Trading Against the Random Expiration of Private Information: A Natural Experiment

• Author: ROBERT J. JACKSON,MOHAMMADREZA BOLANDNAZAR,WEI JIANG,JOSHUA MITTS
• Published date: 01 February 2020
• Date: 01 February 2020
• DOI: http://doi.org/10.1111/jofi.12844

THE JOURNAL OF FINANCE •VOL. LXXV, NO. 1 •FEBRUARY 2020

Trading Against the Random Expiration of

Private Information: A Natural Experiment

MOHAMMADREZA BOLANDNAZAR, ROBERT J. JACKSON JR., WEI JIANG,

and JOSHUA MITTS∗

ABSTRACT

For years, the Securities and Exchange Commission (SEC) accidentally distributed

securities disclosures to some investors before the public. We exploit this setting,

which is unique because the delay until public disclosure was exogenous and the pri-

vate information window was well defined, to study informed trading with a random

stopping time. Trading intensity and the pace at which prices incorporate information

decrease with the expected delay until public release, but the relation between trad-

ing intensity and time elapsed varies with traders’ learning process. Noise trading

and relative information advantage play similar roles as in standard microstructure

theories assuming a fixed time window.

FOR MORE THAN TWO DECADES, the Securities and Exchange Commission (SEC)

has provided investors with access to securities filings containing market-

moving information through its Electronic Data Gathering, Analysis, and Re-

trieval (EDGAR) system, which is available through the SEC’s website. And for

years, unbeknownst to lawmakers and the public, a small group of private in-

vestors were inadvertently given access to these filings before they were widely

released via EDGAR—a government contractor operating a platform known as

∗Wei Jiang is from Columbia Business School. Joshua Mitts is at Columbia Law School and

Columbia Business School. Mohammadreza Bolandnazar is from Columbia Business School.

Robert J. Jackson Jr. is at the U.S. Securities and Exchange Commission. This paper subsumes

a previously distributed working paper titled “How Quickly Do Markets Learn? Private Informa-

tion Dissemination in a Natural Experiment.” The authors thank two anonymous referees, and

Associate Editor, the Editor (Stefan Nagel), Jennifer Arlen, Ian Ayres, Lucian Bebchuk, Emil-

iano Catan, John Coffee, Martijn Cremers, Slava Fos, Merritt Fox, Lawrence Glosten, Jeffrey

Gordon, Kevin Haeberle, Scott Hemphill, Charles Jones, Peter Koudijs, Jonathan Macey, Roberta

Romano, Sarath Sanga, Richard Squire, and Paul Tetlock for helpful discussions and suggestions.

This project has benefited a great deal from feedback at seminars and conferences at the AFA,

Columbia, Harvard, NYU, Peking University, Pittsburgh, and Yale. We are grateful to the Ira M.

Millstein Center for Global Markets and Corporate Ownership for financial support that made our

PDS subscription, and this research, possible. We thank Brian Benvenisty, Like Chen, Yiting Xu,

and especially Cong Liu for excellent research assistance. Jackson is on public service leave from

the New York University School of Law and completed his work on this study before taking office

as a commissioner at the Securities and Exchange Commission (SEC). The views expressed here

are solely his own and do not necessarily reflect those of the other commissioners or the SEC’s

staff. The authors have read The Journal of Finance’s disclosure policy and have no conflicts of

interest to disclose.

DOI: 10.1111/jofi.12844

C2019 the American Finance Association

5

6The Journal of Finance R

the Public Dissemination Service, or PDS, distributed SEC filings to a small

number of paying subscribers moments before they reached the public. In Oc-

tober 2014, the Wall Street Journal exposed this issue.1In response, members

of Congress demanded that the SEC examine the problem. Two months later,

SEC Chairman Mary Jo White pledged to Congress that the Commission would

quickly eliminate PDS subscribers’ advantage.2

There is no evidence that either the SEC or the government contractor acted

opportunistically. Instead, the issue appears to have been due to a lack of engi-

neering coordination between the public website and the PDS feed (see Section

I.A). Nevertheless, the episode provides a rare lab-like setting for studying how

speculators trade on, and the stock market processes, private information that

expires at a random time. Moreover, because the informed traders could form

expectations about the length of the delay based on factors that were exogenous

to traders’ behavior, we are able to identify the causal impact of the expected

delay on trading patterns.

More specifically, this setting contains two features that are typically not

available to researchers. First, we can observe both the arrival time and the

content of private information. While the theoretical literature (pioneered by

Glosten and Milgrom (1985)andKyle(1985)) has developed a standard frame-

work for how securities prices incorporate private information through the

work of informed traders, relatively few studies test these theories based on

private information with well-identified content and timing. This is because

private information is, by definition, not public knowledge and thus neither

the timing of its arrival nor its content are generally observable by econo-

metricians.3In our setting, we observe both the exact time when a small

group of investors receive the information (the time when a filing reaches

PDS subscribers) and the exact content of the information (the filing) that the

investors receive.

Second, we can measure the length of time that the informed traders have the

information before a filing reaches the public. As explained in detail below, the

duration of the “private window” in our setting varies randomly due to technical

limitations of the EDGAR system. Our data confirm that the only factor that

explains a significant portion of the variance in the length of the delay is the

time of day, presumably because it is correlated with the volume of traffic

on the servers. Critically, the delay was beyond the control of both the filers

1AWall Street Journal article published on October 30, 2014, titled “Fast traders are getting

data from SEC seconds early: studies show lag in posting to website” (by Scott Patterson and Ryan

Tracy) was the first to reveal the PDS advantage to the public. Before this issue was revealed,

Jackson and Mitts (see Jackson and Mitts (2014)) subscribed to the PDS service to study the

effects of the early dissemination of market-moving information. The article provides a detailed

analysis of the timing of the delivery of filings through the SEC’s systems.

2Chairman White’s letter,sent in December 2014, specified that, by early 2015, the SEC would

“implement an enhancement to our system . .. to ensure that EDGAR filings are available to the

public on the SEC website before such filings are made available to PDS subscribers.”

3An exception is Cornell and Sirri (1992), who provide a clinical study of an illegal insider

trading case using ex post court records.

Trading Against the Random Expiration of Private Information 7

and the speculators, and is not correlated with the variables of interest. This

unique source of exogenous variation allows us to draw causal inferences about

speculators as well as show how market prices incorporate private information

that is based on the expected length of the delay.

To provide a unified framework for our empirical tests, we develop a model of

informed trading with a random stopping time but a known hazard rate from

an exponential distribution. The model is closely related to Caldentey and Stac-

chetti (2010), with modifications made to fit the institutional features of our

setting. In Caldentey and Stacchetti (2010), an insider has perfect knowledge

of the real-time value of the security, which follows a diffusion process. In our

model, in contrast, features an insider who attempts to learn the fixed liqui-

dation value over time. This feature fits our setting more closely because the

fundamental value of a security is unlikely to change during an interval of a

few minutes and insiders may keep refining their estimation of the true value

by processing the leaked filings, knowing that the private information could be-

come public at a random time over which the insiders can form an expectation.

We next develop several empirical measures that closely track the parame-

ters in our model. Expected delay is the predicted delay, over which time (hour)

of the day is the primary determinant. Trading intensity is captured by both

total trades and directional trades (buying or selling in the “right” direction

based on the dollar value and the trade count). Consistent with the model’s

predictions, we find that when the expected delay is longer, insiders trade less,

in terms of both the dollar value of trades and the number of trades. In addition,

when the expected delay is longer, insiders take more time to submit the first

trade after the leakage. Moreover, both the information content and the ease of

information processing of the different types of filings affect trading speed and

intensity. Schedule 13D filings,4which tend to generate more sizable abnor-

mal returns upon public announcement (as compared to, e.g., Form 4 filings)

and are relatively straightforward to process (as compared to, e.g., Form 8-K

filings), induce the fastest and most intense trading during the private window.

Finally, we take advantage of this unique setting to test the standard insider

trading models. Consistent with Kyle (1985), we find that speculators trade

more aggressively when the value-price divergence is larger, when the filing

entails high information content (measured ex ante or ex post), and when the

market is deeper (measured ex ante or in real time). We also find that trading

via limit orders is preferred to trading via market orders when there is more

time (i.e., when the expected delay is longer) and when market-wide trading

volume is high and volatility is low. Although insiders attempt to smooth out

the price impact of their trading, we find that information disseminates faster

among large-cap, high-turnover stocks and when the overall market experi-

ences more trading activity.

Our study contributes to the vast literature on information transmission and

asset pricing (for comprehensive surveys, see Easley and O’Hara (2003), Biais,

4Schedule 13D provides disclosure of beneficial ownership greater than 5% by investors that

seek to influence corporate policies and control.