When Is It Appropriate to Re-Underwrite an Investment Thesis

This article is authored by MOI Global instructor Gary Mishuris, managing partner and chief investment officer of Silver Ring Value Partners. Gary is an instructor at Best Ideas 2018, the fully online conference featuring more than one hundred expert instructors from the MOI Global membership community.

A value range is an estimate at a point in time about the likely range of business values. As we get additional information, sometimes changing our value range is warranted, while other times we actually gain additional confidence in our initial conclusion. The two extremes, neither of which are optimal, are:

1. Materially changing our value range very frequently based on each incremental data point

2. Anchoring on our initial value range and never changing it regardless of the new evidence

It is also important to differentiate between two types of changes:

1. Changing the base case but leaving the range itself unchanged. Given that the base case is just the point along the confidence interval that represents the most likely scenario, it is more likely that we would be wrong about where the base case lies along the likely range of values than that we are wrong about the range itself.

2. Changing the range itself. I underwrite the range of values to represent ~ 90% of possible outcomes, meaning that if I do a good job the business should be worth less than my worst case no more than 5% of the time. These numbers are meant to be representative of the approximate degree of severity of the worst case that I underwrite to, rather than to reflect any false precision about the probabilities themselves as the exact numbers are impossible to precisely quantify. Therefore, it should take substantially more evidence to change the range than it should to change the point that is the most likely one along that range (base case).

The process through which I update my value estimate for each business is based on Bayesian updating. My initial estimate, which is the result of a rigorous process that attempts to properly weigh the relevant information that is available at the time of the underwriting acts as the Bayesian prior. As new information comes along, some of it causes me to update that prior estimate conditional on the new information. Most of the time the changes are small, and merit immaterial changes. Some of the changes in value are due to the passage of time (e.g. if a company was worth $10 a year ago that means that at that time my future value estimate for it today was $11 using a 10% discount rate). Other changes can occur due to capital allocation actions (e.g. share buybacks or change in the debt levels of a company). These kind of changes tend to be individually immaterial and while it’s good practice to do this sort of housekeeping on a regular basis to keep the value estimates up to date, they are unlikely to move the needle on the value range.

There are three key points when I re-assess the value estimate for a company:

1. Quarterly results. The main question I ask is: What evidence, if any, has become available that either confirms the thesis or informs me that results are tracking either better or worse relative to my long-term expectations? (See the thesis tracker that I discussed earlier in this letter that I use to formalize tracking the output of this process.) Usually little or no thesis-changing information is present in a quarter’s results – some who believe otherwise are likely forgetting that a quarter is shorthand for one fourth of a year and that it is unlikely to have a forecast of many years materially affected by developments over such a short period of time.

2. Material events. These include substantial acquisitions, management changes, material changes in the capital structure and material competitive developments that affect the long-term competitive position of the business. These require a careful reassessment of the thesis and the value range.

3. Three consecutive quarters of results tracking either better or worse than my base case would suggest they should. I have improved my process as a result of instituting my thesis tracker that I described earlier and carefully considering this question. Previously, I would re-underwrite the investment by considering the incremental information in the context of my original thesis for one of the two prior reasons. However, I believe that we are all impacted by behavioral biases, and one of the most prevalent is anchoring on our prior conclusions. It’s easy to let this bias lull us into complacency by disregarding small quarterly deviations from original expectations. On the other hand, we wouldn’t want to overreact by doing a time-intensive re-underwriting of an investment because of a small deviation in a single quarter. I believe that using the rule of doing a mandatory, in-depth re-underwriting following three consecutive quarters of deviation from the base case assumptions is a good way to balance those considerations and guard against behavioral biases.

Some of the steps involved in a full re-underwriting process are:

• Seeking out or, if unavailable, creating the strongest possible opposing thesis

• Building a model from scratch using the full set of currently available data rather than simply incrementally updating an existing model

• Re-assessing each of the three quality ratings (Business, Management and Balance Sheet) to make sure the situation has not changed materially

• Creating a list of all the forces/facts that are making this a structurally better business and a list of those that are making this a lower quality business than it had been historically

• Seeking out additional sources of primary research to confirm/reject concerns about the business