Jeremy Wallace, associate professor of Government at Cornell University, joined CDT to discuss his new book, Seeking Truth and Hiding Facts: Information, Ideology, and Authoritarianism in China. In a wide-ranging conversation, Wallace traced the rise and fall of what he terms the Chinese state’s “limited, quantified vision.” Wallace explains how the state’s desire to increase productive forces after the death of Mao evolved into an obsession with a small subset of numbers, chief among them Gross Domestic Product (GDP), that came to dominate Chinese politics—until, suddenly, they didn’t (although rumors of their demise may be premature.) Other important examples over the years include family planning statistics, coal mine accident rates, air quality readings, and COVID case numbers. The book began as a project on the manipulation of data in China: who manipulates which statistics, and when and why they do so. Wallace explains that the consequences of statistical manipulation are nuanced, and not necessarily all bad. The interview has been edited for length and clarity:
China Digital Times: How did the government approach statistics under Mao’s rule?
Jeremy Wallace: Numbers, like everything else, were radically politicized under Mao. Mao had strong opinions about the correctness of different types of methods of counting. I would highly recommend that those who really want to dive into this topic read Arunabh Ghosh’s Making It Count: Statistics and Statecraft in the Early People’s Republic of China.
Mao liked censuses. He liked investigating particular cases in an almost ethnographic fashion—as opposed to randomized sample surveys or probability sampling that we think of as “modern” or “normal” techniques. This was a real debate pitting socialism versus capitalism. Mao saw surveys as tainted by capitalism; that they framed people’s experiences in particular directions and were pro status quo. Of course, Mao also had very strong ideas about what the answer should be. It wasn’t an open investigation into what was going on. The “correct answer” needed to be provided. This led to many problems throughout the regime, most famously, dramatically, and calamitously the Great Leap Forward and the famine. I think that scarred the generation that followed, at least to some extent.
CDT: At the end of Mao era, there’s a general sense that “maybe this didn’t work so well.” How did that shift away from Maoist ideology—but not the total negation of it—change what the state wanted to know?
Wallace: In some ways the title of the book comes from this. Deng Xiaoping used a classical expression that Mao had used: “Seek truth from facts.” It’s using Mao against Maoism. It’s hard to fight against the political vocabulary that you’re all steeped in and that everyone pledged fealty to yesterday. It’s very hard to immediately transition and be like: “That wasn’t working. I know we all said it was yesterday, but ….” It takes time, and Deng’s use of Mao against Maoism was a particularly savvy technique.
The way this connects to numbers in my mind is that this aphorism is along the lines of pragmatism: “We need to care less about what is correct, theoretically, and focus more on the facts of the matter—outcomes.”
CDT: You describe how, in 1978, Deng called for Party cadres to be rated on their individual performance, which was a step away from Communist collectivism. Just as he used Mao to attack Maoism, he used Lenin to attack Communist collectivism. You quote Deng quoting Lenin: “to refer to collegiate methods as an excuse for irresponsibility is a most dangerous evil.” Why did the Party central feel a need to start evaluating cadres differently than during the Mao era? What was it looking for?
Wallace: This is Deng appropriating authorities in the Communist lexicon to attack what had become Communist practice in the PRC. The phrase that I use throughout the book is “limited, quantified vision.” One of the things Deng truly believed was that the Chinese people had a lot of energy and excitement about improving their lives. Unlike Mao, who thought he knew the correct way to move the population forward, Deng was a little bit more circumspect and thought that what he needed to do was step back. Deng wanted to unleash initiative, to allow individuals to move forward on their own. He thought that giving them targets—giving them individual performance grades and bonuses connected to that—was the way that you would improve performance.
This is his limited and quantified vision. Quantified, because if you don’t count things—if it’s all about Communist spirit—then is that actually going to improve outcomes? That was too flimsy, even if it was rhetorically used. The emphasis was instead on quantified outputs and outcomes. Limited, because it was about not paying as close attention—focusing on particular things rather than having everyone being in everyone else’s business all the time.
China is a very diverse place, and different communities have different ideas about what might work in their communities. So stepping back and allowing individual leaders to initiate and kind of move things as they will, including maybe in policies or actions that could be construed as capitalist. Deng didn’t want to say, “what we need to do is capitalism,” because that would have been a political fight (and maybe he didn’t even believe it). He wanted to step back and not have to specify.
A second level is that the center wouldn’t necessarily know [about events taking place in the provinces]. If there was resistance to capitalistic policies in the center, they would just see what the facts are and then seek truth from there.
The third was to not demand that everyone do [everything] at the same time. The people that were more excited or interested in this direction could move in that direction and we could see the outcomes from it. Then maybe people would change their minds. So if you’re in the northeast of China, and your agricultural production is doing relatively well, you maybe don’t feel as much of a need to immediately shift [how you organize] agricultural production. Whereas in other parts of the country, maybe you do.
For all those reasons, this limiting vision made political and economic sense to Deng. It’s also cheaper. It’s hard for the center to manage huge numbers of people going down and investigating everything.
CDT: In 1978, the Party was experimenting with all sorts of liberalization and modernization. Why did it reject the “fifth modernization,” democracy?
Wallace: Calling for change is an acknowledgment that the status quo has failed, but the Party was in charge of the status quo, too. It highlights the danger of political reform because once you open that can of worms, drawing the line about where change ends is quite difficult. The Four Modernizations versus the Four Cardinal Principles—that’s kind of the ‘78-‘79 story. For me, it points to the contingency of big political moments and political changes. This is something I tried to emphasize again when talking about Xi’s neopolitical turn [discussed further below]. These were moments of huge debate when the shapes of China-to-come were open questions. How far are we willing to go? How far are we not willing to go? Democracy was always a line that the party prevented. It did not want to give up its monopoly on authority.
CDT: Democracy is not how you get promoted within the Party. You get promoted through the cadre evaluation system. What is it?
Wallace: The cadre evaluation system is a central piece of the machinery of China’s limited, quantified vision. It’s the way higher-level officials evaluate lower-level officials. It’s usually done on an annual basis and is based on lots of key performance indicators (KPIs). Especially earlier in the period, these criteria were relatively concrete things like: “Did you improve the growth of industry?” and so forth. Over time, it became gross domestic product (GDP), which became the über statistic.
Cadres do everything they can to perform to those numbers, just like good lower-level employees try to hit KPIs. There’s different aspects of these. There are veto targets like: you can’t have major protests or social instability incidents. For a while, family planning was a major indicator. Cadres worked really hard on those. The things that are not measured are not cared about as much. These are negative externalities. [Take steel for example.] No one is counting the pollution that steel plants produce in your evaluation. They’re looking at how much steel you’re producing—not even how profitable the steel plant is, but solely output.
CDT: So the Party realized: “Okay, we’re building all this steel, but people’s lives are getting worse with smog and all that.” So they try to create new KPIs: Green GDP and PM2.5 are among the most famous. Why did one work and the other not?
Wallace: GDP is this number that tries to encapsulate everything but it clearly doesn’t and everyone knows it doesn’t. Famously, if I look after your kids and you pay me and you look after my kids, and I pay you that will get counted in a way that it wouldn’t if we just look after our own children. Another thing that GDP doesn’t count is the environmental costs of action. Green GDP was a recognition that we need to incorporate sustainability into thinking about GDP as a broad global effort.
It basically failed everywhere, so its failure in China is not particularly surprising. It’s not like Sweden or the United States are using Green GDP. No one uses Green GDP. But I think one of the particular reasons it failed in China was, if you look at some of the pilot [projects], they found that there was no growth in the area if you actually try to think about the environmental costs of what was going on—and that was just unacceptable. GDP was too central to the political game to be completely undermined by trying to think about the environment. It’s also very abstract. GDP itself is really abstract but then to try to shift GDP to think about the environmental costs becomes extremely esoteric, whereas PM2.5 is not. The actual particulate matter that gets deep into your lungs and deep into your blood is really problematic and can be measured very simply. PM2.5 became a topic of conversation after the airpocalypse. It became a convenient number because it was a problem that could be solved as opposed to one that would undermine the whole system, like Green GDP. PM2.5 was a slice of a problem to be solved and so it fit this engineering mentality.
CDT: Limited, quantified vision had major successes but also downsides. It was liable to be manipulated, and didn’t count everything. One example of manipulation you show is that cities cluster slightly above 100 million RMB GDP because it makes them eligible for subway funding. How did limited, quantified vision work? Where did it struggle?
Wallace: Basically, the system worked, right? The reason we’re talking about it, the reason CDT gets the number of hits that it does on its website, the reason that the number of people in the China sphere has grown remarkably over the 40 years of the Reform Era, is that focusing on development was successful. Some of that has Maoist legacies: education helped there. I don’t want to say “Deng caused growth,” but I do think that the focus on modernization was very successful. China was a desperately poor country in 1976. That it is not now is a remarkable success. A large part of that success, I think, is tied to this developmental mindset and in particular, the quantified vision of the state.
In the beginning this was a project about the manipulation of data, GDP manipulation in particular. I think people are inherently interested in manipulation because we live in a world of numbers. Whenever people play games and you can identify people cheating, it’s extremely interesting to humans. Whether that’s kind of an unnatural clustering right above a particular number where you get access to funding or everyone is kinda above average. (China’s national GDP number will actually be lower than almost all the provincial averages because the provinces are all trying to push up their numbers and the center knows that and so pushes down growth.) When you focus on particular numbers, you’re gonna get incentives to manipulate. So this is how I came into the project. Perhaps unsurprisingly, I did find evidence of manipulation.
The difficulty in the project was, if you’re using official data all up and down the project, how do you actually find manipulation? I tried to look for moments when you might expect more manipulation than others: that is, particular moments of political turnover. When one leader is leaving and one is coming in—that’s the time. The data is usually about differences between electricity consumption or production. It’s highly correlated with GDP but not the same as GDP. I can’t claim credit for this. You can see Li Keqiang on Wikileaks explicitly saying he thinks about electricity data, railroad data, and loan data.
What I find is: there is more falsification during times of political turnover in China. This was a period where growth is around eight or nine percent per year for provinces. People weren’t going around doubling or tripling the number. You don’t want to be found out. You want to be just a little bit higher than the other guy. I found manipulation or distortion of about ten percent. So reporting ten- instead of nine-percent growth.
I was never that interested in naming names because in the end, I don’t really know what’s happening in a given year in a given province. The fact that across hundreds of province years you find this pattern that GDP is systematically higher in moments of political turnover [confirmed, to me, that this pattern was real]. Party officials have had to fess up to this, and we find our falsification index is much higher in those localities that have reported falsification themselves. So that gives us real confidence that we are capturing this falsification.
CDT: Is fudging the stats a bad thing?
Wallace: The Center [the central government in Beijing] is not thrilled about the practice of manipulated data at any level. It would prefer it if that were not a possibility. If we could go to a world before the invention of lying, I think they would prefer that. That being said, that’s not the world we’re in. I think that they’re all playing very persuasive political games with each other. Communicative games about: “I understand what you want. I understand the political system. I can get my bureaucrats to do this, but not that.” Li Keqiang knows that this manipulation is happening. People are not necessarily pulling the wool over the eyes of central leaders.
It’s not all fake. There are real limits to falsification because once you really erode trust, in the overall sense of the numbers reflecting reality, then what can you trust? What do you do? Do you make investment decisions? How do you make them? That being said, in very particular moments of political danger, economic expectations and people’s confidence is super important. If you can fudge a number and make things seem a little bit more stable during a particular rough patch, you might find that it’s worth it—even if it does erode trust over the long run! Confidence for trust is something that, at particular moments, you might be willing to trade off. They’re not reporting ninety-nine percent growth every year. There are real limits. It’s not the Sputnik harvests of the Great Leap Forward. Those days are gone.
CDT: One of the things you write about is that you can have both falsification and better outcomes. You specifically wrote about the campaign-style drive for coal mine safety.
Wallace: To be clear, this is me reporting other people’s research. China’s development is a story of success that had real negative externalities. Focusing on performance in particular narrow slices, whether that’s PM2.5 or coal mine accidents and deaths, actually did produce real improved performance—many fewer people die in coal mines now than they used to—but it also produced clear evidence of falsification in that data. There’s vast underreporting. There’s a lot of accidents under the threshold of what is acceptable before you have to report higher up. Once you’re above the threshold, almost nothing happens. Real data doesn’t really work that way. That fits the broader idea that the system was quite successful in what it was performing, but had these problematic side pieces.
CDT: By the time Xi got to power, China was no longer an impoverished backwater governed by Maoist ideology. But there was a sense that Hu Jintao’s years had been a “lost decade”, and Xi Jinping himself thought that all was not well. What is Xi’s “neopolitical turn” as you describe it? And does it spell the end of the limited, quantitative vision of the Chinese state?
Wallace: In 2022, I think a lot of the China-watching set have a pretty strong sense about what I call the neopolitical turn, or Xi’s “new normal.” I think we have a strong sense about the personalization, the Partyization of various things.
One interesting piece that I think is less well understood—and this is something that Susan Shirk says a little bit about in her new book Overreach—is that a lot of the festering problems outside of the limited, quantified vision of the Hu-Wen regime needed to change. So the Party expanded the cadre evaluation system and tried to count more things like environmental data. But they failed. And I think in large part, they failed for reasons that they’re not really that responsible for—because there was a global financial crisis and the sense was “now is not the time to worry about PM2.5, tens of millions of people are out of work!” So they just stimulated and built, running up huge debts. Between 2008-2012, there were real debates about all these problems. Some people saw them as growth problems and the solution as more capitalistic reforms. Some people saw them as political problems and the solution as a strong leader and an anti-corruption campaign. Some people saw these problems and said “democracy.” There were huge, interesting, fascinating debates in that period and a general recognition that problems exist.
When Xi Jinping came into power, he clearly had some ideas that he thought were right. Centralization of authority—they reduced the size of the Politburo Standing Committee to seven instead of nine. It’s a lot easier to get four people on your side than five. There’s anti-corruption campaigns and interesting political movements. That entire first year in office, if you go back and really look at that year as I tried to do, it’s really topsy-turvy. Yes, there’s censorship but there’s also going to Shenzhen. Then the Third Plenum document came out and it was seen as getting the state out of the economy.
That’s not the way it turned out. It turned out that it was actually the “Leading Small Groups” and all this other Party stuff that ended up being important but it didn’t have to be this way. It almost wasn’t. [2013] is a fascinating moment, just like the 1978 story is kind of this jumble where lots of things are happening and it could have turned out very differently. As someone who doesn’t like concentration camps for Uyghurs, the complete decimation of civil society, shutting out of the West, and an aggressive foreign policy … I’m saddened by this.
CDT: Can we talk about COVID? China built a system after SARS to track emerging diseases. Why didn’t it work, initially? What were its blind spots? Because it worked in some ways but not in others—of course, the same story happens in Singapore and, obviously, the United States. How did China see COVID? And how did it miss it?
Wallace: It’s important to remember that the virus is SARS-CoV-2. China literally saw this problem before and prepared for this problem. They say, “You never get the same crisis twice,” and yet China got exactly that. Yet, it failed initially in Wuhan because the local leadership swept it under the rug because it was inconvenient politically. This was the moment the local Party Congresses and local People’s Congress [were in session]. They did not want this story to go anywhere. They thought it was just some pneumonia at a wet market—these things happen—but not like this. The fact [that Party Center] had to send in three central inspection teams before they got the data they wanted shows that even under Xi’s centralization, local officials still hide data.
The center sat on the data for a week. [Xi has stated that he “issued demands during a Politburo Standing Committee meeting on January 7 for work to contain the outbreak,” but the head of the National Health Commission only instructed provincial health officials that human-to-human transmission was occurring on January 14.] It’s hard for me not to think that if they had gone a week earlier, that this might have been SARS II, and then forgotten in the way that SARS was. No one in the United States has a personal experience with SARS. I’m pretty sure South Park made an episode making fun of SARS as a thing that [wasn’t real]. It was a real thing—we got very lucky that time. This time, we didn’t.
China only succeeded for a couple reasons. The timing was very precarious. [It was near Chinese New Year,] generally the biggest annual migration in the world, and that could have gone very, very badly. Instead, it seems like what happened was because people are used to businesses shutting down at that moment, the virus didn’t really spread much beyond the core Wuhan, Hubei area. People just went home and everyone was told to stay in place. Outside of a few institutions like a prison here or there, outside of Hubei there was not a lot of community spread in 2020. There was still huge frustration inside of China politically—about the secrecy, about the whistleblower, especially when [Li Wenliang], the doctor who was a whistleblower, died [of] the virus himself.
The Singapore example is interesting. They don’t see their migrant population. They don’t think about their migrant population as full members [of society]. [Migrant workers] are living in dorms and it becomes a dangerous population for them. Even though the city overall is relatively safe for a long period of time, [COVID] explodes when it is exposed in these areas that they just did not think about as closely. It’s akin to the hukou issues of China: thinking about urbanization in China; thinking about who’s counted, who’s not counted, and so forth; the migrant as a peripheral person, someone who’s not considered fully when governments make decisions.
