Stanford study offers a glimpse into shortcomings in China’s anti-corruption drive

When Chinese President Xi Jinping took office in 2012, he launched the most extensive anti-corruption drive since Maoist rule in China.

For many in China, the campaign was considered a success: it put government authorities under intense scrutiny and curbed corruption. But, as Stanford scholars have discovered, there are some shortcomings in how information about graft was gathered from the public.

In a new paper published in the American Political Science Review, Jennifer Pan, an assistant professor of communication in the Stanford School of Humanities and Sciences, found that citizen complaints of lower management wrongdoing in one Chinese city were routinely concealed from senior authorities.

Hidden in plain sight

Pan’s study, co-authored with graduate student Kaiping Chen, comes from an analysis they conducted of a leaked email archive that came from a city in central China. Although the leaked email archive has received significant press coverage and has remained publicly available, it has not been systematically analyzed, the two scholars noted.

In their investigation of the email archive, Pan and Chen found it contained not just internal communications among lower-tier officials of the city but messages they sent to senior authorities as well. Attached to their messages were more than 600 “Online Sentiment Monitoring Reports” that listed complaints local citizens submitted to their local government over social media.

“It was fascinating because it provided an inside look into how the Chinese government monitors social media and what types of things they pay attention to,” said Pan about the glimpse into how a local government operated.

When combing through 2,768 emails from 2012–14, Pan made a startling discovery. She found that the reports the city officials sent to central management did not include all of the public grievances she saw detailed in the department’s internal communications.

“Not only do we know what they are monitoring but we also get a sense of how they are hiding information,” said Pan about her discovery. Of the 28 percent of complaints related to government wrongdoing, only 31 percent of these were reported upward.

Pan and Chen found that when city officials were accused of misconduct – such as embezzlement, graft and violence – they were less likely to report it upward than other governance issues, such as pollution or education policy.

Pan and Chen explored various reasons that might explain why this information was held back. Did lower-tier officials think some complaints were unreliable? Were some complaints only hearsay? Were the complaints not deemed an important enough issue for superiors to deal with? Or were there political incentives at play?

After cross-checking these different scenarios with the email archive and the public grievances posted online, Pan and Chen found that complaints implicating a city official or a person with political connections to city officials were less likely to be reported.

Complaints about counties with workplace and birthplace ties to city politburo members (the most powerful group of officials in the city, controlling policy, judicial and bureaucratic functions of government) were 11 percent less likely to be reported upward. By comparison, content with the most positive sentiment was 51 percent more likely to be reported upward.

“In this era when machine learning and computational methods are getting better and better at mining large-scale data, there is still someone writing a report to summarize findings, and that’s where there is an opportunity for bias, politics and manipulation,” Pan said about the shortcomings she identified in one anti-corruption effort.

In other words, at the end of the day there is still a person writing a report.

“Even though these complaints were submitted by the public and in plain sight, there are many ways to justify why certain complaints were not reported. There is subjectivity because someone is processing this data, that is is where manipulation comes in,” Pan said about the partial concealment at play.

Accountability in authoritarian regimes

While China has built many channels to gather citizen complaints, Pan’s research shows that there remain shortcomings in how an authoritarian regime gathers information from its public.

“What I found is contrary to an emerging view that accountability can be found in authoritarian countries,” Pan said, noting that public participation is insufficient for accountability in non-democracies. Monitoring the behavior of regime agents will remain a challenge.

Pan’s research illuminates another concern: To what extent can citizens in an authoritarian regime hold non-elected, government officials accountable?

“I think there’s been optimism that even if you don’t elect your officials, you can still somehow get the things that you want from the government,” Pan said. “What I am showing is that there are significant limitations. We can’t assume that because people are allowed to voice their grievances and report on corrupt officials to the government, that this information will make its way to the people who can sanction corrupt officials.”

Pan is an assistant professor in communication. Kaiping Chen, a doctoral candidate in communication, is co-author on the study.