This paper addresses the question of how bureaucratic incentives affect economic performance in China at the provincial level. This paper develops a novel model to measure promotion incentives of the Chinese provincial standing committee members from 1995 to 2015. Using machine learning to incorporate over 200 variables of individual features, this paper derives a predicted probability of political advancement (i.e. prior likelihood of promotion) as a proxy to evaluate officials’ career prospects. The empirical results show that Chinese provincial governments are likely to generate higher economic growth rates when composed of a greater proportion of officials facing intense competition in the promotion tournament. On the other hand, a higher proportion of provincial leaders whose careers are at a dead-end or on a fast track has no effect on the local economic development. Contrary to popular belief, better economic performance stem mainly from the career incentives of the rank-and-file standing committee members, not from those of the supreme leaders of the province, such as the party secretary and the governor.

Dictators initiate elite purges to count threats of coups and mass protests; at the same time, purges vary in frequency, timing, and intensity, even across regimes with similar institutions and survival threats. The dissertation proposes a theory of factional clarity to explain such variations. Factional clarity, the extent to which visible elites can accurately infer individual memberships in factions, influences elites' ability to distinguish allies from enemies and estimate the balance of power between factions. This in turn mediates their willingness to purge, as well as the strategy they use to protect allies from purges.

Factional clarity is endogenous to elite conflicts, yet also diverges across regimes and over time in response to exogenous shocks. I demonstrates this divergence through a comparative analysis of the Chinese and Vietnamese Communist Party. In China, Kuomintang and Japanese offensives led to the creation of isolated Communist factions with membership based on visible geographical and professional ties. In contrast, repeated destruction of the southern revolutionary network by French and American-South Vietnamese forces homogenized Vietnamese Communist Party elites and made these ties irrelevant. I show that periods of high and low purge intensity in both regimes followed those of high and low factional clarity, respectively, and that factional clarity dictated purge initiators’ behaviors during two historical purges, the Gao-Rao Affair in China and the Revisionist Anti-Party Affair in Vietnam.

Using original data on recent disciplinary investigations in China and Vietnam, I then argue that factional clarity also affects the degree and form of political protection during purges. In China, where factions have clear membership, regime leaders engage in ex post protection by intervening in ongoing investigations of officials in localities with ties to them. In contrast, under Vietnam’s opaque factionalism, leaders prevent these officials from being targeted for investigations ex ante.

My findings unify bodies of literature on factionalism, elite networks, and political identities, and provide a framework to analyze authoritarian intra-elite violence. They also open an avenue for future research to explore the role of elite purges---and of factional clarity as a key mediator---on corruption, growth, as well as authoritarian legitimacy.

Overall air quality in China has improved for six consecutive years (Krotkov et al 2015). A little-studied aspect of pollution control in China is that the source of air pollution is dominated by heavy industry and that improvements have varied by sector. In my dissertation, I assess how corporate governance and competition policy has proven to be a useful tool for improving environmental compliance. Through case studies using both large N statistical analysis and in-depth interviews, I provide an empirical story that substantiates that as energy sectors have become more competitive, the Chinese government is able to impose stricter air quality standards and to enforce them more effectively. Monopolies avoid regulation through their control of information and of market supply of an essential commodity. By contrast, if the government breaks the monopoly, suppliers compete not just to sell product but also on compliance in order to gain permits and other government-issued advantages.

How do people living under authoritarian rule organize their beliefs about politics? Are their political preferences still organized along ideological lines? My research focuses on understanding the structure of mass attitudes under autocracy, using new data from China. I show that ideology in China is loosely organized around a left-right economic dimension and an authoritarian-democratic political dimension, and that the most politically sophisticated individuals are the least likely to constrain their ideological preferences to one dimension. Contrary to what we might expect, ideological polarization in China is largely absent at the mass level. In a second paper, I investigate the relationship between ideology and political participation. Using a spatial model of choice, I discover that for most Chinese, perceived government competence is a bigger factor than ideological distance in political participation. The implication of the model is that the diffuse nature of public preferences gives the Communist party a free hand to implement its policy initiatives without alienating key constituencies. The last paper explores the meaning of ideological labels in China, using three national surveys. I find that while many Chinese citizens are willing to locate themselves on a left-right scale, the labels left and right do not carry a consistent programmatic meaning. Further analysis reveals that the partisan and symbolic content of these ideological labels is also limited. I argue that the absence of a shared ideological understanding prevents Chinese citizens from exercising political agency.