Information Complexity and Structural Variation in Managing Crisis Intelligence

• In Event: Environmental Governance in China

Abstract

Crises such as extreme weather events, public health emergencies, and public security incidents threaten to expose incompetence and energize opposition, which leads governments to dedicate vast administrative resources to gather and process information that they need to make accurate predictions and effective preparations. However, the public policy literature has focused on how crises are addressed when they unfold rather than significant intelligence work that takes place in the window between crises, during which the crises being monitored have yet to occur. Not only does the discrepancy fail to reflect the importance of crisis intelligence in public policy, it also means that the existing public policy models are largely confined to the view that information varies in supply and quality, missing out critical variations in properties defined in qualitative terms.

Drawing on the relevant notions from complexity theory, we propose a framework to explain crisis intelligence management as an organizational reflection of those properties. In a systematic comparison of the intelligence work by Chinese policy agencies tracking early signs of crises in public health, public security, and extreme weather, we find that these domains can be distinguished by two types of complexity. Descriptive complexity refers to the level of analytical sophistication required to model a crisis event. For example, extreme weather events such as tropical cyclones can be predicted using a relatively simple model extrapolated from a large body of comparable data points, whereas public health emergencies such as epidemics of unknown origin are descriptively complex not only because it is difficult to collect data due to deliberate concealment but also because modeling new outbreaks based on past events could prove to be challenging when past pathogens often assume divergent properties. The second source of complexity is concerned with interactivity. The incidence of extreme weather is not tied to how much it is being modeled and predicted by communities trying to mitigate its impact. In public security, dissidents contemplating collective action are highly responsive to what the security apparatus knows about their activities, so that the intelligence work itself becomes a factor in crisis incidence. The latter, we argue, is interactively more complex than the former.

We observe that bureaucrats choose to impose more restrictions on information access in some domains but opt for more inclusive practices in others. For some domains, there is a preference to share intelligence with agencies similar in size and status, but partnerships between agencies of dissimilar descriptions are favored. Some policy contexts entail information procured from scattered sources across the government bureaucracy, while others can easily find most requisite data in house or in adjacent domains. We also notice that structural differences exist between the formal arrangements based on China’s tiao/kuai system and the informal patterns arising from actual practices in certain cases. Based on these observations, we develop three sets of propositions that revolve around the process in which bureaucratic choices connect the key informational properties to the aggregate patterns of intelligence exchanges, and then test them with a case study of the crisis intelligence work at Guangzhou, the capital of Guangdong Province.

Guangzhou is a super-large-scale city located in the Pearl River Delta Region (PRD), one of the most disaster-prone areas in Southern China. Over the past few decades, the Guangzhou municipal government has established an intricate network of government agencies to produce and coordinate intelligence essential to preempting various adverse events. These information are accessible in databases on general policy documents and media reports. We first put together a general picture of the local government’s crisis intelligence practices and operations in four selected domains -transportation, meteorology, public health, and public security- from documentary research and in-depth interviews, and then collect quantifiable data on how intelligence flows between government organizations and represent the pattern as networks. Both descriptive and inferential network analysis methods are applied to assess the validity of our propositions about complexity and intelligence practices and to evaluate how much the statistical results are aligned with the observations made by our informants in the qualitative findings from the interviews. The results show that bureaucrats will adjust their behavior strategically when confronted with crisis intelligence of different information properties, which contributes to the aggregate institutional differences across policy domains.

Authors

• Kwan Nok Chan, The University of Hong Kong

• Jiasheng Xiao, University of Hong Kong