Yes, but the differences are going to be slight unless the yield curve is either very steep or very inverted.
But more importantly, it’s a lot of work in a not-so-trivial calculation to add a tiny bit of precision to something that is going to be overwhelmed by (except in unusual situations like ZIRP) much larger random factors and uncertainties in the revenues and profit margins anyway. If your process applies anything like a margin of safety, it’s just not going to make a difference in a decision to pull the trigger.
If you find that the “intrinsic value” is $0.02 less than the price, you’re not likely to lever up and say “Go baby, go!” because you know that there may be lots of unexpected things happening to the underlying business.
In fixed income, you know (at least with straight bonds and simple amortizations) how much is going to be paid, and when, so all of a sudden those big uncertainties disappear. And then, yes, the term structure of interest rates suddenly becomes important. But look at credit spreads, and how much they affect actual bond prices. Then compare those price changes with the fluctuations you get in equities prices, and then you’ll see why people don’t bother with using RFR term structure in their equity DCFs.
And how do you come down with a terminal value if you aren’t going to assume a constant RFR after some point in time. A multiple?? The justifiable change with the RFR. You can’t use a gordon growth model to produce a terminal value without a constant RFR. What year are you going to use? 5? 30? 50? Many stocks have a duration of 50, so maybe the 50y that’s the right one to use. Wait… are there even 50y bonds out there to use?? (not in most countries)
But if you assume that your investor is going to sell the stock in 5 years, then maybe you should use a multiple that reflects the 5y RFR. It just opens so many questions that depend on the investor’s situation that most people will simply say “the average RFR over the long term is (or has been) 2.5 expected inflation + 3% real GDP growth rate (for example). We’ll use that.”
It only makes sense to let the RFR drive your valuation once you’ve figured out what the equity risk premium ought to be. Get that number nailed down to the right number of basis points, and then you can start to sweat over whether you should be using the term structure or an estimated average.
Finally, the right RFR is tuned to the investor’s holding period. If you are going to reevaluate in a month and potentially exit the investment at that time, your holding period is actually 30 days, which would suggest that something like a LIBOR 30d rate is more appropriate than a 30y bond.
You could then ask, “But what if I keep holding this investment for 30 day period after 30 day period for 30 years. Doesn’t that mean I should have used a 30 year rate after all?”
And the answer to that is: It wouldn’t be wrong to use the 30 year bond rate if you know that that’s what you end up doing, but think about what the 30 year bond rate is. For there to be a no-arbitrage condition on the yield curve, the 30 year rate is basically an estimate what the average expected 30 day rate is going to be for all periods between now and 30 years from now, so you get the same answer.”
The practical solution to this is “There’s no perfectly defensible answer.” When you make a calculation that requires a RFR, and you expect others to read it, you simply have to say what RFR you used, and preferably where that number came from. If others want to use a different rate, they will.
If you are reading something that does a calculation that depends on the RFR, then you need to look at what they use, and see if it’s different from what you like to use instead.
What’s not cool is to switch between using short term rates and long term rates depending on what you want the answer to be. If you use short term rates to calcuate stuff, stick with that unless there is real reason to change it (which is typically about expected holding periods and/or reblance frequencies). Switching between short term rates and long term rates to figure out which gives you the best looking answer is manipulation.
Or, if you are on the buy side and want to be extra careful, you can do a scenario analysis using both short and long term rates and see how much of a difference it makes.
You do have a point that - for countries with a ZIRP policy - using short term rates to discount expected earnings 10 and 20 years into the future is disingenuous and implicitly assumes that ZIRP will never end. So perhaps discounting the first 5 years of data with graduated discount rates to reflect the termination of ZIRP makes some sense when the yield curve is highly inclined. We haven’t seen yield curves this steep for this long in recent history (if ever, in the US), so that may make some sense in the current environment if the yield curve is highly sloped one direction or another.
Finally, it’s interesting to query people and figure out what their approach to the RFR is and having the discussion about “why don’t you do it this way” or “let’s see what the differences end up being,” but (as a piece of advice) there’s just not a lot to be gained by pounding your colleague’s heads and insisting that they are doing it wrong.
The best you can do is to do the calculation yourself and compare the prices that come out and ask “do these differences change our investment process and/or results in any significant way.” Depending on how your process works, it may or may not make much of a difference. If you are doing traditional value investing, it most likely won’t, because you’ll be applying a margin of safety, which should be much larger than any differences this will produce.
If you have a quant-driven process, it’s possible that this will affect your portfolio composition in unexpected and nonlinear ways, so it might make a difference and would require substantial backtesting. Even if the portfolio composition is different, it’s still possible that the portfolio’s performance might not be, so that’s a major research project if your organization has the budget for it.