Print
E-MailItay Goldstein is the Joel S. Ehrenkranz Family Professor at the Wharton School of the University of Pennsylvania; Shijie Yang is Assistant Professor at the Chinese University of Hong Kong, Shenzhen; and Luo Zuo is Associate Professor of Accounting at Cornell University SC Johnson College of Business. This post is based on their recent paper.
Modern information technologies have greatly facilitated timely dissemination of information to a broad base of investors at low costs. In our paper entitled The Real Effects of Modern Information Technologies, we exploit the staggered implementation of the EDGAR system from 1993 to 1996 as a shock to information dissemination technologies to examine their effects on the real economy. Our first hypothesis is that the EDGAR implementation leads to an increase in the level of corporate investment through the equity financing channel. This hypothesis follows from the conventional wisdom that greater and broader information dissemination leads to an increase in the amount of total information in the marketplace, which improves the functioning of the financial market and firms’ access to external capital, thereby allowing firms to tap into new investment opportunities.
Our second hypothesis is that the EDGAR implementation affects the sensitivity of corporate investment to stock prices through the managerial learning channel. The idea that prices are a useful source of information goes back to Hayek (1945). Stock prices can reveal traders’ private information that is otherwise not available to managers, and hence can affect managers’ forecasts about their own firms’ fundamentals and their investment decisions. The managerial learning perspective predicts that the investment-to-price sensitivity depends on the extent to which prices reveal new information to managers (i.e., revelatory price efficiency), which can be and is often different from the extent to which prices reflect all available information (i.e., forecasting price efficiency).
Under this perspective, whether the EDGAR implementation enhances or impedes managerial learning depends on its net effect on revelatory price efficiency. Theories predict two opposite effects. On the one hand, the EDGAR system naturally leads to more aggressive trading on information from corporate disclosures, which can reduce uncertainty in trading on other fundamental information and encourage more acquisition and trading of information potentially unknown to managers, resulting in a crowding-in effect. On the other hand, a decline in the cost of accessing corporate disclosures can reduce the equilibrium demand for more precise fundamental signals obtained with a deeper analysis. This crowding-out effect happens because it takes time to develop high precision signals and the trading profits based on these signals are reduced when low precision signals have already been reflected in prices. Given these theoretical tensions, how the EDGAR implementation affects managerial learning and the investment-to-price sensitivity is therefore an empirical question.
To test our hypotheses, we exploit the staggered nature of the implementation of the EDGAR system. On February 23, 1993, the SEC specified a phase-in schedule for registered firms to start filing on EDGAR in ten discrete groups (SEC Release No. 33-6977). Firms in the first and last groups became EDGAR filers in April 1993 and May 1996, respectively. This staggered mandatory implementation of the EDGAR system reduces potential endogeneity concern caused by unobserved firm-, industry-, or market-level shocks or reverse causality. For an omitted variable to confound our findings, it needs to affect different groups of firms at discrete points in time as specified in the phase-in schedule. Using a staggered differences-in-differences (diff-in-diff) research design, we find that the EDGAR implementation leads to a 10% increase in the level of corporate investment but a 20% decrease in the investment-to-price sensitivity. A standard dynamic test shows no difference in pre-trends in investment behavior between the treatment and control groups, supporting the parallel-trends assumption.
We conduct two sets of analyses to understand the underlying mechanisms. First, we examine the equity financing channel. We show that, after a firm becomes an EDGAR filer, the firm’s stock becomes more liquid and less volatile and the firm obtains more equity financing. These results are consistent with our prediction that EDGAR inclusion improves firms’ information environments, access to equity capital, and ability to undertake investment projects.
Second, we examine the managerial learning channel. The observed decrease in the investment-to-price sensitivity suggests reduced managerial learning from the market after EDGAR inclusion. We argue that this reduction in learning happens because greater dissemination of corporate disclosures levels the playing field, discourages private information acquisition, and crowds out some information that is new to managers. We first show that, after a firm becomes an EDGAR filer, it experiences a decrease in ownership by institutional investors, especially those who are more likely to actively acquire and trade on information. This result suggests that the EDGAR implementation provides greater benefits to less-sophisticated retail investors and discourages private information acquisition by more-sophisticated institutional investors.
To provide further empirical support to this argument, we use two measures based on structural market microstructure models to assess the equilibrium level of private information in prices. The first measure is the probability of informed trading based on the Generalized PIN model, and the second measure is the adverse selection component of the bid-ask spread. These two measures are complementary as the former relies on order flows to identify private information arrival while the latter directly measures the extent to which prices are affected by unexpected order flows. We show that the EDGAR implementation leads to a decrease in both measures of private information.
Next, we explore cross-sectional differences between firms to provide a tighter link between investors’ private information and managerial learning. The condition for managerial learning is that investors collectively possess some information that managers do not have. Learning models commonly assume that investors’ information advantage lies in evaluating growth options, which requires analyzing market trends, industry competition, and consumer demand, as well as making comparisons with other firms; investors are unlikely to possess new information about a firm’s assets in place since managers are the ones who put those assets there. Thus, the EDGAR implementation is likely to reduce managerial learning to a greater extent in growth firms than in value firms. Consistent with this cross-sectional prediction, we find that growth firms experience a greater reduction in privately informed trading, institutional ownership, and the investment-to-price sensitivity after the EDGAR shock than value firms.
Lastly, we examine the overall effect of the EDGAR implementation on ex post firm performance. On the one hand, greater dissemination of corporate disclosures and improved stock market liquidity can better incentivize managers (who are the agents of the shareholders) to take value-maximizing actions. On the other hand, reduced managerial learning, especially in growth firms, can hurt firm performance (despite managers’ best intentions). Empirically, we find that, on average, the EDGAR implementation leads to an increase in firm profitability and sales growth in value firms but hurts performance in high-growth firms where managerial learning from the market is particularly important.
Overall, our findings suggest that it is important to consider the tradeoff between financing and learning from prices when evaluating the real effects of modern information technologies. With the rise of FinTech innovation through big data or machine learning techniques, the investing public can now obtain a huge amount of data at relatively low costs. We might reasonably expect the decline in the cost of accessing information to increase forecasting price efficiency. However, our findings suggest that the effect of FinTech innovation on real efficiency is more nuanced as it might dampen investors’ incentives to engage in private information acquisition and reduce managerial learning from prices. Moreover, greater information production and dissemination brought by modern technologies may not necessarily enhance the welfare of investors as they can lead to a reduction in risk-sharing and trading opportunities among investors and an overweight on public signals due to beauty-contest incentives. Evaluating these various tradeoffs brought by FinTech developments is an interesting avenue for future research.
The complete paper is available here.
