Professional Documents
Culture Documents
DOI 10.1007/s11142-013-9273-4
Hal D. White
Abstract In 2009, the SEC mandated that financial statements be filed using
eXtensible Business Reporting Language (XBRL). The SEC contends that this new
search-facilitating technology will reduce informational barriers that separate smal-
ler, less-sophisticated investors from larger, more-sophisticated investors, thereby
reducing information asymmetry. However, if some larger investors can leverage
their superior resources and abilities to garner greater benefits from XBRL than
smaller investors, information asymmetry is likely to increase. Using a difference-in-
difference design, we find evidence of higher abnormal bid-ask spreads for XBRL
adopting firms around 10-K filings in the year after the mandate, consistent with
increased concerns of adverse selection. We also find a reduction in abnormal
liquidity and a decrease in abnormal trading volume, particularly for small trades.
Additional analyses suggest, however, that these effects may be declining somewhat
in more recent years. Collectively, our evidence suggests that a reduction in investors’
data aggregation costs may not have served its intended purpose of leveling the
informational playing field, at least during the initial years after mandatory adoption.
E. Blankespoor
Graduate School of Business, Stanford University, 655 Knight Way, Stanford, CA 94305-7298,
USA
e-mail: blankbe@stanford.edu
B. P. Miller
Kelley School of Business, Indiana University, 1309 E. 10th Street, Bloomington, IN 47405, USA
e-mail: bpm@indiana.edu
H. D. White (&)
Ross School of Business, University of Michigan, 701 Tappan St., Ann Arbor, MI 48109-1234, USA
e-mail: halwhite@umich.edu
123
E. Blankespoor et al.
1 Introduction
In April 2009, the U.S. SEC mandated that companies report their financial
statement data using eXtensible Business Reporting Language (XBRL). The SEC
contends that XBRL has the potential to reduce information asymmetry between
smaller, less-sophisticated investors and larger, more-sophisticated investors by
improving smaller investors’ access to data1:
If [XBRL] serves to lower the data aggregation costs as expected, then it is
further expected that smaller investors will have greater access to financial
data than before. In particular, many investors that had neither the time nor
financial resources to procure broadly aggregated financial data prior to
interactive data will have lower cost access than before interactive data….
Hence, smaller investors will have fewer informational barriers that separate
them from larger investors with greater financial resources.—SEC (2009)
However, even larger investors can benefit from using XBRL by shifting a
significant amount of their resources from data collection to analysis.2 Accordingly,
we investigate the change in information asymmetry, and thus market liquidity, after
the XBRL mandate.
Many argue that XBRL fundamentally changes the way financial data is
communicated to, and processed by, investors. For example, Corey Booth, the
SEC’s chief information officer, states, ‘‘Interactive data represents the logical next
step in the evolution of company disclosure… this move will usher in a quantum
leap in helping companies explain their business to investors’’ (SEC 2009). In
particular, XBRL can (1) reduce acquisition costs through computer automated data
collection, thereby allowing investors to analyze more data, (2) provide a richer set
of data than those provided by data aggregators (e.g., Factset, Compustat, Capital
IQ), which are typically used by larger investors, and (3) allow investors to directly
compare data across firms much more efficiently via a structured taxonomy. Thus,
although XBRL-formatted (interactive) financial statements provide no additional
1
Although there is a broad spectrum of investor types, we characterize investors as either small (e.g.,
retail investors or small institutions) or large (e.g., large institutions) for parsimony, where small investors
are relatively less sophisticated—i.e., they have relatively fewer resources, abilities, or both—as
compared to larger investors. In our empirical analyses, we use trade size as a proxy for investor
sophistication following prior research.
2
For example, Campbell Pryde, CEO and president of XBRL US, indicates that large institutional
investors, such as mutual funds and hedge funds, are using XBRL to fill gaps in their datasets (Merrill
Corporation 2012). This is not surprising given the tremendous amount of resources these institutions
invested in data collection prior to XBRL adoption. As Whalen (2004), co-founder of Institutional Risk
Analytics, states, ‘‘Putting financial data in ready-to-crunch condition might seem a trivial detail until you
consider that Wall Street currently spends 80 % of its time and money in data mining and 20 % on actual
analysis—what data pros call the ‘80–20 rule’.’’ Whalen adds that XBRL could flip the 80–20 rule,
allowing much more time to be spent on analysis.
123
The market impact of the XBRL mandate
data relative to the HTML filings, investors can obtain more data than they would
have otherwise, as a result of more efficient processing.3
To receive these benefits, however, investors must incur nontrivial costs, such as
learning the large and growing U.S. GAAP XBRL taxonomy (with over 15,000
unique tags) and tagging structure. Investors must also develop or modify analysis
software to incorporate the XBRL data into their specific valuation processes. That
is, XBRL can help reduce data aggregation costs, but investors must then
understand how to derive informational benefits from the new data. Accordingly,
investors must weigh these costs against the benefits when deciding whether to
incorporate the new technology.
Given the costs and benefits can vary across investors with different resources, it
is a priori unclear what impact, if any, XBRL will have on information asymmetry
among investors. On the one hand, larger investors may already have similar
(proprietary) technology, so they may receive little to no benefit from implementing
XBRL. To the extent that smaller investors adopt the technology and receive
informational benefits, we should observe a reduction in information asymmetry.
On the other hand, some larger investors may garner greater benefits from XBRL
than do smaller investors, particularly given their superior resources and abilities.
Specifically, while XBRL may reduce the potential for differential data aggregation
costs across investor groups, small investors may still not be able to process and
integrate the new data in the same advantageous manner as larger investors. To the
extent that some larger investors can extend their information advantage either by
leveraging their superior resources and abilities to take advantage of the technology
or because smaller investors are reluctant to adopt the technology or both, we should
observe an increase in information asymmetry.
Another possibility is that XBRL has no impact on trading behavior. First,
investors may simply decide not to implement the new technology and ignore its
potential effects because they believe the net incremental benefits are insignificant.
Second, the trading impact from investors using XBRL (or investors’ concerns
about others using XBRL) may be insignificant.4 Thus the extent to which the
XBRL mandate impacts the market is an empirical question.
To conduct our analyses, we collect the interactive 10-K filings of firms required
to adopt XBRL during the first phase-in year (fiscal periods ending between June 15,
2009 and June 14, 2010) of the three-year phase-in period. During the initial phase-
in period, only a subset of firms is required to comply with the regulation, which
allows us to use a difference-in-difference design to examine the change in
3
We assume investors are constrained in their abilities and resources, so they cannot fully process all
publicly available information (Merton 1987, Hirshleifer and Teoh 2003). As such, processing
efficiencies allow investors to process more information, such as broader peer comparisons or more
detailed fundamental analyses.
4
Consistent with the possibility that the market impact of XBRL is minimal, a Columbia University
survey of 26 analysts and investors indicates that only 8 % of the respondents use XBRL-formatted data
in their models, analyses, or both. The study also notes that ‘‘those using XBRL data are using it for the
perceived informational advantage they have by having interactive access to certain types of data that
they believe they cannot collect elsewhere with the same effectiveness and efficiency’’ (Harris and
Morsfield 2012, p. 61).
123
E. Blankespoor et al.
5
We create abnormal measures of trading activity using a control period to difference out normal trading
activity.
123
The market impact of the XBRL mandate
2.1 Background
The SEC has long recognized ‘‘the vital role of the Internet and electronic
communications in modernizing the disclosure system under the federal securities
laws and in promoting transparency, liquidity and efficiency in our trading
markets’’ (SEC 2008, p. 6). Consistent with this belief, in 1994, the SEC required
filers to make their filings available on the EDGAR system. More recently, the
SEC initiated the XBRL Voluntary Filing Program in 2005 to examine the
benefits of using interactive data for financial filings. Based on feedback from that
program, on August 19, 2008, the SEC announced plans to fully replace the
current EDGAR filing system, which relies on static electronic data, with a new
filing system known as an interactive data electronic applications (IDEA) system
that uses solely XBRL interactive data.6 The overhaul is to take place in stages.
Initially, EDGAR will contain both static (HTML) documents and XBRL
6
See Plumlee and Plumlee (2008) for background information on the SEC’s efforts to incorporate XBRL
into its filing process.
123
E. Blankespoor et al.
interactive data. Then, it will shift to a solely interactive data system at a point in
the future yet to be determined.
In April 2009, the SEC took its first step in this overhaul by enacting the
Interactive Data to Improve Financial Reporting Rule, which mandates that
companies file XBRL documents in addition to their current filings. The rule
institutes a phase-in period, where the initial phase-in period begins with filings for
fiscal periods ending on or after June 15, 2009, and relates solely to large
accelerated filers that have a public common equity float over $5 billion. The second
phase-in period relates to all other large accelerated filers (i.e., public common
equity float over $700 million) and begins with filings for fiscal periods ending on or
after June 15, 2010. All remaining filers must begin filing XBRL statements for
fiscal periods on or after June 15, 2011.
The XBRL mandate requires firms to tag financial statement elements. Tagging is
the process of identifying each financial statement element and linking it to descriptive
information, such as the name (Product Sales), year (2009), currency (USD), and
denomination (millions), as well as more detailed definitions and the relationships
between items (Product Sales ? Service Sales = Total Sales). That is, unlike textual
analysis software, such as PERL, that simply captures textual data in defined locations,
XBRL allows computers to ‘‘understand’’ what a given financial statement number
represents and how it relates to other numbers in the statements. Absent XBRL,
financial statement users, including data aggregators, must incur significant costs to
collect and organize data from all relevant (comparison) filings. By using the XBRL
tagging structure, computer software applications can easily access pertinent
information about data items and process the data with little human intervention.7
In the first year of adoption, the mandate requires firms to tag each financial
statement element (i.e., quantitative amount) in the basic financial statements and
also to tag footnotes as blocks of text. In subsequent years, firms must also
individually tag all quantitative amounts in footnotes.8 The XBRL documents are
made available on EDGAR, and firms are required to post them on their corporate
websites no later than the same day that their statements are filed with the SEC to
ensure that investors are aware of the XBRL filing.
The SEC argues that XBRL can increase investors’ information sets by reducing
their data aggregation costs, providing a richer set of data than those provided by
data aggregators, and allowing investors to directly compare data across firms much
more efficiently via a uniform taxonomy. To receive these benefits, however,
7
Although XBRL is designed to facilitate comparability across firms, it must be applied consistently
across firms. To promote greater consistency, both the FASB and XBRL US have established best
practice suggestions for XBRL tagging. However, to the extent there are inconsistencies across firms,
investors may find XBRL less useful. See the appendix for more discussion of XBRL and its intended
benefits.
8
Since investors’ processing costs are generally higher for acquiring information in footnotes than they
are for obtaining information from the face of the financial statements, one may argue that the benefits of
XBRL are somewhat muted in the initial year of XBRL adoption. However, according to the SEC and the
Committee on Corporate Reporting (which performed extensive outreach on XBRL implementation
issues), even after detailed tagging was made available, financial statement users were ‘‘primarily
interested in tagged information in the basic financial statements’’ (FEI 2011), providing support for the
importance of tagged data in the first year of adoption.
123
The market impact of the XBRL mandate
investors must have the resources and abilities to learn the large XBRL taxonomy,
develop or modify the appropriate software to take advantage of XBRL, and
understand how to incorporate this information into their valuation processes.9
Given the potential that the costs and benefits may differ greatly across investor
groups, it is ex ante unclear how XBRL will impact information asymmetry. For
example, one may argue that larger investors already have similar (proprietary)
technology, so the benefits of implementing XBRL may be minimal for them. To
the extent that smaller investors can adopt the technology and receive informational
benefits, we should observe a reduction in information asymmetry. The intuition is
similar to that developed by Diamond and Verrecchia (1991), where disclosure
reduces information asymmetry by providing uninformed investors access to
information they would not have had otherwise. In our setting, although 10-K filings
are publicly available, any given investor cannot fully process all available
information, as investors have limited attention and processing abilities (Hirshleifer
and Teoh 2003). As such, investors must be selective as to which information they
acquire. Since smaller investors generally have relatively fewer resources, abilities,
or both than do larger investors, they likely process less information than do larger
investors. However, after the mandate, smaller investors may be able to use XBRL
to process information more efficiently, thereby enabling them to more closely align
their information sets with those of larger investors.
Conversely, some larger investors may be able to leverage their superior
sophistication to obtain greater benefits from XBRL than can smaller investors. To
the extent these larger investors can extend their information advantage over smaller
investors, we should observe an increase in information asymmetry. This intuition is
similar to that developed in Kim and Verrecchia (1994), where public disclosure can
increase information asymmetry, as a subset of better processing investors gain an
information advantage from their superior processing ability and the less informed
investors (less capable processors) protect themselves from trading losses by
increasing spreads or, in the case of discretionary liquidity traders, even refraining
from trading around news events. In our setting, XBRL serves as a technological
tool that allows larger investors to increase their information advantage through
more efficient processing of 10-K filings. Thus whether the XBRL mandate has an
impact on information asymmetry is an empirical question.
Because of the recency of the XBRL mandate, there is limited empirical evidence
on the implications of XBRL. The studies that have emerged focus on voluntary
filer characteristics, such as their corporate governance (Premuroso and Bhattach-
arya 2008), the accuracy of their initial XBRL filings (Bartley et al. 2010), and the
information content of their filings (Efendi et al. 2010). Hodge et al. (2004)provide
9
The cost of developing or modifying the appropriate software to take advantage of XBRL during our
sample period is not trivial because of the shortage of quality XBRL analysis software (SEC 2009, XBRL
2011). Although the availability of viewers may increase over time enabling investors to reduce their
collection costs, investors will still need to be capable of continually modifying their valuation procedures
to process the increasing volumes of data available through the XBRL mandate.
123
E. Blankespoor et al.
3 Variable definitions
Although our primary interest is in the effect of the XBRL mandate on information
asymmetry, we also examine the price impact of trade and trading volume to gain
broader insights into the impact of the mandate on the market. To assess the effect
of XBRL adoption on trading behavior, we focus on abnormal levels of bid-ask
spreads, price impact of trade and trading volume around 10-K filings, where
abnormal is defined using the firm as its own control. Specifically, we follow prior
research (e.g., Asthana et al. 2004) and create abnormal market measures using the
5-day window around the 10-K filing (trading day -1 to trading day ?3) as our
event period and the 45 trading days prior to the event window (trading day -49 to
trading day -5) as our nonfiling control period.
Prior research (Cohen et al. 1986; Harris 1990; Lee and Ready 1991) indicates that
the bid-ask spread captures market makers’ and other liquidity suppliers’ (i.e.,
public limit order traders) willingness to trade at a low cost. Leuz and Verrecchia
(2000, p. 99) point out that the bid-ask spread is ‘‘commonly thought to capture
123
The market impact of the XBRL mandate
information asymmetry explicitly. The reason for this common belief is that the bid-
ask spread addresses the adverse selection problem that arises from transacting in
firm shares in the presence of asymmetrically informed investors. Less information
asymmetry implies less adverse selection, which, in turn, implies a smaller bid-ask
spread.’’ Accordingly, we use a firm’s bid-ask spread to proxy for information
asymmetry.10
We measure abnormal spread, ASPREAD, as the event period average daily
percent spread minus the nonfiling average daily percent spread, where daily
percent spread is the daily average of each quote’s spread, calculated as the
difference between the offer price and bid price, divided by the midpoint of the offer
and bid price, all multiplied by 100.11 We winsorize ASPREAD (and our other
market variables) at 1 and 99 % to remove the effect of outliers.12
Following prior research (Amihud 2002; Daske et al. 2008), we create a price
impact of trade measure to serve as our proxy for liquidity. This measure is inspired
by Kyle’s (1985) lambda, as it attempts to capture the amount of price movement
for a given level of trading volume—i.e., it captures the ability of investors to trade
with little price impact. As indicated by Leuz and Wysocki (2008, p. 6), higher
spreads ‘‘reduce the liquidity of share markets, i.e., the ability of investors to quickly
buy or sell shares at low cost and with little price impact’’ (emphasis added).13
Accordingly, we examine the price impact of trade as our proxy for liquidity.
We measure abnormal price impact of trade, AIMPACT, following Hasbrouck
(2009) and Goyenko et al. (2009). Specifically, we estimate Goyenko et al.’s (2009)
model 5, which regresses the stock return on the signed square-root dollar volume
for each 5-min trading interval during the day, and we use the coefficient on volume
as the firm’s price impact. Using TAQ data, we estimate the price impact for each
10
Some studies (e.g., Glosten and Harris 1988; Madhavan et al. 1997) attempt to decompose spread into
an information asymmetry component and a non-information asymmetry component. Van Ness et al.
(2001) provide evidence that spread decomposition methodologies are weak at best, so we focus on
overall spread as our primary measure of information asymmetry. However, we also repeat our tests using
abnormal measures of the information asymmetry component of spread, as estimated following Akins
et al. (2012) and Armstrong et al. (2011). For both estimation methods, we find that the information
asymmetry component of spread increases for XBRL firms relative to those for matched non-XBRL
firms, with three of the four specifications significantly different from zero.
11
We obtain offer and bid prices from TAQ. As recommended by WRDS documentation, we only use
quotes with a positive spread given between 9:30 a.m. and 4:00 p.m. and captured during trading modes.
We also remove quotes with spreads \90 % of the mid-point price.
12
We repeat our main analyses using raw (unwinsorized) variables, as well as eliminating observations
with extreme DFBETAs as defined by Belsley et al. (1980) (i.e., those with DFBETA values [2/Hn)
separately for each regression and winsorizing variables at 1 and 99 %. The results are qualitatively
similar.
13
An alternate liquidity measure is trading volume. However, liquidity relates to investors’ ability to
quickly buy or sell shares at low cost and with little price impact (Leuz and Wysocki 2008). Thus
examining volume without consideration of its relation to price cannot tell us whether the market is
indeed more liquid. As such, we use the price impact measure following Hasbrouck 2009 and Goyenko
et al. 2009, which incorporates both volume and price.
123
E. Blankespoor et al.
To ensure that our results are not attributable to other firm-specific characteristics
that may affect abnormal trading behavior, we include several control variables,
including the presence of information intermediaries, market characteristics,
information content and timing of the filing, and firm specific characteristics. We
discuss the motivation for the inclusion of these control variables next and provide
detailed variable definitions in Table 1.
To control for the effects of information intermediaries on the firm’s information
environment, we include the log of one plus the number of analysts (Lnanalyst) and
the percentage of shares outstanding held by institutions (Inst_hold). We also
include other market characteristics that are likely to affect trading behavior.
Specifically, we control for the prior quarter turnover (Qt1_turn) in the spread and
price impact regressions, prior quarter spread (Qt1_spread) in the price impact and
volume regressions, prior quarter depth (Qt1_depth) in the spread and volume
regressions, and prior quarter stock-return volatility (Qt1_volat) in all three.
123
Table 1 Variable definitions
Variable Variable description Variable definition
name
Abs_abn_ret Market return Absolute value of (FirmCumRetEvent - MktCumRetEvent), using the value-weighted average return for all CRSP firms as the
market return, using data from CRSP
Abs_esurp Forecast error Earnings surprise scaled by price as of the end of the same fiscal year. The forecast error is normally based on the most recent
I/B/E/S consensus forecast prior to the event date. If I/B/E/S data is unavailable, we use the forecast error for the most
recent First Call consensus forecast. If neither is available, we use the seasonal random walk earnings surprise using
Compustat data
AIMPACT Abnormal price impact of Estimated from model 5 in (Goyenko et al. 2009, p. 156) as the slope coefficient k(TAQ) from the regression
trade rn = k(TAQ)…Sn ? un, where, for the nth 5-min period, rn is the stock return; Sn is the signed square-root dollar volume;
and un is the error term. (Goyenko et al. 2009, p. 156). Following Hasbrouck (2009), we multiply this coefficient by 106
The market impact of the XBRL mandate
123
Table 1 continued
123
Non-Filing Non-filing period Measured 49 days prior to the 10-K filing through 5 days before the filing
POST Post period Indicator variable equal to one if the 10-K filing occurs between June 15, 2009, and June 14, 2010, while filings that occur in
the year prior are coded as zero
Qt1_depth Prior quarter depth Log of quarter t - 1 average of daily depths, where depth is the daily average of each quote’s depth, calculated as the sum of
the dollar offer size and the dollar bid size multiplied by 100 for all valid quotes for the firm during the day, using data
from TAQ
Qt1_spread Prior quarter spread Log of quarter t - 1 average of daily spreads, where spread = Avg((OfferPrice - BidPrice)/((OfferPrice ? BidPrice)/
2)) 9 100 for all valid quotes for the firm during the day, using data from TAQ
Qt1_turn Prior quarter turnover Firm Turnover in quarter t - 1 minus market turnover in quarter t - 1, where turnover is the average daily share turnover,
defined as the average daily dollar volume divided by the average market value of shares outstanding for the firm, using
data from CRSP
Qt1_volat Prior quarter stock-return Standard deviation of Log(1 ? Daily Return), annualized by multiplying by H252, measured over quarter t - 1, using data
volatility from CRSP
ROA Performance Operating earnings scaled by assets, defined as Compustat items (ib/at)
SML_AVOL Small investors AVOL measured for small trades—includes trades B$10,000, using data from TAQ
XBRL XBRL adopters Indicator variable equal to one if the firm was mandated to adopt XBRL in the first year of adoption and zero otherwise
E. Blankespoor et al.
The market impact of the XBRL mandate
We also control for the information content and timing of filing. Specifically,
based on Bamber and Cheon (1995) and Bamber et al. (1997), we control for the
absolute value of the market reaction to the 10-K filing (Abs_abn_ret).14
Additionally, we control for the information content of the prior earnings
announcement (Abs_esurp), which we calculate as the absolute value of the
forecast error for the most recent consensus forecast prior to the event date. To
control for the potential effects of the timing of the 10-K, we include both the
number of days after the expected filing date (Daysafter_exp) and the number of
days after the earnings announcement (Daysafter_ea).15
Finally, we control for firm-specific characteristics that are likely to affect trading
activity. We include Lnmv to control for the market value of the firm (Atiase 1985;
Bamber 1987). To control for prior performance, we include the firm’s return on
assets (Roa). We include the ratio of book value to market value of assets (Btm) to
control for a firm’s growth opportunities, and we include the log of the firm’s stock
price (Lnprc) and the log of the number of shareholders (Lnown) to control for other
factors that may lead to differential trading effects.
The SEC elected to phase-in the requirement to file interactive data over 3 years,
where the largest corporate filers are first to adopt the technology. Specifically, all
domestic and non-U.S. large accelerated filers that used U.S. GAAP and had a
worldwide public float above $5 billion were required to file a separate XBRL
exhibit containing tagged financial statement items for the first periodic report
containing financial statements for fiscal periods ending on or after June 15, 2009.
We download all 10-K XBRL documents filed with the SEC starting June 15, 2009,
identifying 447 filings with fiscal periods between June 15, 2009, and June 14,
2010.16
We restrict our investigation to annual 10-K filings because Li and Ramesh
(2009) document that there is no significant stock market and volume reaction for
quarterly filings, after controlling for the concurrent release of earnings information.
To ensure our 10-K filings have significant investor interest, we examine the
average volume and market reactions on the filing days. In untabulated analyses, we
14
Note that bad news could affect our volume and spread results differently than could good news. As
such, in untabulated tests, we control for the sign of the short-window return around the event filing and
find qualitatively similar results.
15
Kross and Schroeder (1984) show that firms delay releasing bad news and the timeliness of that release
affects the market reaction (Chambers and Penman 1984), while Asthana et al. (2004) highlight the
importance of controlling for the delay after the earnings announcement.
16
Note that the SEC allows initial adopters a 30-day grace period in which to file their XBRL
documents. However, for our sample of mandatory XBRL adopters, only one firm filed its XBRL
documents later than its 10-K filing. For this firm, we set the filing date for our tests equal to the date the
XBRL documents were filed (one day later than the day the html 10-K was filed). Results are identical if
this firm’s observations are excluded.
123
E. Blankespoor et al.
Our XBRL sample of 698 firm-filings (349 firms with both a pre-XBRL filing and a
post-XBRL filing) is the starting point for our main tests. We use the year before the
first phase-in year of the mandate—i.e., June 15, 2008 through June 14, 2009—as
the pre-XBRL period, since this period represents the closest pre-XBRL time period
of the same duration. This pre-period enables us to determine whether there was a
shift in trading behavior after the adoption of XBRL. By using abnormal measures
and differencing out the control sample, we can better determine whether
information asymmetry and liquidity are greater or lower around 10-K filings for
17
Given these potential concerns, we do not formally incorporate voluntary filers into our primary
analyses. However, in untabulated tests, we repeat our main analyses after including a voluntary filer
indicator, XBRL_Voluntary, and interact it with Post to determine whether the effects of XBRL adoption
differ across voluntary and mandatory adopters. We find no significant difference between voluntary and
mandatory adopters (i.e., Post*XBRL_Voluntary is not statistically significant across all four of our main
tests—spread, price impact, volume, and large/small volume). However, since there are only 69 voluntary
adopters, it is difficult to conclude whether there really is no difference or whether there is not sufficient
power to determine the difference.
123
The market impact of the XBRL mandate
This table provides details about the sample selection of mandatory XBRL firms. There are several items
to note about sample selection. First, firms were technically required to file XBRL documents for fiscal
periods ending on or after June 15, 2009, starting with 10-Q filings. Therefore, if a firm’s first financial
statement filing after June 15, 2009 was a 10-K filing, the firm was able to delay the XBRL filing until the
first 10-Q filing. Accordingly, these firms are included in the non-XBRL sample since they were not
required to and did not file XBRL documents. Second, the 33 filings dropped for being voluntary adopters
in the mandatory period had market float of \$5 billion for both fiscal year 2009 and 2008. Ninety-three
XBRL-adopting firms in our sample had public float above the $5 billion cutoff for fiscal year 2008 but
not for fiscal year 2009. These firms were required to adopt XBRL for their 10-Q filings for fiscal quarters
ending after June 15, 2009, since fiscal year 2008 market float determined their filing status. When they
continued to file XBRL statements for fiscal year 2009 (even though their market float dropped below $5
billion), we count them as mandatory filers because they were required to file several XBRL 10-Qs prior
to the 10-K and were required to file XBRL 10-Qs and 10-Ks after June 15, 2010, as part of the second
phase of implementers
XBRL adopters than they would have been without XBRL. It is worth noting that
there is an overlap between our pre-XBRL period and the 2007–2008 financial
crisis. Figure 1 indicates that the last 6 months of 2008 have relatively negative
returns and increased volume compared to the 12-month period after XBRL
adoption (i.e., post period). However, only 4 % of the filings of the firms that
adopted XBRL (11 % (8 %) of the filings of the industry-size (industry-analyst)
control firms that did not adopt XBRL) were filed during the latter half of 2008, so
the vast majority of the filings in our sample came after the 2007–2008 financial
crisis and the related erratic market behavior.18
To facilitate identification and further mitigate any potential impact of the
financial crisis, we take advantage of the XBRL mandate’s phase-in period. This
phase-in provides a set of XBRL adopters and a set of non-adopters, allowing us to
implement a difference-in-difference research design, where we compare changes in
trading behavior across pre- and post-adoption regimes for the mandatory XBRL
adopters with those of our control groups. This design choice mitigates the impact
of market-wide non-XBRL related factors that may have occurred during the
transition period.
18
We do not have clear predictions as to any potential bias induced by the financial crisis, especially
given our use of abnormal market measures and a difference-in-difference design. Nevertheless, as a
robustness test, we remove firms with 10-K filings in the latter half of 2008 and repeat our analyses. We
find qualitatively similar results: an increase in abnormal spread, increase in abnormal price impact, and
decrease in abnormal volume for XBRL firms relative to matched sets of non-XBRL firms. In addition,
we remove financial firms (i.e., firms in SIC 6000 thru 6999 industries) from our analyses and find
qualitatively similar results.
123
E. Blankespoor et al.
This table provides details about the industry breakdown of mandatory XBRL firms by one-digit SIC
(Panel A) and descriptive statistics of XBRL and non-XBRL control samples (Panel B). XBRL is the
sample of XBRL adopters in our sample. All Non-XBRL represents all firms with available data that were
not required to adopt XBRL during the first phase-in period and is provided as a basis for comparison.
Industry and Size is the control sample based on the largest market value within an industry. Industry and
Analyst is the control sample based on the highest analyst following within an industry. Market Float is
the aggregate market value of equity held by non-affiliates as of the end of the second quarter of the fiscal
year being reported, as disclosed in the 10-K filing. Market Value is the market value of all outstanding
shares as of the end of the fiscal year. Assets is the total assets as of the end of the fiscal year (Compustat
atq). Number of Analysts is the number of analysts covering the firm, taken from the most recent
consensus analyst forecast measurement date in I/B/E/S (or FirstCall if data is not available in I/B/E/S)
prior to the earnings announcement date
123
The market impact of the XBRL mandate
Fig. 1 S&P pre- and post-period returns and volume. This figure shows volume and returns of the S&P
500 index from June 15, 2008 to June 15, 2010 (Source ^GSPC Yahoo Finance)
value as the selection criteria. Given that the requirement to file XBRL financial
statements is determined by a minimum market float, it is not surprising that the
average size for each industry comparison group in Panel B of Table 3 is smaller
than the XBRL group. However, note that the average matched control firm is still a
large company with an average market float approaching $3 billion, which is much
larger than the market float of all non-XBRL filers ($0.8 billion).
Second, we choose control firms based on industry and high analyst coverage to
obtain control firms with richer information environments. Panel B of Table 3
indicates that the average number of analysts for each of the two industry control
groups (9.7 and 11.9 for the industry-size and industry-analyst groups, respectively)
is close to the average for XBRL firms (15.6), suggesting that the control groups
have information environments that are reasonably similar to the XBRL firms.
In addition to calculating abnormal trading behavior measures and using a
difference-in-difference design, where we match firms on size, industry, and analyst
following, we include both market capitalization and analyst following in our
regressions and a variety of other control variables known to be correlated with
market formation as discussed previously in an attempt to rule out alternative
explanations and spurious correlations.19
To examine the specific effects of XBRL on trading behavior, we estimate the
following OLS regression model:
19
Although we use a difference-in-difference design and abnormal measures as well as include a variety
of control variables, an unobservable effect within industries (our matching group) may remain.
Therefore, in untabulated analyses, we repeat our analyses including indicator variables for each three-
digit SIC (i.e., industry fixed effects) following Cram et al. (2009) and find that our results still hold at
statistically significant levels.
123
E. Blankespoor et al.
X
MKT VAR ¼ a0 þ a1 POST þ a2 XBRL þ a3 POST XBRL þ ai CONTROLS þ e;
ð1Þ
where MKT_VAR is defined as either abnormal bid-ask spread (ASPREAD),
abnormal price impact of trade (AIMPACT), or abnormal trading volume (AVOL).
XBRL is an indicator variable that equals one if the firm was required to adopt
XBRL during the initial phase-in period and zero for control firms. POST is equal to
one if the fiscal year ends on or after June 15, 2009, while filings that occur in the
year prior are coded as zero. The interaction term, POST*XBRL, therefore captures
the impact of XBRL on trading behavior (i.e., spread, price impact of trade, or
trading volume). In addition, we cluster standard errors by firm.20 CONTROLS
represents the control variables defined in Sect. 3.
4.3 Results
Table 4 reports the multivariate regression results for the effects of XBRL on
abnormal bid-ask spreads (ASPREAD). Columns 1 and 2 (3 and 4) report the
coefficient estimates and p values from estimating Eq. (1), where the matched
sample is the industry-size (industry-analyst) control group. The coefficient
estimates on POST*XBRL are significantly positive across both tests (both p values
\0.01). We thus find evidence that XBRL adopters experience higher abnormal
spreads relative to a matched sample of non-adopters.21 We next examine the effect
of XBRL on the abnormal price impact of trade (AIMPACT) in Table 5 and
abnormal trading volume (AVOL) in Table 6. In Table 5, we find that the coefficient
estimates on POST*XBRL across both tests are significantly positive (both p values
\0.01). This evidence shows an increase in our price impact measure relative to
both industry control groups, which suggests that as liquidity decreases, investors’
unwillingness to trade results in greater price pressure—i.e., investors’ ability to
trade at a low cost declines. In Table 6, we further enhance our understanding of
shifts in trading behavior by examining trading volume. The results indicate that
20
We cluster standard errors by firm to control for time-series correlation across a given firm’s two
observations. We do not use firm fixed effects because including firm fixed effects forces us to exclude the
XBRL indicator variable, as the two variables are linear combinations of each other. However, we repeat
our main analyses using firm fixed effects, and the results are qualitatively similar. For cross-sectional
correlation, the POST variable is equivalent to including time fixed effects using a fiscal year indicator
variable, since there are only two years of observations. We do not cluster standard errors by time because
there are only two years of observations and thus too few clusters (Petersen 2009; Gow et al. 2010).
However, we cluster by the filing date as a robustness test, and we find qualitatively similar results.
21
In addition to increasing bid-ask spread, liquidity suppliers can address adverse selection concerns by
adjusting the number of shares they are willing to trade (Leuz and Wysocki 2008; Lee et al. 1993).
Accordingly, we examine firms’ market depths around the 10-K filings. We measure abnormal depth as
the log of the average daily depth during the event period minus the log of the average daily depth during
the nonfiling period, where the daily depth is the daily average of each quote’s depth, calculated as the
sum of the dollar offer size and the dollar bid size. We find that abnormal depths decline for our XBRL
firms, as compared to both of our matched sets of non-XBRL adopters. This result provides further
evidence that adverse selection concerns have increased for XBRL-adopting firms around 10-K filings.
123
The market impact of the XBRL mandate
Table 4 Results of regressing abnormal bid-ask spread around 10-K filing dates on XBRL Adoption
Dependent variable Abnormal bid-ask spread (ASPREAD)
Main variables
Post 0.0015 (0.970) 0.0267 (0.447)
XBRL -0.0812* (0.051) -0.0670** (0.042)
Post * XBRL 0.1581*** (0.001) 0.1212*** (0.002)
Control variables
Lnmv 0.0274 (0.530) 0.0301 (0.422)
Btm 0.0060 (0.953) 0.0673 (0.184)
Inst_Hold 0.0223 (0.619) 0.0003 (0.993)
Daysafter_exp 0.0013 (0.695) 0.0039 (0.155)
Daysafter_ea -0.0015* (0.069) 0.0009 (0.323)
Abs_abn_ret 0.5526** (0.019) 0.7468*** (0.008)
Lnanalyst -0.0687* (0.091) 0.0134 (0.675)
Lnown -0.2675** (0.040) 0.0494 (0.848)
Lnprc -0.0314 (0.540) -0.0261 (0.337)
Qt1_turn -1.6979 (0.374) 0.4454 (0.789)
Qt1_volat 0.0675 (0.257) -0.0030 (0.961)
Qt1_depth 0.0123 (0.869) -0.0266 (0.529)
Abs_esurp -0.3696 (0.110) -0.0043 (0.986)
Roa 0.2165 (0.396) 0.1553 (0.269)
Intercept -0.1270 (0.861) -0.0426 (0.888)
F-test
Post ? Post*XBRL 0.1596*** (0.000) 0.1479*** (0.000)
N 1,380 1,380
Adj. R2 0.020 0.032
This table provides the results of regressing abnormal bid ask spread (ASPREAD) on an XBRL-adopter
indicator, a post-/pre- period indicator, their interaction, and control variables, using 349 mandatory
XBRL-adopting firms and the two non-XBRL control groups (341 industry-size and industry-analyst
matched firms) for 2008 and 2009. Two-sided p values are provided in parentheses to the right of the
coefficients. All variables are defined in Table 1 and are winsorized at the 1 and 99 % level. Standard
errors are robust and clustered by firm
*, **, and *** Significance at the 10, 5, and 1 % or lower levels, respectively
abnormal trading volume decreases for XBRL firms relative to both industry control
groups, since coefficient estimates on POST*XBRL are significantly negative (both
p values \0.01).
As previously discussed, our difference-in-difference design implemented in
Tables 4, 5, and 6 enables us to account for any market-wide changes across the
periods examined. As such, we are primarily interested in the impact of the
regulation on XBRL firms relative to our matched groups (i.e., Post*XBRL). To
123
E. Blankespoor et al.
Table 5 Results of regressing abnormal price impact of trade around 10-K filing dates on XBRL
adoption
Dependent variable Abnormal price impact (AIMPACT)
Main variable
Post -0.3370*** (0.000) -0.5356*** (0.000)
XBRL -0.2588*** (0.002) -0.4219*** (0.000)
Post * XBRL 0.4465*** (0.001) 0.6829*** (0.000)
Control variable
Lnmv -0.0406 (0.393) -0.0571 (0.303)
Btm 0.3394 (0.105) 0.4746*** (0.009)
Inst_Hold 0.0672 (0.525) 0.1355 (0.272)
Daysafter_exp 0.0061 (0.399) 0.0125 (0.185)
Daysafter_ea 0.0043** (0.038) 0.0046 (0.131)
Abs_abn_ret 1.5793 (0.102) 4.4987*** (0.000)
Lnanalyst -0.0239 (0.731) 0.0476 (0.640)
Lnown -0.0317 (0.908) 0.7400 (0.260)
Lnprc 0.0205 (0.692) -0.0112 (0.851)
Qt1_turn -8.9497** (0.049) -2.6312 (0.603)
Qt1_volat 0.3198 (0.218) -0.1089 (0.667)
Abs_esurp -1.1253 (0.243) -1.1797 (0.197)
Roa 0.7666 (0.286) 0.5388 (0.358)
Intercept -0.0516 (0.935) 0.0696 (0.906)
F-test
Post ? Post*XBRL 0.1095 (0.234) 0.1473* (0.070)
N 1,262 1,286
Adj. R2 0.063 0.117
This table provides the results of regressing abnormal price impact (AIMPACT) on an XBRL-adopter
indicator, a post-/pre- period indicator, the interaction of the two indicators, and control variables, using a
sample of 333 mandatory XBRL-adopting firms and each of the two non-XBRL control groups (298
(310) industry-size (industry-analyst) matched firms), each with 2 years represented (2008 and 2009).
These samples are slightly smaller because we require observations to have trading during at least 2/3 of
the 5-min intervals during the trading day. Two-sided p values are provided in parentheses to the right of
the coefficients. All variables are defined in Table 1 and are winsorized at the 1 and 99 % level. Standard
errors are robust and clustered by firm
* Significance at the 10 % level; ** at the 5 % level; and *** at the 1 % or lower level
observe the impact of the regulation on XBRL firms irrespective of the control
firms, we sum the coefficients on Post and Post*XBRL in each of the previously
discussed tables. The results in Table 4 indicate that for both samples there is a
significant overall increase in abnormal spreads after firms adopt XBRL (both
p values\0.01). The sum of the two coefficients in Table 5 are positive, but weaker,
123
The market impact of the XBRL mandate
Table 6 Results of regressing abnormal volume around 10-K filing dates on XBRL adoption status
Dependent variable Abnormal price impact (AIMPACT)
Main variables
Post 0.3602*** (0.000) 0.3898*** (0.000)
XBRL 0.4223*** (0.000) 0.4562*** (0.000)
Post * XBRL -0.5406*** (0.000) -0.5853*** (0.000)
Control variables
Lnmv -0.1104* (0.063) -0.0161 (0.792)
Btm -0.3470 (0.154) -0.2564 (0.155)
Inst_Hold -0.0151 (0.907) 0.0543 (0.681)
Daysafter_exp 0.0025 (0.717) 0.0093 (0.206)
Daysafter_ea -0.0336*** (0.000) -0.0345*** (0.000)
Abs_abn_ret 6.1410*** (0.000) 4.8824*** (0.000)
Lnanalyst 0.1426* (0.053) 0.0907 (0.341)
Lnown -0.2945 (0.613) -0.5777 (0.327)
Lnprc 0.0843 (0.151) 0.0693 (0.298)
Qt1_volat -0.6389*** (0.002) -0.4789*** (0.008)
Qt1_depth -0.0545 (0.450) -0.1397* (0.100)
Qt1_spread -0.0406 (0.323) 0.0071 (0.878)
Abs_esurp 0.7840 (0.687) 0.8708 (0.592)
Roa -0.1514 (0.872) 0.0243 (0.963)
Intercept 1.8850** (0.019) 1.9730*** (0.010)
F-test
Post ? Post*XBRL -0.1804** (0.030) -0.1955** (0.013)
N 1,380 1,380
Adj. R2 0.252 0.216
This table provides the results of regressing abnormal volume (AVOL) on an XBRL-adopter indicator, a
post-/pre- period indicator, the interaction of the two indicators, and control variables, using 349 man-
datory XBRL-adopting firms and the two non-XBRL control groups (341 industry-size and industry-
analyst matched firms) for 2008 and 2009. Two-sided p values are provided in parentheses to the right of
the coefficients. All variables are defined in Table 1 and are winsorized at the 1 and 99 % level. Standard
errors are robust and clustered by firm
*, **, and *** Significance at the 10, 5, and 1 % or lower levels, respectively
for both groups (p values 0.23 and 0.07, respectively) providing evidence of reduced
liquidity for XBRL firms. Finally, the sum of Post and Post*XBRL in Table 6 is
significantly negative across both tests (p values 0.03 and 0.01, respectively), which
indicates that abnormal volume decreased in XBRL firms after adoption. Taken
together, the evidence in Tables 4, 5, and 6 is consistent with an increased concern
of adverse selection and reduced liquidity.
123
E. Blankespoor et al.
5 Additional analyses
123
The market impact of the XBRL mandate
Table 7 Results of regressing abnormal large and small trading volume around 10-K filing dates on
XBRL adoption status
Industry-size Industry-analyst
Main variables
Post 0.3173*** 0.2256** 0.3017*** 0.2865***
(0.001) (0.014) (0.000) (0.002)
XBRL 0.2981*** 0.5934*** 0.3465*** 0.5826***
(0.003) (0.000) (0.000) (0.000)
Post * XBRL -0.2881** -0.7037*** -0.2599** -0.7755***
(0.011) (0.000) (0.017) (0.000)
Diff: v2 statistic (p value) 15.5*** (.00) 27.3*** (.00)
Control variables
Lnmv -0.1075* -0.0039 -0.0153 0.0552
Btm -0.4549** -0.0566 -0.3407** -0.0254
Inst_hold -0.0674 -0.0423 -0.0302 0.0131
Daysafter_exp -0.0020 0.0048 -0.0007 0.0111
Daysafter_ea -0.0193*** -0.0358*** -0.0190*** -0.0355***
Abs_abn_ret 4.1010*** 6.2432*** 3.9941*** 5.1782***
Lnanalyst 0.0327 0.0208 0.0579 -0.0377
Lnown 0.0037 -0.4519 -0.1666 -1.0065
Lnprc -0.0257 0.0837 -0.0777 0.1403**
Qt1_volat -0.4940** -0.7111*** -0.4713*** -0.4861***
Qt1_depth 0.0176 -0.2261*** -0.0832 -0.2114**
Qt1_spread -0.0696** -0.1082*** 0.0101 -0.0522
Abs_esurp 0.7625 0.3313 0.7597 0.3867
Roa 0.0719 0.0783 0.1609 0.1026
Intercept 1.4791** 3.1962*** 1.6275** 2.3589***
v2 test: Post ? Post*XBRL 0.0292*** -0.4781*** 0.0418*** -0.4890***
0.000 (0.000) 0.000 (0.000)
N 1,376 1,378
This table provides the results of using seemingly unrelated regression to simultaneously regress
abnormal large trading volume (LRG_AVOL) and abnormal small trading volume (SML_AVOL) on an
XBRL-adopter indicator, a post-/pre- period indicator, their interaction, and control variables, using a
sample of 348 mandatory XBRL-adopting firms and each of the two non-XBRL control groups (340
(341) industry-size (industry-analyst) matched firms) for 2008 and 2009. These samples are slightly
smaller because firms with zero large trades in the pre-period have an undefined abnormal large trading
volume ratio. If there are zero large trades in the event and pre-periods, we set the abnormal ratio equal to
zero. However, if there are large trades in the event window but zero in the pre-period, we drop that firm’s
observations from the regressions because the ratio is undefined and there is not a reasonable proxy for its
value. Two-sided p values are provided in parentheses beneath the coefficients for variables of interest,
and for parsimony, stars are provided to indicate coefficient significance for the remaining control
variables. Chi squared statistics and related p values are provided for tests of the difference between the
Post*XBRL coefficients for large versus small abnormal trading. All variables are defined in Table 1 and
are winsorized at the 1 and 99 % level. Standard errors are robust and clustered by firm
*, **, and *** Significance at the 10, 5, and 1 % or lower levels, respectively
123
E. Blankespoor et al.
Table 7 provides the SUR results for small and large investors’ trades. We find
small investors are less likely to trade shares of mandatory XBRL adopting firms
around 10-K filings (both p values\0.01). Interestingly, we find some evidence that
large traders are also less likely to trade shares of XBRL adopters relative to the
control sample (p values = 0.01 and 0.02, respectively).26 However, the reduction
in trade is roughly two to three times greater for small trades (all p values \0.01).
This reduction in small trades is consistent with smaller investors having increased
concerns of adverse selection as a result of the new technology.
In our main analyses, we find that XBRL adopting firms experience higher adverse
selection concerns (i.e., higher abnormal spreads) and lower liquidity (i.e., higher
price impact of trade) during the (-1, 3) window around the 10-K filing. To gain
insights into how long these effects persist, we examine the effects over the post-
filing period to determine when they become insignificant. Panels A and B of Fig. 2
provide the coefficient estimates for our difference-in-differences estimator,
POST*XBRL, over our 5-day window (-1, 3) as well as the next four 5-day
windows (4, 8) (9, 13) (14, 18), and (19, 23), with respect to spread and price impact
of trade, respectively. The POST*XBRL coefficient captures the overall effect of
XBRL on adopting firms relative to those that did not adopt the technology, so the
lines in the figures in Panels A and B represent the amount of adverse selection and
liquidity, respectively, associated with XBRL.
As shown in Panels A and B of Fig. 2, the amount of adverse selection and
liquidity declines monotonically over the post-filing period. Specifically, the
increased adverse selection effect becomes insignificant by the (14, 18) window,
and the reduced liquidity effect becomes insignificant by the (19, 23) window,
which suggests that our results are event specific and thus do not appear to represent
a broader fundamental shift in the heterogeneity in investors’ information sets. In
particular, this evidence suggests that XRBL is allowing informed traders to exploit
their information advantage in short-window events, but that any information
advantage dissipates in medium to long windows as smaller investors reduce the
information gap over time. The evidence also indicates that information asymmetry
returns to pre-event levels in the long run, suggesting that XBRL has not (yet)
served its purpose of reducing information asymmetry.
In our main analyses, we focus on the initial year of adoption because it is the
cleanest setting; however, we cannot make inferences about the long-run
consequences of XBRL by looking only at the initial year of adoption, as some
26
Perhaps the increased market frictions (higher spreads) and greater price impact of trading reduce
larger investors’ ability or willingness to trade as actively around 10-K filings of XBRL firms.
123
The market impact of the XBRL mandate
Fig. 2 Duration of Effect Panel A (B) provides the coefficient for Post*XBRL from regressions of
abnormal spread (abnormal price impact) on Post*XBRL and control variables over varying 5-day
windows around the 10-K filing date, using the industry-size and industry-analyst matched control
samples. The actual coefficients and related two-sided p values are provided below the charts. All
variables are winsorized at the 1 and 99 % level. Standard errors are robust and clustered by firm. *, **,
and *** indicate significance at the 10, 5, and 1 % or lower levels, respectively. Panel B Post*XBRL
coefficient from abnormal price impact regression over various 5-day windows
investors may face a learning curve with the new technology, while others may just
be hesitant to use the data initially. To provide at least some additional insights into
the long-term impact of XBRL, we track the initial adopters (i.e., firms adopting in
year one, filing between June 15, 2009, and June 14, 2010, or fiscal 2009) over the
following 2 years (i.e., fiscal 2010 and 2011) to observe whether the abnormal
trading behavior in the short window around 10-K filings persists over the following
years. Specifically, we substitute years two and three of the transition period as the
post-adoption periods for XBRL adopters, using the original control sample and
periods, which allows us to more directly compare the change in trading behavior
for initial adopters across years. Table 8 presents the results of this analysis.
As indicated in Panel A of Table 8, the higher abnormal spreads around the 10-K
filing decline from fiscal 2009 to 2011 across both control groups; in fact, the
123
E. Blankespoor et al.
Panel B—results of regressing abnormal price impact on Post*XBRL and control variables
Industry-size match 0.4465*** 0.5131*** 0.3718***
(0.001) (0.000) (0.001)
Difference from 2009 0.0666 -0.075
P value for difference from 2009 (0.006) (0.068)
Industry-analyst match 0.6829*** 0.7212*** 0.6708***
(0.000) (0.000) (0.000)
Difference from 2009 0.0383 -0.012
P value for difference from 2009 (0.117) (0.781)
This table provides the results of regressing abnormal bid ask spread (ASPREAD) (Panel A) and abnormal
price impact (AIMPACT) (Panel B) on an XBRL-adopter indicator, a post-/pre-period indicator, their
interaction, and control variables, using industry-size and industry-analyst matched control samples.
Column 2009 uses fiscal 2009 as the post-adoption year for the XBRL firms (consistent with all previous
tables), column 2010 uses fiscal 2010 as the XBRL post-adoption year, and column 2011 uses fiscal 2011
as the XBRL post-adoption year. All three columns use the same sample of XBRL firms that first adopt in
2009, and the market conditions for these firms in each of the post-adoption years is compared to the same
XBRL pre-adoption year of 2008, as well as the same pre- and post-adoption years (i.e., 2008 and 2009)
for the non-XBRL adopting firms to provide a consistent baseline. Two-sided p values are provided in
parentheses directly below the coefficients. In addition, two-sided p values for a test of difference from
2009 coefficients are provided below column 2010 and 2011. These p values are a result of performing a
seemingly unrelated regression using both windows and testing for a difference in the coefficients. All
variables are defined in Table 1 and are winsorized at the 1 and 99 % level. Standard errors are robust and
clustered by firm
*, **, and *** Significance at the 10, 5, and 1 % or lower levels, respectively
coefficient estimates in fiscal 2011 are roughly half those in fiscal 2009. Panel B of
Table 8 shows a reduction in the price impact of trade from fiscal 2009 to 2011
using the industry-size matched sample; however, there does not appear to be a
decline in the price impact of trade using the industry-analyst matched sample.
Collectively, the evidence in Table 8 is consistent with the impact of XBRL being
more than an initial year phenomenon but also supports the notion that some
investor learning may be taking place, leading to lower adverse selection concerns
in the latter part of the adoption period.
123
The market impact of the XBRL mandate
0.2 0.2
0.15 0.15
0 0
(-1,+3) (4,8) (9,13) (14,18) (19,23) (-1,+3) (4,8) (9,13) (14,18) (19,23)
-0.05 -0.05
-0.1 -0.1
Post * XBRL Coefficient
(-1 ,+ 3 ) (4 , 8 ) (9 ,13 ) (1 4 , 1 8 ) (1 9 , 2 3 ) (-1 ,+3 ) (4 , 8 ) (9 , 1 3 ) (1 4,1 8) (1 9 , 2 3 )
2009 0.1581 *** 0.1305 *** 0.1090 * 0.0811 0.0367 2009 0.1212 *** 0.1616 *** 0.0923 * 0.0376 0.0217
(p-value) (0.001) (0.007) (0.059) (0.147) (0.462) (p-value) (0.002) (0.001) (0.095) (0.427) (0.648)
2010 0.1100 ** 0.1749 *** 0.2029 *** 0.1578 *** 0.1213 ** 2010 0.0650 * 0.1742 *** 0.1741 *** 0.1166 ** 0.0983 **
(p-value) (0.012) (0.000) (0.001) (0.008) (0.022) (p-value) (0.078) (0.000) (0.003) (0.026) (0.046)
2011 0.0858 ** 0.0698 0.0246 0.0168 -0.0537 2011 0.0641 * 0.0901 * -0.0056 -0.0251 -0.0654
(p-value) (0.033) (0.115) (0.636) (0.749) (0.276) (p-value) (0.084) (0.056) (0.906) (0.572) (0.138)
0.8
0.6
0.7
0.5
0.6
0.4 2009 0.5 2009
2010 2010
0.3 0.4
2011 2011
0.3
0.2
0.2
0.1
0.1
0 0
(-1,+3) (4,8) (9,13) (14,18) (19,23) (-1,+3) (4,8) (9,13) (14,18) (19,23)
Post * XBRL Coefficient
(-1 ,+ 3 ) (4 , 8 ) (9 ,13 ) (1 4 , 1 8 ) (1 9 , 2 3 ) (-1 ,+3 ) (4 , 8 ) (9 , 1 3 ) (1 4,1 8) (1 9 , 2 3 )
2009 0.4465 *** 0.5140 *** 0.4535 *** 0.2821 *** 0.0482 2009 0.6829 *** 0.7409 *** 0.6474 *** 0.3283 *** -0.0119
(p-value) (0.001) (0.000) (0.000) (0.009) (0.681) (p-value) (0.000) (0.000) (0.000) (0.007) (0.925)
2010 0.5131 *** 0.6336 *** 0.5968 *** 0.3544 *** 0.0629 2010 0.7212 *** 0.8425 *** 0.7357 *** 0.3713 *** -0.0127
(p-value) 0.000 (0.000) (0.000) (0.002) (0.615) (p-value) (0.000) (0.000) (0.000) (0.003) (0.922)
2011 0.3718 *** 0.5352 *** 0.4634 *** 0.2985 *** 0.0687 2011 0.6708 *** 0.8039 *** 0.6786 *** 0.3420 *** 0.0187
(p-value) (0.001) (0.000) (0.000) (0.002) (0.508) (p-value) (0.000) (0.000) (0.000) (0.005) (0.884)
Fig. 3 Duration of effect, subsequent years. Panel A (B) of this figure provides the coefficient for
Post*XBRL from regressions of abnormal spread (abnormal price impact) on Post*XBRL and control
variables over varying 5-day windows around the 10-K filing date for 2009, 2010, and 2011, using the
industry-size and industry-analyst matched control samples. The actual coefficients and related two-sided
p values are provided below the charts. All variables are winsorized at the 1 and 99 % level. Standard
errors are robust and clustered by firm. *, **, and *** indicate significance at the 10, 5, and 1 % or lower
levels, respectively. Panel B Post*XBRL coefficient from abnormal price impact regression over various
5-day windows and subsequent years
For completeness, we also examine the duration of the market impacts for fiscal
2010 and 2011 (in addition to fiscal 2009) to determine whether there are any trends
over subsequent years. As shown in Panel A of Fig. 3, the duration of the increased
adverse selection concerns declines from 2009 to 2011. In particular, the effect
123
E. Blankespoor et al.
becomes insignificant at the (4, 8) window using the industry-size matched sample
and (9, 13) window using the industry-analyst matched sample in 2011, compared to
the (14, 18) window in 2009. However, Panel B of Fig. 3 indicates that the duration
of the increased price impact of trade appears to remain constant over the 3 years.
These results provide further evidence that, while the effects persist, they may be
getting muted over more recent years.
To rule out alternate explanations related to differences across XBRL and non-
XBRL firms during our sample period that are unrelated to XBRL adoption, we
conduct our main analyses on a different, but related, information event that does
not incorporate XBRL: earnings announcements. If the difference between the
XBRL firms and the non-XBRL control firms is attributable to something other than
XBRL, then we should observe a similar pattern around this alternate information
event during the same period.
For the firm-year observations in our sample, we identify the related earnings
announcement date and re-estimate the market variables using the (-1, ?3) window
around this date as the event period.27 We find that, in stark contrast to our main
results, only one of the Post*XBRL coefficients is significantly different from zero
across the abnormal spread, abnormal price impact, and abnormal volume
(including large and small trades) falsification tests around earnings announcements,
consistent with the notion that our main results are attributable to XBRL adoption.
To address any additional concerns that our results may be attributable to firm size,
we provide a falsification test in which we restrict our sample to XBRL adopters
only and examine changes in investor trading activity for the largest XBRL firms
relative to the smallest XBRL firms. If our results are attributable to size rather than
XBRL, we would expect to see differential market frictions between large and
small XBRL firms, similar to that which we observed between XBRL and non-
XBRL firms in our main analyses—i.e., higher abnormal spreads, higher abnormal
price impact, and lower abnormal trading volume. To test this conjecture, we repeat
our main analyses using large XBRL firms (those with market floats above the
median XBRL market float) as the treatment sample and small XBRL firms (those
with market floats below the median XBRL market float) as the control sample.
Although we find a significant difference in size across the two groups—mean
market float of $23.4 billion ($6.5 billion) for the large (small) XBRL firms—we do
27
To ensure that the earnings announcement event period was not affected by the XBRL filing, we drop
observations where the filing date fell during the earnings announcement event window. After this
change, the earnings announcement sample is approximately 1,000 observations or 80 % of the original
filing sample.
123
The market impact of the XBRL mandate
not find a difference in abnormal volume and price impact, nor do we find a
significant decrease in abnormal spread for the large XBRL firms compared to the
small XBRL firms. These results provide additional evidence consistent with the
notion that the increase in market frictions is related to XBRL adoption and not
market float.
In our main analysis, we attempt to control for the difference in the information
environments of XBRL and non-XBRL firms by matching on industry-size and
industry-analyst and by including firm size and information intermediary control
variables in our tests. However, there may be other factors that are more directly
associated with float that are related to a firm’s information environment. For
example, firms with more than $700 million float are considered ‘‘large accelerated
filers’’ by the SEC and are subject to a 60-day (rather than 75-day or 90-day) 10-K
filing deadline. Accordingly, as an alternate test, we select our non-XBRL industry-
matched sample using only firms that have a public market float of at least $700
million. For symmetry, we also remove XBRL firms with a market float above $9.3
billion to bound the range to the $5 billion XBRL cutoff plus or minus $4.3 billion.
These differences are statistically significant for all our dependent variables (i.e.,
abnormal spread increases, abnormal price impact increases, and abnormal volume
decreases).
We also perform two similar tests that use a subsample of XBRL and non-XBRL
firms. For the first test, we examine the smallest XBRL firms and the largest non-
XBRL firms based on market float. Specifically, we keep XBRL firms with floats
below the median XBRL firm float and select our non-XBRL industry-matched
sample from the largest non-XBRL firms, resulting in a mean market float of $6.5
billion ($3.6 billion) for the XBRL (non-XBRL) firms. For the second test, we
examine the XBRL firms with the smallest analyst coverage and the non-XBRL
firms with the largest analyst coverage by keeping XBRL firms with analyst
following below the median XBRL firm number of analysts and selecting our non-
XBRL industry-matched sample from the non-XBRL firms with the largest analyst
following. After these partitions, the average analyst following is 11.6 (12.0) for the
XBRL (non-XBRL) firms. Although both of these tests reduce our sample by
approximately 50 %, our results still hold for both alternate sample cuts at
statistically significant levels.
123
E. Blankespoor et al.
controls for the quantity and complexity of the information in the 10-K. Specifically,
we control for the log of the total number of words in the report, after implementing
cleaning techniques similar to those described by Miller (2010), to capture the
amount of data in text format, the log of the sum of the number of words, and the
number of table cells as a broader measure of the amount of information provided in
the report (Miller 2010), and the FOG score (as described in 2008) to capture the
complexity or readability. Our results are nearly identical.28
In our main analyses, we cluster standard errors by firm to account for correlation in
the residuals across firm observations. We do not cluster by year because there are
only 2 years in our sample, and to create consistent estimates of standard errors, the
use of at least 10–50 clusters (i.e., years, in this case) is recommended (Petersen
2009; Gow et al. 2010). However, because our tests focus on market activity on
filing dates and filing dates tend to cluster together over the year, there could be
correlation in the residuals of firms filing on the same dates. Therefore, we repeat
our abnormal spread, price impact, and volume analyses using standard errors
clustered by firm and by filing date (two-way clustering), and we find that our results
still hold at statistically significant levels.
7 Conclusion
In April 2009, the SEC mandated that companies report their financial statement
data using XBRL. The SEC asserts that this new search-facilitating technology will
improve the usability of financial disclosures, thereby reducing information
asymmetry. Given investors’ differential abilities to take advantage of the new
technology, the impact of the regulation is unclear. Although the SEC argues that
XBRL provides more benefits to small investors than it does to large investors, the
potential for large traders to use their superior processing capabilities to leverage the
new technology for informational gains may actually disadvantage small investors.
As a result, we investigate the effects of the initial phase-in of the mandatory XBRL
filing requirement on market behavior—bid-ask spread, price impact of trade, and
trading volume—around 10-K filings.
Using a difference-in-difference design to examine market effects resulting from
mandatory XBRL adoption in the initial year, we find greater market frictions for
mandatory XBRL adopters relative to two matched sets of firms. In particular, we
find evidence of greater abnormal bid-ask spreads, increased abnormal price impact
of trade, and lower abnormal trading volume for adopting firms than for non-
adopting firms after XBRL implementation. Further, we find that the reduction in
abnormal trading volume is much larger for small traders.
28
We also repeat the tests including all three disclosure variables simultaneously and find very similar
results.
123
The market impact of the XBRL mandate
We also examine spread and the price impact of trade (i.e., our main measures of
adverse selection and liquidity, respectively) over a longer window to determine
how long the effects we document in our main analyses persist outside the short-
term window around the 10-K. We find that these effects diminish and become
statistically insignificant over the 23-trading day period following the filing.
Extending our analyses to the following 2 years, we find generally similar results to
those in the initial adoption year, except the higher abnormal spreads for XBRL
firms in the short window are significantly smaller relative to the initial year and the
duration of the effect is much shorter in the third year.
Our results are consistent with the impact of XBRL being more than an initial
year phenomenon, but they are also consistent with the notion that some investor
learning may be taking place leading to lower adverse selection concerns in the
latter part of the adoption period. This suggests that perhaps the incorporation of
XBRL into the market has not reached equilibrium yet. Although we do not find
evidence of a reduction in information asymmetry during the first few years
following the mandate, this might change in future years, as the technology becomes
more accessible. We leave the investigation of the long-term consequences of
XBRL adoption, including the impacts of nonstandard tagging and advancements in
XBRL viewers on investor data aggregation efficiencies, to future research.
Acknowledgments We are grateful to two anonymous referees, Salman Arif, Daniel Beneish, John
Core, Anna Costello, Patricia Dechow (Editor), Paul Fischer, Max Hewitt, Charles Lee, Laureen Maines,
Greg Miller, Mike Minnis, Joe Piotroski, Marlene Plumlee, Tianshu Qu, Cathy Schrand, Cathy
Shakespeare, Nemit Shroff, Dan Taylor, Jim Vincent, Chris Williams, Teri Yohn and workshop
participants at Indiana University, Ohio State University, Santa Clara University, University of
California—Berkeley, University of Chicago, University of Florida, University of Miami, University of
Michigan, University of North Carolina, University of Pennsylvania (Wharton), University of Southern
California, University of Utah, Washington University and the 2011 Stanford Summer Camp for helpful
discussions. We would also like to thank Bob Rand at the SEC for assistance with identifying XBRL
filings. Elizabeth Blankespoor gratefully acknowledges financial support from the Deloitte Doctoral
Fellowship, Brian Miller gratefully acknowledges financial support from the Arthur Weimer Faculty
Fellowship, and Hal White gratefully acknowledges financial support from Ernst and Young.
Financial statements are comprised of numbers that can provide information about a
firm’s performance and value. To understand the basic meaning of a number,
though, users (including data aggregators) must read the contextual information
surrounding the number. For example, a textual analysis language such as Perl,
could find the number ‘‘167’’ in a set of financial statements, but the program would
not know whether that number represents the firm’s sales, assets, number of
employees, mailing address, or even the page number of the document. For a
computer to understand the meaning of ‘‘167,’’ the user would have to program it to
look before, after, above, and/or below for the description (e.g., ‘‘Product Sales’’),
the year (e.g., 2009), the currency (e.g., USD), the denomination (e.g., millions),
etc. Because firms use a variety of formats, it is often difficult to automatically
identify the relevant descriptions or ‘‘understand’’ the number. In addition,
comparing data items across firms can be difficult if firms use slightly different
123
E. Blankespoor et al.
Advantages of XBRL
The SEC discusses several advantages of XBRL in its 2009 Final Rule, including
(1) more efficient processing, (2) a more comprehensive set of data than provided by
aggregators, and (3) improved comparability across filings.29
First, the data is available in a less costly and timelier fashion. Once the setup
costs have been incurred, the costs of processing the data in XBRL filings should be
greatly reduced. Specifically, instead of hand-coding the information from html
filings, investors can rely on computer-automated tools to access and compare data
across many filings in a fraction of the time, leaving more time for analysis.
Christopher Whalen (2004), co-founder of Institutional Risk Analytics, claims that
‘‘Putting financial data in ready-to-crunch condition might seem a trivial detail until
you consider that Wall Street currently spends 80 % of its time and money in data
mining and 20 % on actual analysis—what data pros call the ‘80–20 rule’.’’ Whalen
adds that standardized data format could flip the 80–20 rule, allowing analysts to
spend most of their time assessing rather than gathering financial data.
Second, the level of detail available to investors is more flexible and much richer
with XBRL than with data aggregators. In fact, even during the first year of adoption
when detailed tagging of footnotes was not fully implemented, many useful items
29
An additional benefit of XBRL lauded by the SEC is that the data should be more accurate than that
provided by data aggregators or manually input. This benefit has been challenged by some noting that
there are errors in some of the XBRL tags. According to McCann (2010), these errors affect about 1 % of
the tags, but the vast majority of the errors are easily fixable with a proper algorithm (e.g., incorrect sign).
Although the cost is likely minor, the existence of these tagging errors is an additional cost for investors
implementing the new technology.
123
The market impact of the XBRL mandate
Name: SalesRevenueGoodsNet
Label: Products Sales per
Data Aggregator
Name: SalesRevenueGoodsNet
Label: Sales
Sales per
Data Aggregator
tagged in the financial statements were not available through data services (e.g.,
Factset, Bloomberg, etc.). For example, revenue breakdown (e.g., products,
services, financing), cost of revenue breakdown (e.g., cost of products, cost of
services, financing interest), receivables breakdown (e.g., trade, financing, short and
long term), and breakdowns of provision for bad debt and provision for inventory
are available with XBRL but not available through traditional data services. Since
data aggregators do not capture all information in the filings, investors using XBRL
filings are at a relative advantage. Below we illustrate the benefits of this additional
detailed data by comparing Hewlett-Packard (HP) and International Business
Machines (IBM) financial disclosures.
For an investor interested in comparing HP and IBM for forecasting and
valuation purposes, the first item of interest is typically revenue. If the investor uses
information from a common data aggregator, he or she would be limited to total
revenue in 2009 of $114,552 million and $95,758 million for HP and IBM,
respectively (see Fig. 4). However, using XBRL tagging, investors can identify the
revenue on the Consolidated Statements of Operations in the 10-K filing broken
down by source (Goods, Service, and Financing). This information is useful to
investors as they can now observe that 65 % of HP’s revenues are from selling
goods, while 40 % of IBM’s revenues are from selling goods. This is but one detail
provided by XBRL that is unavailable from data aggregators. In fact, the most
recent U.S. GAAP taxonomy (2011) includes over 15,000 tags (and growing),
compared to the several hundred that are typically captured by third-party
aggregators (FASB 2011).
In addition to quicker processing and a more comprehensive set of data items, the
standardized tagging structure of XBRL provides a way to compare information
123
E. Blankespoor et al.
across firms. Given the precision and standardization of the tagging structure,
investors no longer have to guess how to map items from one firm’s financial
statements into others’ financial statements. For example, firms may sometimes
differ in the naming convention of certain line items. As shown in Fig. 4, HP
identifies income earned from the sale of goods as ‘‘Products’’ revenue, while IBM
labels it ‘‘Sales’’ revenue. The XBRL tag ‘‘SalesRevenueGoodsNet’’ reduces the
uncertainty about what is included and allows for easy comparison. As another
example, firms may group line items together differently (e.g., depreciation may be
included in operating activities for some firms but not in others). The XBRL tagging
structure allows investors to quickly access comparable items without having to
manually adjust data groupings across firms for better comparisons—it has been
done for them already. Allowing investors to acquire and process consistent
financial information for all firms in an industry enables them to reliably compare
across the firms as well as across time.
References
Akins, B., Ng, J., & Verdi, R. S. (2012). Investor competition over information and the pricing of
information asymmetry. The Accounting Review, 81, 35–58.
Amihud, Y. (2002). Illiquidity and stock returns: Cross-section and time-series effects. Journal of
Financial Markets, 5, 31–56.
Armstrong, C. S., Core, J. E., Taylor, D. J., & Verrecchia, R. E. (2011). When does information
asymmetry affect the cost of capital? Journal of Accounting Research, 49, 1–40.
Asthana, S., Balsam, S., & Sankaraguruswamy, S. (2004). Differential response of small versus large
investors to 10-K filings on EDGAR. The Accounting Review, 79, 571–589.
Atiase, R. K. (1985). Predisclosure information, firm capitalization and security price behavior around
earnings announcements. Journal of Accounting Research, 23, 21–36.
Bamber, L. S. (1987). Unexpected earnings, firm size, and trading volume around quarterly earnings
announcements. The Accounting Review, 62, 510–532.
Bamber, L., Barron, O., & Stober, T. (1997). Trading volume and different aspects of disagreement
coincident with earnings announcements. The Accounting Review, 72, 575–597.
Bamber, L., & Cheon, Y. (1995). Differential price and volume reactions to accounting earnings
announcements. The Accounting Review, 70, 417–441.
Barber, B., Odean, T., & Zhu, N. (2009). Do retail trades move markets? The Review of Financial Studies,
22, 151–186.
Barclay, M., & Warner, J. (1993). Stealth trading and volatility: which trades move prices? Journal of
Financial Economics, 34, 281–305.
Bartley, J., Chen, Y., & Taylor, E. (2010). Avoiding common errors of XBRL implementation. Journal of
Accountancy, 209, 46–51.
Belsley, D. A., Kuh, E., & Welsch, R. E. (1980). Regression diagnostics: Identifying influential data and
sources of collinearity. New York: John Wiley and Sons.
Bhattacharya, N. (2001). Investors’ trade size and trading responses around earnings announcements: an
empirical investigation. The Accounting Review, 76, 221–244.
Bhattacharya, N., Black, E., Christensen, T., & Mergenthaler, R. (2007). Who trades on pro forma
earnings information? The Accounting Review, 82, 581–619.
Blankespoor, E., Miller, G. S. and White, H. (2014). The role of dissemination in market liquidity:
Evidence from firms’ use of twitter. The Accounting Review, 89(1).
Bloomfield, R. (2002). The ‘incomplete revelation hypothesis’ and financial reporting. Accounting
Horizons, 16, 233–243.
Bushee, B., Matsumoto, D., & Miller, G. (2003). Open versus closed conference calls: The determinants
and effects of broadening access to disclosure. Journal of Accounting and Economics, 34, 149–180.
123
The market impact of the XBRL mandate
Campbell, J., Ramadorai, T., & Schwartz, A. (2009). Caught on tape: Institutional trading, stock returns,
and earnings announcements. Journal of Financial Economics, 92, 66–91.
Chambers, A., & Penman, S. (1984). Timeliness of reporting and the stock price reaction to earnings
announcements. Journal of Accounting Research, 22, 21–47.
Cohen, K. J., Maier, S. F., Schwartz, R. A., & Whitcomb, D. K. (1986). The microstructure of securities
markets. Englewood Cliffs, NJ: Prentice-Hall.
Cram, D. P., Karan, V., & Stuart, I. (2009). Three threats to validity of choice-based and matched-sample
studies in accounting research. Contemporary Accounting Research, 26, 477–516.
Daske, H., Hail, L., Leuz, C., & Verdi, R. (2008). Mandatory IFRS reporting around the world: Early
evidence on the economic consequences. Journal of Accounting Research, 46, 1085–1142.
Diamond, D., & Verrecchia, R. (1991). Disclosure, liquidity, and the cost of capital. The Journal of
Finance, 46, 1325–1359.
Efendi, J., Park, J.D. and Subramaniam, C. (2010). Do XBRL Reports have incremental information
content? An empirical analysis. Working paper, The University of Texas at Arlington and Towson
University.
Financial Accounting Standards Board. (2011). FASB US GAAP financial reporting taxonomy technical
guide: Version 2011. http://www.fasb.org/cs/ContentServer?site=FASB&c=Document_C&
pagename=FASB/Document_C/DocumentPage&cid=1176157270813. Accessed July 5, 2012 .
Financial Executives International. (2011). FEI—CCR XBRL rulemaking request. Retrieved July 5, 2012
from the FEI’s website: http://www.financialexecutives.org/KenticoCMS/Knowledge-Center/
Comment-Letters/FEI-CCR-XBRL-Rulemaking-Request.aspx.
Glosten, L. R., & Harris, L. E. (1988). Estimating the components of the Bid/Ask spread. Journal of
Financial Economics, 21, 123–142.
Gow, I. D., Ormazabal, G., & Taylor, D. J. (2010). Correcting for cross-sectional and time-series
dependence in accounting research. The Accounting Review, 85, 483–512.
Goyenko, R. Y., Holden, C. W., & Trzcinka, C. A. (2009). Do liquidity measures measure liquidity?
Journal of Financial Economics, 92, 153–181.
Grossman, S., & Stiglitz, J. (1980). On the impossibility of informationally efficient markets. American
Economic Review, 70, 393–407.
Harris, L. (1990). Liquidity, trading rules and electronic trading systems. New York University
Monograph Series in Finance and Economics, Monograph No. 1990-4.
Harris, T., and S. Morsfield (2012). An evaluation of the current state and future of XBRL and interactive
data for investors and analysts. White Paper, Columbia Business School.
Hasbrouck, J. (2009). Trading costs and returns for U.S. equities: Estimating effective costs from daily
data. The Journal of Finance, 64, 1445–1477.
Hirshleifer, D., & Teoh, S. H. (2003). Limited attention, information disclosure, and financial reporting.
Journal of Accounting and Economics, 36, 337–386.
Hodge, F., Kennedy, J., & Maines, L. (2004). Does search-facilitating technology improve the
transparency of financial reporting? The Accounting Review, 79, 687–703.
Kalay, A. (2011). Investor sophistication, disclosure and the information environment of the firm.
Dissertation, The University of Chicago.
Kim, O., & Verrecchia, R. E. (1994). Market liquidity and volume around earnings announcements.
Journal of Accounting and Economics, 17, 41–67.
Kross, W., & Schroeder, D. (1984). An empirical investigation of the effect of quarterly earnings
announcement timing on stock returns. Journal of Accounting Research, 22, 153–176.
Kyle, A. (1985). Continuous auctions and insider trading. Econometrica, 53, 1315–1335.
Lee, C. (1992). Earnings news and small traders: An intra-day analysis. Journal of Accounting and
Economics, 15, 265–302.
Lee, C., Mucklow, B., & Ready, M. J. (1993). Spreads, depths, and the impact of earnings information:
An intraday analysis. The Review of Financial Studies, 6, 345–374.
Lee, C., & Radhakrishna, B. (2000). Inferring investor behavior: Evidence from TORQ data. Journal of
Financial Markets, 3, 83–111.
Lee, C., & Ready, M. J. (1991). Inferring trade direction from intraday data. The Journal of Finance, 46,
733–746.
Leuz, C., & Verrecchia, R. E. (2000). The economic consequences of increased disclosure. Journal of
Accounting Research, 38, 91–124.
123
E. Blankespoor et al.
Leuz, C. and Wysocki, P. (2008). Economic consequences of financial reporting and disclosure
regulation: A review and suggestions for future research. Working paper, University of Chicago and
University of Miami.
Li, F. (2008). Annual report readability, current earnings, and earnings persistence. Journal of Accounting
and Economics, 45, 221–247.
Li, E. X., & Ramesh, K. (2009). Market reaction surrounding the filing of periodic SEC reports. The
Accounting Review, 84, 1171–1208.
Madhavan, A., Richardson, M., & Roomans, M. (1997). Why do security prices change? A Transaction-
level analysis of NYSE stocks. The Review of Financial Studies, 10, 1035–1064.
McCann, D. (2010). 18,000 tagging errors in XBRL filings so far: But that’s better than expected, say the
SEC and XBRL US. Retrieved July 5, 2012 from the CFO.com website: http://www.cfo.com/article.
cfm/14529555.
Merrill Corporation. (2012). Where does the market obtain XBRL data? Retrieved July 26, 2012 from
Merrill disclosure solutions’ website: http://merrillcompliancesolutions.wordpress.com/2012/07/26/
where-does-the-market-obtain-xbrl-data/.
Merton, R. C. (1987). A simple model of capital market equilibrium with incomplete information. The
Journal of Finance, 42(3), 483–510.
Meulbroek, L. (1992). An empirical analysis of illegal insider trading. The Journal of Finance, 47(5),
1661–1699.
Miller, B. (2010). The effects of reporting complexity on small and large investor trading. The Accounting
Review, 85, 1487–1519.
Petersen, M. A. (2009). Estimating standard errors in finance panel data sets: Comparing approaches.
Review of Financial Studies, 22, 435–480.
Plumlee, D. R., & Plumlee, M. A. (2008). Assurance on XBRL for financial reporting. Accounting
Horizons, 22, 353–368.
Premuroso, R. F., & Bhattacharya, S. (2008). Do early and voluntary filers of financial information in
XBRL format signal superior corporate governance and operating performance? International
Journal of Accounting Information Systems, 9, 1–20.
Puckett, A., & Yan, X. (2011). The interim trading skills of institutional investors. The Journal of
Finance, 66(2), 601–633.
Securities and Exchange Commission. (2005). XBRL voluntary financial reporting program on the
EDGAR system. Release No. 33–8529. Retrieved August 13, 2013 from the SEC’s website: http://
www.sec.gov/rules/final/33-8529.pdf.
Securities and Exchange Commission. (2008). Commission guidance on the use of company websites.
Release No. 34-58288. Retrieved July 5, 2012 from the SEC’s website: http://www.sec.gov/rules/
interp/2008/34-58288.pdf.
Securities and Exchange Commission. (2009). Interactive data to improve financial reporting. Retrieved
July 5, 2012 from the SEC’s website: http://www.sec.gov/rules/final/2009/33-9002.pdf.
Van Ness, B. F., Van Ness, R. A., & Warr, R. S. (2001). How well do adverse selection components
measure adverse selection? Financial Management, 30, 77–98.
Whalen, C. (2004). Opening the books: A new computer format for financial reporting could improve
investment analysis. Barron’sOnline. (January 19, 2004). http://online.wsj.com/news/articles/
SB107429714251690100.
XBRL U.S. GAAP Financial Taxonomy. (2011) Retrieved July 5, 2012 from the XBRL.US website:
http://xbrl.us/taxonomies/Pages/US-GAAP2011.aspx.
Yohn, T. (1998). Information asymmetry around earnings announcements. Review of Quantitative
Finance and Accounting, 11, 165–182.
123