Merrill Corporation - Dimensions eNewsletter - August 2012
An Interview with Suzanne Morsfield, CEASA
The Center for Excellence in Accounting and Security Analysis (CEASA) at Columbia Business School recently completed a detailed study on how investors, analysts, and other data-seekers in the market are using XBRL-tagged financial data. In Part 1 of an interview for Dimensions, which appeared in the July issue, CEASA Director of Research Suzanne Morsfield outlined some of the study's findings and recommendations. Here in Part 2, Dr. Morsfield gives details of CEASA's discussions with analysts and investors.
In Part 1 of this interview, you noted that CEASA began its discussions with investors and analysts by asking them about the numerical information-other than data from the body of the financial statements-that they used for their analyses. Would you provide our readers with more background on your analysis?
XBRL, as mandated in the United States, requires that companies tag all numbers in the body of the financial statements as well as in the footnote disclosures. We believed that before analyzing if and how end users are utilizing XBRL, we should first step back from the narrow XBRL question and look at the bigger issue: namely, what numerical data are required in general by investors and analysts? If they do not really need or want footnote data, for example, then this would provide one explanation for why they were not using XBRL detail-tagged data-assuming that one can measure accurately who is or is not using XBRL. This latter point about the ability to measure who is using XBRL is an important one, and it was one of the reasons for going directly to a varied sampling of investors and analysts to find this out. We also wanted to gather evidence on the sources and timing requirements of the data. Again, if end users do not require data on a real-time basis, then this aspect of XBRL data's perceived attraction (or not!) to investors and analysts would be better understood.
Asking about if and how analysts and investors were using or were interested in using XBRL data was a natural next step in the design of our study. XBRL has been advertised, so to speak, as data that will be more timely, more accurate, and more complete than the data that can be purchased from the large financial data providers/aggregators. We thought it was worth knowing whether more of this is, in fact, what investors and analysts need or want.
You have stated in other settings that it is important that any discussion or study of how analysts and investors use data should be careful to delineate the findings by type of investor/analyst. How did you implement this in your own study?
The front-end of the survey was focused on identifying the type of investor/analyst and each type's respective research/investment profile. We learned where they work; whether they represent buy-side or sell-side; what their roles entail (for example, research, portfolio management, modeling, or a combination of these); what type of sample of firms they analyze (that is, one firm over time peers in the same industry; or an entire population of firms, such as the S&P 500); and, if buy-side, what kind of investment strategy their fund pursues. Gathering this information is a very important component of any such study, in our opinion, because we know firsthand that buy-side vs. sell-side analysts, for example, can have very different information and timing requirements for a variety of reasons, including that the investment horizons of their clients can be very different. Hence, rather than referring to investors and analysts homogenously, it is good to state which type of analyst or investor.
Sell-side analysts generally follow a particular set of companies within a specific sector. Their analysis horizon is often the fiscal year of the company they follow and is updated quarterly after each earnings release. While the annual 10-Ks and quarterly 10-Qs are the basis of sell-side financial models, the press releases prior to issuance of 10-Ks or 10-Qs are the basis for updating sell-side quantitative models and publishing initial research reports that comment on-and potentially revise-the related share price targets and buy/sell/hold calls of the covered companies.
On the other hand, buy-side analysts and portfolio managers often have a longer analysis and investment horizon than that of the sell side (if they are not high-frequency or quantitative-only traders). Of course, the exact breadth and depth of buy-side analysis varies by the investment interests and horizons of the investment house. Buy-side research is often focused on making an investment in a company versus making a trade; as a result, they are generally less likely to pay attention to whether a company beats forecasted EPS by one cent, for example, than sell-side analysts would be.
The exact financial information requirements of buy-side analysts and portfolio managers overlap with those of the sell-side participants in many instances, especially where basic financial-statement elements are concerned, but they differ in at least a few key areas. All sell-side analysts require earnings press-release data, whereas most of the buy-side analysts we spoke with stated that earnings releases are less important than 10-Q or 10-K information, if important at all. Footnotes and trends in these accounting metrics are key data points for virtually all buy-side research and portfolio management, as well as any changes in the content of the MD&A from quarter to quarter. These items may or may not be as important to sell-side analysts, depending on their analysis style and industry.
We eventually spoke with scores of investors and analysts, in a relatively formalized manner, concluding with a structured survey of two dozen end-user representatives-ranging from portfolio managers in larger and smaller investment houses to research analysts of smaller firms and larger banks. I should also be clear, however, that we only included in our findings the results of discussions with or surveys completed by analysts and investors who utilize some type of financial-statement analysis when making their investment conclusions. While high-frequency traders and other quantitative-only investors may be of interest to study as well, their behavior is beyond the scope of our current study.
Would you share with us some of the key findings, then, of your study with respect to the use of numerical information by investors and analysts?
Sure, with the caveat that we encourage your readers to contact us for a copy of the full report, which we will issue this summer. The full report will be the best basis for learning what our stated position is on the current state of XBRL usefulness and usability to investors and analysts, because it will provide analysis across a number of dimensions additional to our survey of end users. Having said that, I am happy to share a number of findings from that survey. On a high level, the evidence gathered from our conversations indicates several generalities about the use of numerical data in their analyses:
- There is no one-size-fits-all set of data that investors and analysts require. We could not find evidence that all analysts and investors utilized the same 300 financial metrics, for example, nor that they only wanted basic financial-statement data for the most part.
- Virtually all investors and analysts use detailed numerical data from the footnotes and MD&A sections of the financial statements, noting that analyses of those items comprise more than 50% of their total analysis. However, the specific data items required vary by investor or analyst in many cases, unless their analysis covers the same industry or sector.
- Almost all investors and analysts utilize information that is not directly comparable with the peers of the company they are analyzing. They often are looking at trends, or at those one-time or noncomparable events that could signal something about the future performance of the company they are analyzing.
- Sell-side analysts as a whole require earnings announcement details.
- Approximately 60% of buy-side analysts or investors utilize nonfinancial information, as well as financial information gathered from other sources than SEC filings. Some of those we interviewed expressed interest in machine-readable nonfinancial data that could be combined with financial-statement information.
You have also noted in various public settings that the analysts and investors you have spoken with or surveyed who know something about XBRL have some serious reservations about using XBRL in its current state. Would you elaborate on this?
Most of those we talked to do not trust the underlying quality or stability of the XBRL data, based on what they have seen or heard. They acknowledge that the large data aggregators have data quality issues as well, but all of the end users we spoke with
had methods in place for addressing those. The concerns they have with XBRL data are driven, in part, by concerns that they would have to disrupt their current workflow to accurately identify and address some or all of the following XBRL data quality problems:
- Unnecessary or excessive use of extensions (filer-specific tags), instead of using standard tags provided by the FASB when available.
- Tagging or data errors, such as numerical items that have a different value in the related EDGAR HTML filing (and in the rendering or XBRL Excel file on the SEC's website, by the way).
- Absence of external auditor assurance on the XBRL filing, and the tags themselves.
Did your discussions cover the topic of XBRL data consumption tools?
Yes. Again, given that most of those we spoke with were concerned by the thought of any disruption to their current work flow, especially within the current investment and analysis environment, these questions about consumption of the XBRL data were brought up:
- Some noted the need for training to be able to understand how to consume XBRL data, given that one cannot use these data without having quite an in-depth knowledge of the language and architectural features of XBRL itself.
- Some stated that there are not enough consumption tools that do more than provide a rendering of a specific firm's financials.
- The issue brought up most frequently was that they could not easily gather the values associated with an accounting concept across a large sample of firms across a time series (for example, one of the analysts and investors we interviewed did a search for repatriated foreign earnings for the S&P 500 using the standard XBRL tag available; the search produced only two filers' data, even though many more companies had reported it in their EDGAR HTML filings).
A notable improvement in the area of the visibility of consumption tools, from my perspective, occurred when XBRL US and others sponsored a competition that increased the visibility about this, as well as encouraged the development of new consumption tools. Calcbench was the winner of the competition, and its creators are working steadily with regulators and end users to address the specific needs reflected in the concerns of analysts and investors we spoke with, for example. Other data consumption providers are stepping up as well, and XBRL US will be co-sponsoring another consumption-tool development competition in the upcoming months-both of which can only be a good thing for XBRL usage.
Any concluding thoughts for filers and preparers?
The demand for a complete set of machine-readable footnote and MD&A data is clearly there for most investors and analysts. But, as mentioned in the previous part of this interview, this demand is likely to remain pent-up unless and until the data quality and data consumption issues are addressed.
Filers and preparers do not have control over the data consumption tools that are being developed, but they do have control over many aspects of the quality of their XBRL-tagged data. The data quality issues I listed in my previous answer are the main reasons that the investors and analysts we spoke with are not using XBRL data at present. Improving data quality is a critical step toward seeing markedly increased usage of data by end users. And, as a reminder, filers likely should not measure usage of their XBRL filings data by consulting the number of hits to it on their corporate websites.