According to at least one opinion, there was more to the mortgage meltdown than the result of ordinary people defaulting on houses they couldn't afford. The other half of the fiasco was the result of well-heeled financial institutions being duped into buying these worthless mortgages.

Some believe that was thought of as "innovative financing" should have been recognized as the "creative accounting" of sub-prime mortgages. Large groups of mortgages, including sub-prime mortgages, were bundled together in packages through these "innovative" investment products. The financial institutions bought these products because they were rated as good, sound investments by highly respected credit rating service companies. How does a group of poor investments get rated as good?

There were a few problems with how the rating services came up with the rating. They relied on a new mathematical model that was supposed to predict the failure rate of the mortgages. They collected large fees for rating these products and collected the fee each time a new package was put together for sale.

The problem with the sophisticated mathematical model was that it was so complex few people really understood it, and not enough time had passed to test whether the results were, in fact, accurate.

The problem with the high fees was that they fostered an incentive for giving good ratings so that each package could more easily be sold. Because this became a new major profit center for the rating service companies, nobody wanted to be a "wet blanket" and therefore discourage the sale of these packages - the better the ratings, the more packages sold… until the mortgages started to default.

What can we as cost estimators learn from this?

Many of us rely on cost data and models for cost estimating and analysis. A model can be defined as an abstract mathematical representation of a process or concept using variables to represent inputs, outputs, and formulas to describe the interaction. When you calculate a square foot estimate, you are using a simple model: Total Building Cost = Square Footage of Building * [Unadjusted Costs * Size Adjustment Factor * Location Adjustment Factor * Cost Escalation Factor * addition contingencies and other adjustments]. When you are using CostWorks, an estimating program or a spreadsheet, you are using a computer based model.

Some suggestions…

1. Don't be drawn, even subconsiously, to the conclusion that because something is done with the help of computer, it must therefore be accurate. I remember that when estimates first appeared, dot matrix printed on continous computer paper instead of being handwritten. We had to remind ourselves of "GIGO" - garbage in, garbage out. Then again, when the output was sharp laser print on crisp, white, letter-size paper - GIGO. I fear that there may be some of that going on with the advent of BIM, with its amazing animated graphics.
2. Know where the underlying data came from. Know what the assumptions of the model are based on. See if the data has been vetted.
3. Understand the model and know what's behind the black box. There is a reason why you had to learn how to solve problems in school, when you could instead have just used a pocket calculator to get the results. I'm not suggesting that you go into the computer code, but know what the fundamental basis of the model is.
4. Test the results. Do reality checks before you accept the results. Apply some rules of thumb. Does it sound right based on your experience? At the same time, don't try to force the results to fit in with your past experience.
5. Realize that the results are not always the last word, but a good place to start.

With regard to points 2 & 3, the high quality of the RS Means underlying cost data is well known throughout the industry. The Means engineers do more than simply collect data - they understand it, which allows them to verify, test and monitor the cost data. The basis and methodology of Means cost data (used in CostWorks and in software using Means data) is easy to find, and is clearly explained in "How the Book is Built" and "How To Use the Book." "Square Foot & Cubic Foot Costs, Estimating Tips" explains how these costs were compiled and provides additional insight into using the data. Along with the median, the data includes the cost for which 1/4 of the projects had lower costs and 3/4 of the projects had higher costs. This is important in order to see the amount of variation in costs from the median. Remember, we are talking about the median, not the mean, or average. When determining the median, a sequence of numbers is arranged from smallest to largest: The middle number, for example, in the sequence 100, 110, 150, 230, 240, the median is 150. The assumptions and methodology used for adjustment factors can be found in the "Historical Cost Index" and "Square Foot Project Size Modifier" sections.