If I wanted to find the current equity risk premium, would the calc just be something like the average return over the past year of the S&P 500 - average return of a gov’mt bond over the same year?
I’m not the most qualified person to answer this for you, but I was recently asked to do a valuation and used 5.1% based on this: http://www.cxoadvisory.com/equity-premium/the-2010-equity-risk-premium-from-practitioners/
Thanks but aside from taking other’s estimates, how would I manually calculate it?
Morningstar calculates it as “large company stock total returns minus long-term government bond income returns” and they use the 20 yr Treasury Coupon Bond as the “long-term government bond”. Damodaran has the data he uses in his calc somewhere on his page: http://pages.stern.nyu.edu/~adamodar/
But for the current premium, would I use large company stock returns for the past year minus lt government bond income returns?
Oh and the Damodaran paper is good, what I’ve read so far at least, thanks for that
Well, there is some confusion about what to use for the RFR. Basically, you want the treasury security with the same maturity as your holding period. So it’s not truly objective, because it depends on what your holding period is. Long-term asset allocations studies typically use the 10-year yield because the data is available and 10 years is a relatively long holding period, and stocks are supposed to be held (as an asset class) over a long period. In academic settings I usually use this, and get a MRP of around 4.9% for US Equities. In practice, if you are rebalancing a lot, and may change your allocation on a monthly or quarterly basis, then taking 90d T-bills as the RFR is a common practice. I actually do this most often in my practitioner work. You can defend using 1y bills as well. Normally, the difference isn’t all that great, but the yield curve is pretty steep right now… one result is that Sharpe ratios look really great these days because the 90d tbill is so low.
A common approach from bankers and other valuation shops I’ve worked with is to use a very long-run average when calculating the excess return of equities over the risk-free rate. In practice, this means rather than focusing on the excess return for the current year, looking at it as an average going back to the mid-1920’s. This data is available from Morningstar/Ibbotson.
depends on what you mean by “current” premium and for what purpose. it sounds like you are interested in the historical average as of today rather than implied risk premium. if that is the case, you need to keep in mind a couple of things - 1) geometric vs arithmetic average of returns - i think in practice most people would prefer geometric. it is the more conservative estimate if nothing else 2) long-term bonds or t-bills - depends on the purpose, but if its for measuring cost of equity in CAPM use long-term bonds 3) period of averaging - do not just take last year, go as far back as your data is available. do all of this if you are writing an academic paper or preparing a glossy consultant report to impress your firm’s client. otherwise, plug in 5% and call it a day
I meant current as in this very second. However if you do that, i.e. s&P 500 vs. 3 month T-Bill you’ll get a current risk premium of 15% which is absurdly high. I’m really at a loss for what to do though…
Pretend you’re an investor. If the 3m T-Bill is yielding 1%, what additional yield would you require to invest instead in SPDR? That’s the spot ERP. Don’t make this more complicated than it has to be.
DarienHacker Wrote: ------------------------------------------------------- > Pretend you’re an investor. If the 3m T-Bill is > yielding 1%, what additional yield would you > require to invest instead in SPDR? > > That’s the spot ERP. Don’t make this more > complicated than it has to be. Great, but the key is analyze the ERP implied by the mkt and compared it with your own estimate.
MissCleo Wrote: ------------------------------------------------------- > I meant current as in this very second. However if > you do that, i.e. s&P 500 vs. 3 month T-Bill > you’ll get a current risk premium of 15% which is > absurdly high. I’m really at a loss for what to do > though… What are you error bars? (You don’t really think you’re calculating an exact value, do you?)
I guess what I’m getting from this discussion is that the historical ERP is very easy and obvious to calculate. But for the current premium, you pretty much have to use an estimate. Is this correct?
> the key is analyze the ERP implied by the mkt and compared it with your own estimate Maybe, maybe not. Given the lack of visibility into Cleo’s issue, it’s hard to advise her well here. If say she just wants to compare price to her calculation of value (putting her in the same camp as Graham, Dodd, Buffet, etc.), there’s no need to compare her own ERP with a value implied by some sort of market comparison. But we’re all speculating until she lets us know how she wants to use the ERP. > (You don’t really think you’re calculating an exact value, do you?) +1 > plug in 5% and call it a day It’s extremely hard to argue that any other approach would perform better than this.
historical ERP is easy to calculate sure, but even with 60-70 years of data there will be a fairly large standard deviation around your mean, and different people may still disagree about what the mean historical ERP is because of issues 1 - 3 I meantioned above. if you want to use implied ERP you need forecasts (your own, or consensus from analysts)… so of course it is an estimate, and different people will likely have different estimates even if they apply the same methodology. bottom line, 5%.
I’m sure someone has looked to see how ERP is related to the business cycle. I wonder what they’ve found. Certainly there are studies of P/E and other multiple expansions as a function of the business cycle, and that’s been on my research to-do list for a while.
It’s important to understand what the ERP is and isn’t. Here are three frequently confused terms: 1. historical return. E.g. “the average return over the past year of the S&P 500 - average return of a gov’mt bond over the same year” 2. expected return. E.g. “the ERP implied by the mkt” (I assume – most market-implied returns are trying to solve for an expected return) 3. required return. E.g. “additional yield would you require to invest” This is the CAPM-specified ERP. It’s not observable, which leads practitioners to use proxies like #1 or #2. Here’s a thought experiment. Say the Dow rises from 7,000 to 14,000 over a few years, and market P/E is 22. Note that… a. the recent historical return is very high b. the expected return is low (assuming prices mean revert) c. the required return is unperturbed by this recent price rally This isn’t to say that required return doesn’t change–it certainly does–but the “pricing of risk” is driven by more factors than just historical or expected return, and is certainly not synonymous with either.
MissCleo Wrote: ------------------------------------------------------- > I guess what I’m getting from this discussion is > that the historical ERP is very easy and obvious > to calculate. But for the current premium, you > pretty much have to use an estimate. Is this > correct? No, none of that is correct, except for the using an estimate part.
DarienHacker Wrote: ------------------------------------------------------- > Here’s a thought experiment. > Say the Dow rises from 7,000 to 14,000 over a > few years, and market P/E is 22. > Note that… > a. the recent historical return is very high yep > b. the expected return is low (assuming prices > mean revert) nope, prices don’t revert. maybe you meant PEs revert? however a PE can go down in two ways. > c. the required return is unperturbed by this > recent price rally the required return is probably higher.