Ownership and Media Slant during Sino-US Trade Conflict

Ownership, Partisanship, and Media Slant: Evidence of the U.S. Media during the Sino-US Trade Conflict (Job Market Paper) I explore what determines media slant towards foreign nations using the 2018-2019 Sino-U.S. trade negotiation as a testing ground. Using an event study design and coverage by local U.S. newspapers, I analyze how stories about China respond to shifts of U.S. policy towards China, and how this media reaction is determined by owners' partisan affinity, controlling for readers' characteristics. I find that local newspapers with Republican-leaning owners increase the intensity of negative coverage following a shift towards hostile trade policies relative to papers of nonpartisan owners, and they decrease this slant following a conciliatory shift; the opposite is true for Democratic-leaning media owners. To address the potential endogeneity of diplomatic events, I select events that induced significant abnormal price fluctuations of trade-war-related financial securities. I further establish a causal effect of owners' preferences by exploiting mergers and acquisitions among national conglomerates as a source of variation in political orientation of owners. These findings imply a spillover from domestic policy in the formation of citizens’ sentiment towards other nations: the media, as their lens to view the world, is colored by domestic political polarization. Checks and Balances: Media Criticism in China I study how political competition among provincial officials affects media criticism in China. I collect news reports of local mouthpiece outlets operated by local provincial governments that at least point out the weakness of local governance from 2004 to 2017. By exploiting the semirandomness of the pairing of the provincial governor and the party sectary, based on an established fact that bureaucrats are likely to be promoted in their third or fourth year (hereafter referred to as the promotion examination period), I prove that higher competition induces media criticism. My empirical findings expand in three dimensions. First, if a pair is assigned such that their promotion examination periods overlap, then during the examination period of the secretary: 1) mouthpieces increase the number of critical reports; 2) this increase of media criticism is mainly driven by the criticism on economic affairs, as opposed to public affairs; 3) mouthpieces increase their reports on local achievements. Second, pairs assigned to expect an overlapped promotion examination have a higher GDP growth rate, a common performance measure. Third, the correlation between media criticism and secretaries’ promotion is positive for secretaries in pairs expecting to be examined together, especially when the GDP growth rate is mediocre, and is negative otherwise. I construct an adverse selection model to illustrate how competition can be generated by promotion pressure. Intuitively, when both officials go through promotion examination, the economic signal from individuals cannot be observed separately, and media content serves as an additional signal sent by the secretary to increase the chance of promotion. These results suggest that the checks and balances embedded in the bureaucratic system allow the government-led media outlets to sometimes serve effectively as a watchdog than a lapdog. State Ownership and R&D Efficiency: Evidence from Chinese Listed Firms (Submitted) This paper empirically investigates how state-owned firms differ from non-state-owned firms on their R&D efficiency. We estimate the economic value of invention patents granted to Chinese publicly listed firms using the stock market’s responses to patent issuances, following the methodology proposed in Kogan, Papanikolaou, Seru and Stoffman (2017). We measure the return 3 of R&D by dividing the future patent value by current R&D expenditure, and find that the stateowned firms’ R&D efficiency is higher with very low R&D intensity, and is lower for medium and high R&D intensity. This finding is robust across different specifications, with both non-parametric and parametric models. 4