August 14, 2013
When analyzing the data you collect to develop CI, sometimes you have to make assumptions. Actually, you have to make a lot of assumptions a lot of times. Some of them a very small and it is usually safe to make them. Others are not so small, and potentially more dangerous.
Usually when dealing with assumptions, you know when you are assuming something. You have a gap between two sets of facts or you have a set of facts and you are trying to determine what they mean. There you understand that you are making assumptions, so you are usually careful.
But there is a special problem dealing with assumptions: serial assumptions. By serial assumptions, I mean assumptions that connect with other assumptions or assumptions are somehow dependent on a former assumption. These can be difficult to manage.
Let me illustrate what I’m talking about by reference to a recent article in Bloomberg BusinessWeek. In an article discussing adjustable-rate mortgages (ARMs), the authors noted that, while it is likely that home prices will probably continue to rise, is difficult to predict that by state or by region. That means potential buyers are usually making that assumption, without qualification, when deciding to buy a home with an ARM.
However many homebuyers, according to the article, applying for ARMs also make another assumption, perhaps unstated, about their income. [Note: unstated assumptions are perhaps the most dangerous of all.] Their assumption is that their income will be higher by the end of the loan’s fixed payment period, some 3 to 5 years out. To them, that means that they would be able to handle bigger mortgage payments, even if they cannot sell the house. So they take the ARM and buy the house.
The article makes very clear the danger of these serial (and unstated) assumptions: “[S]ays Henry Savage, president of PNC mortgage, ‘When you start making those calculations, you’re playing golf in the dark’”.