A colleague of mine at the University of Nebraska at Omaha needs Say-On-Pay vote results. The disclosure of these votes is relatively structured. This will allow us to write some Python code to extract and normalize the results for a large number from the filings. However, there are significantly diminishing returns to writing code to get every last one. For those that the code writing effort is too bothersome we will have our RA (a very diligent graduate student) review the documents and transcribe the data. We try to be conscious of the tedium of this type of work and reduce as many steps as we can. Our SmartBrowser offers an efficient way to review the relevant documents so she can focus on the data collection rather than making too many mouse-clicks.

The first step of this process is to extract the relevant documents from the platform. Below is an image of the results of the first search for this sample.

The search string was (DOCTYPE contains(8K*)) and (ITEM_5.07 contains(YES)) and compensation. The first parameter limits the search to 8-K or 8-K/A since we have not seen many results where the disclosure is in an exhibit. The second parameter limits the search to ITEM 5.07 as that is the item code for Submission of Matters to a Vote of Security Holders. The final parameter is to make sure that the word compensation is mentioned in the document. Clearly this is not a perfect search – it provides us a place to get started.

We want to extract the actual 8-K – I want to take advantage of the HTML structure to help better parse the votes so we are using the DocumentExtraction feature. An explanation/example of extracting documents and them moving them to your local computer (so you can share them with your RA or so you can write code to manipulate them) is available from this Video (DocumentExtraction and transfer are illustrated beginning at about 2:58). I will observe that you might want to also do a SummaryExtraction to generate a CSV file with all of the metadata about each document. In addition I would generate a listing of all of the documents that are extracted using the FileListing tool from the Utilities menu

This file is useful because we have to change the name of the files extracted so as to prevent name collision and to make sure you can map the document back to the filing. If the files in the selected folder have the directEDGAR file name convention the utility will parse apart the name components (CIK, RDATE, . . .).

We wrote some code to parse out and normalize the votes for those cases where we have a very high level of certainty about the disclosure. In this sample there were about 300 specific filings where it is easier to manually collect the data. Now we are ready to use the SmartBrowser. The stand-alone version of the SmartBrowser offers a convenient way to review a large number of htm/txt files that are named per our convention (CIK-RDATE-CDATE-etc). It has all of the features of the version embedded in the platform.

Here is a screenshot of Boston Beer’s disclosure:

Like the version in the platform the stand-alone version of the SmartBrowser has many of the features of a standard web browser that are available from the right-click menu. In addition you can increase/decrease font size (CTRL + (either +/-)). A CTRL+F allows you to search within the document that is displayed. Remember that this is the entire 8-K so being able to search for Compensation and then jumping to that location will reduce the workload considerably.

The stand-alone SmartBrowser installs on Windows 7+. It does not require any license validation and has no embedded security – other than it will not display documents without our naming convention. (You can get around that by renaming your documents if you wanted to).

The bottom-line is that this allows us to get our research assistant to help without having to tie them to significant training on the use of the platform. We provide them a USB drive with the installer for the SmartBrowser, the documents, an Excel file with the metadata they need (CIK etc) and they can review these documents and focus on what is important – collecting the voting results.

This is the accounting faculty member in me speaking. I figure it takes anywhere from 30 seconds to 2 minutes to take a CIK to EDGAR, find the right 8-K filing, open the document and find the disclosure. We are saving them somewhere between 150 to 600 minutes (2.5-10 hours) of key strokes/mistakes and boredom on this sample. If you don’t have access to the SmartBrowser installer please send an email to support@ . . .