Valuing Businesses Using Regression Analysis

CHAPTER 8Selection of Transactions in the Sample

The old axiom “garbage in, garbage out” is especially true when selecting transactions for samples to be used in regression analysis. It only takes one or two outlier transactions in a sample to skew the results inappropriately. This can be the case when using medians or averages of a sample but even more so with regression. In appraisal school, our instructors suggested we build a sample by selecting every comparable transaction within the subject company's SIC (Standard Industrial Classification) code. The explanation was that in doing so, opposing experts could not challenge us for having cherry‐picked transactions for our sample. Cherry‐picking was a challenge occasionally used by attorneys to discredit an appraiser's results.

This is false logic because the result would be a sample with companies generating $500,000 in revenue alongside companies earning $5,000,000 in revenue. Business appraisers need to take a page from the real estate appraisers' playbook. A real estate appraiser would never consider comparing a 5,000 sq. ft. house to a 1,000 sq. ft. house. The two houses have completely different construction costs, quality of materials, and are in totally different markets. A 5,000 sq. ft. house is likely to be in a very wealthy neighborhood; not so much for a 1,000 sq. ft. house. Likewise, in business appraisals there is no way a $5,000,000 transaction has any similarity to a $500,000 transaction. Their balance sheets ...