The information content of 10‐K file size change

• Published date: 01 December 2021
• Author: Quan Gan,Buhui Qiu
• Date: 01 December 2021
• DOI: http://doi.org/10.1111/irfi.12324

ORIGINAL ARTICLE

The information content of 10-K file size change

Quan Gan | Buhui Qiu

Discipline of Finance, University of Sydney,

Sydney, New South Wales, Australia

Correspondence

Quan Gan, Discipline of Finance, University of

Sydney, Sydney, New South Wales, Australia.

Email: quan.gan@sydney.edu.au

Funding information

University of Sydney Business School

Abstract

Change in 10-K file size robustly and negatively predicts future

stock returns. The documented return predictability reflects

mainly information content of 10-K file size change on future

cash flow news. We examine whether this return predictability

derives from managers' risk disclosures or disclosure obfusca-

tion. We find that the return predictability increases in future

return horizon. It is driven by positive file size changes and is

stronger for firms with high information asymmetry. It is stron-

ger when managers have more “skin in the game.”It persists

even when investor attention is high. It derives from the com-

ponent of file size change that cannot be explained by business

fundamentals. Although word count changes in specific 10-K

items can predict future stock returns, word count change in

Item 1A, the risk disclosure section, has no return predictabil-

ity. Our findings are broadly consistent with the managerial

disclosure obfuscation explanation and suggest that lengthier

10-K disclosure, to the extent it is used by managers to obfus-

cate bad news, does not necessarily benefit shareholders.

KEYWORDS

10-K file size change, disclosures, future stock returns, readability

JEL CLASSIFICATION

G12; G14; G18

1|INTRODUCTION

The median length of the annual 10-K report that provides comprehensive disclosure of business operations includ-

ing audited financial statements has more than doubled over the past decade (Dyer, Lang, & Stice-Lawrence, 2017).

More disclosure can make stock prices more informative and facilitate efficient asset allocation (e.g., Fishman &

Received: 16 February 2020 Revised: 8 May 2020 Accepted: 1 July 2020

DOI: 10.1111/irfi.12324

© 2020 International Review of Finance Ltd. 2020

International Review of Finance. 2021;21:1251–1285. wileyonlinelibrary.com/journal/irfi 1251

Hagerty, 1989; Grossman & Stiglitz, 1980; Haggard, Martin, & Pereira, 2008). However, increased disclosure length

and complexity may impede information dissemination and depress firm value (e.g., Blankespoor, Miller, &

White, 2014; Hwang & Kim, 2017; Loughran & McDonald, 2014; You & Zhang, 2009). In this article, we empirically

investigate the following research question: Does an increase in 10-K disclosure length benefit shareholders?

Figure 1 shows future annual portfolio returns and future earnings surprises for quintile stock portfolios formed

based on recent 10-K file size change. Specifically, we sort U.S. stocks into quintile portfolios at the end of June in

each year of 1994–2014, according to their most recent 10-K file size changes. We then calculate the time-series

averages of the equally weighted quintile portfolio returns over the next 12 months (from July to the following June)

and portfolio future earnings surprises. Interestingly, future stock returns are initially flat for the first three quintile

portfolios but significantly decrease for the last two quintile portfolios (which have large 10-K file size changes).

Similar pattern can be observed for future portfolio earnings surprises. Thus, the figure shows that decreases in 10-K

file size are not related to future stock returns and cash flow news while increases in 10-K file size are related to

both lower future stock returns and lower future cash flow news.

Consistent with Figure 1, our in-depth empirical analysis further reveals that although 10-K file size per se has

no return predicting power, 10-K file size change significantly and negatively predicts future stock returns, with the

predictive power increasing in the future return horizon. The cross-sectional stock return predictability of 10-K file

size change is robust to controlling for firm characteristics commonly used in the literature to predict cross-sectional

stock returns as well as to controlling for industry fixed effects.

Why can 10-K file size change negatively predict future stock returns? We posit managerial disclosure obfusca-

tion as an important driver of the uncovered stock return predictability of 10-K file size change. Extant literature

conjectures that managers' career and compensation concerns constitute incentives for exercising substantial discre-

tion in information disclosure. Managers tend to release good news in a timely manner and hide bad news in vague

FIGURE 1 Future stock returns and earnings surprises for quintile portfolios formed on past 10-K file size

change. This figure shows future annual portfolio stock returns and future earnings surprises for quintile stock

portfolios formed based on the most recent 10-K file size change [Log(filesize)]. Specifically, we sort U.S. stocks into

quintile portfolios at the end of June in each year of 1994–2014, according to their most recent 10-K file size

changes. We then calculate the time-series averages of the equally weighted quintile portfolio returns over the next

12 months (from July to the following June) and portfolio future earnings surprises (ES2). Portfolio returns are shown

on the right axis while portfolio earnings surprises are shown on the left axis. The number close to each bar indicates

the average 10-K file size change of each quintile portfolio. Definitions of the variables are provided in Table A1 of

the Appendix

1252 GAN AND QIU

and noisy disclosures (e.g., Dye, 1985; Jin & Myers, 2006; Kothari, Shu, & Wysocki, 2009; Verrecchia, 1983). If man-

agers strategically increase 10-K file length to reduce readability and thereby bury signals of bad news in large

amounts of distracting information, 10-K file size change may contain important information about future cash flow

news, and disclosure obfuscation can explain why 10-K file size change has stock return predictive power.

Consistent with this conjecture, we find changes in 10-K file size to strongly and negatively predict future cash

flow news as proxied by earnings surprises and changes in return-on-assets (ROA). That is, large 10-K file size

changes usually precede bad cash flow news. When we control for next period cash flow news, the return predictive

power of 10-K file size change becomes negligible at the 3- and 6-month horizons and substantially weaker at the

1-year horizon. These findings clearly indicate that 10-K file size change forecasts stock returns mainly because of

its information content about future cash flow news.

1

The findings support the notion that increasing disclosure file

length obfuscates negative cash flow news, thereby delaying its incorporation in stock prices.

Disclosure obfuscation further implies that the return predictability of 10-K file size change is asymmetric.

Return predictability should wax with the resulting delay in stock price reactions when longer 10-Ks are used to

obfuscate bad news, and wane when stock prices respond quickly and fully to good news clearly disclosed in

10-Ks. Consistent with this conjecture (and the pattern documented in Figure 1), when we distinguish positive from

negative file size changes, we find the predictive power of file size change on future stock returns and cash flow

news to be associated mainly with positive file size changes.

Moreover, the return predictability of 10-K file size change should be stronger when the information environ-

ment is opaque, because it is less costly and more effective to obfuscate disclosures in environments with high

information asymmetry (in which obfuscation is less likely to be detected) than in transparent environments with low

information asymmetry. Investigating the return predictability of 10-K file size change in different information

environments by interacting file size change with a set of commonly used information asymmetry proxies

(e.g., Fang & Peress, 2009) revealed that 10-K file size change exhibits greater return predictability in more opaque

information environments.

An alternative explanation, however, is that managers expand 10-K length to disclose potential risks associated

with business operations and elaborate on negative developments, which, in combination with investor inattention

to such risk disclosures, results in the documented return predictability of 10-K file size change. Nevertheless, our

finding that the return predictive power of 10-K file size change increases in future return horizon is inconsistent

with such a risk disclosure explanation, as it would imply that investors, on average, need a long time to react and

incorporate negative information from expanded risk disclosures in stock prices. We perform multiple empirical tests

to further distinguish between the disclosure obfuscation explanation and the alternative risk disclosure explanation.

First, examining the return predictability of text word count change in each 10-K item, we find that although

word count changes in specific 10-K items can predict future stock returns, word count change in the risk disclosure

section (Item 1A) has no return predictive power. It being arguably the most important risk disclosure section in the

10-K file,

2

the finding that word count change in Item 1A has no return predictability casts significant doubt on the

risk disclosure explanation. By contrast, we find that the word count increase of the Management Discussion and

Analysis section (Item 7: MD&A) not only is substantial but also strongly and negatively predicts future returns. This

result is consistent to the findings of Brown and Tucker (2011) that MD&A informativeness declines in the past

decade even as its length has become longer, lending further support to the disclosure obfuscation explanation.

Second, if 10-K files are lengthened at least in part to obfuscate bad news, the incentive to do so should be

stronger for managers with more “skin in the game.”Thus, the return predictability of 10-K file size change should be

stronger when managerial share ownership is greater. By contrast, if managers lengthen the 10-K file simply to elab-

orate on risks and negative developments, the return predictability should be unrelated to managerial share owner-

ship. Consistent with disclosure obfuscation, we find that the return predictability of 10-K file size change is indeed

stronger when managerial share ownership is greater.

Third, risk disclosure in 10-Ks should be more likely to be quickly incorporated in stock prices when investor

attention is high. Thus, under high investor attention, if managers simply lengthen 10-K files to communicate risks

GAN AND QIU 1253