Since 2008, Japanese companies have been required to prepare the financial statement section of their securities filings in XBRL by the FSA (Financial Service Agency) and the TSE (Tokyo Stock Exchange). However, users have experienced some unexpected problems.
Previous to the advent of XBRL tagged filings, the users who were maintaining the financial content database in evaluating company values or monitoring risks, were manually transcribing data from securities filings. These were typically over 100 pages, released by as many as 800 or more companies on a single day.
With the mandate of XBRL filings in Japan, they anticipated and prepared their new work flow systems for gathering XBRL data and updating databases automatically. Because the facts in an XBRL tagged filing are machine-readable characteristics, it was assumed that XBRL would relieve them of all their manual data entry tasks.
However, once users started to review the XBRL-tagged filings made by companies, some unexpected results occurred. The facts, although tagged with XBRL, were reading as incorrect, and unknown error messages were being sent.
XBRL needs perfect digits for all figures.
The issue here is the “perfect” nature of XBRL.
The content might have errors even when there are no problems with the financial statements' presentation on a computer screen or printed page. Such errors are presumably attributable to a failure to adequately check the XBRL content.
When XBRL content is printed on paper or displayed on a screen, it is formatted in a prescribed layout. Such formatting sometimes omits certain information, such as a number’s last three or six digits. Income statements printed on paper typically present individual line items in thousands or millions of yen. With XBRL, however, numbers must be inputted in units of one yen, but they are sometimes not inputted accurately.
Should we change the way we audit?
When preparing financial statements in XBRL format, not only issuers but also auditors should review the XBRL itself, not just printed or displayed on the screen version.
If this erroneous data had been automatically processed by a computer without the user noticing the error, it can lead to a misevaluation of the company.
In the case of Japan, many XBRL users still have to check the data out of concern about inadequate checking or a lack of understanding by the companies that prepare the data.
Without auditors’ eyes on all XBRL data before filing, the benefits of XBRL's adoption are unlikely to be realized.
Tags have to be strictly unique — more than issuers expected.
The second issue is the risk that financial statements' line-item names fail to accurately convey their intended meaning.
With corporate operations becoming increasingly diverse, the task of compiling financial data in comparable forms has become more difficult, even when the data is correctly reported.
When securities filings were published on paper, users comprehensively evaluated the information in the filings and classified it based on their interpretation of the line-item names' meaning. To automate this evaluation step with computer systems, the requisite information must be accurately communicated in XBRL.
Examination of securities filings reveals that companies use the same or similar names for line items even if the line items have different roles, strictly speaking.
To calculate growth rates or profit margins for the purpose of inter-company comparisons, users typically make assumptions regarding how revenue derived from the company's main business activities is designated on the income statement.
Revenue is generally reported as Net Sales in the manufacturing sector, and Operating Revenues in the financial sector.
However, with companies becoming increasingly diverse operationally within the same sector, this general rule does not hold true universally.
For example, one manufacturing company reported Net Sales, but because it is a holding company, it also reported the sum of its Net Sales and other revenues as Operating Revenues. Although the company is a manufacturer, it may be undervalued by the market if financial statement users assume that its revenue consists of Net Sales alone.
Automatic capturing of XBRL could not be aware of those differences. Before XBRL, companies weren’t aware that their line items should be unique throughout whole companies. If line items are different from other companies, they will be easier to notice for users. The first action of users would be automated screening. But there is no chance to correct it, even if they read each figure the wrong way.
XBRL filings could have helped users to recognize these differences. XBRL tags are supposed to be selected based on their prescribed definitions. If two line items have different definitions, they should be tagged differently. Even if they share the same name. XBRL tags indicate not only differences in line items' definitions, but also the names of their parent line items. Tags also have defined relationships in terms of how line-item values are calculated.
XBRL can therefore communicate accounting terminology's predefined meaning to users' IT systems if it is selected carefully and edited correctly.
Companies need to train their staff to understand the distinctions between accounting concepts, and to be able to accurately choose the correct tags for each line item.
Only XBRL tells the facts, it cannot prevent their misrepresentation.
Let me share another example to explain this issue:
One day, our team found three similar, yet different, XBRL data items used in three different companies’ filings. All of which were listed as “Allowance for Doubtful Accounts” on their balance sheets.
They had used the same shorter names (labels) to save space on the display.
However, they had associated different tag names with different definitions with the label. Strictly XBRL speaking, they are different line-items, as the ‘tag’ is the determining metadata for the fact, not the label, which is for display purposes only.
1. One Allowance for Doubtful Accounts belongs to “Long Term Loans Receivable”.
2. Another Allowance for Doubtful Accounts belongs to “Investment Other Asset” and group deduction.
3. Another Allowance for Doubtful Accounts belongs to “Claims in Bankruptcy Reorganization Claims and Other”.
In the case of Japanese companies, we would never have read this information because of our traditional customs before XBRL.
It might cause misunderstanding when users manually select data without directly questioning the company about the validity and true meaning of each line item.
Let us think about the best way to correctly make the meaning of line items unique for all companies.
If we were able to create that, and auditors could successfully assure it, then XBRL could be a global environment for communication between companies and investors. My assumption is this dream system could be successful, if users were to provide tags for the purpose of evaluating companies. Then XBRL could be a workable system of delivery for company information. XBRL can be a more interactive tool for the market. It can effortlessly help users to access detailed information. Also, companies would benefit from knowing the reason why users select specific line items and how they make investment judgments.
How can we improve the XBRL value proposition for companies?
Companies (issuers) should consider changing their reporting process to be more informative and efficient. Then XBRL can be a more communicative tool between issuers and investors. It would be beneficial not only to understand ways that users can use data, but also to consider and understand the reasons why users care about those figures in order to evaluate companies.
Though XBRL is only one technology, the tags also have an important role in sharing information between issuers and investors. Through organizing and structuring information of company value, these tags can help us greatly understand a company’s information. If we could use this technology in the correct way, we would be able to get more valuable results from it.
by Chie Mitsui
Leave a comment