Koch Industries gets an enthusiastic thumbs-up on two issues. First, Lloyd Boyd actually makes a habit of doing research before making a business case. As basic as that sounds, it is remarkably rare. Most business cases are created from thin air where the numbers are chosen not for their realism, but as a delicate balancing act between credibility and getting funding for something someone has already decided they want. Second, Koch at least attempts to account for uncertainty?a major component of any IT investment?by applying what Boyd calls “risk factors.”
I suggest that Boyd use his research skills to determine how most actuaries and statisticians have been modeling risk for many decades. He would discover that in the fields of decision theory and quantitative analysis, they don’t apply a subjective series of risk factors to each number.
Unless we can answer a question like, What is the probability of a negative ROI? we are not actually doing risk assessment. For decades, statisticians, economists and others have been modeling risk with the Monte Carlo method. The models are based on ranges or “probability distributions,” which are then used to generate thousands of scenarios so that we can determine, for example, the probability of a negative ROI. Those ranges are still initially subjective but the approach has advantages over Boyd’s approach. It turns out that people are pretty lousy at intuitively assessing risk?they are typically overconfident about the risks they take. However, there is a lot of data about how to modify these ranges accordingly. What strikes me about Boyd’s factors is that they are relatively small compared with the adjustments typically necessary for IT estimates. When I ask IT execs for ranges and to track the outcomes, their ranges should be three times wider.
Finally, how uncertainties add up is not intuitive. This particular investment seems like a small one for Koch. The relative risk of small or large investments is computed with an equation often used by financial institutions. Koch may find with this equation that the risk is negligible because the investment is such a small share of the portfolio. Only the math will tell.