Here at Calcbench, we pride ourselves on helping clients to use data more intensively, which lets your people on payroll work more efficiently. After all, time is money.

So recently we aimed our database superpowers at a common but frustrating question: What has changed from one annual report to the next?

This question can drive financial analysts nuts. Sometimes old items from previous reports disappear; sometimes new items arrive without fanfare. Either way, analysts generally can find the answer only through painstaking, manual labor.

Calcbench uses a different approach. We start by going into our databases and constructing a query to catalog the all the XBRL tags in a company’s most recent 10-K. Remember, XBRL tags are code that identify what a piece of financial data is, so computer programs (like ours) know how to interpret a numerical value they find: “Oh, this XBRL tag says , so the value next to that tag is the revenue number. I should display it at the top with the word ‘revenue’ next to it.” That sort of thing.

So first we catalog all the XBRL tags from a company’s most recent 10-K. Then we compare that result with all the XBRL tags used in the prior year’s 10-K— and from there, it’s a short exercise to identify which old tags were discontinued, which new tags were added, and which tags appear in both years (and therefore can be ignored, since we’re only looking for what has changed).

We ran this exercise for the 30 companies in the Dow Jones Industrial Average, and the results are as follows.

In other words, the financial statements of the DJIA were slightly less complex this year compared to 2016: on average, they introduced 47.6 new tags, but removed 53.8 tags—a reduction of 6.2 tags.

In absolute terms that reduction isn’t much; most companies use a host of tags, and most tags endure from one year to the next. (Don’t forget that academic paper we discussed earlier this year, arguing that the total number of XBRL tags in a financial statement is a good measure of accounting complexity, and by extension the probability of errors or financial restatements.)

Anyway, back to the financial analysts. Once you know what tags have changed from one year to the next, you can ask the CFO the next logical question during an earnings call: Why has the accounting changed? Did two lines of business become so similar they were consolidated? Was an old line of business shut down forever? Are the emerging markets growing so rapidly that you want to assign new tags for those revenue streams?

We at Calcbench won’t necessarily know the answer. Our gig is to give you precise information so you can ask the probing questions that help you make better decisions. That, we can do in spades.

Right now, we can do this year-by-year accounting churn analysis upon request. It doesn’t take us long, but we do need to work with subscribers to target specific companies. In the future we’ll publish a longer churn analysis for the S&P 500, so stay tuned. And if you do have a specific request, let us know at info@calcbench.com.