We think this is one of the most consequential improvements we have made to the directEDGAR platform.
For years, researchers have used directEDGAR to search specific SEC document types and filter by CIK at scale. Finding the right disclosure is an essential first step. Collecting the amounts, dates, and other details for a research dataset can still require considerable work. Our new feature searches financial statement notes and exports their text and tagged facts, taking on much of that next step. If you’re ready to put it to work, the Knowledge Base guide covers how to search, export results, and use the data.
The idea grew out of a client’s request for structured data from financial statement footnotes. It also brought me back to my dissertation, when I searched microfiche copies of 10-Ks for acquisition disclosures and recorded purchase price allocations on my first laptop. I know how much time can go into finding and transcribing those observations.
The new tool searches notes curated from iXBRL 10-K filings and exports the matching disclosures as HTML snippets. It also collects their tagged facts into separate CSV files for tables and narrative text. Each fact retains its tag, value, unit, available period and dimension information, and a reference back to its source snippet. Narrative facts also retain the sentences identified as containing them.
Finding the useful facts within a matching note
For a test, I searched for purchase price AND allocation AND goodwill. One match was Iron Mountain’s summary of significant accounting policies, which spans almost 15 pages in its 10-K filed in 2021. The exported snippet supplied 504 tagged fact rows from tables and another 80 from text. The disclosure contained my search terms and some useful facts, but its broad coverage also brought in many facts outside my data collection goal.
That is where the individual fact tags help. I can distinguish goodwill acquired in an acquisition from goodwill impairment, and filter out facts about leases, depreciation, or other subjects. The search brings me to the disclosure; the tags help me identify useful observations within it. The source HTML remains available so I can check what the amounts mean.
One sentence, five tagged facts
The same search found the business combinations note in Quest Diagnostics’ 2020 Form 10-K, filed in 2021. It describes Quest’s acquisition of the remaining 56% interest in Mid America Clinical Laboratories on August 1, 2020. The note reports the preliminary allocation in this sentence:
Based on the preliminary purchase price allocation, which may be revised as additional information becomes available during the measurement period, the assets acquired and liabilities assumed consist of $84 million of goodwill (of which $47 million is tax-deductible), $74 million of intangible assets, $11 million of working capital and $11 million of property, plant and equipment.
Each of the five amounts was individually tagged, so the text facts file provides five rows, with the complete sentence alongside each one. The underlying XBRL value for goodwill is reported in dollars as 84000000, while the displayed 84 is preserved separately.
The fact’s XBRL context also supplies the acquisition date and identifies the acquisition. Those details are available alongside the amount, even though the allocation sentence itself does not state the date.
This is your research. We preserve the original sentence, displayed values, and source HTML so you can check the amounts against the disclosure and understand their meaning as you integrate the data into your models.
For example, the tag on the $47 million row identifies the amount as goodwill expected to be tax-deductible:
us-gaap:BusinessAcquisitionPurchasePriceAllocationGoodwillExpectedTaxDeductibleAmount
The sentence confirms that this amount is included in the $84 million of total goodwill, rather than another allocation amount to add to it.
Try it with your research question
We expect this to shorten the path from finding a disclosure to assembling a research dataset. You will still need to select the relevant facts and review their context, but much of the collection work is already done. The initial databases cover 2021 and 2022, and we are adding coverage incrementally.
For access instructions, search examples, continuation steps, and descriptions of the CSV fields, see Search Financial Statement Notes and Export Tagged Facts in the Knowledge Base. Please share your experiences and suggestions—we expect the tool to evolve as you put it to work.