«Elizabeth Blankespoor University of Michigan Stephen M. Ross School of Business blankbe January 2012 Abstract This paper examines the ...»
First, with XBRL, less time, money, and effort is necessary to acquire financial information in an electronic format, ready for manipulation into statistics. Currently, investors must either hand-collect information from various parts of the filings (i.e. spend time and effort) or pay someone else to do this work (i.e. pay and wait for information from data aggregators). Each incremental piece of information is costly to collect, and much of the footnote information is not available via data aggregators. With XBRL filings, the information is already in electronic format, ready for transfer Specifically, XBRL is required for domestic and foreign public companies preparing financial statements in accordance with U.S. GAAP and foreign private issuers preparing financial statements in accordance with IFRS.
into spreadsheets or valuation software. Investors receive more information at a lower cost, with the saved time and resources available for additional processing and integration of the information.
Second, XBRL facilitates comparison of data across time and across firms because of its uniquely identified, standardized data tags. The FASB is responsible for maintaining an XBRL taxonomy (or dictionary of tags) for U.S. GAAP, and firms are strongly encouraged to use tags from within this taxonomy whenever possible, making it easier to compare financial information across time and firms.12 This increased comparability decreases investors’ costs of acquiring the information as well as the costs of integrating the information to arrive at a final decision.
Third, information within the tags such as the item’s organizational or mathematical relation to other data items, descriptions of the amount being captured, or references to relevant accounting standards, can provide investors with contextual information that would have required additional searching and studying of non-XBRL filings. This contextual information can lower the processing costs of making in-depth evaluations of financial information.
An important assumption of this study is that firms believe XBRL will reduce investor processing costs. Both prior literature and SEC statements provide support for this assumption. Consistent with the expected benefits listed above, Hodge, Kennedy, and Maines (2004) find experimentally that investors provided with an XBRL-enhanced search engine are better able to both acquire and integrate information.13 The SEC has spent significant resources to promote the use of XBRL and interactive data because it believes in the benefits for investors.14 In addition, the SEC has highlighted the cost savings its own staff would realize in reviewing disclosures (XBRL 2009). Thus, I expect firms to anticipate a reduction in processing costs for market participants because of XBRL.
Firms are also allowed to create their own tags (called “extensions”) when the standard tags are not appropriate.
Also, Blankespoor, Miller, and White (2011) examine the market impact of XBRL adoption in the first year of basic tagging and find results suggesting at least the perception that some investors are using and benefitting from XBRL. In their study, they look at the way non-XBRL-using investors respond to the perception that a subset of investors are using XBRL, and this study examines the way managers respond to the perception that market participants – including analysts, media, and regulators, as well as investors – will use XBRL to scrutinize firm disclosure.
2.4 XBRL and Disclosure Choice: Predictions Main Prediction Based on disclosure theory, my hypothesis is that firms will increase disclosure when investor processing costs decrease. Consistent with this hypothesis, I predict that when firms adopt XBRL, they will respond to the expected reduction in investor processing costs by increasing their disclosure, or specifically, the quantitative disclosures in the footnotes.15 Prediction 1: The number of quantitative footnote disclosures increases upon firms’ adoption of XBRL detailed tagging requirements.
I choose to focus on quantitative footnote disclosures and XBRL implementation of “detailed tagging” in the second year of firms’ adoption (fiscal year 2010 for Tier 1 firms) for several reasons.
First, although items in the footnotes are relevant for understanding firm performance, they are more likely to impose high processing costs on investors and thus receive less investor attention.
DeFranco, Wong, and Zhou (2011) show evidence of investors using footnotes to make accounting adjustments when valuing the firm, and many papers show the value relevance of footnote amounts (e.g. Barth, Beaver, and Landsman 1992, Ely 1995). However, the disclosure versus recognition literature suggests that investors do not fully incorporate items disclosed in the footnotes, potentially because of higher processing costs (Harper, Mister, and Strawser 1987, Davis-Friday, et al. 1999, Schipper 2007). Supporting the assertion that footnote disclosures are difficult to process but still relevant for valuation, Li, Ramesh, and Shen (2011) find a market reaction to newswire filing alerts containing highlighted footnote items, even though the SEC filings were already publicly available.
Second, the inherent benefits of XBRL – ease of obtaining multiple data items, standardization of data structure, and increased ability to compare across firms – are helpful in processing detailed An alternative is that firms respond to the adoption of XBRL by decreasing their disclosure, because they want (1) to avoid the increased investor attention or (2) to reduce the costs of tagging. However, for reason (1), it is not clear why these firms were voluntarily disclosing information they didn’t want investors to process. For reason (2), the fixed costs of implementing XBRL tagging are larger than the variable costs of adding one more tag, making it unlikely that the costs of tagging would cause an average decrease in disclosure.
footnote disclosures. In promoting the use of XBRL, the SEC has focused on the increased ease of pulling facts out of text (i.e., numbers out of footnotes). Per former-SEC Chairman Cox, “The result … is that investors, using standard software, will be able immediately to pull up the information the way they want it, without having to slog through pages and pages of dry text” (SEC 2007).
Finally, the level of discretion available to firms is arguably higher for footnote disclosures than for amounts on the face of the financial statements. This flexibility exists to allow firms to tailor their communication to their investors. The SEC Advisory Committee on Improvements to Financial Reporting alluded to the disclosure of additional information by stating that “disclosure guidance generally establishes a ‘floor’ for communication between companies and investors, rather than a ‘ceiling’.” (SEC 2008). A result of this flexibility, though, is that there can be inconsistencies in disclosures across firms. For example, in a study of firms’ securitization disclosures, the FASB found a wide variety in the amount and level of detail provided (FASB 2001). More recently, the Investors Technical Advisory Committee (ITAC) asked the FASB to create a disclosure framework, saying that current “disclosures are (unfortunately) inconsistent and incomplete” (ITAC 2007). The combination of inherently higher investor processing costs, footnote-specific benefits of XBRL, and more flexibility in footnote disclosure choices results in the “detailed tagging” of footnotes being a strong setting to examine the relation between investor processing costs and firm disclosure choice.16 Cross-Sectional Predictions: Variations in Information Processing Costs If the increase in firm disclosure upon implementation of XBRL detailed tagging is driven by an anticipated reduction in investor processing costs (rather than an alternate story), I should see the Although I focus on footnote disclosures, the change in investor processing costs could affect firm disclosure beyond the footnotes. If investors can more easily assess the relevant information available to management, firms might increase all their disclosure (e.g. press releases, earnings forecasts). However, the most likely disclosure change would be to the quantitative footnote disclosures because of the nature of XBRL. First, XBRL reduces investor processing costs only for a very specific type of disclosure (quantitative financial statements and footnotes).
Although investors’ overall understanding of the firm improves with XBRL, their comparative advantage is in understanding the quantitative information disclosed in financial statements and footnotes. Second, because XBRLtagged footnotes have lower processing costs, investors are more likely to demand new information in the same format. Intuitively, investors comparing tagged footnote disclosures across firms will be less satisfied by a firm that provides one of the disclosure items in a press release, rather than in the XBRL format conducive to comparison.
increase in disclosure vary in settings where investor processing costs are either more likely or less likely to be a binding constraint on firms’ disclosure choice. Therefore, I examine several crosssectional settings where firm or investor characteristics cause variation in investor processing costs.
In the first cross-sectional test, I examine firms with information that is inherently more costly to process. Following Cohen and Lou (2010), I call these firms “complicated” to capture the idea that they have more detailed information environments with many factors driving performance, making summary statistics such as earnings less helpful for decision-making and detailed information more pertinent. Specifically, I identify complicated firms as those operating in multiple industries, those with more volatile earnings, and those with more disperse analyst earning forecasts. I choose the multiple industries measure because a greater amount and variety of detailed information is necessary to evaluate firms operating in multiple industries. Prior literature supports this assertion, showing that firms operating in multiple industries impose higher processing costs on investors (Cohen and Lou 2010). I choose the earnings-related measures because detailed information is again more helpful in valuing firms with more volatile or uncertain earnings. Consistent with this statement, prior literature shows that firms with less informative earnings are more likely to provide additional information to investors by including balance sheet information and pro forma amounts in earnings announcement press releases (Chen, DeFond, and Park 2002, Lougee and Marquardt 2004) or voluntarily holding conference calls (Tasker 1998).
For complicated firms, detailed information is more helpful in assessing firm value, but this detailed information is costly to process. If the cost of processing detailed information declines, investors are more likely to increase their processing of detailed disclosures for the firms for which detailed information is more pertinent, i.e. complicated firms. Therefore, complicated firms are more likely to increase their disclosure in response to a reduction in investor processing costs. Specifically,
I predict the following:
Prediction 2: The increase in quantitative footnote disclosures upon adoption of XBRL detailed tagging requirements is larger for firms with information that is inherently more costly to process.
In the second cross-sectional test, I turn to characteristics of a firm’s investor base. Some investors or stakeholders have more expertise, resources, and/or ability, enabling them to process disclosures at a lower cost than other investors. Firms with a higher proportion of these sophisticated investors already face increased investor attention and processing of disclosures (Dye 1998, Fishman and Hagerty 2003). Consistent with this theory, several papers show that analysts and institutions are associated with (and at least partially motivate firms to provide) better disclosure (Lang and Lundholm 1996, Healy, Hutton, and Palepu 1999, Ajinkya, Bhojraj, and Sengupta 2005, Hollander, Pronk, and Roelofsen 2010). Firms followed by more sophisticated investors are more likely to already be held accountable for their disclosure and are thus less likely to respond to changing investor processing costs.
In addition, one of the SEC’s goals in implementing XBRL is to level the playing field by reducing the costs for non-professional investors relative to professional investors. If the reduction in costs is not proportional across investors, professional investors could see a smaller reduction in their processing costs and thus would not increase demands for disclosure as much. In either case, a decrease in investor processing costs is less likely to motivate a change in disclosure for firms with
more sophisticated investors. Therefore, I predict the following:
Prediction 3: The increase in quantitative footnote disclosures upon adoption of XBRL detailed tagging requirements is smaller for firms with more sophisticated investors.
3. Sample Selection, and Variable Definitions
3.1 Sample Selection In my tests, I compare pre- and post-XBRL detail filings for Tier 1 firms, the first group required to file detailed XBRL documents (i.e. tag all quantitative amounts in the financial statements and footnotes). I then obtain a control group of filings – statements from Tier 2 and Tier 3 firms that have not yet adopted detailed XBRL requirements, filed over the same time periods as the Tier 1 firms – to estimate a baseline non-XBRL-related change in disclosure. Using these two groups, I am able to implement a difference-in-difference design to isolate the effect of XBRL on disclosure.
To create my sample of firms, I download all 10-K documents filed with the SEC for fiscal years 2006 through 2010 (i.e. fiscal periods between June 15, 2006 and April 30, 2011).17 Matching the filings to Compustat and requiring firm-level data yields a sample of 18,721 10-K filings for 4,877 firms over the five years. Using Perl, I separate out the financial statement footnote section and count the numbers, or quantitative disclosures, within the footnotes. Since there is variation in the structure and section titles in filings, the Perl code is not able to identify the footnote section for all filings.