Information intermediaries provide or transmit useful information to other parties in the capital market, either because (i) it has not been publicly released, or (ii) it has not been widely disseminated. Therefore, information intermediaries influence a firm’s information environment. For example, institutional investors and analysts help incorporate more earnings-related information into prices during the pre-announcement period, leading to smaller investor reactions to earnings surprises during the announcement period. More recently, a growing stream of research has investigated the role of other types of information intermediaries, including the business press, social media, data providers, and the internet. Collectively, this research finds that intermediaries are associated with greater price efficiency.
In our paper, we ask whether employees function as information intermediaries in conjunction with their professional networks. This question is motivated by the idea that employees and their professional networks satisfy the following three conditions for information intermediaries: (i) access to private information, (ii) ability to disseminate information widely, and (iii) trading incurred by disseminated information. First, information and knowledge are widely dispersed within organizations, and employees of all levels have privileged access to value-relevant information. For instance, prior studies find that employees’ aggregate stock purchases and option exercises predict future returns. Second, employees have expansive professional networks, consisting of their direct and indirect (i.e., friends of friends) professional connections, allowing them to distribute their information outside the firm and ultimately into stock prices. Accordingly, value-relevant information can be transmitted from employees to numerous market participants within a few steps (the small world problem by Stanley Milgram). Finally, the nascent “word of mouth” literature finds that investors (especially retail investors) frequently make investment decisions based on information shared through their connections. In sum, employees have value-relevant information about their firms and transmit it through their professional connections, thereby spurring sufficient trading that helps impound the information into stock prices.
To address our research question, we exploit a unique and proprietary dataset on interfirm employee connections from a dominant business card management application in Korea (“Remember”). As in many Asian countries, it is a pervasive and entrenched cultural practice in Korea to exchange business cards with a new professional contact during the first in-person interaction. This type of exchange is essential for building professional relationships. Remember’s dominant market position provides a reliable and precise way to identify a near-comprehensive set of meaningful professional connections in Korea for both executive and non-executive employees. With these data, we create employee-specific measures of first-order (i.e., direct) connections and second- and third-order (i.e., indirect) connections. Since the ability of a network to transmit information decays when the information must travel through more people, we accordingly discount higher-order connections relative to the first-order connections. We finally use the average number of connections per employee to calculate time-varying firm-level connection measures.
Employee Connections and Market Reactions around Earnings Announcements
We first examine whether employee connections are associated with lower stock price and trading volume reactions to earnings surprises. We focus on earnings announcements because they are well-defined information events for all firms and allow us to control the expected earnings level. If value-relevant information about upcoming earnings is transmitted through employees’ professional connections, then firms with more-connected employees would have smaller market reactions around earnings announcements. Our results show that the second- and third-order (although not first-order) employee connection measures are negatively associated with the magnitude of the earnings response coefficient (ERC), the sensitivity of stock price reaction to earnings surprises. The associations are economically and statistically significant. For example, in one specification, the ERC decreases almost 23% if a firm moves from the 25th to 75th percentile of the second-order connection measure. Our results also show, conditional on the announcement period abnormal returns, abnormal trading volume is negatively associated with the second- and third-order connection measures. However, employee connections are only negatively associated with the abnormal trading volume of retail investors; we find no significant association between the connection measures and the abnormal trading volume of institutional investors. These findings are consistent with our expectations that most “connection-based” trades are by retail investors since their trades are more likely to be influenced by their individual connections than those of institutional investors.
It is important to note that employees’ direct connections are not associated with weaker market reactions to earnings surprises. First, the insignificant results for the direct connections emphasize the importance of considering the expansive nature of professional networks in capturing employees’ ability to spread information. To illustrate the expansive nature, consider a star network of 100 employees, each at a different firm, in which one employee is connected to all other 99 employees, and these 99 employees are only connected to that employee. The average first-order connection is 1.98 = (99⨯1+1⨯99) / 100, and the average second-order connection is 97.02 = (0⨯1+98⨯99) / 100. Therefore, the second-order connection increases exponentially because more-connected employees have a disproportionately larger influence on others’ second-order connections. Second, our results are unlikely to be affected by omitted factors correlated with connection measures and market reactions since any omitted variables (e.g., the characteristics of the firms and/or their employees) are most likely to be reflected in the first-order connection measure. For example, if business professionals prefer to connect with employees in more popular firms, those firms will have more highly connected employees. In addition, investors may be more focused on or attracted to more popular firms, and therefore, these firms will have weaker market reactions to earnings announcements. However, the higher-order measures are less likely to be subject to this concern than the first-order measure. Thus, combining the insignificant results for the first-order connections and the significant ones for the higher-order connections partially alleviates the concern.
Further analyses show that our return and volume results hold for executive and non-executive connections, indicating that non-executives and their connections act as information intermediaries. Our results are also stronger for firms with weaker information environments. Overall, our results suggest that earnings-related information is transmitted through employees’ professional networks and that this information is incorporated into prices via the trades of retail investors before earnings are announced. Together, these findings are consistent with employees’ professional connections collectively acting as information intermediaries that increase the level of price efficiency around earnings announcements.
Employee Connections and Intra-period Timeliness over Quarterly Earnings Cycles
Our earnings announcement results suggest that more earnings-related information is impounded into stock prices during the pre-announcement period when firms are more connected. If correct, prices should reflect earnings-related information earlier during the quarter when employees are more connected. We provide evidence about this conjecture by examining the association between employee connections and the intra-period timeliness (IPT) of earnings news. IPT captures how quickly information is impounded into prices by holding constant price response and information content. Intuitively, IPT increases when the higher portion of the quarter’s returns is realized earlier because it indicates faster price discovery.
Figure 1. Speed of Price Discovery and Employee Connections
Notes: The figure presents the percentage of 63-day buy-and-hold market-adjusted abnormal returns for each day from 60 trading days before the earnings announcement date to two trading days after it. We partition firm-quarter observations into three portfolios based on the tercile of second-order employee connection measure and plot the percentage for the highest and lowest terciles. The solid (dashed) line represents high (low) employee connection portfolios.
We examine the association between employee connections and the speed of price discovery by constructing High Connection and Low Connection portfolios based on the tercile of the second-order employee connection measure (2ndOrder). In Figure 1, we plot for each day in the quarterly earnings cycle the cumulative buy-and-hold market-adjusted abnormal returns, scaled by the cumulative buy-and-hold abnormal return for the entire 63-trading day period. Each point captures the percentage of the entire quarter’s abnormal return realized up to and including a particular day (i.e., the area under the cumulative price change curve over a given window). On the last day of the period, the plot equals 100% by construction since 100% of the quarter’s abnormal returns must be realized by then. Figure 1 presents the High Connection (solid line) and Low Connection (dashed line) portfolios, indicating that price discovery is faster when a firm’s employees are more connected. Starting about six days after the beginning of the period, IPT is always higher for the more-connected portfolio than for the less connected portfolio. A large gap between the lines begins around day −40 and generally persists to about day −16 when IPT for high-connection firms is almost 100%. After that, the gap between the two lines narrows until the two lines converge on day +2 (by construction). This evidence further supports the idea that employees act as information intermediaries through their professional connections.
Employee Connections and Return Predictability of Investors’ Trading Activities
Our results on trading volume around earnings announcements indicate that trades by retail traders reflect more earnings-related information when they trade in firms with more-connected employees. These results suggest that (at least some) retail traders receive information through employees’ connections and use this information when making their trading decisions. In this case, retail trades in more-connected firms are more likely based on private information. To examine this implication, we examine how retail order imbalances (i.e., the difference between retail buy volume and retail sell volume scaled by total retail trading volume) in one week are associated with abnormal returns in the following week. We find that the informativeness of retail order imbalances is increasing with employee connections. Furthermore, we find that retail order imbalances only predict future returns among firms with more-connected employees. Together, these results further support the notion that employees act as information intermediaries through their professional connections.
One caveat to our study is that while we can precisely identify an employee’s actual professional connections, we cannot directly observe whether employees pass on value-relevant information to their connections and, if so, what information is transmitted. Thus, we can only determine whether employees and their professional connections act as information intermediaries based on the associations between employee connections and price efficiency. As such, our results should be interpreted accordingly.
Altogether, we provide interesting and novel evidence that the employees, in conjunction with their professional networks, function as independent information intermediaries and are an important factor in increasing stock price efficiency for earnings-related information. Understanding the role of employees as intermediaries is important because it affects stock price efficiency (and hence the efficient allocation of capital) and because it represents private disclosures beyond the firm’s control.
DuckKi Cho is an Assistant Professor of Finance at the Peking University HSBC Business School.
Lyungmae Choi is an Assistant Professor of Accounting at the City University of Hong Kong.
Steve Hillegeist is an Associate Professor of Accounting at the Arizona State University W.P. Carey School of Accountancy.
This post is adapted from their paper, “The Role of Employees as Information Intermediaries: Evidence from Their Professional Connections,” available on SSRN.
