When you talk, I remain silent: Spillover effects of peers’ mandatory disclosures on firms’ voluntary disclosures

Matthias Breuer* The University of Chicago Booth School of Business

Katharina Hombach† Frankfurt School of Finance and Management

Maximilian A. Müller‡ WHU – Otto Beisheim School of Management

This Draft: March 2018

Abstract

We predict and find that regulated firms’ mandatory disclosures crowd out unregulated firms’ voluntary disclosures. Consistent with information spillovers from regulated to unregulated firms, we document that unregulated firms reduce their own, voluntary disclosures in the presence of regulated firms’ mandatory disclosures. We further find that unregulated firms reduce their voluntary disclosures more the greater the strength of the regulatory information spillovers. These results hold for a several voluntary disclosure measures. Our findings suggest that a substitutive relationship between regulated and unregulated firms’ disclosures attenuates the effect of financial reporting regulation on the market-wide information environment.

JEL classifications: M41, G38 Keywords: Mandatory disclosure; Voluntary disclosure; Information spillovers; Crowding out Acknowledgements: We thank Phil Berger, Jannis Bischof, Pietro Bonetti, Ulf Brüggemann, Willem Bujink, Hans Christensen, Joachim Gassen, Igor Goncharov, Ole-Kristian Hope, Martin Jacob, Laurence van Lent, Christian Leuz, Garen Markarian, Mike Minnis, Steven Monahan, Maximilian Muhn, Peter Pope, Nemit Shroff, Harm Schütt, Thorsten Sellhorn, Jeroen Suijs, Rodrigo Verdi, Robert E. Verrecchia, and seminar participants at Tilburg University, the University of Iowa, the University of Mannheim, the University of Göttingen, Stockholm School of Economics, the WHU – Otto Beisheim School of Management, the Frankfurt School of Finance and Management, the EAA Doctoral Colloquium 2014, the AFAANZ Doctoral Symposium 2014, and the 2017 FARS Midyear Meeting for helpful comments and suggestions on an earlier version of this paper. Breuer gratefully acknowledges the financial support of the Accounting Research Center at the University of Chicago Booth School of Business and the Doctoral Fellowship of the Deloitte Foundation. Müller gratefully acknowledges financial support by the German Academic Exchange Service (DAAD). * mbreuer@chicagobooth.edu; The University of Chicago Booth School of Business, 5807 South Woodlawn Avenue, Chicago IL 60637, United States of America. † k.homach@fs.de; Frankfurt School of Finance and Management, Adickesallee 32-34, 60322 Frankfurt, Germany. ‡ maximilian.mueller@whu.edu; WHU – Otto Beisheim School of Management, Chair of Financial Reporting, Burgplatz 2, 56179 Vallendar, Germany.

1

1. Introduction

Financial reporting regulation is frequently motivated by the desire to strengthen the market-

wide information environment (e.g., Coffee, 1984; Dye, 1990; Easterbrook and Fischel, 1984; Fox,

1999; Leuz and Wysocki, 2008). The extent to which this end is achieved, however, is unclear

because regulated firms’ public disclosures may crowd out other market participants’ information

production (e.g., Goldstein and Yang, 2017). In particular, theoretical work by Admati and

Pfleiderer (2000) predicts that regulating the public disclosures of some firms can crowd out the

public disclosures of unregulated firms.

We test this prediction by Admati and Pfleiderer (2000), investigating how regulated firms’

mandatory disclosures affect unregulated firms’ voluntary disclosures.1 If regulated firms’

disclosures are useful for evaluating unregulated firms (information spillovers), we expect that

unregulated firms reduce their own disclosures, free-riding on the benefits conferred by regulated

firms’ disclosures and economizing on disclosure costs (e.g., proprietary costs) (e.g., Admati and

Pfleiderer, 2000; Baginski and Hinson, 2016). If, by contrast, regulated firms’ public disclosures

shift market participants’ attention toward the regulated firms (competition spillovers), we would

expect that unregulated firms increase their own disclosures, counteracting the loss of market

participants’ interest (e.g., Fishman and Hagerty, 1989; Grossman, 1981; Grossman and Hart, 1980;

Merton, 1987; Ross, 1977). Absent spillovers, we would not expect that regulated firms’ public

disclosures affect unregulated firms’ disclosures.

To test the effect of regulated firms’ disclosures on unregulated firms’ disclosures

empirically, we exploit the size-based financial reporting regulation applying to private limited

1 In line with the theoretical predictions of Admati and Pfleiderer (2000), the disclosure dimension we are interested in is the precision of information. In our empirical tests, we approximate the precision of information by the amount of information (i.e., number of characters in firms’ disclosures) (e.g., Chen et al., 2015).

2

liability firms in Germany. Under this regulation, firms are classified as “small” (“medium”) if they

fall beneath (exceed) any two of three regulatory thresholds related to firm (total assets:

approximately €5 million; sales: approximately €10 million; employees: 50) for two consecutive years.

“Small” firms need only disclose an unaudited and very simple balance sheet including short notes.

By contrast, “medium” firms must disclose an audited and detailed balance sheet, an income

statement, extended notes, and a management report. In our tests, we consider “small” firms

around the regulatory thresholds as effectively unregulated and “medium” firms around the

regulatory thresholds as effectively regulated. The idea behind this approach is that, as a result of

the discontinuous disclosure requirements, firms around the regulatory thresholds reporting as

“small” firms face low disclosure requirements relative to their actual firm size, whereas firms

around the same thresholds reporting as “medium” firms face high disclosure requirements relative

to their actual size.2

In our empirical tests, we then investigate whether unregulated (“small”) firms’ disclosures

are affected by spillovers from regulated (“medium”) firms’ disclosures. For this test, we need to

approximate unregulated firms’ counterfactual disclosures in a voluntary regime—a setting without

differential disclosure requirements. To this end, we use the disclosure behavior of a set of

benchmark firms representing firms disclosing in a voluntary regime. These benchmark firms are

the largest firms in our setting, whose firm sizes by far exceed the regulatory thresholds. We

consider these largest firms’ disclosures as unaffected by their own and other firms’ disclosure

requirements. For one, the largest firms have the greatest voluntary disclosure incentives, likely

2 Our empirical evidence supports the idea that, around the thresholds, “medium” firms are more constrained than “small” firms by the regulation. We find that a switch from the “small” to the “medium” category leads to a sizable increase in firms’ disclosures. We discuss and provide empirical evidence on these necessary conditions in Section 7.1.

3

exceeding their own minimum disclosure requirements. For another, the largest firms are unlikely to

benefit much from and free-ride on (comparably small) “medium” firms’ disclosures.3

From these largest firms, we derive a size-adjusted benchmark for “small” firms’

counterfactual disclosures in a voluntary regime. Our size adjustment takes into account that our

benchmark firms differ from the “small” firms in terms of firm size and, hence, disclosure

incentives. To adjust for size-related differences in public disclosure incentives, we scale firms’

disclosures (measured as the characters in firms’ filings) by the number of clicks received for their

filings on the official publication platform (similar to SEC Edgar in the United States). We refer to

this measure as the “disclosures-per-click ratio.” As the number of clicks approximates the number

of stakeholders interested in firms’ financial statements (Breuer et al., 2017), the disclosures-per-click

ratio essentially captures how much a firm discloses per interested stakeholder.4 The largest firms’

disclosure-per-click ratio, thus, constitutes a size-adjusted benchmark of firms’ disclosures (per

stakeholder) in a voluntary regime. We use this benchmark to approximate “small” firms’

disclosures in a voluntary regime. We essentially assume that, in a counterfactual voluntary regime,

unregulated “small” firms would provide the same disclosures per stakeholder as the largest firms. For

this to hold, the key identifying assumption underlying our approach is that firms’ voluntary

disclosures, on average, increase at an approximately constant rate with the number of their

stakeholders.

In our main test, we compare the disclosures-per-click ratios for “small” and “medium”

firms observed in the mandatory regime to our benchmark ratio for the voluntary regime. We refer

3 We discuss and provide empirical support for the necessary condition that our benchmark firms are widely unaffected by the disclosure requirements in Section 7.1. 4 We discuss the intuition underlying our approach in Section 5.2 and provide analytical examples in Section A “Analytical examples” in the online appendix. For now, we note that the idea that firms’ public disclosure net benefits increase in the number of stakeholders is consistent with the robust correlation between voluntary disclosure and firm size documented in prior literature (e.g., Allee and Yohn, 2009; Breuer et al., 2017; Lang and Lundholm, 1993).

4

to the difference between firms’ observed disclosures-per-click ratio and our benchmark ratio as

“abnormal disclosures.” If “medium” firms are directly affected by the regulation, we expect them

to disclose more (per stakeholder) than they would in a voluntary regime. Accordingly, we expect

“medium” firms’ observed disclosures-per-click ratio to exceed the benchmark ratio. If “small”

firms are indirectly affected by the regulation due to positive (negative) information spillovers from

“medium” firms, we expect “small” firms’ observed disclosures-per-click ratio to fall short of

(exceed) the benchmark ratio. For example, “small” firms could reduce the level of disaggregation

provided in their balance sheets or reduce additional information they would otherwise provide

voluntarily in their notes (such as depreciation schedules, supplemental information on the maturity

of their liabilities, or further information on their employees).5

Our main result supports regulated firms’ mandatory disclosure crowding out unregulated

firms’ voluntary disclosures. Consistent with “medium” firms around the regulatory threshold being

effectively regulated, we find that these firms provide around 21% more disclosures per stakeholder

than predicted in a voluntary regime (amounting to an increase of about 1.3 pages of information).

By contrast, we find that similar “small” firms exhibit negative abnormal disclosures around the

regulatory thresholds. In particular, we find that they provide 36% fewer disclosures per stakeholder

than predicted in a voluntary regime (amounting to a decrease of about 0.8 pages of information).

Importantly, this finding suggests that “small” firms exhibit lower net benefits of disclosure per

stakeholder in the presence of regulated firms compared to a voluntary regime, consistent with

regulated firms’ disclosures crowding out unregulated firms’ own disclosures (Admati and Pfleiderer,

2000).

5 For illustrative examples on the heterogeneity in unregulated (“small”) firms’ disclosures, see Section B “Heterogeneity in “small” firms’ mandatory disclosures” in the online appendix.

5

In cross-sectional tests, we document that the crowding-out effect varies with the strength of

the information spillover (Admati and Pfleiderer, 2000). We expect the spillover effect to be

stronger when firms’ fundamentals are more highly correlated, when there are more peers in the

same local industry, and when there are more mandatory disclosures provided by regulated peers in

the same local industry. Consistent with our expectations, we find that the gap between “small” and

“medium” firms’ abnormal disclosures around the regulatory threshold is greater in industries where

firms’ fundamentals are highly correlated. We further find that the disclosure gap is larger for local

industry peer groups in which firms have more peers and in which regulated peers provide more

mandatory disclosures.6 These cross-sectional tests allow us to relax the reliance on our disclosures-

per-click benchmark derived from the largest firms, allaying concerns that our results are merely due

to a flawed large-firm benchmark.

In supplemental tests, we validate the necessary conditions underlying our empirical

approach. For example, we document that “medium” firms are substantially more regulated than

“small” firms, and that the largest firms provide de facto voluntary disclosures. In placebo tests, we

further find that we cannot identify crowding-out effects in two alternative (“placebo”) settings

where either all firms are effectively unregulated or all firms are effectively regulated. These tests

suggest that our main findings are plausibly due to regulated firms’ disclosures crowding out

unregulated firms’ disclosures.

Our paper contributes to the literature on financial reporting regulation (for a review, see

Leuz and Wysocki, 2016), stressing that the effect of mandatory financial reporting on the market-

wide information environment is muted due to crowding-out effects. This sort of effect features 6 Our peer group definition at the local industry level is based on recent studies showing that firms share information commonalities when they operate in the same industry and geographic area (Engelberg et al., 2017; Ma, 2017). For our small and medium-sized sample firms, especially, the main input and output markets are likely rather local. Hence, information about economic conditions of suppliers and competitors in these local markets should provide the greatest information spillovers.

6

prominently in theoretical work on firms’ public disclosure in the presence of endogenous

information acquisition and provision of other market participants (e.g., Admati and Pfleiderer,

2000; Beyer et al., 2010; Goldstein and Yang, 2017; Kurlat and Veldkamp, 2015). This line of work

highlights that, as a result of crowding-out effects, financial reporting regulation can have an

ambiguous effect on the market-wide information environment and resource allocation efficiency.

Recent empirical work by Breuer et al. (2018) and Breuer (2017) likewise provides suggestive

evidence that financial reporting regulation deters banks’ private information acquisition and firms’

proprietary information generation. Extending this nascent literature, our paper provides direct

evidence about a further channel through which the effect of financial reporting regulation on

market-wide information is muted: regulated firms’ disclosures crowd out unregulated firms’

disclosures.

Our paper echoes recent evidence on crowding-out and displacement effects of regulation in

the economics literature (e.g., Crépon et al., 2013; Duguay et al., 2017; Rotemberg, 2017). Our work

suggests that financial reporting regulation benefits unregulated firms through information spillovers

of regulated firms (e.g., Badertscher et al., 2013), while imposing costs (e.g., proprietary costs) on

regulated firms. As unregulated and regulated firms tend to compete in the same product markets

(e.g., Bens et al., 2011), financial reporting regulation may contribute to the displacement of

regulated firms by unregulated firms, rather than improving aggregate outcomes. In this vein, recent

work by De Fontenay (2017) suggests that the bifurcated financial reporting regulation in the United

States (regulating public but not private firms) contribute to the declining attractiveness of U.S.

public capital markets (e.g., Chaplinsky et al., 2017; Doidge et al., 2017; Gao et al., 2013). Consistent

with this argument, Breuer (2017) documents that firms enter public capital markets at lower rates if

a greater share of firms in a given product market is exempted from financial reporting regulation.

7

The remainder of this paper is structured as follows. In Section 2, we briefly summarize the

related literature. In Section 3, we describe the theoretical predictions underlying our work, before

presenting the institutional background and our data in Section 4. We then discuss our identification

strategy in Section 5. In Sections 6 and 7, we present our main and supplemental results,

respectively. We conclude in Section 8.

2. Related Literature 2.1. Spillovers of financial reporting

Firms’ financial reporting can provide information relevant for peer firms (e.g., Foster, 1981;

Savor and Wilson, 2016). For example, Durnev and Mangen (2009), Beatty et al. (2013), and

Badertscher et al. (2013) document that firms’ financial reporting is used by peers in deciding on

their investing activities. In addition, Garmaise and Natividad (2016), Shroff et al. (2017), and

Berger et al. (2017) show that capital providers incorporate firms’ financial reporting in pricing peer

firms. These spillovers lower the estimation risk premium required for financing peer firms and

reduce the demand for peer firms’ own financial reporting. Consistent with reduced demand for

peer firms’ own financial reporting, Baginski and Hinson (2016) show that, in response to the

cessation of firms’ provision of voluntary management forecasts, peer firms start providing such

forecasts on their own.7

2.2. Financial reporting of private firms

Financial reporting of private firms in the United States is generally unregulated. Allee and

Yohn (2009) and Lisowsky and Minnis (2016) document that, in the absence of regulation, few U.S.

private firms produce (accrual-based) financial reports, obtain an audit, or provide public

7 Our paper differs from Baginski and Hinson (2016) in that we examine spillovers resulting from regulated firms’ mandatory disclosure of financial statements, whereas Baginski and Hinson (2016) focus on the disclosure of a specific item–management earnings forecasts–absent regulatory differences across firms. Conceptually, as discussed in Section 3, a voluntary equilibrium as the one documented by Baginski and Hinson (2016) presents the benchmark for our incremental regulatory effect.

8

disclosures. It is mainly larger firms and those seeking access to financing that tend to invest in their

financial reporting. Accordingly, the revealed preference of U.S. private firms suggests that

incentives (e.g., size and financing needs) matter for firms’ financial reporting and that, on average,

private firms’ financial reporting incentives are low. However, the weak financial reporting

incentives may, at least in part, reflect private firms’ ability to free-ride on information spillovers

provided by their public counterparts (e.g., Badertscher et al., 2013).

In contrast to the United States, financial reporting of private firms (in particular, limited

liability firms) in Europe is generally regulated. Although regulation constrains European private

firms’ financial reporting choices, Ball and Shivakumar (2005), Burgstahler et al. (2006), Bernard et

al. (2016), and Breuer et al. (2017) document that the (remaining) cross-sectional variation in

European private firms’ financial reporting is nevertheless mainly driven by firms’ size and financing

needs, similar to the U.S. private (and public) firm evidence.8 Thus European private firms’

observed financial reporting is not merely due to regulation. In this vein, Bernard (2016) and Breuer

et al. (2017) document that, during a low enforcement period, several firms, especially those with

numerous stakeholders (e.g., as approximated by online clicks on firms’ filings) and high financing

needs engaged in de facto voluntary public disclosure of their financial reports in Germany (see also

Shroff, 2016, p. 330).

Although private firms’ incentives matter and some private firms engage in voluntary public

disclosure even absent regulation, Bernard (2016), Bernard et al. (2018), and Breuer et al. (2018) also

document that the regulation, if enforced, strongly affects the financial reporting outcome (in

particular, public disclosure) of private firms around the regulatory thresholds. In particular, Breuer

et al. (2018) show that the greater disclosure requirements applying to “medium” firms starkly 8 U.S. public firms’ reporting choices are similarly constrained by reporting requirements. Nonetheless, a substantial body of research exploits variation within this group of regulated firms which originates from firms exceeding the minimum requirements, e.g., in terms of the level of disaggregation in their financial statements (e.g., Chen et al., 2015).

9

increase these firms’ public disclosures relative to the levels of otherwise similar “small” firms which

are subject to less stringent requirements. (For a discussion of the regulatory requirements, refer to

Section 4.) Exploiting this stark regulatory intervention, we test whether the strongly regulated

“medium” firms’ public disclosures crowd out otherwise similar “small” firms’ own public

disclosures.9

3. Economic theory and predictions

In a voluntary regime, a firm decides on its own disclosure by weighing benefits of reduced

adverse selection discounts against costs such as proprietary information loss (e.g., Admati and

Pfleiderer, 2000; Jung and Kwon, 1988; Verrecchia, 1983, 1990). The firm’s disclosure decision

takes peer firms’ disclosure decisions into account if peer firms’ disclosures provide correlated

information relevant for the firm and its stakeholders (and vice versa) (e.g., Admati and Pfleiderer,

2000; Dye, 1990).

Disclosures of financial statements provide correlated information allowing for direct and

specific cross-firm learning (e.g., Foster, 1981). In the presence of these information spillovers,

disclosures by peers reduce stakeholders’ uncertainty about the firm and hence the firm’s marginal

benefit of own disclosure. At the same time, the firm’s own disclosure is useful to peers and their

stakeholders increasing the firm’s proprietary costs and hurting its competitive position. Thus, own

and peer firm disclosures are substitutes in the presence of information spillovers.

Figure 1 presents the substitutive relationship between own and peer firm disclosures for the

case of two correlated firms—one regulated and one not—in two scenarios—a voluntary and the

9 This is similar to asking whether public firms’ disclosures crowd out otherwise similar private firms’ disclosures in the United States. In our setting, however, we argue that the starkest difference in disclosure requirements is between otherwise similar “small” and “medium” firms and not between private and public firms as is the case in the United States. Consistent with this argument, Badertscher et al. (2013) document that public firms provide substantial information spillovers to private firms in the United States, but not in the United Kingdom.

10

mandatory regime—following the disclosure game of Admati and Pfleiderer (2000). In the

voluntary regime, both firms take each other’s disclosure into account leading to an equilibrium in

which neither would prefer to increase or decrease its disclosure.

In the mandatory regime, the disclosure mandate pushes the regulated firm’s disclosure

above its disclosure in a voluntary regime. This allows the unregulated firm to reduce its disclosure

below its level in a voluntary regime for two reasons. First, the greater disclosure by peers reduces

stakeholders’ uncertainty about the firm and, hence, the firm’s benefit of own disclosure. Second,

the regulated peer firm cannot react to the unregulated firm’s reduction of own disclosure because

the regulated firm’s best response function is constrained from below. The result is an equilibrium

in which the unregulated firm can reduce its own disclosure and associated proprietary cost without

suffering from an increase in its stakeholders’ uncertainty that would arise if the regulated peer firm

could respond upon its disclosure reduction by reducing disclosures as well.

Based on the disclosure game, we predict that if spillovers from regulated firms are present,

unregulated firms reduce their disclosure in the mandatory regime below what they would have

provided in a voluntary regime. In the cross-section, we expect this substitutive relationship to be

stronger the greater the information spillovers.

Our cross-sectional prediction is based on two further results in Admati and Pfleiderer

(2000). First, we expect greater information spillovers for highly correlated firms. In this case, firms

and their stakeholders can learn more about their own fundamentals from peer firms’ disclosures.

Second, we expect greater information spillovers when the regulation leads to greater increases in

firms’ information environment. Specifically, we expect greater information spillovers when firms

have more peers. This follows from the finding in Admati and Pfleiderer (2000) that firms’

voluntary disclosure levels decrease as the number of their peers increases. With lower voluntary

11

disclosure levels, the additional information that regulated firms are required to disclose (relative to

their voluntary levels) becomes larger. As the direct effect of the regulation (on regulated firms)

increases in the number of peer firms, we expect the indirect effect (on unregulated firms) to

increase with the number of peer firms as well.

In contrast to our predicted negative correlation between a firm’s own and its peers’

disclosures in the presence of information spillovers, signaling theories would suggest a positive

correlation due to competition spillovers (e.g., Fishman and Hagerty, 1989; Grossman, 1981;

Grossman and Hart, 1980; Ross, 1977). For example, greater disclosures by one firm can increase

its visibility (e.g., Merton, 1987) and act as a quality signal (e.g., Hail et al., 2014; Leland and Pyle,

1977; Ross, 1977). Although this may benefit the disclosing firm through an increased number of

stakeholders (e.g., customers, suppliers, and investors) and a reduced adverse selection discount,

such competitive changes incentivize other firms to likewise increase their transparency to limit

market share (or investor attention) losses and differentiate from the remaining pool of lower quality

firms. Our prediction differs from the signaling prediction because we expect that the content of

firms’ signals is correlated (i.e., there are information spillovers). By contrast, firms’ signal contents

(but not their disclosure decisions) are independent in the signaling theories.

4. Institutional background and data

Our empirical setting is the EU disclosure regulation for limited liability firms as

implemented in Germany. The regulation requires distinct minimum levels of disclosure for three

firm size groups (“small”, “medium”, and “large”). Table A.1 Panel B summarizes the disclosure

requirements for each group. “Small” firms must disclose an unaudited, highly aggregated balance

sheet with brief notes only. By contrast, “medium” firms must provide audited financial statements

including a disaggregated balance sheet, an income statement, extended notes, and a management

12

discussion. “Large” firms must additionally disclose a number of further line items and notes.

Across all size groups, firms must publicly disclose their financial statements via a central repository,

the Federal Gazette, which is similar to EDGAR in the United States.

Our focus is on “small” and “medium” firms because the difference in disclosure

requirements is starkest between these two groups. Firms are classified into either regulatory size

group based on three size criteria. They are classified as (“small”) “medium” if they (do not) exceed

any two out of three firm size thresholds—related to total assets (€4.84 million), sales

(€9.68 million), and employees (50)—for two consecutive years (cf. Panel A for the medium-large

threshold values).

In our main tests, we use data on the disclosures by a comprehensive set of German firms

between 2006 and 2011. Importantly, during our sample period, firms’ must publish their

disclosures online via the Federal Gazette (Bundesanzeiger), and firms’ compliance with disclosure

requirements is close to perfect. The former ensures the public dissemination of firms’ disclosures

to all interested stakeholders and allows tracking stakeholders’ demand for such disclosures (via

online click statistics).10 The latter implies that the stark differences in disclosure requirements

between “small” and “medium” firms are actually enforced.

We obtain data about firms’ disclosure requirements (e.g., size group) and behavior (e.g.,

number of characters in a filing) for fiscal year-ends from 2006 to 2011, and stakeholders’ disclosure

demand (aggregated clicks) for filings published between December 2010 and February 2013 (i.e.,

10 The Federal Gazette is the only comprehensive and public platform providing the firms’ regulatory filings. This ensures that we capture a comprehensive measure of (online) stakeholder demand. This also pertains to demand for information of the largest firms as these are still mostly private firms which do not usually employ investor relation departments or provide financial statements on their company websites. Although this ensures an arguably comprehensive measure of stakeholder demand, we stress that our identification approach does not hinge on such a comprehensive measure. In particular, our approach uses the relative number of characters to measured clicks (not total demand). Hence, our approach merely requires that the proportion of stakeholder demand captured by online clicks does not vary starkly across firm sizes.

13

fiscal year-ends 2008 to 2011) from the Federal Gazette. We enrich this data with financial data

about our sample firms from Bureau van Dijk’s dafne database. A full list of our variables, their

definitions, and corresponding sources is provided in the variable appendix. Table A.2 provides

descriptive statistics.

5. Identification

5.1. Approach

We hypothesize that regulated firms’ disclosures crowd out unregulated firms’ disclosures.

To test this hypothesis, we exploit that “medium” firms’ minimum disclosure requirements are

substantially greater than those of otherwise similar “small” firms. Around the regulatory

thresholds, we consider “medium” firms as “regulated,” because we expect that their disclosure

requirements, on average, exceed their voluntary disclosures. By contrast, we consider otherwise

similar “small” firms around the regulatory thresholds as “unregulated,” because we expect that their

disclosure requirements, on average, are not binding and, therefore, fall short of their voluntary

disclosures.11 (We summarize and provide empirical support for the necessary conditions underlying

our approach in Section 7.1)

In testing our hypothesis, we face the following identification challenge: We want to

compare unregulated (“small”) firms’ disclosures in the presence of regulated firms’ mandatory

disclosures (mandatory regime) with unregulated (“small”) firms’ disclosures in the absence of

regulated firms’ mandatory disclosures (voluntary regime). At any given time, however, we only

observe unregulated (“small”) firms’ disclosures in one of these regimes. In our case, we only

11 Importantly, this requirement relates to the intensive margin of firms’ disclosures. We do not require that “small” firms would voluntarily provide public disclosures absent their disclosure requirements. We merely require that, conditional on a mandate to disclose at least some basic information (i.e., abstracting from the extensive margin), firms’ disclosure incentives along the intensive margin exceed the required minimum level of disclosures. (See Breuer et al. (2017) for a discussion of the extensive versus intensive margin of firms’ public disclosure decisions.)

14

observe unregulated (“small”) firms’ disclosures in a mandatory regime. Thus, we need to

approximate the counterfactual disclosures of these firms in a voluntary regime.

To uncover unregulated (“small”) firms’ disclosures in a voluntary regime, we use other

unregulated firms’ disclosures observed in a de facto voluntary regime. In particular, we consider the

largest firms in our setting as unaffected by their own and other firms’ regulatory minimum

disclosure requirements (including “medium” firms’ requirements). The idea is that the largest firms

exhibit substantial voluntary disclosure incentives that exceed their minimum disclosure

requirements (e.g., Breuer et al., 2017; Lang and Lundholm, 1993). Moreover, the largest firms tend

to provide information spillovers to other (“small” and “medium”) firms rather than receive

substantial information spillovers from smaller firms’ disclosures (e.g., Arif and De George, 2016;

Bonsall et al., 2013). Accordingly, the relationship between disclosure incentives and disclosure

outcomes of these largest firms should be widely unaffected by the regulation, making them a

natural benchmark for unregulated firms’ disclosures in a voluntary regime.

The largest firms, however, differ substantially from our unregulated (“small”) firms in terms

of size. Accordingly, we do not use the level of the largest firms’ disclosures as the benchmark for

unregulated firms’ disclosures in a voluntary regime. Instead, we use the largest firms’ average

disclosures per click ( /L LQ N , where Q denotes the disclosure quantity, N denotes the number of

clicks, and the subscript L refers to the largest firms) as a size-adjusted benchmark. (We discuss the

economic motivation and interpretation of this ratio in Section 5.2.) We essentially test whether

unregulated (“small”) firms in a mandatory regime disclose, on average, less per click than predicted

by the disclosures-per-click ratio of unregulated firms in a voluntary regime (i.e., the largest firms).

In accordance with this logic, we can approximate the counterfactual level of (unregulated)

firms’ disclosures in a voluntary regime by multiplying the disclosures-per-click ratio observed for

15

the largest firms ( /L LQ N ) with the number of clicks observed for (unregulated) firms in a

mandatory regime ( MN , where the superscript M refers to the mandatory regime):

( )ˆ /V ML LQ Q N N= , (1)

where ˆVQ denotes the predicted disclosure quantity ( Q̂ ) in a voluntary regime (superscript V ).

The following example illustrates our approach. Our benchmark firms disclose, on average,

about 19,000 characters ( LQ ) and receive about 52 clicks ( LN ), so their disclosures-per-click ratio is

365 ( /L LQ N ).12 Hence, we predict that an unregulated firm having received, for example, eight

clicks (

MN ), would disclose 2,920 characters in a voluntary regime ( ( ) 365 8V ML LQ̂ Q / N N= = × ).

In our empirical tests, we compare these predicted disclosures with the unregulated firm’s observed

disclosures (in a mandatory regime).

5.2. Interpretation

Our empirical approach to approximating unregulated firms’ counterfactual disclosures in a

voluntary regime via the disclosures-per-click ratio permits an accessible economic interpretation.

Intuitively, the disclosures-per-click ratio captures the (average) net disclosure benefit per

stakeholder interested in firms’ financial statements. We expect that spillovers from regulated firms’

mandatory disclosures affect unregulated firms’ net disclosure benefits per stakeholder by lowering

the (net) benefits of unregulated firms’ own disclosures. The lower (net) benefits of unregulated

firms’ own disclosures in the mandatory regime relative to the counterfactual voluntary regime

would manifest in a lower disclosures-per-click ratio in the mandatory regime than predicted for the

voluntary regime.

12 As benchmark firms, we use all firms in a given size range far above the regulatory threshold (with total assets of about €66 million, corresponding to a logarithm of 18; see also Figure 2). By doing so, we select our benchmark firms based on their disclosure incentives (varying with firm size, measured by total assets), but do not restrict them to a particular regulatory size class (determined by multiple firm size dimensions and over two years).

16

Our interpretation is based on the notion that firms weigh benefits and costs in deciding on

their disclosures in a voluntary regime. Firms’ disclosure benefits arise from reduced information

frictions vis-à-vis stakeholders transacting with them. For example, firms may benefit from more

favorable payment terms granted by suppliers as a result of more extensive disclosures. Firms’

disclosure costs comprise direct production and publication costs as well as indirect costs such as

the cost of proprietary information loss associated with more extensive disclosures. As firms

provide one level of public disclosure to all stakeholders, their disclosure benefits increase in the

number of stakeholders transacting with them, whereas their costs do not or at least do so to a lesser

extent (For a detailed discussion, see Breuer et al., 2017). Accordingly, the breadth of firms’

transacting stakeholders is a chief (and approximately linear) determinant of firms’ total net benefit

of disclosure and, thus, their public disclosures.13

We use this insight to adjust for firm-size-related differences in disclosure incentives and

levels between our unregulated firms of interest and our benchmark comprised of the largest firms,

normalizing firms’ disclosures with their number of transacting stakeholders. Although we do not

directly observe the number of transacting stakeholders, we rely on a closely related empirical proxy:

the number of online clicks.

Clicks capture the number of all stakeholders interested in firms’ public disclosures. These

stakeholders comprise transacting stakeholders, but also non-transacting ones who, if anything, have

a negative effect on firms’ net disclosure benefits (e.g., neighbors or competitors causing privacy or

proprietary costs). Importantly, the clicks from non-transacting stakeholders do not invalidate our

use of the disclosure-per-click ratio to capture firms’ average net disclosure benefit per stakeholder.

The inclusion of non-transacting stakeholders in our deflator decreases our ratio by inflating the 13 The non-rivalry property of public goods such as public disclosure suggests that the benefit per average stakeholder transacting with the firm is approximately constant. Thus, the (net) benefit of disclosure increases approximately linearly with transacting stakeholders. In the online appendix, we provide a stylized analytical example illustrating this point.

17

denominator (clicks) and potentially reducing the nominator (disclosure) relative to the “true” ratio

of “net disclosure benefits” per transacting stakeholder. However, we are not concerned with

measuring this “true” ratio of “net disclosure benefits” per transacting stakeholders. Instead, we aim

to compare disclosures-per-click ratios across firms. Thus, the fact that our clicks also capture non-

transacting stakeholders (e.g., the interested public) is not a problem so long as the relative

composition of transacting and non-transacting stakeholders contributing to the clicks does not

differ significantly across firm sizes.14

5.3. Specification

We define the difference between firms’ disclosures in a mandatory regime and their

counterfactual disclosures in a voluntary regime as “abnormal” disclosures:

ˆM VQ Q− . (2)

In our first test, we investigate the average of firms’ abnormal disclosures across the firm-

size distribution and focus on otherwise similar regulated and unregulated firms just around the

small-medium threshold. We calculate the average abnormal disclosures conditional on firm size

using a smoothed kernel estimator. We expect that regulated firms (to the right of the threshold)

exhibit positive abnormal disclosures, because the regulatory requirements push these firms’

disclosures above their disclosures in a voluntary regime. By contrast, we expect that unregulated

firms (to the left of the threshold) exhibit negative abnormal disclosures, because their own

disclosures in the presence of regulated firms’ spillovers are less (net) beneficial than in a voluntary

regime.

14 Breuer et al. (2017) provide empirical evidence that competitor-related clicks do not dominate the total observed variation in the number of clicks. Consistent with this finding, they document that controlling for competition proxies does not notably affect the strongly positive linear relation between the number of clicks and firms’ disclosures.

18

The key benefit of this non-parametric approach is that we obtain an estimate of the average

impact of any (regulated) peers’ spillovers on unregulated firms’ disclosure for a given firm size. In

particular, we do not need to pre-specify which firms are (regulated) peers providing spillovers and

which are not. In essence, our estimates capture the average impact of all possible (regulated) peers

on unregulated firms’ disclosures without imposing any assumptions on the correlation structure

among firms. The key drawback of this parsimonious approach is that we rely on the identifying

assumption that the disclosures-per-click ratio of the largest firms provides an accurate benchmark

for our unregulated firms’ disclosures absent spillovers from regulated peers’ disclosures. That is,

we assume that our benchmark disclosure-per-click ratio is size invariant.

Our assumption implies that the marginal (net) benefit per click equals the average (net)

benefit per click, rendering the disclosures-per-click ratio independent of firms’ scale (e.g., firms’ size

and their number of clicks). If anything, we expect that firms’ marginal (net) benefit of an additional

click decreases in the number of clicks (e.g., because the additional clicks disproportionally stem

from the interested public rather than suppliers or creditors). This would imply that the largest

firms’ disclosures-per-click ratio underestimates unregulated (“small”) firms’ disclosure incentives in

a voluntary regime, biasing against finding support for greater disclosures in a voluntary than a

mandatory regime.15 (For a corresponding derivation, refer to Section A “Analytical Example” in

the online appendix.)

Although we regard our identifying assumption as plausible, we relax the reliance on our

disclosures-per-click benchmark in further tests. In particular, we test whether the difference in

15 Although we cannot exactly test our identifying assumption, we can gage its plausibility by estimating the correlation of firms’ (mandatory) disclosures-per-click ratio with firm size (total assets). While (mandatory) disclosures and clicks increase strongly in firm size, in Table A.3 in the online appendix, we find that firms’ (mandatory) disclosures-per-click ratio, if anything, appears to decrease in firm size. Similarly, firm size explains little variation of the (mandatory) disclosures-per-click ratio. These results should provide some comfort that our approach to adjusting for scale differences by using firms’ disclosures-per-click ratios appears to work reasonably well.

19

disclosure outcomes between unregulated and regulated firms of the same size varies with the

expected strength of the information spillovers. In this vein, we estimate the following specification

via ordinary least squares:

, , 1 , 2 , , , , , , ,( ) ( )M Vi t i t i t i t c j t c j t i t i tLog Q Log Q Unregulated Unregulated Spillover fβ β a f e− = + × + + + , (3)

where i , t , c , and j denote the firm, year, county, and industry classification, respectively. Our

dependent variable, M Vi ,t i ,tLog( Q ) Log( Q )− , is a firm’s abnormal disclosure in a given year (defined as

the difference between the logarithms of its mandatory and voluntarily disclosure). i ,tUnregulated

is an indicator variable taking the value of one for firms classified as “small”, and zero for firms

classified as “medium” in a given year. c , j ,tSpillover is a proxy for the expected strength of

information spillovers within a peer group. c , j ,ta represents the fixed effect for a given county-

industry-year combination. i ,tf is a control function including (log) firm sizes (total assets, sales, and

employees) centered at the regulatory threshold and (log) firm age. i ,te is the error term.

1β captures the difference in regulated and unregulated firms’ abnormal disclosures for firms

with no information spillovers. 2β , our coefficient of interest, captures the incremental impact of

the expected strength of the spillover effect in a given county, industry, and year on the difference in

regulated and unregulated firms’ disclosure outcomes. If the regulation has positive information

spillovers, we expect 2β to be negative as unregulated firms reduce their abnormal disclosures

relative to their regulated peers.

To capture variation in the strength of information spillovers, we employ three proxies

based on the analytical predictions in Admati and Pfleiderer (2000) (as described in Section 3). Our

first proxy, 2 jHigh _ R , reflects the strength of comovement in firms’ fundamentals in a given

20

industry. Following Guiso and Parigi (1999) and Badertscher et al. (2013), we first obtain the R-

squared from industry-specific regressions of firms’ standardized asset growth on year-fixed effects.

This measure reflects how much of firms’ variation in year-over-year asset growth can be explained

by factors shared by all industry peers.16 We then define 2 jHigh _ R to be one for industries in the

top quartile of the distribution of our R-squared measure, and zero otherwise.

Our further proxies capture variation in firms’ information environment stemming from

their regulated peers. Our second proxy, c , j ,tPeers , reflects that regulatory spillovers and associated

free-rider concerns increase in the number of peers. We define peer groups at the local industry

level and calculate c , j ,tPeers as the (logarithm of) the number of firms operating in the same county,

industry, and year. This peer group definition is based on prior literature documenting information

commonalities among firms operating in local industry clusters (Engelberg et al., 2017; Ma, 2017)

and takes into account that our private sample firms likely compete in local product and labor

markets. Our third proxy, c , j ,tAggr _ Discl , captures the total amount of information disclosed by

regulated peers (Shroff et al., 2017). We define c , j ,tAggr _ Discl as the logarithm of the total

number of characters disclosed by regulated firms in a given county, industry, and year.

The key benefit of our cross-sectional approach is that it relaxes the reliance on the

disclosure-per-click benchmark by focusing on the difference between regulated and unregulated

firms as a function of peer firms. As both, regulated and unregulated firms’ abnormal disclosures

are determined relative to the largest firms’ disclosures-per-click benchmark, taking the difference

16 Our industry-specific R-Squared is correlated with the number of firms operating in a given industry. In particular, industries with only few firms will exhibit a higher R-squared. To purge the variation in R-squared of variation in the number of peer firms (the role of which we investigate separately using our second proxy), we residualize our R-squared measure with respect to industry size (measured by the number of firms in a given industry).

21

between regulated and unregulated firms’ abnormal disclosures cancels the benchmark out. Hence,

we do not rely on the specific benchmark ratio derived from the largest firms being correct.

The key drawback of this specification is that we need to pre-specify a peer group (i.e.,

assume a correlation structure) and only capture relative, rather than absolute, differences in

disclosure due to spillovers. By defining unregulated firms’ peers as the firms within the same

industry and county in the same year, for example, we assume that these within-county-industry

firms are, on average, more strongly correlated with the unregulated firms than other firms would

be. Although we expect this assumption to hold on average, we note that any other spillovers not

originating from within-county-industry firms are neglected in this design.17

6. Results

6.1. Spillover effects

Figure 2 presents firms’ disclosures in the mandatory vis-à-vis a voluntary regime across firm

sizes. In the right tail, observed disclosures in the mandatory regime are generally undistinguishable

from predicted disclosures in the voluntary regime for a wide range of firm sizes among the larger

firms. In part, this pattern emerges by construction because we use larger firms’ disclosure patterns

as benchmarks for disclosures in a voluntary regime (i.e., scaled at total assets of about €66 million,

corresponding to a logarithm of 18 on the x-axis in Figure 2). Yet, observed and predicted

disclosures continue to overlap even for large firms of smaller size than our benchmark firms (with

the firm size region of the overlap approximately extending to firm sizes about €15 million total

assets, corresponding to a logarithm of 16.5 on the x-axis in Figure 2). These patterns support the

17 More precisely, any information spillovers extending across our pre-specified county-industry combinations are implicitly captured by the “control group” (i.e., other county-industry combinations), thereby reducing our effect sizes (due to a violation of the stable unit treatment value assumption, e.g., Glaeser and Guay, 2017; Leuz and Wysocki, 2016; Rubin, 1974).

22

validity of our benchmark: for firms presumably unaffected by regulatory information spillovers, the

disclosures-per-click ratio does not appear to strongly vary with firm size.

Another pattern supporting the validity of our benchmark is found in the left tail, among the

smallest firms. According to our benchmark, the smallest firms’ public disclosures in the mandatory

regime exceed those expected in a voluntary regime. This pattern is consistent with prior evidence

that the vast majority of the smallest firms, in a prior period with low enforcement, did not make de

facto voluntary disclosures (e.g., Bernard, 2016; Breuer et al., 2017). Moreover, this pattern is

consistent with regulatory action taken in 2012, after our sample period, exempting the smallest

firms (below the micro-small threshold) from “excessive” financial reporting requirements (e.g.,

Breuer et al., 2017; European Commission, 2011).

Turning to the patterns around the small-medium threshold, we find that disclosures in the

mandatory regime significantly deviate from disclosures in the voluntary regime around this

threshold. In particular, firms just above the threshold—predominantly regulated firms—exhibit on

average slightly greater disclosures in the mandatory than in the voluntary regime. By contrast, firms

just below the threshold—predominantly unregulated firms—exhibit on average lower disclosures in

the mandatory than in the voluntary regime. These patterns are confirmed in a graph of abnormal

disclosures across firm sizes (Figure 3) and consistent with crowding-out effects due to information

spillovers.18

To further explore whether the differences between firms’ disclosures in the mandatory and

the voluntary regime are driven by regulation, Figure 4 provides abnormal disclosures separately for

regulated and unregulated firms around the small-medium threshold. Regulated firms at the small-

18 In our sample, firms are assigned to the regulated group (i.e., classified as “medium”) when they exceed two out of three size criteria over two years. Accordingly, there is some overlap of regulated and unregulated firms with similar total assets, and the local averages depicted in our figure compose observations from both regulated and unregulated firms. With average firm size increasing above the total asset threshold, the share of regulated firms increases.

23

medium threshold exhibit abnormal disclosures of about 4,000 characters or around 1.3 pages of

information (21% of their voluntary disclosure levels), suggesting that the disclosure mandate indeed

pushes their disclosures in the mandatory regime above their disclosures in the voluntary regime.19

By contrast, unregulated firms at the small-medium threshold exhibit abnormal disclosures of about

-2,500 characters or -0.8 pages of information (-36% of their voluntary disclosure levels), suggesting

that they reduce their disclosures in the mandatory regime relative to the disclosures provided in the

voluntary regime. Consistent with our prediction, these patterns suggest that firms reduce their own

disclosures when other firms are forced to increase their disclosures. Similar patterns emerge for

alternative disclosure measures (i.e., publication lag and voluntary disclosures of sales information),

supporting our conclusion (Figure A.3).

We highlight that the documented patterns are unlikely to be unduly confounded by

selection of firms around the small-medium threshold (“selection effect”). Prior literature

documents that firms with low net benefits of additional disclosures (e.g., those with high

proprietary costs) try to avoid additional disclosure requirements by manipulating their size (e.g.,

Bernard et al., 2018; Gao et al., 2009). Hence, firms just before the threshold could exhibit low

disclosure levels due to sorting based on, e.g., proprietary cost of disclosure, rather than spillover

effects. Although this is a valid concern, two pieces of evidence refute that the selection effect

(primarily) accounts for our results.

First, we document that unregulated firms over an extended range (from €1.2 million up to

the small-medium threshold at €4.84 million of total assets) before the small-medium threshold

provide less disclosures than expected in a voluntary regime. It is unlikely that this pattern can be

explained by local selection behavior around the threshold. Second, Breuer et al. (2018, Table 3)

19 A page usually contains around 3,000 characters.

24

document that regulated and unregulated firms just around the threshold do not exhibit substantial

covariate imbalances after accounting for the three regulatory size dimensions (total assets, sales, and

employees). This suggests that selection—even right around the threshold—should not be a major

concern.

6.2. Cross-sectional variation in strength of information spillovers

The preceding analysis suggests that unregulated firms reduce their disclosures in the

mandatory regime compared to the voluntary regime, whereas regulated firms are pushed above

their disclosure levels of the voluntary regime due to the regulation. As a result, there is a gap

between regulated and unregulated firms’ abnormal disclosures. If this gap is due to information

spillovers, we expect unregulated firms to widen the disclosure gap further when information

spillovers are stronger. Based on the analytical results of Admati and Pfleiderer (2000), we expect

unregulated firms’ to widen (decrease) the gap (their disclosures) more, compared to regulated firms,

when firms’ fundamentals are more highly correlated, when there are more peers, and when there

are more mandatory disclosures made by regulated peer firms.

Table 1 presents our cross-sectional results. In Panel A, we show how the disclosure gap

between regulated and unregulated firms varies with respect the strength of the comovement in

firms’ fundamentals. We find that the difference between regulated and unregulated firms’

abnormal disclosures is larger in industries with a high comovement in asset growth, as indicated by

the negative and significant coefficient on 2i ,t jUnregulated High _ R× (Column 1 of Panel A). This

finding supports the notion that firms and their stakeholders can learn more from peers’ disclosures,

and hence benefit less from firms’ own disclosures, when firms’ fundamentals are highly correlated

(Admati and Pfleiderer, 2000, p. 499).

25

In Panel B, we show how the disclosure gap varies with strength of the peer information

environment. In a voluntary regime, firms’ disclosure decrease in the number of peers (Admati and

Pfleiderer, 2000, pp. 500-503). Accordingly, the regulation pushes regulated firms more strongly

above their voluntary level when firms have more peers, leading to greater spillover effects.

Consistent with this prediction, we find that the abnormal disclosure gap between regulated and

unregulated firms increases with the number of peers, as reflected by the negative and significant

coefficient on i ,t c , j ,tUnregulated Peers× (Column 1 of Panel B). Similarly, we find that the

abnormal disclosure gap increases when regulated peers provide, in total, greater mandatory

disclosures (Column 2 of Panel B). These findings are consistent with unregulated firms reducing

their disclosures to a greater extent when their regulated peers’ disclosures enhance the market-wide

information environment.20

We find similar patterns in the gap between regulated and unregulated firms’ mandatory

(“raw”) disclosures. Compared to regulated firms, unregulated firms reduce their mandatory

disclosures more when firms are highly correlated (Column 2 of Panel A), when they have more

peers (Column 3 of Panel B), and when their regulated peers provide, in total, greater disclosures

(Column 4 of Panel B). These results suggest that the widening of the abnormal disclosure gap is

not merely driven by higher disclosure demand for unregulated firms’ disclosures (e.g., due to more

peers), but rather due to unregulated firms adjusting their actual disclosures downward when their

regulated peers disclose more information.

20 We assess the robustness of our results with respect to different peer group definitions and aggregated disclosure measures. We report these additional specifications in Table A.4 in the online appendix. Our inferences remain unchanged when we define peer groups at the industry level (rather than county-industry), when we use the number of regulated peers (rather than all peers), and when we use regulated peers’ aggregated abnormal disclosures (rather than mandatory disclosures) as cross-sectional variables. The latter finding is consistent with the idea that unregulated firms reduce their disclosures more when the regulation pushes their peers more strongly above their voluntary levels.

26

Finally, we document that unregulated firms increase the disclosure gap measured in terms

of two alternative voluntary disclosure proxies when spillovers from regulated firms are stronger. In

particular, we show that unregulated firms—relative to regulated firms—delay the publication of

their disclosures (Litjens and Suijs, 2014) and are less likely to make voluntary sales disclosures

(Dedman and Lennox, 2009) when firms are highly correlated (Columns 5–6 of Panel A) and when

they operate in a richer information environment due to their regulated peers (Columns 7–8 of

Panel B). Notably, these alternative voluntary disclosure proxies—unlike the number of

characters—are not directly affected by differential disclosure requirements between regulated and

unregulated firms.

Taken together, our cross-sectional patterns allay concerns that our findings are primarily

due to both, regulated and unregulated firms merely supplying the minimum disclosure levels

prescribed in our setting (“mechanical effect”). Consistent with our main findings, such a

mechanical effect would suggest that we would observe lower disclosure levels for unregulated than

for regulated firms. The mechanical effect, however, suggests that such differences are due to a

direct effect of differential disclosure requirements rather than an indirect crowding-out/spillover

effect. Inconsistent with the mechanical effect, but consistent with the crowding-out/spillover

effect, the cross-sectional evidence suggests that our main effect varies with proxies for the spillover

strength and holds for alternative disclosure outcomes not directly affected by minimum disclosure

requirements.21

21 Our cross-sectional results do not ipso facto establish interactions between firms’ disclosure behaviors. Our cross-sectional tests investigate how the difference between regulated and unregulated firms varies with the spillover strength. This difference, however, comingles the regulations’ direct effect (e.g., on the regulated firm) with its indirect effect on unregulated firms. Hence, we cannot directly support our main prediction with our cross-sectional specifications. Instead, our cross-sectional tests complement our main result by showing that it is not merely due to a mechanical effect.

27

7. Supplemental results

7.1. Necessary conditions

In this section, we briefly discuss the four (main) necessary conditions (NCs) underlying our

research design and test their empirically validity. Although we provide evidence in support of each

of these necessary conditions, we also stress that violations of our conditions would generally work

against finding evidence supporting our hypothesis with our research design.

NC 1: Around the regulatory thresholds, “medium” firms’ disclosures are, on average,

regulated; that is, “medium” firms’ disclosure requirements exceed their voluntary disclosures.

If this condition were violated, “medium” firms’ disclosures would not be constrained by

their disclosure requirements, preventing “small” firms from free-riding on regulated firms’

mandatory disclosures.22

Our graphical evidence discussed in Section 6.1 already suggests that “medium” firms

around the small-medium threshold appear to exhibit disclosures in a voluntary regime that fall short

of their disclosures in the mandatory regime (Figure 4). We provide further evidence on the

constraining effect of the regulation on firms’ disclosures by investigating disclosure changes

observed for firms switching from the “small” to the “medium” regulatory size class. Column 1 of

Table 2 Panel B documents that a switch from the “small” to the “medium” size class increases

firms’ disclosures by about 116% ( exp(0.772) 1 116%− ≈ ). This evidence strongly supports that

“medium” firms’ disclosures are pushed above their voluntary levels by their disclosure

requirements.

22 To illustrate this, we investigate an alternative (placebo) setting in which all firms, not just the less regulated or unregulated ones, exhibit voluntary disclosures exceeding the respective disclosure requirements in the mandatory regime: firms around the medium-large threshold (Section 7.2).

28

NC 2: At least some “small” firms’ disclosures are unregulated; that is, these “small” firms’

voluntary disclosures exceed their minimum disclosure requirements.

If this condition were violated, all “small” firms would merely provide the required

minimum in the mandatory regime, preventing us from detecting incremental effects of “medium”

firms’ mandatory disclosures on “small” firms’ disclosures in a mandatory regime.23

Although we cannot directly test this condition, we note that our main results support its

validity. For one, Figure 4 documents that, conditional on our identifying assumption, “small’ firms

around the small-medium threshold appear to exhibit disclosures in a voluntary regime that exceed

their disclosures (and requirements) in the mandatory regime. For another, our cross-sectional

results in Table 1 document that “small” firms’ disclosures in the mandatory regime vary predictably

with the strength of information spillovers. We would not observe such cross-sectional variation if

all “small” firms were only providing the uniform minimum required disclosures (see also Section

7.2). In sum, our main results support the validity of our second necessary condition.

NC 3: The largest firms’ disclosures are generally unconstrained by their minimum

disclosure requirements and unaffected by spillovers from other (“medium”) firms’ disclosure

requirements.

If this condition were violated, the largest firms’ disclosures would not constitute de facto

voluntary disclosures. If the largest firms’ observed disclosures were pushed above their voluntary

levels through their own disclosure requirements, we would overstate “small” firms’ disclosures in a

voluntary regime. If, by contrast, the largest firms’ observed disclosures were reduced through

23 To illustrate this, we investigate an alternative (placebo) setting in which all firms, not just the more regulated ones, exhibit voluntary disclosures below the respective disclosure requirements in the mandatory regime: firms around the micro-small threshold (Section 7.2).

29

spillovers from other (“medium”) firms’ disclosure requirements, we would understate “small”

firms’ disclosures in a voluntary regime.

To test whether the largest firms’ disclosures are constrained by the regulation, we

investigate disclosure changes observed for firms switching from the “medium” to the “large”

regulatory size class. Column 2 of Table 2 documents that this switch does not significantly increase

and, in fact, slightly decreases firms’ disclosures by an immaterial 1.5%. This evidence suggests that

the disclosures observed for the largest firms exceed their minimum disclosure requirements,

consistent with our third necessary condition. Importantly, this evidence alleviates concerns that we

overstate unregulated (“small”) firms’ disclosure incentives in a voluntary regime by using the largest

firms’ disclosures-to-click ratio as a benchmark.

NC 4: Firms’ number of clicks is not substantially affected by firms’ own or other firms’

minimum disclosure requirements.

If this condition were violated, we could not use the observed number of clicks in the

mandatory regime as a proxy for firms’ disclosure incentives in a voluntary regime. A potential

concern is that the number of clicks observed for unregulated (“small”) firms’ disclosures in a

mandatory regime could be reduced by informational spillovers from regulated (“medium”) firms.

Thus, we would underestimate unregulated (“small”) firms’ disclosures in a voluntary regime by

multiplying the disclosures-to-click ratio with a too low number of clicks. Similarly, we would

overestimate regulated (“medium”) firms’ disclosures in a voluntary regime if the more extensive

disclosure requirements applying to these firms would per se attract more clicks (e.g., by shifting

stakeholders’ attention). (Note that these patterns, if anything, would work against our findings.)

As described in Section 6.1., our main results show that the number of clicks smoothly

increases in firm size, which does not support the idea that the disclosures of unregulated firms

30

around the threshold are affected by regulated firms’ disclosures.24 We further test for the influence

of firms’ own minimum disclosure requirements on the number of clicks by investigating changes in

the number of clicks for firms switching from the “small” to the “medium” regulatory size class.

Column 3 of Table 2 Panel B documents that the number of clicks increases only marginally for

firms switching into the “medium” mandate ( exp(0.034) 1 3%- ≈ ; elasticity with respect to

disclosure changes: 3% 116% 0.03/ ≈ ). Consistent with our fourth necessary condition, this

evidence suggests that the number of clicks is generally unaffected by direct effects of minimum

disclosure requirements.

Notably, however, this evidence does not suggest that regulated (“medium”) firms’

disclosures do not affect unregulated (“small”) firms’ disclosure “demand.” It merely suggests that

the number of interested stakeholders is unaffected. It does not imply that the (net) benefit per

stakeholder (as measured by the disclosures-per-click) remains unaffected. To the contrary, our

main results documents that regulated (“medium”) firms’ disclosures reduce unregulated (“small”)

firms’ (net) benefit per stakeholder (as measured by characters provided per click).

7.2. Placebo tests

Our main analysis focuses on firms around the small-medium threshold. We argue that this

is a feasible setting to examine our research question because of the stark difference in mandatory

minimum requirements applying to “small” and “medium” firms. While this differential regulation

constrains “medium” firms’ disclosures from below, “small” firms around the threshold are, on

average, unconstrained in that their voluntary disclosures exceed the low minimum requirements

(satisfying NC 1 and NC 2).

24 Figure A.1 in the online appendix further supports the idea that the number of clicks is largely unaffected by informational spillovers. The figure shows that the smooth increase in the number of clicks along the firm-size dimension closely resembles voluntary disclosure rates observed during a low-enforcement period.

31

In this section, we provide placebo tests exploiting two alternative thresholds: the micro-

small threshold and the medium-large threshold. Around these two thresholds, all firms are either

constrained or unconstrained by the regulation. Accordingly, these thresholds fail to provide for a

setting where some, but not all firms are forced to disclose more than they would in a voluntary

regime. We neither expect to find nor find evidence that “regulated” firms’ disclosures crowd out

“unregulated” firms’ disclosures in these placebo settings.

A constrained setting: the micro-small threshold

In 2012, after our main sample period, disclosure requirements have been relaxed for so-

called “micro” firms based on claims that disclosure costs exceed these very small firms’ disclosure

benefits (European Commission, 2011). “Micro” firms can abridge the balance sheet, avoid

preparing notes, and opt to restrict public access to their disclosures (Table A.1). Firms are

classified as “micro” if they do not exceed any two out of the following three firm size thresholds

for two consecutive years: total assets (€0.35 million), sales (€0.7 million), and employees (10).

The micro-small firm setting fails to provide unregulated firms with substantial leeway and

interest to incorporate other firms’ disclosure decisions into their own disclosure decisions. Our

abnormal disclosure measure implies that most firms of “micro” size are pushed above their

voluntary disclosures during our sample period (Figure 5, Panel A) and continue to exhibit

mandatory disclosures exceeding their voluntary disclosures even after the introduction of the

reduced disclosure requirements (Figure 5, Panel B). Thus, the micro-small firm setting lacks

unregulated firms who can freely choose their disclosures (Admati and Pfleiderer, 2000).

In the absence of unregulated firms, we do not detect spillover effects of “small” firms’

mandatory disclosures on “micro” firms’ disclosures. In particular, our cross-sectional results in

Table 2 do not replicate for “micro” and “small” firms around the micro-small threshold. We do

32

not find that “micro” firms around the micro-small threshold reduce their disclosures more than

comparable “small” firms when facing more peers, when their regulated peers provide more

disclosures, and when they operate in industries with high asset comovement (Panel A of Table 3).

Hence, that “micro” firms’ disclosures do not vary with the expected strength of a spillover effect,

consistent with these firms’ disclosures being constrained by the (even low) minimum requirements.

An unconstrained setting: the medium-large threshold

The medium-large firm setting fails to provide regulated firms that are substantially

constrained by their disclosure mandate. In particular, we document that “large” firms’ disclosures

in the mandatory regime are not constrained by their minimum disclosure requirements (Columns 2

and 4 of Table 2). Thus, the medium-large firm setting lacks regulated firms that provide mandatory

disclosures in excess of their voluntary disclosures (Admati and Pfleiderer, 2000).

In the absence of regulated firms, we do not detect any spillover effects of “large” firms’

disclosures on “medium” firms’ voluntary disclosures. In particular, the cross-sectional estimates in

Table 3 Panel B suggest that the gap between “medium” and “large” firms’ disclosures does not vary

with the expected strength of the spillover effect as proxied by the number of peer firms, the

aggregate disclosures provided by regulated peers, and by our measure of high asset comovement in

a given industry.

8. Conclusion

We document that mandating some firms’ disclosures appears to reduce other firms’

voluntary disclosures. This crowding-out effect is consistent with theoretical work on the impact of

disclosure regulation on firms’ disclosures (Admati and Pfleiderer, 2000) and adds to recent

empirical work on the interaction of firms’ and information intermediaries’ information production

efforts (e.g., Baginski and Hinson, 2016; Balakrishnan et al., 2014; Breuer et al., 2018).

33

Our evidence has three immediate implications. First, forcing some firms in an industry to

provide more disclosures does not one-for-one improve the industry-wide information environment

because the unregulated firms respond reducing their own disclosures. Second, unregulated firms’

disclosures observed in a mandatory regime are not necessarily equivalent to their disclosures in a

voluntary regime. This follows as unregulated firms’ disclosures are indirectly affected by regulated

firms’ mandatory disclosures. Third, financial reporting regulation taxes regulated firms and

subsidizes unregulated peer firms, potentially distorting competitive positions and contributing to

displacement effects.

These implications caution against using the observed disclosures of unregulated firms with

regulated peers to infer the value of public disclosures for these firms. For example, our evidence

suggests that the opacity of even larger private firms in the United States can reflect their free-riding

on the rich public information provided by their listed peers (e.g., De Fontenay, 2017).

34

Variable Appendix

Variable Definitions Variable Name Source Definition

Dependent Variables Mandatory Disclosure Amount

Federal Gazette Number of characters in a filing, observed under the mandatory disclosure regime

Disclosure Demand Federal Gazette Number of online views of a filing during the twelve months after its publication

Voluntary Disclosure Amount

Federal Gazette Disclosure amount in a voluntary regime, calculated as Disclosure Demand multiplied by a conversion factor

Abnormal Disclosure Amount

Federal Gazette Difference between mandatory disclosure amount and voluntary disclosure amount (in number of characters)

Publication Lag Federal Gazette Number of days between fiscal year-end and publication date

Voluntary Sales Disclosure

Federal Gazette, Bureau van Dijk

Indicator variable taking the value of 1 if sales information is voluntarily reported, 0 otherwise

Restricted Access Federal Gazette Indicator variable taking the value of 1 if the firm uses the option available to “micro-entities” to restrict access to its financial statements, 0 otherwise

Disclosure Classification Variables

Unregulated Federal Gazette Indicator variable taking the value of 1 for firm-years in which a firm is classified by the Federal Gazette as "small", 0 otherwise

Micro Federal Gazette Indicator variable taking the value of 1 for firm-years in which a firm is classified by the Federal Gazette as "micro", 0 otherwise

Medium Federal Gazette Indicator variable taking the value of 1 for firm-years in which a firm is classified by the Federal Gazette as "medium", 0 otherwise

Size-Related Control Variables

Total Assets Federal Gazette, Bureau van Dijk

Total assets

Sales Federal Gazette, Bureau van Dijk

Sales

Employees Federal Gazette, Bureau van Dijk

Number of employees

Age Bureau van Dijk Number of years between incorporation and fiscal year-end

35

Cross-Sectional Variables High_R2 Own

calculations Indicator variable taking the value of 1 for industries with high comovement in asset growth, 0 otherwise. To measure comovement in asset growth, we first obtain the R-Squared from regressions, by industry, of firms’ standardized asset growth on year indicators following Guiso and Parigi (1999). We then residualize the R-Squared with respect to the number of firms operating in the industry, and define high-comovement industries as those in the upper quartile of the distribution of residualized R-Squareds.

Peers Own calculations

Log of 1 plus the number of small, medium, and large firms in a given industry, county, and year

Aggr_Discl Own calculations

Log of 1 plus the aggregate number of characters disclosed by regulated peers in a given county, industry, and year

36

References

Admati, A.R., Pfleiderer, P., 2000. Forcing Firms to Talk: Financial Disclosure Regulation and Externalities. Review of Financial Studies 13, 479-519.

Allee, K.D., Yohn, T.L., 2009. The Demand for Financial Statements in an Unregulated Environment: An Examination of the Production and Use of Financial Statements by Privately Held Small Businesses. The Accounting Review 84, 1-25.

Arif, S., De George, E.T., 2016. Financial Reporting Frequency and Earnings Information Spillovers around the World. Working paper; Available at SSRN: www.ssrn.com/abstract_id=2900988.

Badertscher, B., Shroff, N., White, H.D., 2013. Externalities of public firm presence: Evidence from private firms' investment decisions. Journal of Financial Economics 109, 682-706.

Baginski, S.P., Hinson, L.A., 2016. Cost of Capital Free-Riders. Accounting Review 91, 1291-1313.

Balakrishnan, K., Billings, M.B., Kelly, B., Ljungqvist, A., 2014. Shaping Liquidity: On the Causal Effects of Voluntary Disclosure. Journal of Finance 69, 2237-2278.

Ball, R., Shivakumar, L., 2005. Earnings quality in UK private firms: comparative loss recognition timeliness. Journal of Accounting and Economics 39, 83-128.

Beatty, A., Liao, S., Yu, J.J., 2013. The spillover effect of fraudulent financial reporting on peer firms' investments. Journal of Accounting and Economics 55, 183-205.

Bens, D.A., Berger, P.G., Monahan, S.J., 2011. Discretionary Disclosure in Financial Reporting: An Examination Comparing Internal Firm Data to Externally Reported Segment Data. Accounting Review 86, 417-449.

Berger, P.G., Minnis, M., Sutherland, A.G., 2017. Commercial Lending Concentration and Bank Expertise: Evidence from Borrower Financial Statements. Journal of Accounting and Economics Forthcoming.

Bernard, D., 2016. Is the risk of product market predation a cost of disclosure? Journal of Accounting and Economics 62, 305-325.

Bernard, D., Burgstahler, D., Kaya, D., 2016. Disclosure Incentives and Data Availability for Private Firms: Implications for Comparisons of Public and Private Firm Financial Reporting Quality. Working paper; Available at SSRN: www.ssrn.com/abstract_id=2749747.

Bernard, D., Burgstahler, D., Kaya, D., 2018. Size Management by European Private Firms to Avoid Audit and Disclosure Costs. Journal of Accounting and Economics Forthcoming.

Beyer, A., Cohen, D.A., Lys, T.Z., Walther, B.R., 2010. The financial reporting environment: Review of the recent literature. Journal of Accounting and Economics 50, 296-343.

Bonsall, S.B., Bozanic, Z., Fischer, P.E., 2013. What Do Management Earnings Forecasts Convey About the Macroeconomy? Journal of Accounting Research 51, 225-266.

37

Breuer, M., 2017. How does financial reporting regulation affect market-wide resource allocation? Working paper; Available at SSRN: www.ssrn.com/abstract_id=3063320.

Breuer, M., Hombach, K., Müller, M.A., 2017. The Economics of Firms' Public Disclosure: Theory and Evidence. Working paper; Available at SSRN: www.ssrn.com/abstract_id=3037002.

Breuer, M., Hombach, K., Müller, M.A., 2018. How does financial reporting regulation affect firms' banking? Review of Financial Studies 31, 1265-1297.

Burgstahler, D.C., Hail, L., Leuz, C., 2006. The Importance of Reporting Incentives: Earnings Management in European Private and Public Firms. The Accounting Review 81, 983-1016.

Chaplinsky, S., Hanley, K.W., Moon, S.K., 2017. The JOBS Act and the Costs of Going Public. Journal of Accounting Research 55, 795-836.

Chen, S., Miao, B., Shevlin, T., 2015. A new measure of disclosure quality: The level of disaggregation of accounting data in annual reports. Journal of Accounting Research 53, 1017-1054.

Coffee, J.C., 1984. Market Failure and the Economic Case for a Mandatory Disclosure System. Virginia Law Review 70, 717-753.

Crépon, B., Duflo, E., Gurgand, M., Rathelot, R., Zamora, P., 2013. Do Labor Market Policies have Displacement Effects? Evidence from a Clustered Randomized Experiment. Quarterly Journal of Economics 128, 531-580.

De Fontenay, E., 2017. The Deregulation of Private Capital and the Decline of the Public Company. Hastings Law Journal 68, 445-502.

Dedman, E., Lennox, C., 2009. Perceived competition, profitability and the withholding of information about sales and the cost of sales. Journal of Accounting and Economics 48, 210-230.

Doidge, C., Karolyi, G.A., Stulz, R.M., 2017. The U.S. listing gap. Journal of Financial Economics 123, 464-487.

Duguay, R., Minnis, M., Sutherland, A.G., 2017. Regulatory Spillovers in Common Audit Markets. Working paper; Available at SSRN: www.ssrn.com/abstract_id=3020316.

Durnev, A.R.T., Mangen, C., 2009. Corporate Investments: Learning from Restatements. Journal of Accounting Research 47, 679-720.

Dye, R.A., 1990. Mandatory versus Voluntary Disclosures: The Cases of Financial and Real Externalities. Accounting Review 65, 1-24.

Easterbrook, F.H., Fischel, D.R., 1984. Mandatory Disclosure and the Protection of Investors. Virginia Law Review 70, 669-715.

Engelberg, J., Ozoguz, A., Wang, S., 2017. Know thy neighbor: Industry clusters, information spillovers and market efficiency. Journal of Financial and Quantitative Analysis (forthcoming).

European Commission, 2011. Financial Reporting: burden reduction for micro-entities – Frequently Asked Questions. Press release available under: http://europa.eu/rapid/press-release_MEMO-11-911_en.htm?locale=en, accessed 29 February 2016.

38

Fishman, M.J., Hagerty, K.M., 1989. Disclosure Decisions by Firms and the Competition for Price Efficiency. The Journal of Finance 44, 633-646.

Foster, G., 1981. Intra-industry information transfers associated with earnings releases. Journal of Accounting and Economics 3, 201-232.

Fox, M.B., 1999. Retaining Mandatory Securities Disclosure: Why Issuer Choice Is Not Investor Empowerment. Virginia Law Review 85, 1335-1419.

Gao, F., Wu, J.S., Zimmerman, J., 2009. Unintended Consequences of Granting Small Firms Exemptions from Securities Regulation: Evidence from the Sarbanes-Oxley Act. Journal of Accounting Research 47, 459-506.

Gao, X., Ritter, J.R., Zhu, Z., 2013. Where Have All the IPOs Gone? Journal of Financial and Quantitative Analysis 48, 1663-1692.

Garmaise, M.J., Natividad, G., 2016. Spillovers in Local Banking Markets. Review of Corporate Finance Studies 5, 139-165.

Glaeser, S., Guay, W.R., 2017. Identification and Generalizability in Accounting Research: A Discussion of Christensen, Floyd, Liu, and Maffett (2017). Journal of Accounting and Economics Forthcoming.

Goldstein, I., Yang, L., 2017. Information Disclosure in Financial Markets. Annual Review of Financial Economics 9.

Grossman, S.J., 1981. The Informational Role of Warranties and Private Disclosure about Product Quality. Journal of Law & Economics 24, 461.

Grossman, S.J., Hart, O.D., 1980. Disclosure Laws and Takeover Bids. Journal of Finance 35, 323-334.

Guiso, L., Parigi, G., 1999. Investment and demand uncertainty. The Quarterly Journal of Economics 114, 185-227.

Hail, L., Tahoun, A., Wang, C., 2014. Dividend Payouts and Information Shocks. Journal of Accounting Research 52, 403-456.

Jung, W.-O., Kwon, Y.K., 1988. Disclosure When the Market Is Unsure of Information Endowment of Managers. Journal of Accounting Research 26, 146-153.

Kurlat, P., Veldkamp, L., 2015. Should we regulate financial information? Journal of Economic Theory 158, Part B, 697-720.

Lang, M., Lundholm, R., 1993. Cross-Sectional Determinants of Analyst Ratings of Corporate Disclosures. Journal of Accounting Research 31, 246-271.

Leland, H.E., Pyle, D.H., 1977. Information Asymmetries, Financial Structure, and Financial Intermediation. Journal of Finance 32, 371-387.

39

Leuz, C., Wysocki, P.D., 2008. Economic Consequences of Financial Reporting and Disclosure Regulation: A Review and Suggestions for Future Research. Working paper; Available at SSRN: www.ssrn.com/abstract_id=1105398.

Leuz, C., Wysocki, P.D., 2016. The Economics of Disclosure and Financial Reporting Regulation: Evidence and Suggestions for Future Research. Journal of Accounting Research 54, 525.

Lisowsky, P., Minnis, M., 2016. Accounting Choices and Capital Allocation: Evidence from Large Private U.S. Firms. Working paper; Available at SSRN: www.ssrn.com/abstract_id=2373498.

Litjens, R., Suijs, J., 2014. Supply and Demand of Proprietary Information. Working paper.

Ma, M., 2017. Economic links and the spillover effect of earnings quality on market risk. Accounting Review 92, 213-245.

Merton, R.C., 1987. A Simple Model of Capital Market Equilibrium with Incomplete Information. Journal of Finance 42, 483-510.

Ross, S.A., 1977. The Determination of Financial Structure: The Incentive-Signalling Approach. Bell Journal of Economics 8, 23-40.

Rotemberg, M., 2017. Equilibrium Effects of Firm Subsidies. Working paper.

Rubin, D.B., 1974. Estimating causal effects of treatments in randomized and nonrandomized studies. Journal of Educational Psychology 66, 688-701.

Savor, P., Wilson, M., 2016. Earnings Announcements and Systematic Risk. Journal of Finance 71, 83-138.

Shroff, N., 2016. Discussion of “Is the risk of product market predation a cost of disclosure?”. Journal of Accounting and Economics 62, 326-332.

Shroff, N., Verdi, R., Yost, B.P., 2017. When Does the Peer Information Environment Matter? Journal of Accounting and Economics forthcoming.

Verrecchia, R.E., 1983. Discretionary disclosure. Journal of Accounting and Economics 5, 179-194.

Verrecchia, R.E., 1990. Information quality and discretionary disclosure. Journal of Accounting and Economics 12, 365-380.

40

Figures & Tables

Figure 1

This figure illustrates the disclosure game and the predicted effect of the disclosure regulation on the regulated and unregulated firms’ disclosure amount (or precision). The figure is based on Figure 3 of Admati and Pfleiderer (2000). The lines depict the firms’ best response functions (i.e., a firm’s disclosure amount as a function of the other firm’s disclosure amount). The black dot represents the equilibrium disclosure amounts of the two disclosure games. The dotted line marks the mandatory disclosure amount applying to the regulated firm in the mandatory regime. In the voluntary regime (left side), both firms provide a disclosure amount of 5.3. In the mandatory regime (right side), the best response function of the regulated firm

0

1

2

3

4

5

6

7

8

9

10

Disc

losu

re A

mou

nt o

f Reg

ulat

ed F

irm

0 1 2 3 4 5 6 7 8 9 10Disclosure Amount of Unregulated Firm

Regulated FirmUnregulated FirmEquilibrium Amount

Voluntary Regime

0

1

2

3

4

5

6

7

8

9

10

Disc

losu

re A

mou

nt o

f Reg

ulat

ed F

irm

0 1 2 3 4 5 6 7 8 9 10Disclosure Amount of Unregulated Firm

Mandatory Regime

Disclosure Game

41

is constrained from below due to the mandatory disclosure of 7. As a consequence, the regulated firm provides a disclosure amount of 7 and the unregulated firm provides a disclosure amount of 4.8 in the mandatory regime. The parameters chosen for the best response functions correspond to the values in Figure 3 of Admati and Pfleiderer (2000). The only exception is the correlation parameter between the two firms for which we choose a higher value (0.95) to better illustrate our prediction. Note that we do not consider any discontinuities in the best response functions as the alternative non-disclosure equilibrium derived by Admati and Pfleiderer (2000) is ruled out in our setting where all firms are subject to a general disclosure mandate (with varying minimum disclosure requirements).

42

Figure 2

This figure plots the local averages of voluntary and mandatory disclosures as a function of firm size (measured by the logarithm of total assets). The black line presents firms’ mandatory disclosures, measured by the number of characters observed in firms’ filings. The gray line presents firms’ voluntary disclosures, measured by the predicted number of characters using our approach described in Section 5.1. Local averages are calculated using a kernel regression with an Epanechnikov kernel. The shaded gray areas present 95% confidence bands. The vertical line presents the regulatory threshold relating to total assets for “small” and “medium” firms.

0

10,000

20,000

30,000N

umbe

r of C

hara

cter

s

8 10 12 14 16 18Log of Total Assets

VoluntaryMandatoryCI 95%

Firms' Mandatory and Voluntary Disclosures

43

Figure 3

This figure plots the local average of firms’ abnormal disclosures as function of firm size (measured by the logarithm of total assets). The black line presents firms’ abnormal disclosures, measured by the difference in the number of characters of firms’ disclosures observed in the mandatory regime and firms’ disclosures predicted for the voluntary regime. Local averages are calculated using a kernel regression with an Epanechnikov kernel. The shaded gray areas present 95% confidence bands. The vertical line presents the regulatory threshold relating to total assets for “small” and “medium” firms.

-3,000

-2,000

-1,000

0

1,000

2,000N

umbe

r of C

hara

cter

s

8 10 12 14 16 18

Log of Total Assets

Abnormal DisclosuresCI 95%

Abnormal Disclosures

44

Figure 4

This figure plots the local average of abnormal disclosures separately for regulated and unregulated firms around the regulatory threshold. Unregulated firms are firms classified as “small” under the regulation and regulated firms are firms classified as “medium” under the regulation. Note that the classification of a firms as “medium” is based on multiple criteria over two years so that we observe both unregulated firms with total assets above and unregulated firms with total asset below the threshold (see Table A.1 for details). Local averages are calculated using a kernel regression with an Epanechnikov kernel. The vertical line presents the regulatory threshold relating to total assets for “small” and “medium” firms.

-2,000

0

2,000

4,000

6,000N

umbe

r of C

hara

cter

s

14.5 14.75 15 15.25 15.5

Log of Total Assets

Unregulated firmsRegulated firms

Abnormal Disclosures Around Regulatory Threshold

45

Figure 5

Panel A

Panel B

Panel A shows firms’ mandatory and voluntary disclosures as a function of firm size. Next to the regulatory threshold distinguishing “small” and “medium” firms, Panel A includes as vertical lines the regulatory threshold distinguishing “medium” and “large” firms, as well as the regulatory threshold distinguishing “micro” and “small” firms. Note that the regulatory category of “micro” firms was introduced only after our main sample period. Panel B shows the mandatory and voluntary disclosures of “micro” firms (during the post-sample period) around the micro-small regulatory threshold. The sample comprises the 2012 financial statements by firms classified as “micro” firms by the Federal Gazette. Mandatory disclosures are measured by the number of characters observed in “micro” firms’ 2012 financial statements. Voluntary disclosures are predicted using the approach described in Section 5.2 and using the number of clicks on the prior year’s financial statements (i.e., before any relaxations applied to “micro” firms). Local averages are calculated using a kernel regression with an Epanechnikov kernel.

Micro-Small Threshold

Medium-Large Threshold

Small-Medium Threshold

0

10,000

20,000

30,000

Num

ber o

f Cha

ract

ers

8 10 12 14 16 18Log of Total Assets

VoluntaryMandatoryCI 95%

Mandatory and Voluntary Disclosures

Micro-Small Threshold

500

1,000

1,500

2,000

Num

ber o

f Cha

ract

ers

8 8.5 9 9.5 10 10.5 11 11.5 12 12.5 13

Log of Total Assets

MandatoryVoluntary

Disclosures Around Micro-Small Threshold

46

Table 1

Firms’ Disclosures and Strength of Information Spillovers Panel A: Firms’ Disclosure and Asset Growth Comovement

Abnormal Disclosure

Mandatory Disclosure

Publication Lag

Voluntary Sales Disclosure

(1) (2) (3) (4) Unregulated -1.032*** -1.121*** -0.046*** 0.129*** (-25.52) (-64.71) (-7.09) (8.68) Unregulated * High_R2 -0.098*** -0.025** -0.002 -0.019** (-3.07) (-2.23) (-0.23) (-2.26) Observations 54,657 54,657 285,017 640,178 # Clusters 396 396 398 398 Adjusted R-Squared 0.194 0.793 0.101 0.126 Panel B: Firms’ Disclosure and the Peer Information Environment

Abnormal Disclosure

Mandatory Disclosure

Publication Lag

Voluntary Sales Disclosure

(1) (2) (3) (4) (5) (6) (7) (8) Unregulated -0.762*** -0.089 -1.057*** -0.879*** -0.081*** -0.196*** 0.268*** 0.620***

(-10.52) (-0.42) (-33.62) (-33.62) (-5.81) (-4.89) (8.67) (5.49)

Unregulated * Peers -0.053***

-0.013**

0.006**

-0.028***

(-4.42)

(-2.38)

(2.56)

(-3.88)

Unregulated * Aggr_Discl

-0.068***

-0.020***

0.010***

-0.035***

(-4.58)

(-2.98)

(3.66)

(-4.14)

Observations 54,657 54,657 54,657 54,657 285,017 285,017 640,178 640,178 # Clusters 396 396 396 396 398 398 398 398 Adjusted R-Squared 0.194 0.194 0.793 0.793 0.101 0.102 0.127 0.127 This table presents evidence on the cross-sectional variation in firms’ disclosure with respect to the strength of the information spillover. Abnormal Disclosure is the difference between the logarithm of firms’ observed disclosures less the logarithm of firms’ disclosures predicted for the voluntary regime. Mandatory Disclosure is the logarithm of firms’ observed disclosures. Publication lag is the log of the publication lag (measured in terms of number of days between fiscal year-end and publication dated). Voluntary Sales Disclosure is an indicator taking the value of one if a firm voluntarily discloses sales information, and zero otherwise. In Panel A, we show how the disclosure gap between regulated and unregulated firms varies with the

47

comovement of fundamentals in a given industry. We regress “small” and “medium” firms’ disclosure outcomes on Unregulated (an indicator variable that takes the value of 1 if the firm is classified as “small” in a given year, and zero otherwise) and an interaction of Unregulated with High R2, a proxy for the comovement of fundamentals in a given industry. To construct High R2, we first obtain the R-squared from industry-specific regressions of firms’ standardized, year-over-year asset growth on year fixed effects, and residualize the R-squared with respect to the number of firms operating in the same industry. We then construct High R2 as taking the value of one for industries in the top quartile of the R-squared distribution across industries, and zero otherwise. In Panel B, we show how the disclosure gap between regulated and unregulated firms varies with the richness of the peer information environment. We regress “small” and “medium” firms’ disclosure outcomes on Unregulated (an indicator variable that takes the value of one if the firm is classified as “small” in a given year, and zero otherwise) and an interaction of Unregulated with proxies for the amount of information provided by firms’ (regulated) peers. Specifically, we use the log of the number of firms operating in the same county-industry-year (Peers) and the total disclosures provided by regulated peers operating in the same county-industry-year (Aggr_Discl). All specifications include county-industry-year fixed effects and legal form fixed effects. We further include a control function including all three regulatory size criteria (total assets, sales, employees) and age. We report t-statistics based on standard errors are clustered by county in parentheses. *, **, and *** denote statistical significance levels below 10%, 5%, and 1%, respectively.

48

Table 2

Disclosure Regulation, Firms’ Mandatory Disclosures, and Number of Clicks Panel A: Regulatory Class Changes

Mandatory Disclosures

(Changes) Number of Clicks

(Changes) (1) (2) (3) (4) Up (“Small”-“Medium”) 0.781***

-0.029

(8.97)

(-0.41) Down (“Medium”-“Small”) -0.619***

0.097

(-6.56)

(0.91) Up (“Medium”-“Large”)

-0.015

0.016

(-0.66)

(0.29)

Down (“Large”-“Medium”)

-0.021

0.115

(-0.53)

(1.18)

Sample Firms Small, Medium Medium, Large Small, Medium Medium, Large Observations 9,478 2,341 9,478 2,341 # Clusters (Counties) 342 239 342 239 Adjusted R-Squared 0.236 0.023 0.032 0.033 Panel B: Fixed Effects

Mandatory Disclosure Number of Clicks

(1) (2) (3) (4)

“Medium” Dummy 0.719***

0.020

(13.19)

(0.41)

“Large” Dummy

0.029***

-0.006

(3.19)

(-0.21)

Sample Firms Small, Medium Medium, Large Small, Medium Medium, Large Observations 42,320 11,697 42,320 11,697 # Clusters 398 396 398 396 Adjusted R-Squared 0.195 0.165 0.044 0.087 This table presents the results from first differences (Panel A) and firm fixed effects (Panel B) analyses. Mandatory Disclosure is the logarithm of the number of characters in a firm’s filing. Number of Clicks is the logarithm of one plus the number clicks a firm’s filing received in the 12 months after its publication. The regressions in the Columns 1 and 3 (Columns 2 and 4) are restricted to “small” and “medium” (“medium” and “large” firms). In Panel A, Up (“Small”-“Medium”) (Up “Medium”-“Large”) takes the value of one for firms switching up in their regulatory size class from “small” to “medium” (from “medium” to “large”), and Down “Small”-“Medium” (Down “Medium”-“Large”) takes the value of one for firms switching down from the “medium” to the “small” (from the “large” to the “medium”) regulatory size class. In Panel B, “Medium” Dummy (“Large” Dummy) takes the value of one when a firm is classified as “medium” (“large”), and zero when it is classified as “small” (“medium”). In Panel A, we include legal form and county-industry-year fixed effects, and control for changes in firm characteristics including a firm’s total assets (log), sales (log), number of employees (log), age (log), number of

49

owners (log), institutional ownership (%), number of banks (log), cash (in % of total assets), and fixed assets (in % of total assets). In Panel B, we include legal form fixed effects, firm fixed effects and year fixed effects and control for the levels of same firm characteristics as in Panel A. Standard errors are clustered by county. *, **, and *** denote statistical significance levels below 10%, 5%, and 1%, respectively.

50

Table 3

Firms’ Disclosure and Strength of Information Spillovers: Placebo Tests Panel A: Constrained Setting Mandatory Disclosure (1) (2) (3) Micro -0.432*** -0.405*** -0.453***

(-14.48) (-4.81) (40.33)

Micro * Peers -0.003

(-0.54)

Micro * Aggr_Discl

-0.003

(-0.51)

Micro * High_R2

0.017

(1.09)

Observations 136,817 136,817 136,817 # Clusters 398 398 398 Adjusted R-Squared 0.212 0.212 0.212 Panel B: Unconstrained Setting

Abnormal Disclosure Mandatory Disclosure Publication Lag

(1) (2) (3) (4) (5) (6) (7) (8) (9)

Medium -0.039 -0.102 0.010 -0.060*** -0.025 -0.045*** 0.022 0.000 0.029***

(-0.78) (-0.58) (0.29) (-2.67) (-0.32) (-3.85) (1.63) (0.00) (3.71)

Medium * Peers 0.005

0.003

0.000 0.001

(0.53)

(0.74)

(0.00) (0.59)

Medium * Aggr_Discl

0.007

-0.001

0.002

(0.51)

(-0.25)

(0.67)

Medium * High_R2

-0.056

-0.001

(-1.62)

(-0.15)

Observations 21,345 21,345 21,345 21,345 21,345 21,345 76,253 76,253 76,253 # Clusters 385 385 385 385 385 385 394 394 394 Adjusted R-Squared 0.309 0.309 0.309 0.412 0.412 0.412 0.125 0.125 0.125 This table presents evidence on the cross-sectional variation in firms’ disclosure with respect to the strength of the information spillover in two placebo settings. Abnormal Disclosure is the difference between the logarithm of firms’ observed disclosures less the logarithm of firms’ disclosures predicted for the voluntary regime. Mandatory Disclosure is the logarithm of firms’ observed disclosures. Publication lag is the log of the publication lag (measured in

51

terms of number of days between fiscal year-end and publication dated). Voluntary Sales Disclosure is an indicator taking the value of 1 if a firm voluntarily discloses sales information, and zero otherwise. In Panel A, we show results on the disclosure gap between regulated and unregulated firms in a constrained setting, i.e., using a sample of firms classified as “small” or “micro.” Micro is an indicator variable taking the value of 1 if a firm is classified as “micro” in a given year, and zero otherwise. Peers is the number of peers operating in the same county-industry-year. Aggr_Discl is the total disclosures provided by “small” firms operating in the same county-industry-year. High_R2 is a proxy for the comovement of fundamentals in a given industry. In Panel B, we show results for an unconstrained setting, i.e., using a sample of firms classified as “medium” or “large”. Medium is an indicator variable taking the value of 1 if a firm is classified as “medium” in a given year, and zero otherwise. Aggr_Discl is the total disclosures provided by “small” firms operating in the same county-industry-year. All other variables are defined as in Panel A. In Panel A, all specifications include county-industry fixed effects (given that we only have a cross-section of one year) and legal form fixed effects. In Panel B, all specifications include county-industry-year fixed effects and legal form fixed effects. We further include a control function including all three regulatory size criteria (total assets, sales, employees) and age. We report t-statistics based on standard errors are clustered by county in parentheses. *, **, and *** denote statistical significance levels below 10%, 5%, and 1%, respectively.

Online Appendix (For online publication only)

When you talk, I remain silent: Spillover effects of peers’ mandatory disclosures on firms’ voluntary disclosures

Matthias Breuer

The University of Chicago Booth School of Business

Katharina Hombach Frankfurt School of Finance and Management

Maximilian A. Müller WHU – Otto Beisheim School of Management

Table of contents

- A. Analytical examples - B. Heterogeneity in “small” firms’ mandatory disclosures - Figure A.1: Number of disclosing firms in pre- and post-enforcement period - Figure A.2: Mandatory and voluntary disclosure amounts of regulated and unregulated firms - Figure A.3: Publication lag and voluntary sales disclosures around small-medium threshold - Figure A.4: Mandatory and voluntary disclosure amounts around small-medium and medium-

large thresholds - Table A.1: Regulatory size thresholds and minimum disclosure requirements - Table A.2: Descriptive statistics of additional (firm-level) variables - Table A.3: Mandatory disclosure, clicks, disclosures-per-click and firm size - Table A.4: Cross-sectional determinants of online views

1

A. Analytical examples

In this section, we provide two stylized analytical examples to illustrate the theoretical

foundation and assumptions underlying our empirical strategy. The examples are based on the

general cost-benefit trade-off of firms’ public disclosure decision. As discussed in detail in

Breuer et al. (2017), firms decide on their public disclosure quantity q (along the intensive

margin) by maximizing their net benefit of disclosure:

{ }( )

*

1arg max ( ) ( ) arg max ( ) ( ) ( )

N q

jq qjq p q q C q N q p q q C q

=

= − = − ∑ ,

where *q is the optimal (intensive margin) disclosure quality, jp ( q ) is the marginal “shadow

price” (e.g., reduced adverse selection discount) paid by transacting stakeholder j given

disclosure quantity q , N( q ) is the number of transacting stakeholders given disclosure quantity

q , and C( q ) denotes the total disclosure cost given disclosure quantity q . The disclosure

benefit can be represented as the product of the number of transacting stakeholders, their

average marginal “shadow price” per quantity, and the disclosure quantity ( N( q )p( q )q ). Thus,

the benefit of firms’ public disclosure increases in the number of transacting stakeholders,

whereas disclosure costs are independent of the number of transacting stakeholders. This

follows because one quantity is provided to all transacting stakeholders (instead of separately to

each of them) and can be consumed by all transacting stakeholders (as a result of the non-rivalry

property of public goods). Accordingly, the net benefit of firms’ public disclosure, as formulated

above, is a linear function of firms’ number of transacting stakeholders.

Using specific functional form assumptions for the benefit and cost functions, we explore

the implications of this relation between firms’ net benefit of disclosure and their number of

transacting stakeholders for the disclosures-per-click ratio (*

*( )q

N q) used in our empirical strategy.

Example 1: Constant marginal benefit, increasing marginal costs

In the first analytical example, we assume, for simplicity, that the number of transacting

stakeholders is unaffected by the firm’s disclosure quantity ( q ): N( q ) N= . That is, a firm’s

additional disclosures do not attract new stakeholders (along the extensive margin). Our

evidence in Table 2 supports this simplifying assumption. We further assume that the (average)

marginal benefit of disclosure per transacting stakeholder ( p( q ) ) is independent of the firm’s

disclosure quantity ( q ): p( q ) p= . Given these simplifying assumptions, the firm’s marginal

benefit per disclosure quantity is given by: Np .

We assume that the firm’s cost of disclosure is given by the following cost function:

21( )2

C q cq f= + ,

where 12

cq is the variable cost per disclosure (following the standard quadratic marginal cost

function) and f is the fixed cost of disclosure. Given these parameterizations, the firm’s net

benefit of public disclosure is given by:

21( )2

q Npq cq fp = − − .

The first order condition yields:

( ) 0q Np cqq

p∂= ⇔ =

∂.

Hence, the firm’s optimal disclosure quantity is:

* Npqc

= .

The marginal impact of the number of transacting stakeholders on the disclosure quantity

is:

*

0q pN c∂

= >∂

.

Notably, the disclosures per marginal stakeholder is constant (i.e., independent of N ):

2 *

2 0qN∂

=∂

.

Thus, the disclosures per marginal stakeholder and the average disclosures per

stakeholder are identical:

* *q p qN c N∂

= =∂

.

This analysis makes two important points. First, the disclosures-per-stakeholder ratio can

be interpreted as the net benefit per stakeholder, increasing in the (marginal) disclosure benefit

( p ) and decreasing in its (marginal) cost ( c ). Second, the firm’s disclosures-per-stakeholder

ratio of larger firms ( LN ) corresponds exactly to the disclosures-per-stakeholder ratio of smaller

firms ( SN , where L SN N> ), supporting our use of a common disclosures-per-click ratio.

Although this exact result is derived under admittedly strong and stylized assumptions, it

illustrates the analytically founded idea of our approach and identifies the relevant assumptions.

Example 2: Decreasing marginal benefit, constant marginal costs

In the second example, we again assume that the number of transacting stakeholders is

unaffected by the firm’s disclosure quantity ( q ): N( q ) N= . In contrast to our first example,

however, we now assume that there are diminishing marginal benefits of disclosure per

transacting stakeholder: 1ˆp( q ) p( vq )= − , where p̂ denotes the average price intercept and v

denotes the inverse price elasticity. This is an arguably more realistic assumption than constant

marginal benefits of public disclosure per stakeholder.

We further assume the following linear functional form for the cost function:

( )C q cq f= + .

Given these parameterizations, the firm’s net benefit of disclosure is given by:

2ˆ ˆ ˆ( ) (1 )q Np vq q cq f Npq Npvq cq fp = − − − = − − − .

The first order condition yields:

( ) ˆ ˆ0 2q Np Npvq cq

p∂= ⇔ − =

∂.

Hence, the optimal disclosure quantity is:

* ˆ 1ˆ ˆ2 2 2

Np c cqNpv v Npv−

= = − .

The marginal impact of the number of transacting stakeholders on the disclosure quantity

is increasing, as before, and given by:

*

2 0ˆ2

q cN N pv∂

= >∂

.

Notably, the marginal impact is decreasing in the number of transacting shareholders:

2 *

2 3 0ˆ2

q cN N pv∂

= − <∂

.

This means that additional stakeholders increase firms’ disclosure quantity, but at a

decreasing rate given that additional disclosures’ values decline. Notably, this suggests that the

marginal impact of a transacting stakeholder is lower than its average impact:

* *

2

ˆˆ ˆ2 2

q c p c qN N pv Npv N∂ −

= < =∂

.

Accordingly, this second example suggests that, given the plausible assumption of

diminishing benefits of public disclosure per stakeholder, our common disclosures-per-click ratio

derived from the largest firms, if anything, likely understates the disclosures-per-click ratio

expected for smaller firms. This understatement would work against identifying our

hypothesized effect of lower than predicted disclosures-per-click ratios for unregulated (“small”)

firms. The understatement may, however, not be too severe given use of average disclosures-per-

click ratios. Deviations in average ratios will be less stark than deviations in marginal disclosures-

per-click ratios as a result of declining marginal disclosure benefits (esp., given the relatively

narrow range of number of “clicks” (in contrast to the range in total-asset firm sizes): 8 clicks on

average for “small” firms versus 95 clicks on average for “large” firms).

1

B. Heterogeneity in “small” firms’ mandatory disclosures

This section contains anecdotal evidence of the heterogeneity in “small” firms’

mandatory disclosures. Specifically, we present three different filings by “small” firms,

illustrating their distinct disclosure choices. The filing in Example 1 merely provides the

minimum level of information required by the regulation, whereas the filing in Example 2

voluntarily provides a finer level of financial statement disaggregation than required, and the

filing in Example 3 voluntarily provides a depreciation schedule). All of the following filings are

provided by “small” firms operating in the same industry (manufacturing) and covering the same

fiscal year.

Example 1: “Small” firm providing minimum disclosures

This firm provides mandatory disclosures which do not exceed the minimum

requirements.

Example 2: “Small” firm providing additional disclosures (disaggregation)

This “small” firm provides a finer disaggregation in its balance sheet and more extensive

notes, including, e.g., further information on the nature and maturity of its liabilities, than

required.

Example 3: “Small” firm providing additional disclosures (depreciation schedule)

This “small” firm provides additional disclosures in its notes, including a detailed depreciation schedule.

9

Figure A.1

Panel A

Panel B

This figure compares our disclosure demand measure (Panel A) to firms’ voluntary disclosure of their financial statements during a low-enforcement period pre-dating our sample period (Panel B). Disclosure demand is measured by the number of online views a filing receives on the official publication platform twelve months after its publication. We identify voluntary disclosure of financial statements in the low enforcement period based on whether or not annual financial information as of fiscal year-end 2004 is available in Bureau van Dijk’s dafne database (as of 2013). The lines present local averages of the number of online views (Panel B) and the fraction of firms disclosing financial statements (Panel B) conditional on firm size calculated using a Kernel regression with an Epanechnikov kernel to weigh local observation. The shaded gray areas present 95% confidence bands. The vertical lines present the total asset values based on which the disclosure regulation classifies firms as “small”, “medium”, or “large” (see Table A.1 for details).

Small-Medium Threshold

Medium-Large Threshold

0

20

40

60

80

Disc

losu

re D

eman

d (N

umbe

r of O

nlin

e V

iew

s)

8 10 12 14 16 18Log of Total Assets

CI 95%Avg. Number of Views

Disclosure Demand

Small-Medium Threshold

Medium-Large Threshold

0.1

.2.3

.4V

olun

tary

Disc

losu

re (F

ract

ion

of F

irms)

8 10 12 14 16 18Log of Total Assets

Voluntary DisclosureCI 95%

Voluntary Disclosure in Pre-Period

10

Figure A.2

Panel A

Panel B

This figure plots disclosures in the mandatory and voluntary regime by groups of firms. Panel A (Panel B) shows disclosures of unregulated (regulated) firms. Firms are considered unregulated (regulated) when they are classified as “small” (“medium”) under the regulation. The black lines represent disclosures observed under the mandatory regime and measured by the number of characters in firms’ filings. The gray lines present firms’ disclosures (in number of characters) predicted for the voluntary regime using our approach described in Section 5.1. The vertical line presents the threshold distinguishing “small” and “medium” firms. Local averages are calculated using a Kernel regression with an Epanechnikov kernel. Note that the disclosure regulation assigns disclosure requirements based on three size criteria (sales, employees, total assets) over two years, giving rise to some overlap of regulated and unregulated firms around the small-medium total asset threshold.

4,000

5,000

6,000

7,000

Num

ber o

f Cha

ract

ers

14.5 14.75 15 15.25 15.5

Log of Total Assets

VoluntaryMandatoryCI 95%

Unregulated Firms' Disclosures Around Regulatory Threshold

12,000

14,000

16,000

18,000

20,000

Num

ber o

f Cha

ract

ers

14.5 14.75 15 15.25 15.5

Log of Total Assets

VoluntaryMandatoryCI 95%

Regulated Firms' Disclosures Around Regulatory Threshold

11

Figure A.3

Panel A

Panel B

This figure plots alternative measures of voluntary disclosures around the regulatory threshold separately for regulated and unregulated firms. Firms are considered unregulated (regulated) when they are classified as “small” (“medium”) under the regulation. Panel A (Panel B) shows local averages of firms’ publication lag (voluntary sales disclosure). Local averages are calculated using a Kernel regression with an Epanechnikov kernel. Note that the disclosure regulation assigns disclosure requirements based on three size criteria (sales, employees, total assets) over two years, giving rise to some overlap of regulated and unregulated firms around the small-medium total asset threshold.

360

365

370

375

380

Num

ber o

f Day

s

14.5 14.75 15 15.25 15.5

Log of Total Assets

UnregulatedRegulated

Publication Lag Around Regulatory Threshold

.36

.38

.4

.42

Vol

unta

ry D

isclo

sure

14.5 14.75 15 15.25 15.5

Log of Total Assets

Unregulated firmsRegulated firms

Voluntary Sales Disclosure Around Regulatory Threshold

12

Figure A.4

Panel A

Panel B

This figure plots local averages of voluntary and mandatory disclosure amounts as a function of size. Local averages are calculated using a Kernel regression with an Epanechnikov kernel. The black lines present firms’ mandatory disclosures, measured by the number of characters observed in firms’ filings. The gray lines present firms’ voluntary disclosures, measured by the predicted number of characters using our approach described in Section 5.1. Panel A (Panel B) side shows local averages calculated separately for observations below and above the “small”-“medium” (“medium”-“large”) total assets threshold. The shaded gray areas present 95% confidence bands. The vertical lines present regulatory thresholds distinguishing between “small” and “medium”, and “medium” and “large” firms, respectively.

Small-Medium Threshold

Medium-Large Threshold

0

5,000

10,000

15,000

20,000

25,000

Num

ber o

f Cha

ract

ers

8 10 12 14 16 18

Log of Total Assets

VoluntaryMandatoryCI 95%

Mandatory and Voluntary Disclosures

Small-Medium Threshold

Medium-Large Threshold

0

5,000

10,000

15,000

20,000

25,000

Num

ber o

f Cha

ract

ers

8 10 12 14 16 18

Log of Total Assets

VoluntaryMandatoryCI 95%

Mandatory and Voluntary Disclosures

13

Table A.1

Regulatory Size Thresholds and Mandatory Public Disclosure Panel A: Thresholds implemented in German company law

Fiscal Year Classification Total Assets (million EUR)

Sales (million EUR) Employees Statutory Source

Since 2008 Small X ≤ 4.84 X ≤ 9,68 X ≤ 50 s. 267 German Commercial Code Medium 4.84 < X ≤ 19.25 9,68 < X ≤ 38.5 50 < X ≤ 250 Large X > 19.25 X > 38.5 X > 250

Since 2012 Micro X<0.35 X<0.7 X<10 s. 267a German Commercial Code Panel B: Reporting requirements

Balance sheet Income statement Notes Management Report Audit Small Abbreviated (22) None Major exemptions No No Medium Condensed (39) Condensed (20/25) Minor exemptions Yes Yes, by chartered bookkeeper Large Full (63) Full (27/31) Full Yes Yes, by statutory auditor Micro Abbreviated (10) None None No No This table summarizes the regulatory size thresholds and associated mandatory disclosure requirements. Panel A of this table presents the threshold values for the assignment into one of the three regulatory size categories as implemented in Germany during our sample period. A firm is classified as medium-sized or large if it exceeds the thresholds of any two of the three size criteria in two consecutive years. Panel B of this table displays the differential reporting requirements applying to the three regulatory size categories. The numbers in brackets in the balance sheet and income statement column refer to minimum number of single-line items that need to be disclosed. For medium-sized and large firms, the number of positions in the income statement reflect the number of positions required under function of expense and nature of expense method, respectively.

14

Table A.2

Descriptive Statistics Panel A: Small Firms N Mean SD P25 P50 P75 Variables used in non-parametric tests Mandatory Disclosure (Number of Characters) 2,045,395 3,468 1,731 2,444 3,195 4,136 Clicks 2,045,395 8 16 1 4 9 Voluntary Disclosure (Number of Characters) 2,045,395 2,761 5,838 367 1,468 3,303 Abnormal Disclosure (Number of Characters) 2,045,395 707 5,870 -17 1,603 2,729 Voluntary Sales Disclosure (Indicator) 2,045,395 0.30 0 0 0 1 Publication Lag (Number of Days) 2,045,395 376 130 346 375 397 Total Assets (million Euro) 2,045,395 2.00 87.10 0.05 0.21 0.75 Total Assets (Logarithm) 2,045,395 12.26 1.91 10.90 12.26 13.53

Variables used in cross-sectional tests Abnormal Disclosure (Logarithm) 45,537 0.94 2.11 -0.22 0.52 1.38 Mandatory Disclosure (Logarithm) 45,537 8.38 0.48 8.07 8.33 8.60 Publication Lag (Logarithm) 45,537 5.85 0.33 5.82 5.92 5.98 Sales (million Euro) 45,537 1.99 2.48 0.39 1.00 2.56 Sales (Logarithm) 45,537 13.75 1.36 12.87 13.82 14.76 Employees 45,537 16 20 3 8 20 Employees (Logarithm) 45,537 2.28 1.06 1.39 2.20 3.04 Peers (Logarithm) 45,537 5.37 1.69 4.34 5.37 6.39 Aggr_Discl (Logarithm) 45,537 10.87 3.94 10.72 11.75 12.89 Resid_R2 45,537 -0.0004 0.0070 -0.0061 -0.0020 0.0053

Further covariates Age (Years) 35,346 15.77 15 6 12 20 Age (Logarithm) 35,346 2.46 0.92 1.95 2.56 3.04 Owners 35,346 1.84 1 1 2 2 Owners (Logarithm) 35,346 0.47 0.50 0.00 0.69 0.69 Institutional Owner 35,346 0.03 0 0 0 0 Banks (Indicator) 35,346 1.17 1 1 1 2 Banks (Logarithm) 35,346 0.69 0.42 0.69 0.69 1.10 Cash 35,346 0.17 0.21 0.01 0.09 0.26 Tangibility 35,346 0.20 0.23 0.03 0.11 0.29 Leverage 35,346 0.58 0.30 0.31 0.61 0.86 Return on Assets 35,346 0.03 0.21 -0.02 0.03 0.11 Loss Indicator 35,346 0.29 0 0 0 1

15

Panel B: Medium Firms N Mean SD P25 P50 P75 Variables used in non-parametric tests Mandatory Disclosure (Number of Characters) 49,577 22,051 8,555 16,597 20,310 25,466 Clicks 49,577 52 46 24 42 68 Voluntary Disclosure (Number of Characters) 49,577 19,106 16,948 8,808 15,414 24,956 Abnormal Disclosure (Number of Characters) 49,577 2,945 18,433 -4,543 5,034 12,858 Voluntary Sales Disclosure (Indicator) 49,577 0.42 0 0 0 1 Publication Lag (Number of Days) 49,577 373 105 343 376 404 Total Assets (million Euro) 49,577 18.00 99.00 6.13 9.19 14.90 Total Assets (Logarithm) 49,577 16.12 0.83 15.63 16.03 16.51

Variables used in cross-sectional tests Abnormal Disclosure (Logarithm) 17,743 0.53 0.93 -0.07 0.46 1.06 Mandatory Disclosure (Logarithm) 17,743 10.04 0.35 9.80 10.02 10.26 Publication Lag (Logarithm) 17,743 5.86 0.29 5.79 5.92 6.00 Sales (million Euro) 17,743 21.60 15.10 11.70 17.70 27.30 Sales (Logarithm) 17,743 16.68 0.67 16.28 16.69 17.12 Employees 17,743 109 91 53 85 138 Employees (Logarithm) 17,743 4.38 0.87 3.99 4.45 4.93 Peers (Logarithm) 17,743 4.82 1.89 3.43 4.83 6.09 Aggr_Discl (Logarithm) 17,743 12.16 1.38 11.19 11.98 13.00 Resid_R2 17,743 0.0020 0.0089 -0.0044 0.0001 0.0053

Further covariates Age (Years) 6,974 24.24 19 11 19 31 Age (Logarithm) 6,974 2.96 0.75 2.48 3.00 3.47 Owners 6,974 2.17 1 1 2 3 Owners (Logarithm) 6,974 0.60 0.57 0.00 0.69 1.10 Institutional Owner 6,974 0.06 0 0 0 0 Banks (Indicator) 6,974 1.88 1 1 2 3 Banks (Logarithm) 6,974 0.97 0.44 0.69 1.10 1.39 Cash 6,974 0.12 0.16 0.01 0.05 0.16 Tangibility 6,974 0.27 0.26 0.05 0.18 0.43 Leverage 6,974 0.90 0 1 1 1 Return on Assets 6,974 0.04 0.12 0.00 0.03 0.08 Loss Indicator 6,974 0.18 0 0 0 0

16

Panel C: Large Firms N Mean SD P25 P50 P75 Variables used in non-parametric tests Mandatory Disclosure (Number of Characters) 14,099 32,482 15,524 23,014 29,226 37,876 Clicks 14,099 95 93 39 71 121 Voluntary Disclosure (Number of Characters) 14,099 34,825 34,227 14,313 26,057 44,407 Abnormal Disclosure (Number of Characters) 14,099 -2,343 35,573 -14,229 4,553 16,601 Voluntary Sales Disclosure (Indicator) 14,099 0.00 0 0 0 0 Publication Lag (Number of Days) 14,099 356 120 298 364 404 Total Assets (million Euro) 14,099 166.00 828.00 25.80 43.60 94.80 Total Assets (Logarithm) 14,099 17.70 1.44 17.07 17.59 18.37

Variables used in cross-sectional tests Abnormal Disclosure (Logarithm) 11,249 0.15 0.90 -0.46 0.11 0.69 Mandatory Disclosure (Logarithm) 11,249 10.31 0.37 10.06 10.29 10.53 Publication Lag (Logarithm) 11,249 5.84 0.32 5.73 5.91 6.00 Sales (million Euro) 11,249 136.00 219.00 46.10 71.90 130.00 Sales (Logarithm) 11,249 18.19 0.99 17.65 18.09 18.69 Employees 11,249 404 482 125 268 471 Employees (Logarithm) 11,249 5.41 1.26 4.84 5.59 6.16 Peers (Logarithm) 11,249 4.57 1.93 3.14 4.50 5.95 Aggr_Discl (Logarithm) 11,249 12.28 1.30 11.35 12.10 13.09 Resid_R2 11,249 0.0025 0.0091 -0.0024 0.0001 0.0059

Further covariates Age (Years) 4,723 26.49 22 11 20 37 Age (Logarithm) 4,723 3.00 0.83 2.48 3.04 3.64 Owners 4,723 2.09 1 1 2 3 Owners (Logarithm) 4,723 0.57 0.55 0.00 0.69 1.10 Institutional Owner 4,723 0.07 0 0 0 0 Banks (Indicator) 4,723 1.89 1 1 2 3 Banks (Logarithm) 4,723 0.97 0.45 0.69 1.10 1.39 Cash 4,723 0.09 0.14 0.00 0.03 0.12 Tangibility 4,723 0.26 0.26 0.04 0.18 0.42 Leverage 4,723 0.90 0 1 1 1 Return on Assets 4,723 0.03 0.10 0.00 0.01 0.06 Loss Indicator 4,723 0.15 0 0 0 0 This table presents descriptive statistics separately by size class. Resid_R2 is the R2 from regressions by industry of firms’ standardized asset growth on year fixed effects, residualized with respect to the number of firms operating in the industry. All other variables are defined as in the variable appendix.

17

Table A.3

Mandatory Disclosure, Clicks, Disclosures-Per-Click and Firm Size Panel A: Small, Medium, and Large Firms

Mandatory Disclosure Clicks

Mandatory Disclosures-Per-Click

Total Assets (Log) 0.248*** 0.232*** -0.155***

(75.47) (46.43) (-17.76)

Observations 1,675,511 1,675,511 1,675,511 # Clusters 399 399 399 Within R-Squared 0.061 0.054 0.024 Panel B: Medium and Large Firms

Mandatory Disclosure Clicks

Mandatory Disclosures-Per-Click

Total Assets (Log) 0.323*** 0.212*** -0.025

(28.24) (18.39) (-1.63)

Observations 50,615 50,615 50,615 # Clusters 396 396 396 Within R-Squared 0.104 0.045 0.001 Panel C: Large Firms

Mandatory Disclosure Clicks

Mandatory Disclosures-Per-Click

Total Assets (Log) 0.393*** 0.266*** -0.105***

(20.73) (11.04) (-4.43)

Observations 7,712 7,712 7,712 # Clusters 304 304 304 Within R-Squared 0.154 0.071 0.011

This table presents evidence on the relation between (mandatory) disclosures, clicks, disclosures-per-click and firm size. Mandatory Disclosure is the number of characters provided in firms’ mandatory filings. Clicks is the number of online clicks firms’ mandatory filings received. Mandatory Disclosures-Per-Click is the number of characters provided in firms’ mandatory filings scaled by the number of online clicks firms’ mandatory filings received. Total Assets (Log) is a proxy for firms’ sizes calculated as the natural logarithm of firms’ total assets. In Panel A, we estimate the relation between disclosures, clicks, disclosures-per-click and firm size for the entire sample of firms, including firms in the small, medium, and large regulatory size category. In Panel B, we estimate the relation between disclosures, clicks, disclosures-per-click and firm size only for firms in the medium, and large regulatory size category. In Panel C, we estimate the relation between disclosures, clicks, disclosures-per-click and firm size only for firms in the large regulatory size category. All specifications include county-industry-year fixed effects and legal form fixed effects. We report standardized coefficients, within-R-squared values (purged of variation explained by fixed effects), and t-statistics based on standard errors are clustered by county in parentheses. *, **, and *** denote statistical significance levels below 10%, 5%, and 1%, respectively.

Our preferred specification is shown in Panel B. The exclusion of small firms is warranted as the smallest firms’ mandatory disclosures are strongly pushed above their voluntary disclosures (Figure 2). Accordingly, their inclusion strongly biases toward a negative correlation between firms’ (mandatory) disclosures-per-click ratio and firm size. This negative correlation, however, is induced by the mandate. Thus, it is not representative of the relation between disclosures-per-click and firm size in a voluntary regime (which we are actually interest in and essentially assume to be flat). The results in Panel B suggest that firm size does not explain much of the variation in the disclosures-per-click ratio and does not appear to be strongly positively

18

or negatively related to the disclosures-per-click ratio. The results in Panel C provide comparable inferences with respect to the explanatory power of firm sizes for the variation in disclosures-per-click. The relation between disclosures-per-click and firm, however, is significantly negative for this subsample of firms. This estimate may hint at diminishing returns to scale (clicks), especially given the extensive range of firm sizes covered by the large firms. This estimate may, however, also be driven by few outliers (given the lower sample size, the extensive size range, and the low explanatory power of the linear fit).

19

Table A.4

Firms’ Disclosure and the Peer Information Environment: Alternative Specifications Panel A: Firms’ Disclosures and Number of Peers

Abnormal Disclosure Mandatory Disclosure Publication Lag Voluntary Sales Disclosure

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)

Unregulated -0.415** -0.882*** -0.361 -1.027*** -1.061*** -1.010*** -0.103*** -0.082*** 0.072 0.271*** 0.289*** -0.115**

(-2.23) (-9.62) (-1.13) (-15.47) (-26.95) (-10.04) (-3.03) (-4.65) (1.54) (7.21) (8.46) (-2.36)

Unregulated * Peersj -0.051***

-0.008

0.004*

-0.012***

(-3.49)

(-1.61)

(1.70)

(-3.64)

Unregulated * Reg_Peersc,j,t

-0.068***

-0.021*

0.012**

-0.056***

(-2.86)

(-1.89)

(2.09)

(-4.27) Unregulated * Reg_Peersj

-0.074**

-0.012

-0.012**

0.025***

(-2.23) (-1.23) (-2.55) (4.76) Observations 54,657 54,657 54,657 54,657 54,657 54,657 285,017 285,017 285,017 640,178 640,178 640,178 # Clusters 396 396 396 396 396 396 398 398 398 398 398 398 Adjusted R-Squared 0.194 0.194 0.194 0.793 0.793 0.793 0.101 0.101 0.101 0.126 0.127 0.126 Panel B: Firms’ Disclosures and Aggregated Peer Disclosures Abnormal Disclosure Mandatory Disclosure Publication Lag Voluntary Sales Disclosure (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Unregulated 0.521 -0.580*** -0.640*** -1.015*** -1.075*** -1.134*** 0.057 -0.059*** -0.116*** 0.064 0.188*** 0.423***

(1.08) (-8.97) (-7.12) (-7.01) (-54.35) (38.61) (0.93) (-8.46) (-9.87) (0.81) (11.78) (23.15)

Unregulated * Aggr_Disclj -0.082***

-0.006

-0.005*

0.003

(-3.29)

(-0.80)

(-1.70)

(0.75)

Unregulated * Aggr_Abn_Disclc,j,t -0.330***

-0.036*** 0.008***

-0.046***

(-9.18)

(-3.22)

(2.85)

(-12.25)

Unregulated * Aggr_Abn_Disclj

-0.326***

0.004

0.054***

-0.249***

(-4.81)

(0.18)

(6.38)

(-21.35)

Observations 54,657 54,657 54,657 54,657 54,657 54,657 285,017 285,017 285,017 640,178 640,178 640,178 # Clusters 396 396 396 396 396 396 398 398 398 398 398 398 Adjusted R-Squared 0.194 0.195 0.194 0.793 0.793 0.793 0.101 0.101 0.102 0.126 0.127 0.129 This table presents evidence on the cross-sectional variation in firms’ disclosure with respect to the strength of the information spillover using alternative specifications of our cross-sectional variables. Abnormal Disclosure is the difference between the logarithm of firms’ observed disclosures less the logarithm of firms’ disclosures predicted for the voluntary regime. Mandatory Disclosure is the logarithm of firms’ observed disclosures. Publication lag is the log of the publication lag (measured in terms of number of days between fiscal year-end and publication dated). Voluntary Sales Disclosure is an indicator taking the value of 1 if a firm voluntarily discloses sales information, and zero otherwise. In Panel A, we show how the disclosure gap between regulated and unregulated firms varies with the richness of the peer information environment. Peersj is the logarithm of the number of peers in the same Fama-French industry group. Reg_Peersc,j,t is the number of regulated (“medium”) firms operating in the same county-industry-year. Reg_Peersj is the number of regulated (“medium”) firms operating in the same Fama-French industry. In Panel B, Aggr_Disclj is the logarithm of the total mandatory disclosure provided by regulated

20

(“medium”) firms operating in the same industry. Aggr_Abn_Disclc,j,t (Aggr_Abn_Disclj,) is the abnormal disclosure provided by regulated firms operating in the same county-industry-year (operating in the same industry), where abnormal disclosure is measured by the difference between the logarithm of the total mandatory disclosure and the logarithm of the total voluntary disclosure. All specifications include county-industry-year fixed effects and legal form fixed effects. We further include a control function including all three regulatory size criteria (total assets, sales, employees) and age. We report t-statistics based on standard errors are clustered by county in parentheses. *, **, and *** denote statistical significance levels below 10%, 5%, and 1%, respectively.