# Logic of discounting forward price or EV?

Was reading an old article / interview where investment manager Murray Stahl briefly described a relative valuation (screening) method and I was confused about certain things.

How generally do you approach valuation?

MS: We estimate what we think earnings can be four to five years out, apply what we consider to be a reasonable multiple on those earnings, and then discount the result back to today using a 20% annual rate. If that’s less than the current price, implying a discount rate in excess of 20%, that’s something we’ll look at closely.

So I would think that looks something like...

``````Say EBIT in year 0 is e0
We project straightline annual growth of rate: p
So: e5 = e0 x (1 + p)^5
Say we pick some reasonable EV/EBIT multiple by whatever means to be: m
So we have EV @ year 5 being projected to be: e5 x m (say E5)
Then by how I interpret the interview, we do:
E5 / (1 + 0.2)^5
and we JUST want to see if LESS than the current price (or in this case EV)
``````

I'm am confused about the logic of the last sentences. Is what I showed what they mean by "discounting back"? After you discount the price back to the present, why would you want to see it less than the current price (wouldn't that mean it's overvalued presently)? How does doing this "imply a discount rate of 20%"? Anything else I appear to be confused about here?

I will explain MS's response as I understand it.

1. We estimate what we think earnings can be four to five years out

This literally means, come up with yearly earnings estimates. Think through what the company might go through in the next 5 years, and how that will impact earnings.

So somehow come up with e1, ..., e5. Whether that's by linearly extrapolating the past growth rate or some other means (MS does not say).

Perhaps also come up with e1_best, e1_worst..., to find what earnings can be as MS says, not just a point estimate, and repeat further steps for the best/worst cases.

1. apply what we consider to be a reasonable multiple on those earnings

In order to transform "earnings" (a flow of money; money per year) to "value" or "expected value" (a fixed amount of money), you have to allow for a "unit change" factor. The company isn't worth exactly what its total earnings are the next 5 years. Whether it's less or more, however, I don't know.

I would guess you have to account for costs and taxes (if they are high, m should be lower than 1 I suppose), but also for the chance of outliving the next 5 years (say, if you expect it to have a total lifetime of 10 years, maybe multiply m by 2, because we're only looking at the 5 next years).

So now we have 5 numbers, m x e1, ..., m x e5.

1. then discount the result back to today using a 20% annual rate.

Money earned in the future is worth less than money currently in one's pocket. Here, MS wants a 20% gross return. So, we discount the future values by 20% yearly, because they aren't worth as much to us.

The numbers are now: m x e1 x (1 - 0.2), ..., m x e5 x (1-0.2)^5

Then, sum them up to get future value, FV.

Note that (1-0.2)^5 = 0.8^5 ~= 0.33; money earned 5 years from now is only worth a third of the money earned today, to Mr. Stahl.

1. If that’s less than the current price, implying a discount rate in excess of 20%, that’s something we’ll look at closely.

Here I agree with you. You'd want FV > market capitalization for you to get more "value" for your money. I think MS meant to say "if that's more than the current price", but made a mistake.

A higher expected discount rate would mean a lower FV; this would mean even fewer companies would have FV > market cap.

• The "multiple on earnings" is typically a price-to-earnings (P/E) ratio for that company's industry. It's used to convert the earnings into some future share price. May 8, 2021 at 14:18
• That makes sense! This helps compare apples to apples (companies in the same industry). I guess in my answer, m = [1/5 x industry P/E] of it, if you're summing it up for 5 years. May 8, 2021 at 14:22