Discount Rate

I have found that most equity portfolio managers do not use CAPM to derive the discount rate used in DCF. I found that the either use some arbitrary hurdle rate or the base the discount rate on the actual volatility of the company’s earnings. Is there a good way to go about deriving the discount rate based on the volatility of earnings or is this a subjective process? I tend to use a high hurdle rate like 15%. This allows me a greater margin of safety but I am interested in other people’s process.

I agree that CAPM is not frequently used in practice. I’m not sure if your question is directed specifically at the cost of equity of the WACC, but I tend to start with a 10% WACC for my initial screens and come up with a more company specific figure as I learn more about a company. I would describe my approach as more of a build-up method, where I start with say a 12% cost of equity and add (or subtract) for risks (or moats).

1 Like

Yeah, I was referring to the cost of equity. I tend to build FCFE models with a 15% cost of equity. Probably too high, but I like to have that margin of safety.

What you should be doing is choosing a WACC that reflects your view on the cost of capital surrounding what you think is the optimal capital structure. For some companies, this closely matches the WACC you’d pull up on Bloomberg; in other cases, it doesn’t (therefore, there could be a potential opportunity to discover a mispricing in the market). That’s why it’s up to a savvy analyst to choose the right WACC to figure out what the appropriate valuation is.

To put some numbers around things, most of the companies I look at have WACC ranging from 9-12% and I generally look at companies in $800mm-5bn market cap.

Interesting. Numi, how do you go about determining the WACC? Do you use CAPM for the cost of equity or do you use a build up method?

Simply put, the discount rate is an arbitrary decision, and if you think about it, it will not have a large impact on your decision making.

Think of the discount rate as your “required rate of return”. What kind of return are you happy with? When I invest, I’d like to earn a return of at minimum 10%, so usually, that is my discount rate. You can use the discount rate to capture volatility, but you don’t have to. You can model that volatility by messing around with the cash flows (which is what I prefer).

The link between the discount rate and the required return is subtle, but there is a distinction. And I disagree that whatever method you choose is ultimately irrelevant.

The discount rate is simply that, the rate that you use to dicount future cash flows and value in a valuation or exercise. What would P be if we set the discount rate to 50%? Ah, ok, that’s what it would be. Valuation at 50% rate now complete.

The required return is the expected return that you require (i.e. think is reasonable) in order to make an investment. The flip side of this is your valuation decision… if the asset is expected to deliver more than is reasonable, it’s underpriced, if it’s delivering less, it’s overpriced. The two are intrinsically linked, just like bond prices and bond yields.

So, in a valuation, most people set the discount rate equal to the required return in order to determine a fair price (many don’t even know that this is why they are setting the discount rate to that number, they’ve just always done it that way, and think “that’s how you do it.”). But the price is only fair if the discount rate is appropriate, and setting the discount rate is where lots of noise and subjectivity can enter the calculation. Lots of analysts just confidently proclaim “yes, this is fairly/over/under-priced” full of bluster and pseudo-confidence. But if you don’t know what assumptions they make, you have no idea if it’s true or not.

But what return should you require? This really boils down to 1) what the investor’s risk tolerance is (if it’s high, then it doesn’t need to require as much), and 2) what the actual risks of the investment are. Those two things are often pretty hard to measure.

So some people just set a hurdle rate of 10% or 15% for their discount rate and say anything that is expected to deliver 10% is good enough for me. Personally, I don’t like that, because I think it sets you up to accept a lot of high-risk investmets that may be overpriced (maybe it’s so risky it should deliver 20%, but it’s only priced to deliver 15%, i.e. it’s overpriced, but that still passes the hurdle rate, so you buy it anyway and hope nothing goes wrong - but people who do that often don’t even know that that’s what they’re doing, and in good years, their returns look good simply because they are > 10%).

I don’t personally believe in CAPM, but I do like the index model, which reduces to the same security market line equation as CAPM, with a different set of assumptions. What it does is it says that the required return must increase with the risk, which to me is an essential element of valuation, and index/CAPM measures risk as the covariation with the market (the likelihood that it will be down when everything else is also down, which is basically related to the likelihood that you’ll need the money while the asset is down). I much prefer this to an arbitrary hurdle rate.

Still, what is a reasonable return to expect? Well, if you run a CAPM or index model type of regression, the regression line basically tells you “in the past, X level of risk (measured as covariance) has delivered Y level of return, on average.” Considering that you can’t just say “market, you MUST do better than that, now perform, or I shall be very, very angry, and I will write a letter - in the style of John Hussman - telling you just how angry I am,” then it seems to me that the past performance of a given risk level is not such a bad approximation of what a “reasonable expected return” should be for that kind of risk.

Now, there are undoubtedly risks that are separate from covariance with the market, and you can either expand the number of factors (say an APT model, which boils down to a quantified version of a build-up model), or you can just try to judge them qulaitatively and call it alpha (though your clients may not see it as alpha if they take your returns and stick them into an APT-based attribution model).

WACC is another approach, but it still requires a “cost of equity” figure. “Cost of equity” is bascally what an average investor would require for taking on the level of risk in the company, because if they can’t get it, it makes more sense to invest money elsewhere. Since cost of equity is basically a weighted average of most investors’ required returns across the marketplace (as opposed to your own, personalized requirements), it’s actually a lot more reasonable to use a CAPM-like equation or an APT model when computing the cost of equity. Another advantage of WACC approaches is that given that the cost of equity estimation can have a lot of noise in it, that noise is effectively shrunk somewhat by the fact that equity is usually just a portion of the full capital structure, and there’s probably less noise in the other portions.

1 Like

In my previous company, we used a cost of capital of 15% regardless of what project we were valuing, but knowing that it was a loose figure. So a 14% IRR was not necessarily a deal killer.

In private equity, I think fixed hurdle rates may make more sense (though more from a business perspective than from an investment one) because the liquidity premium is so large and difficult ot measure. People are not willing to lock up their money for long periods of time for only market level rates of returns. There’s a lot of extra risk, particularly in venture, and it makes sense to set the rate extra high for that reason. It’s also harder to get good quantitative data on private equity companies, so a single high hurdle rate may be the best that one can do.

1 Like

I didn’t say that “discount rate” and “required rate” are the same. My point was in the context of investment valuation, you should set k = required rate. There is no rule that high risk investments need to have a higher discount rate to compensate. That high risk can easily be modeled in your cash flow projections, and may well be easier to do so. Furthermore, even with a high risk investment, the returns can be broken up into low and high risk flows and can be valued at the same discount rate. And if you go the conservative route you can assume a baseline growth rate and treat any high growth as a bonus.

The discount rate is broadly defined to be “opportunity cost”. In theory, that should be the rate of return on stocks, as that is the index return. However, being an investor you are aiming for a higher rate of return than the index, so you have to artificially define an opportunity cost, and by setting k = required rate , you actually earn this rate assuming your model estimates happen as you predicted.

Sure, you can figure out your cost of equity through CAPM. However, all assumptions need to be adjusted based on your view of the capital markets and the company you’re analyzing. For example, if you’re looking at a company on Bloomberg that has a beta of 1, but you think that its capital structure is underlevered (and more importantly, you have some thought or evidence that the company will take on more debt as % of equity in the foreseeable future), you should adjust beta upward to reflect leverage. That will cause beta to go up. Also, if a company has experienced more near-term volatility because of company-specific instabilities, yet Bloomberg tells you that the two-year beta is 1, then chances are you want to use a higher beta if your forward view is that these instabilities could continue.

Now, as far as how your cost of equity impacts WACC, you then need to work through your calculations/assumptions to figure out what the ideal cost of debt is and what portion of the capital structure it should comprise. Again, there is a function for this on Bloomberg, but I encourage a skilled analyst to come up with their own calculation for WACC too. Your WACC again should reflect your view on the optimal capital structure. Normally, debt financing has a lower financing cost of capital than equity because of its tax deductability. However, you also need to adjust for things such as bankruptcy risk, cost of financial distress, and agency costs, not to mention potential breach of covenants on that debt. This is exactly why you can’t have a firm that’s financed with 100% debt even if the WACC formula tells you that an all-debt capital structure leads you to your lowest WACC. Also, in situations where you have a changing capital structure (i.e. highly leveraged transaction) then you can either adjust for WACC at each time interval in your model as debt gets paid down, or consider using something such as APV (adjusted present value) which is more dynamic, though less commonly used.

In most cases, you can use the company’s current capital structure as optimal, and probably should if you don’t have any evidence that it might change in the future. However, if the company has postured towards a share buyback or recapitalization or is in an M&A process, then you definitely should adjust your WACC because a skilled analyst will probably notice this before your typical sell-side equity research person. I have no shame in saying this since I myself spent several years in equity research – I can readily admit that the sell-side doesn’t understand capital structure or pay enough attention to it as it should.

Sometimes people will say that maybe WACC or APV or whatever doesn’t really matter, and it’s more important in theory, but I can tell you that it’s underutilized in practice and the reason it’s not utilized more often is because people don’t know how to use it correctly. I routinely see sell-side models that have some f*cked up assumptions about discount rates that don’t make any sense, but people go off that stuff blindly. That creates opportunities for skilled analysts such as the ones on this forum to take advantage of things, if they’re willing to go through the motions. And when we do, we exploit inefficiencies in the market and make money.

Granted, developing a view on optimal capital structure is easier when you’re in private equity or have some other way to actually influence the capital decisions of a company. When I worked in private equity/LBO’s I was extremely thoughtful about proper valuation methodology. But now even as an analyst at a long/short equity fund, where I don’t have any actual say in the capital structures of the companies we invest in as a minority shareholder, I still see benefit to developing valuation correctly. Using the right cost of capital / discount rate enables you to identify mispricings in the market. Moreover, if you believe that most corporate managers will act in the best faith towards maximizing shareholder value (obviously this is debatable, but I will save this debate for another day), then you’d still expect that the current capital structure will converge towards the optimal structure over time, just as how current valuation will converge to “correct valuation” if your analysis about a company is correct.

Hope this helps.

Also you may find a couple of these MBA slide decks helpful. I took a quick read of them to ensure “quality control” and they look solid.

My apologies, I wasn’t saying you did. But a lot of people don’t make that distinction and felt that this was a good place to start. I should have been clearer about that.

No, but they do need to RETURN more. That’s the distinction between required return and discount rate again.

That’s true. I’d forgotten that aspect, of lowering expected cash flows to compensate for risk, which would then imply that you could lower the discount rate by some amount to correspond to it if you wanted, but what that appropriate amount is is still kinda difficult to figure out.

Very true. And this is why in a goldilocks economy, the method works so well. If everything goes smoothly, you get that higher k. But maybe higher k is simply because the investments are levered to the economy (effectively a higer beta). At the end of the day, money is money, but if you’re trying to figure out if all that investment analysis really made you richer than sticking stuff in a index fund levered to the same beta, you have to chew on that question a little more.

I’m not sure I understand what you’re thinking. Are you talking about scenarios to produce a range, or are you talking about taking CFs from high risk projects and low risk projects, discounting their average by some subjectively determined value in the numerator to reflect your estimate of project-level risk, then adding those cash flows and discounting at whatever hurdle rate you’ve chosen?

I often don’t get into that level of detail and so prefer to group things together into a single discount factor. But I do more macro-oriented stuff, so it tends to makes more sense to do it that way. I can see how that might work, and it’s not unreasonable.

I agree that high risk investments need to return more in order to be compensated for the risk you take on. That being said, I think it is *easier* to adjust the cash flows downwards rather than modify the discount rate, which as you know, is very sensitive. But of course it is a matter of preference, hence my point that the method itself may not be that important, at least in my opinion.

For example, Apple’s future cash flows are full of risk. We have no idea what product they’re bringing out next. So one way to do it would be to bump up the discount rate to account for that, however, what we can also do is use the same discount rate, and run different scenarios for cash flows. Undoubtedly you know this, but just for the sake of clarity:

Say we come up with 4 scenarios.

Scenario 1: iThing cash flows grow moderately - base case

Scenario 2: Scenario 1 cash flows + high growth rate due to iPad Mini 2

Scenario 3: Scenario 2 cash flows + iTV sales

Scenario 4: iCook screws up iPhone 6, iTV is a flop etc.

I personally would pick Scenario 1, as it would strip away the “risk”, and focus on the “sure thing” cash flows, as a “baseline”, in theory should have a similar effect to picking a high discount rate to compensate for risk. Essentially both approaches are conservative.

Ok, I see where you’re coming from. I have issues with the “one hurdle rate to rule them all” approach because I think it lures people into forgetting that higher risks need to be compensated with higher returns, and can lead people into overvalued risky things if they are not careful.

But your point is that if you incorporate the risks directly into your cash flow estimates by being conservative in your projections, and even more conservative for higher risk projections, you can get more or less the same effect. And I agree. This is a reasonable way of addressing the concern I just talk about.

I don’t use it very much, but I’ve always liked the principle behind the franchise value approach that Bruce Greenwald advoces. Your approach, as described in your Apple example, has a lot in common with that, too.

Damodaran claims the MRP swings between 4-6%. Add that to the RFR and you’re good.

Interesting idea, I’ll have to give Greenwald a read sometime.

You’d think the k if based on CAPM should be adjusted for the timing of each cash flow. The beta for one year, the risk free rate the same year, the beta 2 years out, the rfr 2 years out etc.

If stocks had bond-like levels of volatility, then that approach would make more sense. However, the noise from year to year very likely overwhelms these small adjustments.

There is a debate about whether it is better to use the cash rate or the 10y note as the RFR. the spread is typically around 4%. I usually use 3m T-bills for the RFR, because we will review our estimates and rebalance within that time frame, but if you are talking about long holding periods, there’s a case to be made for using 10y Treasury notes as the risk-free alternative. To some extent, using the longer-term note, because of its spread, will compensate for not readjusting beta annually or adjusting the RFR to the exact structure of the yield curve.

Some very detailed answers here but in practice I think it usually comes down to a rule of thumb like double the cost of debt or the risk free rate +5% or whatever.

Ultimately the cost of equity you input into your model is personal to you and your risk tolerance. There is no right or wrong answer. Of course in reality, people will come up with convoluted models to demonstrate to their boss/client how the COE figure was derived, but I think almost always the analyst will start with a number in mind and work the model backwards to arrive close to that inital figure.