was published, a group of Chinese researchers, Qian Yang et al. (2023), conducted a replication of their results in China across 10 Chinese cities with 496 wallets, with and without money. They also added new elements such as undercover observers (who watched what recipients did after receiving a wallet), follow-up surveys of recipients, and a national online survey of public attitudes. Additionally, they employed three metrics of civic honesty: email response rates (the original method utilized by Cohn et al.), wallet recovery rates (i.e., was the wallet actually kept safe and returned?), and complete wallet contents recovery (i.e., was everything, including money, returned?). They were able to replicate the low email response rate of Cohn et al.—22.2% for wallets without money and 32.7% for wallets with money. However, with wallet recovery rates, it was 77% for wallets without money and 78.6% for wallets with money. With complete wallet contents recovery rate, it was 75% for wallets without money and 65.7% for wallets with money.

The discrepancy between email response rates and the latter two measures meant that many wallets were safeguarded until the owner could come collect them, even if no one emailed. The authors speculated that civic honesty was interpreted differently in more collectivistic societies, supporting this notion with the finding from their survey that a significantly greater percentage of respondents endorsed safekeeping as part of civic honesty compared to emailing/contacting the owner—84.59% vs. 61.52%.

The results of this replication study prompted a back and forth between the original authors and the Chinese researchers. I will try to summarize the arguments covered below, starting with the former in Tannenbaum et al. (2023):

• Yang et al. claimed that in China, collectivism predicts wallet “safekeeping” but not emailing. However, Tannenbaum et al. argue that this is a statistical artifact caused from including both city fixed effects and city-level collectivism scores in the same regression, which created perfect multicollinearity. When corrected for by removing city fixed effects, collectivism no longer predicted safekeeping. They also ran 4,400 alternative model specifications and never found collectivism to be a significant predictor.

• Yang et al. argued that safekeeping aligns better with honesty norms than emailing in China. But Tannenbum et al. claim that Yang et al.’s own nationally representative survey demonstrated the opposite. They recoded the survey and found that more than twice as many respondents endorsed emailing as a moral norm compared to safekeeping.

• Yang et al.’s “total recovery” measure was biased, since wallets without money can’t “lose” money while wallets with money can. Using simulations, they demonstrate that even if there was no true difference between wallets without money and wallets with money, Yang et al’s coding rules was basically guaranteed to produce a negative effect.¹

A portion of the Chinese researchers then wrote a reply in Zhang et al. (2023) and made the following points:

• They clarified that they used provincial-level collectivism, not city-level, which reduces the risk of perfect multicollinearity. They also stress that since their study was a relatively small-sample replication, limited statistical power makes “nonsignificant” results less decisive.

• The “twice as many” result in the recoded survey is due to Tannenbaum et al’s narrower way of coding, as they defined “emailing” as endorsing both obligations: not stealing from the wallet and contacting the owner. “Safekeeping”, in their view, only required the first. Additionally, Yang et al. pointed out how in 81% of cases, multiple employees shared responsibility for the lost wallet, suggesting honesty in China is often collective rather than individual.

• Tannenbaum et al.’s coding assumed independence across wallet conditions, which generated logically impossible scenarios. For example, their simulations sometimes included cases where a no-money wallet “lost money”, or where a whole wallet was missing but some items were intact. A whopping 25.9% of the simulated data points were theoretically impossible.

Despite all this, the crucial issue of the choice of communication method remained. Was the original “email only” method unfairly stacked against China? A study published this year, Hung et al. (2025), also contradicts the results of Cohn et al. As it turns out, emails are not commonly used in China, so regardless of the individualism-collectivism debate, one would naturally expect a lack of common usage to drive low response rates. In their field experiment, Hung et al. replaced the email addresses used in Cohn et al.’s global study with WeChat QR codes. Each wallet carried a unique QR code sticker that, when scanned, opened a WeChat mini-program. From this page, finders could either 1) make a phone call to the wallet owner with one tap, or 2) send a text message through WeChat. Reporting was measured by whether participants scanned the QR code, initiated a phone call, or sent a message through the mini-program. The result? The reporting rate in the study was 59.3%.² Taking the results of this study and the one by Yang et al. together, it would seem that the civic honesty in China is actually quite high, perhaps even on par with European nations.³

Despite being an ancient civilization, China is not very well understood in the West. Even genuine efforts to research Chinese society often fall short, and this is in large part due to shortcomings with the methods employed. The lost wallet study demonstrates this with regard to civic honesty, but what about self-reported trust? Consider the “generalised trust” item from the World Values Survey. Here are the results by country as shown by Our World in Data:

On this measure, China appears to be one of the world’s most trusting societies—well ahead of even most Western nations. The result seems almost too good to be true. Are they?

In 2011, a team of sociologists led by Jan Delhey noticed that the item’s wording might not mean the same thing everywhere. In individualistic cultures, “most people” usually refers to strangers. In collectivist cultures, it tends to refer to one’s in-group—family members and close friends. Delhey et al. (2011) re-analyzed data from the fifth wave of the survey to correct for this difference in meaning (do however, also see van Hoorn, 2014 for a critique of the study, and Delhey et al., 2014 for their reply).

They utilized the items measuring in-group and out-group trust, as well as the generalized trust item. By comparing the answers, they were able to estimate out how wide the “trust radius” is in each country. Their results suggest that very low trust is actually the norm throughout the world. In fact, Sweden is the only country on this list to have an adjusted trust score above 50%. After adjusting for the “trust radius” of each country, China’s ranking plunged, though it remained slightly higher than Taiwan and South Korea.

However, the authors caution that China’s adjusted results might still be an overestimate because there were high rates of non-response or “don’t knows” on out-group trust items. But this raises another question: what kind of distrust are we really measuring here? The “out-group trust” scale doesn’t just cover strangers you meet on the street. It also includes foreigners and religious outsiders, and Based on the author’s framing, it appears that China’s non-response is disproportionately concentrated on the two “xenophobia-sensitive” items, since they mentioned that respondents from China and Ukraine “had particular problems with questions about people of other religions and nationalities” (p. 791).

But the problem goes even deeper. In their online supplement, Delhey et al. explicitly note that the “people you meet for the first time” item—the one meant to capture trust in strangers—did not behave consistently across countries (it was not measurement invariant). As a consequence, they had to relax its factor loading for 18 countries to get acceptable model fit. Hence in those countries, the out-group trust factor was driven more by the other two indicators—trust in people of another religion and of another nationality. This tilts the construct away from measuring civic openness to strangers and toward measuring tolerance of outsiders (i.e., xenophobia). Since the authors never reported which countries were affected, we cannot know whether the “people you meet for the first time” item item worked properly for China specifically. In a noble effort to correct for bias in the measurement of generalized trust, the authors wound up introducing new bias into the comparison instead.⁴

Another realm of trust is that of political trust, which has not fared well in Western countries over time, but results from China seem to indicate the Chinese strongly trust their government and national institutions. Skeptics often attribute this to various sorts of social bias in the way the Chinese answer surveys. Researcher Noah Carl examines this possibility, finding that the bias explanation came up less fruitful than what skeptics would have liked.

The high political trust among Chinese citizens cannot simply be dismissed as being caused by reference group bias, because when the European Council on Foreign Relations asked Chinese respondents which country comes closest to having a real democracy, the overwhelming majority answered “my own”. In fact, only six percent chose the United States.

On social desirability bias, Carl points to a study conducted by researcher Erin Carter and her colleagues using a list experiment. Instead of directly asking, “Do you support Xi Jinping?”, respondents were divided into two groups. One group was shown three neutral statements and asked how many they agreed with. The other group was shown the same three, plus one sensitive statement—“I support Xi Jinping.” The difference between the groups’ average scores revealed the hidden level of support.

While the result suggested that ordinary surveys do overstate government approval (support for Xi was 20 to 40 points lower when measured indirectly), this “de-biased” approach had limitations. In the list experiment, respondents could not express degrees of agreement. Instead, someone who “slightly agreed” in a normal survey might have been forced to say “no” here, artificially widening the gap. And as Carl noted, even taking the lower estimate at face value, about 75 percent of respondents still supported Xi—a higher share than Western citizens express toward their own leaders.

It is possible that China’s high political trust is just caused by government brainwashing and the Great Firewall. But, as Carl points out, poor political approval is common in nondemocratic regimes as well:

Yet high levels of satisfaction with government are far from universal in authoritarian countries. Although Nicaragua, Venezuela and Egypt have similar scores to China on the Economist Democracy Index, only 17% of Nicaraguans, 14% of Venezuelans and 8% of Egyptians have confidence in the national parliament (compared to 93% of Chinese). The difference, arguably, is that Chinese political elites are competent. They enjoy the “mandate of heaven”.

Beyond what was covered in his piece, there is also the argument that China’s unusually high trust in the national government might just be due to top-down scapegoating. The logic goes: Beijing pins their failures on local governments, and in doing so, they create an illusion of competence so that citizens end up trusting the national government. If that were true, we’d expect survey data to show significant distrust for local authorities among the Chinese. Yet the numbers don’t support that. Consider the findings from Wu (2021):

The 2018 World Values Survey reported that 95 percent of Chinese citizens said that they have a great deal or quite a lot of trust in national government. Comparatively, about 69 percent felt the same way about their local government.

Since the Chinese government already enjoyed very high levels of trust from its citizens before the pandemic, did this trust increase? Our surveys asked about trust in government at five different levels — the township, county and city level as well as the provincial and national levels.

The data show that Chinese citizens’ trust in their national government increased to 98 percent. Their trust in local government also increased compared to 2018 levels — 91 percent of Chinese citizens surveyed now said they trust or trust completely the township-level government. Trust levels rose to 93 percent at the county level, 94 percent at the city level and 95 percent at the provincial level. These numbers suggest that Chinese citizens have become more trusting in all levels of government.

…

Among the 44 percent of respondents who have placed more trust in some levels more than others, the mean level of trust is 89 percent. The fact that trust is high among Chinese citizens who look at government performance with a critical eye suggests that high government trust in China during the pandemic reflects Chinese citizens’ true satisfaction with their government performance.

The knee-jerk response here would be that Chinese citizens simply became too scared to speak their minds after the Covid-19 lockdowns. But even if that were true, it doesn’t explain why local governments already enjoyed majority support in 2018 prior to the lockdowns, and this is a majority that was achieved in spite of allegedly being used as scapegoats by Beijing for their screw-ups whenever convenient, if that explanation is to be believed.

In both the cases of general and political trust, the same lesson can be illustrated: bias in social research rarely goes one way. In China’s case, methodological quirks employed by researchers in an attempt to better understand their society often exaggerate their national vice, even if well-intended.

Turning beyond social and everyday life, in fields like science and technology, China’s output has grown at a remarkable speed. The conventional wisdom is that China simply copies everything the West does and produces nothing novel of its own, but is that still the case? As Anatoyl Karlin argues, one of the best ways to compare countries with respect to innovation is using the Nature Index:

What can we use as a proxy? Nobel Prizes in the sciences lag real world accomplishments by 20-30 years. Measures of individual eminence, such as Pantheon, only become crisp in long-term retrospect, and moreover, the Human Accomplishment database only runs to 1950. Total number of articles published, patents granted, R&D personnel, or R&D spending don’t adjust for quality. University rankings may be biased due to reputational and “brand” name factors, such as the worldwide prestige enjoyed by Oxbridge and the Ivy League. What can we then use instead?

The Nature Index (natureindex.com) bypasses almost all of these problems. This index measures the amount of publications in the 82 most prestigious scientific journals in the natural sciences. While they account for less than 1% of natural science journals in the Web of Science database, they produce almost 30% of all citations in this sphere. Every year, every research institution and country that contributed to these journals gets a score on the Nature Index measuring its research output (there is also a “running total” for the past year that covers Dec 2017-Nov 2018 as of the time of writing). This makes the Nature Index an ideal source of crisp, up-to-date, quantitative data on the production of elite level science.

Something else to note is that since the index is based on prestigious scientific journals, it’s less likely to be gamed by a country simply producing a ton of academic junk (though attempting to game publication-based metrics through paper milling does not seem to work that well in practice).⁵ The index has also been validated against other measures of innovation. I was able to update the fractional count of the countries listed on the most recent version of the index, and here are the results:

Back in 2018, China’s per capita rate was 10.9, Taiwan’s was 29.7, and Japan’s was 40.1. In 2024, China is now at 35.54, Taiwan at 35.71, and Japan at 40.48. In other words, China is now as scientifically productive per capita as Taiwan and has narrowed most of the gap with Japan. In line with this trend, China has entered the top ten of the Global Innovation Index, which comprises 78 indicators. The country’s record suggests that its gains are not merely the product of scale or imitation. In medicine, for instance, China has rapidly increased its share of active drug development, and these drugs are now being licensed by Western pharmaceutical firms—evidence that they meet international standards of quality.

The Economist (2025)

Energy tells a similar story. In nuclear power, an industry often praised by the right, China has achieved what the United States and France could not: steadily declining construction costs. While Western projects have become paralyzed by regulatory inflation and safety anxieties since the 1970s, China’s costs have continued to fall. As Lui et al. (2025) explained, the country is on track to surpass the United States as the world’s largest producer of nuclear energy by 2030.

The pattern repeats in automation and electric vehicles. In 2021, China overtook the United States in robot installations per 10,000 manufacturing workers. But as noted by Atkinson (2023), robot adoptions are partially a function of wealth.

the decision to install and run a robot is usually based on the cost savings that can be achieved when a robot can perform a task instead of a human worker—and those cost savings are directly related to the compensation levels of manufacturing workers. It should therefore come as no surprise that high-wage Germany has a higher penetration rate of robots than low-wage India. But the interesting question is how national economies perform in robot adoption when controlling for wage levels, given that the payback time for a robot gets shorter as manufacturing labor costs increase.

This is important because China is still significantly poorer than the United States when it comes to the average standard of living of its citizens. When wage levels are controlled for, the difference between the United States and China in robot adoption increases to twelve times.

Actual robot adoption rate as a share of expected robot adoption rate

In the electric vehicle sector, China’s BYD has overtaken Tesla in global sales and is seeing adoption by neighboring countries. Domestically, Chinese electric vehicles have become significantly more affordable to citizens, so much so that the price differential between electric vehicles and regular gasoline cars has vanished.⁶

China is by no means a perfect country. It still lags far behind the West in living standards and, like its Northeast Asian neighbors, faces a severe fertility collapse. It also misallocates much of its human capital, leaving immense potential unrealized. Yet none of this justifies caricaturing China as hopelessly backward or its people as inherently dishonest. In public debate, the country is cast in contradictory terms. On one hand, it is portrayed as a menacing juggernaut—a disciplined, calculating power whose state is ruthlessly efficient, so capable that it threatens to overwhelm the West. On the other hand, it is dismissed as a hollow “paper dragon”, crippled by its own incompetence and corruption—hardly a rival that warrants aggressive countermeasures. These two narratives coexist comfortably because both flatter the same conviction: that China cannot be both competent and legitimate at once.

We should see China for what it really is: a nation that survived the immense upheavals of the twentieth century and is now straining to propel itself into modernity. Its progress has been hard-won, and while its flaws are undeniable, its achievements should also be acknowledged. The country has traveled an extraordinary distance from the depths of its historical low point. As its economy matures, it is also witnessing the return of its ethnic talent from abroad—those who once left to seek opportunity in the West. That flow itself is a measure of recovery.

The argument that the Chinese are “anti-white” and therefore all this rhetoric against them is justified doesn’t even make any sense. No non-white nation will ever be “pro-white” by default; that is not a standard they can be held to. The obvious counterexamples some point to—Japan or South Korea—are not exceptions so much as countries still shaped by American occupation.⁷ The Japanese didn’t suddenly become a different people genetically after 1945, but you wouldn’t know it from the way rhetoric toward them flipped once they were firmly under U.S. control.⁸

A depiction of an anti-white bugman, commonly known as a Chinese Japanese.

With Imperial Japan crushed and no longer positioned to rival the West, China has now inherited the role of primary target for “legitimate” racial resentment (Russians, after all, are white). In the minds of many on the right, the Chinese and Japanese are not kindred peoples with an extensive shared heritage, but practically different species—one sanitized and “acceptable”, the other cast as the eternal bugman.⁹

Blind resentment toward China is not analysis but consolation. The constant insistence that Chinese accomplishments are stolen or illusory functions as psychological reassurance: a way for the right to soothe itself in the face of Western decline. It is easier to dismiss China’s progress than to imagine a world where the West is no longer the axis of civilization. Yet denial does not change reality.

China will not vanish because internet schizos spam “Three Gorges Dam” memes, nor will its demographic challenges erase its gains overnight. By sheer population size alone, it has several decades to adapt before anyone can safely count it out. Hatred of China will not halt mass immigration, restore (non-dysgenic) fertility, or undo the cultural and social decay hollowing out Western nations. It is easier to laugh at “Chicom bugs” than to confront the failures of our own elites. China’s rise should be a wake-up call: a reminder that nations which take themselves seriously can still claw their way back from disaster. We can resent that, or we can learn from it. But burying our heads in cope won’t change the balance of power, and it won’t save us.

1. If we were to accept the validity of this criticism, it would mean that the results for the complete wallet recovery rate for wallets with money in Yang et al. is downwardly biased, implying that civic honesty in China is even higher than what they found.

2. Cohn et al. did acknowledge the possibility that email was an uneven measure across cultures. In their supplementary materials, they restricted the analysis to hotels on the grounds that hotel staff should not only be familiar with email, but more likely to use it given their frequent dealings with foreign tourists and businesses. They also adjusted country rankings by national email penetration. Neither check meaningfully changed the results, which led them to conclude that email usage was not a major source of bias. But Hung et al.’s WeChat-based replication suggests this assumption was overly optimistic. Even in internationally oriented hotels, staff in China evidently did not regularly use email for this kind of interaction, meaning the hotel-only analysis could not actually neutralize cultural bias. This is further vindicated by Yang et al.’s survey results of the employees themselves, finding that only 39.19% considered not contacting the owner to be civically dishonest whereas it’s 80.91% for retaining the wallet.

3. Hung et al. caution that their reporting rates cannot be directly compared with Cohn et al.’ global study because of design differences (e.g., WeChat vs. email, wallet contents, and intervention framing). Unlike the original lost-wallet study, their field experiment used only wallets containing personal items and no money. The direction and magnitude of any resulting bias are unclear. Cohn et al. found that even in China, wallets containing money were more likely to be reported than moneyless wallets. Hung et al.’ online experiments, which varied wallet contents (“BigMoney” = 586 RMB + key, “Money” = 93 RMB + key, “NoMoney” = 0 RMB + key, and “NoKey” = 93 RMB without key), likewise showed that money and personal items evoke different psychological responses that shape reporting intentions.

Because the designs differ on several dimensions, it is not possible to determine whether Hung’s field design would yield higher or lower reporting than the money-containing wallet conditions in the original study. What can be said with confidence is that the original 7% estimate for China was almost certainly deflated by relying on an unfamiliar communication method. Once this design mismatch is removed, measured civic honesty in China rises substantially—even though the exact corrected rate cannot be inferred.

4. And of course, perceptions of trust is itself a different question from actual trustworthy behavior. To take just one example: this paper shows South Korea and South Africa, or Taiwan and India, ending up with comparable adjusted trust scores. But no serious observer would claim that how people actually conduct themselves in daily life is essentially the same across those societies just because their survey scores line up.

5. Just to be clear, I am not claiming that academic fraud isn’t an issue in China, it most certainly is, but simply that the country has made real gains in spite of it.

6. And this is not masked by Chinese electric vehicles being particularly skewed towards higher price ranges. Here’s earlier data for 2024 showing the distribution of price ranges for electric and combustion cars in China, Europe, and the United States:

7. And even then, it’s not clear if those countries today are genuinely “pro-white” in the racial sense beyond just being culturally Westernized on the surface, since both Japan and Korea had a greater share of respondents answering they would have voted Democrat than Republican in the 2024 U.S. presidential election.

8. The friendly attitudes toward Japan after World War II was also partially conditional on the perception that the country would not have any serious capabilities to challenge the West. During Japan’s economic boom in the 1980s, public sentiment in the U.S. towards the country began to sour was even regarded by some as dangerous. For instance, from a Gallup article (emphasis in bold is mine):

For most of the post-World War II era through the 1980s, U.S.-Japanese relations were primarily characterized by U.S. efforts to contain communism and its commitment to defend Japan. However, in the 1980s, surging Japanese economic power brought the country into competition with the U.S. Gallup's first reading on Japan, in 1989, found 69% of Americans holding positive opinions. However, this rapidly declined to 48% by November 1991, as the U.S. struggled with a substantial recession. In a separate question Gallup asked that month, 77% of Americans identified Japan as an economic threat to the United States.

From 1991-1995, roughly half of Americans continued to express positive views of Japan. This may have been related to both continued concerns about the economic threat Japan posed and a series of policy disputes centered on continued U.S. military presence there.

In 1996, President Bill Clinton began an effort to reaffirm positive relations with Japan, including expanding defense cooperation to meet the growing threats in Asia. While this effort was underway, the percentage of Americans viewing Japan positively jumped to 65%. As the U.S. economy surged in the '90s, positivity toward Japan continued to rise, reaching 69% in 1999. Past U.S. worries about the Japanese economy also waned, and by 2016, 24% of Americans registered concern about the country as an economic threat (Reinhart, 2021).

9. The discourse wasn’t always this way either. Sir Francis Galton had once proposed to populate Africa with Chinese people, arguing that they are “endowed with a remarkable aptitude for a high material civilization” as well as their ability to endure the continent’s climate.