Valuing Businesses Using Regression Analysis

CHAPTER 9Regression 2.0

The regression template and all the discussion on the regression methodology thus far has been oriented toward the basic procedure I have used for many years with great results. In the last few years I decided to take it up a notch and fine‐tune the process. The four regression methodologies discussed in this book all work on the assumption that a transaction in a sample that was more than one standard error from the trend line would be considered an outlier. The first‐level regression we ran would identify the outliers and the second‐level regression with the outliers removed produced the trendline formula used to calculate the multipliers for the subject.

If you were able to wrap your mind around all that math and statistics, you may be ready for the advanced course—Regression 2.0. If the fog hasn't lifted yet, go to the next chapter, and bypass this one.

In this chapter we will explore three additions to the regression methodology designed to fine‐tune the final conclusion of value:

1. Rather than two levels of regressions, we will consider running up to four regressions with outliers being removed at each successive level. By removing more transactions that are considered outliers, R² will increase, and the resulting regression may be a more accurate indication of where the market is.
2. In our original two‐level regression methodology introduced in Chapter 3, we arbitrarily chose the break point for determining outliers at 1.0 times the standard error ...