Monday, April 16, 2012

An Unusual Source for Competitive Intelligence

I stumbled upon a new way to gather information about manufacturing processes. In this particular example, the information is about a number of Dow Chemical's processes - actually Union Carbide's processes, who is owned by Dow. Want details? Well it appears that they
  • looked at using an Amoco anticoking process in one of their steam crackers
  • injected sodium borohydride into their caustic scrubber to reduce acetaldehyde in their butadiene which they sold to Shell
  • and used phenylenediamine as a inhibitor in their C3 column.
And where does all this information come from? The proceedings of a trial before the Federal Tax Court. The missing link between a trial over taxes and this technology disclosure is that US corporations can take a tax credit for R & D efforts (only the IRS calls it R & E, with the E being for experimentation). Dow/Union Carbide claimed some R & E credits over this work and the IRS didn't think there was enough "researchiness" to the effort so they denied the claimed credit, which then led to the courtroom battle. Dow lost the case and is now appealing the decision. A news story about the appeal sent me looking for the original case (just so you don't think that I routinely prowl over the Federal Tax Court records.)

There are some caveats with the information provided above. It's all from the mid- and late-90's (yes, the hands of justice move quite slowly) and you have to cull through a 299-page document to find some of this. These items noted above were from the background section of the court's decision, so you know that transcripts from the trial would have even more details. However, I seriously doubt that much of this information is too proprietary, but nonetheless, it still makes for some curious and somewhat revealing reading. I'm just surprised that all these details were needed in the trial, as I would think that the judges would only be interested in a high level view of the research, not specific chemical entities and customer's names. I don't recall running across any of the technical details that were wrong, something that the Supreme Court could use a few lessons in (2 + 2 is somewhere between 3 and 5.

No comments: