As fans of Calcbench already know, XBRL is the technology behind our database superpowers. XBRL “tags” financial data so computer programs (like ours) can understand what a particular piece of data is. That, in turn, lets those computer programs deliver more sophisticated, comprehensive services to consumers of financial information.

Today we catch up with two researchers, Rani and Udi Hoitash (yes, they’re brothers), using XBRL to measure the complexity of a company’s operations and accounting. The Hoitash brothers just published a paper, “Measuring Accounting Reporting Complexity Using XBRL.” They argue that the more XBRL tags a filer has in its statements, the more complex the accounting is—and therefore, the more risk a filer has of restatement or material weakness. Without XBRL it would not have been possible to use this simple and elegant solution for measuring accounting complexity.

We asked Rani (accounting professor at Bentley University) and Udi (accounting professor at Northeastern University) about their work. Excerpts are below.

CB. Why did you do this paper?

Udi: Mainly because there are no good measures of accounting complexity. It’s hard to assess whether a company’s accounting is complex or not, and where they stand on the complexity scale—and that’s very important to many stakeholders.

CB. Was that a surprise, that we have no good measures for complexity? After all, regulators have been talking about complexity in financial reporting for years.

Udi: It actually was, because complexity is discussed all over the place. People have used operating complexity as a proxy for accounting complexity, and indeed if you have many operating or geographic segments, that complicates your accounting. But those concepts are fairly crude, and don’t directly capture accounting complexity.

Another measure is linguistic complexity: how hard is it for consumers of a financial report to understand the written words? For example, these measures capture how many words are used in the financial report. But again, those aren’t really good measures for accounting complexity, because they’re not focused on the accounting context. They’re focused on the narrative.

CB. So what’s the overall goal for a good measure of complexity?

Rani: A good measure of accounting complexity should take into account all of the accounting disclosures that firms make. Firms can be simple along certain dimensions and complicated along others. For example, some firms may report capital leases, others may report inventory, and some may report both. The thesis we proposed was that the number of accounting concepts [XBRL tags] a firm discloses captures the accounting concepts in the financial reports more completely.

CB. So the more XBRL tags are used, the more complex the firm and its accounting are, and the greater risk of misstatement or material weakness we have?

Rani: Yes. The reason is that every XBRL tag relies on an accounting standard or regulation. So as companies use more XBRL tags, essentially they need to know more of those accounting standards in-depth.

Remember, standards are complicated. If I’m an auditor looking at the financial reports of a firm with a lot of tags, I need to know more standards. That suggests I’m less likely to have an in-depth understanding of all those standards. I may not apply them accurately. Misstatement is definitely one possible outcome.

CB. So audit firms could use this metric, yes? It could help them determine the time, manpower, and evidence necessary to perform a thorough audit?

Rani: I think so. Until now, audit firms in the negotiation process might have had a harder time demonstrating to the client why they need to perform more work, or why some firms are more risky than others. The measure captures overall complexity of firms. One potential application is that audit firms can use it to plan their level of effort in an engagement.

Udi: That’s especially true if the audit firm is new to the engagement, and not familiar with the firm’s operations.

Q: OK, that’s the audit firms. How would investors or analysts use this metric?

Udi: This measure can be particularly useful for analysts or investors who have less intimate knowledge of all firms. It can tell them where in the “spectrum of risk” a specific company stands.

Q: For example, then, a board considering merger targets—it could use this metric to weigh the complexity of one target relative to another?

Udi: Yes, that’s possible. Knowing that detail might help a board or a CFO assess the integration costs of a particular merger target, sure.

Rani: I could also envision a corporate controller using this metric to analyze his or her own accounting complexity. Then you could present arguments to the CFO or board’s audit committee, explaining why you might need to invest more resources into your accounting systems. This metric can quantify and compare your complexity against other firms.

Q: Now that the ideas are flowing, couldn’t regulators like the PCAOB or SEC use this measurement for risk analysis? Or academics use it to assess other research?

Udi: We don’t see why not. The crucial point is that the metric lets you assess complexity over time, comparing the number of tags this year to prior years; or it can be used to benchmark complexity of Firm A to Firms B, C, and D. So yes, all these stakeholders can benefit from a better measurement of complexity.

CB. You also mention in the paper that complexity should fluctuate with a firm’s transactions. Excellent point; tell us more.

Udi: This goes back to the measurement issue, and it’s not rocket science: if you have more economic events, then your accounting should be more complex. So a good measure of accounting complexity should capture this, and fluctuate with the level of economic activity.

What we find is that traditional measures—linguistic complexity or operating complexity—don’t really fluctuate with economic events. Our measure captures a significant portion of that change in economic events such as an M&A, a restructuring, a spin-off, etc.

CB. Extension tags, that companies create themselves because the taxonomy of GAAP-endorsed tags doesn’t quite fit—do they raise any special concerns?

Rani: In our study we didn’t concentrate on extensions too much. Our measure just counts any tag, whether it’s an extension or within the GAAP taxonomy. We treat them equally; a tag is a tag, as a measure of complexity.

Udi: Also remember that complexity isn’t driven by extensions. They add to complexity, certainly, but it’s not just extensions—all tags refer to accounting standards and therefore increase complexity. The best measure is the combination of extensions and taxonomy tags together.

CB: Thank you for your time.