We investigate whether and how environmentally inclined politicians (EIPs), i.e., politicians with prior environment-related work experience, affect local environmental performance in China. We find that firms located in cities with EIPs have lower levels of sulfur dioxide (SO2) emissions. The effect is attenuated when the politician is in his/her second term and among firms that are economically important. Firms in cities with EIPs commit fewer environmental violations, receive more green subsidies from the local government, and choose to establish new polluting subsidiaries in cities without EIPs. We also find that the economic performance of cities governed by EIPs is not statistically different from that of cities governed by non-EIPs. Furthermore, the promotion likelihood of EIPs is negatively related to local emission levels. The findings overall suggest that local officials strategically leverage their expertise in environment protection to allocate more effort on environmental causes.

We develop a machine learning (ML) approach to establish new insights into how memory affects ffnancial market participants’ belief formation processes in the field. Using analyst forecasts as proxies for market beliefs, we extract analysts’ mental contexts and recalls that shape forecasts by training an ML memory model. First, we find that long-term memories are salient in analysts’ recalls. However, compared to an ML benchmark trained to fit realized earnings, analysts pay more attention to distant episodes in regular times but less during crisis times, leading to recall distortions and therefore forecast errors. Second, we decompose analysts’ mental contexts and show that they are mainly shaped by past earnings and forecasting decisions instead of current firm fundamentals as indicated by the ML benchmark. This difference in contexts further explains the recall distortion. Third, our comprehensive memory model reveals the significance of specific memory features and channels in analysts’ belief formation, including the temporal contiguity effect and selective forgetting.

During the past decade, Wechat-public-accounts (WPAs) have gained increasing popularity as a novel tool for voluntary disclosure among Chinese public firms. This paper examines whether WPA disclosures provide value-relevant information to the market. Using a topic model to process over 1.6 million WPA articles during 2012-2020 and an event study design, we find that the stock market reacts strongly following a WPA disclosure event and the magnitude varies with the topic and textual feature of the WPA articles. We further present evidence that firms use their WPAs to provide new information rather than reinforce information that is already presented in other channels. Moreover, financial analysts, journalists, and retail investors rely on corporate WPAs for their information production. Collectively, our findings indicate that corporate WPAs are an economically significant source of new information for market participants that supplement traditional disclosure channels considered in prior studies.

This paper investigates the effect on A–H premiums of the China Stock Connect, which allows the Mainland to invest in H-shares in Hong Kong (Southbound) and overseas to invest in A-shares in the Mainland (Northbound). It removes barriers to investor trading all crosslisted A- and H-shares but leads to them an enlarged premium. We develop the differential demand hypothesis of Stulz and Wasserfallen (1995) in China and identify the elasticities of Stock Connect relying on the demand asset pricing of Koijen and Yogo (2019). We ffnd that the average elasticity of Northbound (A-shares) is 0.18, and that of Southbound (H-shares) is 0.66, implying that A-H shares have different levels of substitute effect for investors on each side of Stock Connect, leading to the long-term premium. On a univariate basis, they explain 20% of the variation of the A-H premium and remain highly signiffcant when controlling other variables. We also estimate the cross-listed and time-varying elasticities of Stock Connect. They illustrate the strong positive spillover effect of A-H shares and check the robustness of our results.

Using an extensive panel of Chinese firms from the Annual Tax Survey and relying on labor unrest as shock to local social stability, we show that state-owned enterprises (SOEs) react to nearby labor unrest by creating additional employment at the expense of firm performance. Each SOE exposed to unrest hires 3% more employees, which is a sizeable aggregated effect. This effect is larger when labor unrest occurs in the same industry as the exposed SOEs, when local governments have sound fiscal budgets, and when governing mayors have stronger promotion incentives. SOEs obtain more fiscal benefits when they absorb additional labor. In contrast, non-SOEs do not react to labor unrest, and their performance is unaffected. Similar effects are detected when we use the population of Chinese listed firms. This paper provides evidence that SOEs internalize the goal of maintaining social stability and contribute to the growth of the non-state sector.

We uncover a signiffcant relationship between the persistence of marketing and investment skills among U.S. mutual fund companies. Using regulatory filings, we calculate the share of marketing-oriented employees to total employment and reveal alarge heterogeneity in its level and persistence. A framework based on costly signaling and learning helps explain the observed marketing decision. The model features a separating equilibrium in which fund companies’ optimal marketing employment share responds to their past performance differently, conditional on the skill level. We confirm the model prediction that the volatility of the marketing employment share negatively predicts the fund companies’ long-term performance.

Large technology firms have substantial advantages in data, a key non-rival input for developing AI technology. We argue that investments by large technology firms stimulate innovation by AI startups through the sharing of data, bringing more than money to the startups. We assemble a unique dataset containing (nearly) the universe of AI-inventing firms in China to examine the innovation effects of these investments. Our difference-in-differences estimation shows that, after receiving investments from large technology firms, AI startups increase the number of AI patent applications by 62% and the number of software products by 56%, relative to their mean values prior to the investments. Using a triple-differences strategy, we further find that the innovation impact of investments by large technology firms is stronger than that of investments by other firms without data advantages. We confirm these findings using an instrumental variables approach based on recent investments by large technology firms in peer startups. Finally, we provide novel evidence that the innovation effect works mainly through sharing non-rival data by leveraging our rich information on non-AI data-related patent applications and data-related online job postings.

We investigate the employment consequences of earnings management. Using audit firm consolidation as an exogenous shock impacting earnings management, we find a positive casual effect of firm-level earnings management on employment growth. The effect is concentrated in privately owned enterprises and firms with higher operational risk, consistent with earnings management affecting labor dynamics by influencing employees’ perceptions of job security and subsequent career decisions. We further document a crowding out effect in local labor market, where a firm’s earnings management negatively influences the employment growth of local peer firms.

Using unique data on Alipay users' investment accounts, we find that, in addition to generating better performance than investors’ self-directed portfolios, robo-advising has a positive spillover effect on its adopters in terms that it improves their investment behaviors. Investors have more diversified portfolios and exhibit fewer behavioral biases in portfolio management and fund choices in their self-directed accounts after adopting robo-advising. The spillover effect is more prominent for adopters who interact with the service more actively and who were less sophisticated before adopting the app. We also find that adopters learn from the robo-advisor by simply imitating its portfolios or strategies. Collectively, this study provides large-sample, non-laboratory evidence that robo-advising effectively plays a role in educating investors through repeated interactions with its adopters and setting investment models that are easy to follow.

Using staggered exogenous shocks to mandatory CSR disclosure, we examine the effect of mandatory CSR disclosure on employment growth. We find that CSR reporting firms have a higher employment growth following the mandate than non-CSR reporting firms. With respect to potential channels, we document that mandatory CSR disclosure promotes employment growth by improving firms’ CSR performance on employee welfare. In cross-sectional tests, we find that the employment effect is more pronounced for state-owned enterprises, firms in hazardous industries and firms in high-tech industries. We also find that cities most impacted by the mandate exhibit higher aggregate employment growth. While mandatory CSR disclosure promotes employment growth of mandated firms, it has a crowding out effect on employment growth of non-mandated local peer firms. Our paper offers novel evidence on the impact of mandatory CSR disclosure on labor resource allocation.