Bias in Valuation

I kicked off the valuation class with a discourse (actually, more of a rant) about the bias in the valuation process and how it skews numbers. I firmly believe that bias is, by far, the biggest enemy of good valuations and that it is pervasive. By bias, I am referring not only to the preconceptions we bring into the valuation of a company but also to the payoff to doing the valuation. He who pays the prices sets the bias in motion, and the valuation reflects it. Thus, if I were an investment banker hired to value a target company in an acquisition, I am going to bias my value upwards, since I make money if the deal goes through but not if it does not. I am sure that Pfizer has an investment banking valuation of Wyeth, right now, that backs up their bid.  In class today, I set up 12 different valuation scenarios and asked for guesses on the direction of the bias. Here is the link to the list:

http://www.stern.nyu.edu/~adamodar/pdfiles/eqnotes/biastests.pdf

Make your best guesses... Visit my web site for the answers.

So, what is the cure for bias? I am afraid that there isn't one. However, we can reduce bias by changing the process by which we pay for valuations. 

1. No deal maker should ever be asked to analyze whether the deal makes sense. This is what we ask of investment bankers in acquisitions. The process will create bad valuations. I think that acquirers should pay a third party (one that makes money only for doing appraisals) for the valuation.

2. In legal processes, we should have less adversarial valuation, where the expert for one side comes in with a high number and the other side with the low number and the court splits the difference.

3. Trust, but verify. Even the biggest names in the business - Goldman, McKinsey - are susceptible to bias. 

As someone who looks at other peoples' valuations all the time, what experience has taught me is that the most critical questions to ask in assessing  a valuation are: Who paid the appraiser to do the valuation? What are the potential sources of bias? That tells me a great deal more than perusing the numbers.