Has anyone run regressions to find model EBITDA multiples by industry?

December 15, 2021

by a searcher from Central European University in Atlanta, GA, USA

I have been trying to gather a dataset of recent lower-middle market transactions in the US from sources like Pitchbook, but I am running short of a big-enough sample to run a statistically significant regression to be able to gauge a pattern of sales. If anyone has access to more databases or even simple entries (with a few features like: EBITDA, multiple, revenue, date, location, size of company etc.) of recent and verified lower-middle market acquisitions in the US, I would greatly appreciate you sharing them. Thank you in advance!

2 Likes

13 Replies

176 Views

Replies

Reply by an intermediary
from Naval Postgraduate School in Bellevue, WA, USA

A "multiple" is a ratio based on historical performance. The 'value" is based upon a realistic projection of future earnings. Accordingly, one company with steady historical earnings without significant annual growth should be considered as less valuable than a company that is making the same earnings but there is a realistic expectation of growth. Statistically significant data on lower middle market transactions, in my opinion, do not exist since the databases do not include any data related to future expectations. Further the lower risk component associated with larger companies naturally suggests higher multiples.; This tend to reduce the revenue range of appropriate "comps" to a small number. Whenever I am provided a list of comps, I always use the internet to examine whether or not a comp is accurately listed. In a recent list of 20 comps, I reduced the sample to +/- $10M of revenue and then did a search on the balance (about ten). When complete there were only three comps that were useful. A very poor sample for statistical analysis. The value is always what a willing seller and willing buyer agree to. .

Reply by an intermediary
from The University of Chicago in Chicago, IL, USA

I do not your objective. As an academic exercise doing regression is your choice.
X1 and X2 make the same product. Their financials are exactly the same. Forecast is the same. Yet, they could have vastly different multiples.
X1 is sold. Its multiple in different databases can be different.
Few years ago, I analyzed a large sample of multiples in one SIC code for transaction value <$100 M. Sliced it into 5 revenue buckets, 2 transaction structures (Asset vs. Stock) and 2 buyer-types (Private vs. Public buyer). Hence, 20 boxes. Removed outliers and top 10% and bottom 10%. The result: If average multiple was 5x, then 2/3 of multiples (i..e. Std. deviation) ranged from 3x to 7x for each of the 20 boxes.
Let us take ABC Mfg. Co. Its EBITDA is going to be different depending on who you ask. Even the broker who submits the data to the database company has one EBITDA for selling ABC, and one for submitting to the database. Even the price ABC got sold for is different depending on how you look at it.
I am stunned by the over-dependency by buyers on multiples. Hope it is not the B-school teaching.

+11 more replies.

Join the discussion