Equity Risk Premium Levels Suggest March Lows Will Hold

Last quarter, we dusted off the “Fed Model” to look at historical risk premiums and their relationship to earnings. At the commencement of earnings season we shall once again return to the Fed Model and examine the market reaction to spikes in risk premium.

Robert Shiller of Yale University has done extensive work on the P/E ratio of the S&P 500. Moreover he is gracious enough to post online his raw historical data on both the S&P 500 and long term bond yields. It is from this data that we have drawn the following analysis.

As a refresher, the term “Fed Model” was coined by Ed Yardeni when he referenced research the Federal Reserve conducted on the equilibrium between the yield on 10 year Treasuries and the earnings yield of the S&P 500 index. Simply stated, the Fed model suggests that the yield on 10-yr Treasuries should be equal to the earnings yield on the S&P 500. The earnings yield is simply the P/E ratio upside down or earnings divided by price.

While rarely at “equilibrium” the Fed model can be used to compare the relative value of Treasuries to equities. When the earnings yield on the S&P 500 is greater than the 10 year yield, the Fed model suggests investors should sell Treasuries and buy Equities. The difference between the yields is referred to as the risk premium. We used the data from Robert Shiller to construct the risk premium from 1881 to 2009.

Click to enlarge:

Risk Premium (Earnings Yield – 10 Year Treasury Yield)

The yellow highlighted circles represent periods of extreme risk premiums, that is to say a “peak” in risk premium. Since 1881 there have been 10 major peaks in risk premium; 9 out of the 10 peaks resulted in a major market rally at the 3, 6 and 12 month time horizon.

The peaks and subsequent S&P 500 returns are presented in the table below.

Click to enlarge:

It is clear to see that with the exception of the December 1920 peak, the S&P 500 experienced significant appreciation over the next 3, 6 and 12 months.

The accuracy and subsequent rise in the S&P 500 requires our undivided attention to this indicator. Since the March 2009 peak in risk premium the S&P 500 has risen 34.44%. Based on historical evidence we would expect the S&P 500 to continue its rise over the next 9 months.

Interestingly, the period that experienced the largest percentage gain was 1932-1933. The resemblance of today’s markets to that period is uncanny. What’s more is 1932-1933 was the only period in which the 6 month return was less than the 3 month return.

This fits with our technical take that the current leg up was the first 1/3 of this correction within a bear market. Furthermore, this would suggest the downtrend that began on June 11th will not break the March 2009 lows.

Disclosure: I am long SDS, for now…

Comments

[sheeple123jump:]

.interesting as this data is...is it wise to use this single indicator as a definitive proof that we wont see a double bottom..... given that so many other important market conditions are very different between 1930 and 2009....the industrial base we had then is gone now. the value of the dollar then and now, etc etc.... ,seems like a very complex analysis is necessary in order to 'predict' the next direction.

Jul 07 05:48 AM

[Moon Kil Woong:]

Interesting, but do you think the massive one time accounting gains the banks and financial institutions reported are real and should be factored into such a study? Personally, I don't think so. Of course reversing them out is such a pain. That's why most chartists tend to revert to oversimplification destroying the basis for their charting all together.

There are probably tons of inconsistancies that should be taken into account with such studies and I have yet to see very many charts where the analyst has attempted to rectify the varagries between different time periods. They tend to make the same argument all the time, "It all cancels each other out." Somehow I don't think that passes muster in class, so why should it get a pass in the real world?

Jul 07 06:11 AM

[chap08:]

Your logic is flawed. How can you say that we at a peak? Answer: you can't, it could keep rising. How can you say that we at an extreme? Answer: you can't, infact it looks from the chart as though we are close to the historical average.

It makes more sense to ask yourself: how do you personally judge the risks now relative to other points in history? For example, it looks as though we are close to the 1974 level. How do you view the current situation relative to 1974?

But remember, this is a relative graph and any peak could be resolved either by strong returns from equities - or disastrous returns on treasuries.

Jul 07 07:09 AM

[Gerry Sullivan:]

chap08 is correct. A peak can only be identified well after it occurs.
We could be at a peak right now or half way to a peak. There are points on the chart where someone might have called a peak only
to be early. Both in time and in pain.

When deflation is causing the pain, as in the 1930s and now, the Treasury yields are much lower and the peaks much higher.

Jul 07 08:41 AM

[doubleshortetf:]

More of "green shot" wishful thinking based on ONE of hundreds of models.

Granted "Fed Model" is more "reliable" but this market is no ordinary market.

Jul 07 10:33 AM

[robert.b.fe...:]

I hope you are correct. If the March low holds we could actually be at THE bottom. If not hold on to your hats boys this could be a wild ride. I'm still betting DIA @ 8000 at the end of the week.

Jul 07 10:34 AM

[Andrew Butter:]

WOW one other lonely guy on SA not predicting the end of the world.

I agree with this analysis although I am suspicious of P/E ratios either trailing (as a proxy for future) or forecast.

In my analysis I simply figure well (a) the stock market is representative of the economy as a whole (b) earnings long term reflect economic value added; so a GDP projection (and historical for modeling is a better proxy), and the value of those can be discounted by long term interest rates.

That's not a pointer on if you should be in equities or Treasuries, that's a valuation using International Valuation standards.

That gives you the "equilibrium" value which should be the "Price" if the market is not mispriced.

From there you can look at the degree of mispricing which goes in long waves where mispricing of slumps can be predicted from the mispricing of the previous "up wave".

That said in January 2009 that the S&P would bottom at 660 , I was out on the timing (figured out the mistake - I do real estate don't know nuffunk about stock markets - that was my first attempt at a prediction), in March I looked again and figured that was it and that depending on nominal GDP growth should end up about 1,000 by the end of the year (didn't look hard since).

See seekingalpha.com/artic... and other stuff I wrote.

Jul 07 11:02 AM

[RiskAverseA...:]

One problem with these risk premium comparisons with data over the past 100+ years...

When did any of the aforementioned peaks coincide with a complete breakdown of an infinite credit creating machine (i.e. structured finance) after it had produced a mountain of liabilities atop an utterly gutted economic capacity to generate revenue necessary to service these liabilities?

Don't get me wrong. Your technical take could be correct. However, given the unprecedented nature of present circumstance, you might be quite mistaken thinking the market's bounce off March bottom is only 1/3 complete.

Jul 07 11:20 AM

[Larry House:]

Could be. I hope. I am very leery of anything that is based on "historical" data. I just question whether it fits this time. Interesting read. A crash is coming, say some, or you say we are 33% through a rally. Oh well. Nothing is easy.

Jul 07 11:41 AM

[Stone Fox C...:]

If this period resembles '32, then why were the 3 month returns just half of those in that period. '09 seems to be lagging big time at this point of the rally.

Jul 07 11:49 AM

[Yohei:]

The Feds are not going to be able to maintain the current low return rate on Treasuries and other government guaranteed debt. Even using the present estimated Federal deficits rising rates are inevitable (and probably sooner than commonly expected).

The current official Federal deficit projections should be taken with plenty of salt in any case; they are based on expenditures not increasing above what is "planned", overestimated economic growth and subsequent increased Federal income from taxation.

Notice the steadily increased pressure for another "stimulus package" of late. (Got to buy them votes, eh.) If the coming bailout of State and local finances hits first "health care reform" is toast; California will get Federally guaranteed loan issuance rights within weeks and many or all States will soon follow. No way to say how this will turn out.

All this is common knowledge. That is why "health care reform" is being pushed through at maximum speed; the "window of opportunity" is very short. Once "health care reform" is in place it is expected to be politically impossible to curtail the resulting spending growth. (Like Social Security on steriods.)

All is proceeding according to plan.

Jul 07 11:58 AM

[manuel:]

this peak could be as high as the highest peak, nobody knows, how do you know this peak will not get as high as the highest? On your graph only one peak was lower than this one, so I would be careful, 2 were the same, and the others were higher

Jul 07 01:33 PM

[manuel:]

historically bear markets can go down to Schiller P/E 8 during recessions, it seems that we are about 13 14 now, so marchs lows could get retested and your the peak could be higher if prices drop. it is the market that defines when you have the peak, not you deciding which is the peak

Jul 07 01:40 PM

[manuel:]

why did you highlight only 10 peaks? there are some that could be higher peaks than the lowest and they are not highlighted

Jul 07 02:14 PM

[manuel:]

you may could define i that way: once you identify a potential peak after a rally 9 out of 10 the really lasts or at least the subsequent bear does not make the peak go higher, it peaks at the same level, but it seems that the last part of the graph could fit in the first part, so this could be the 1 out of 10......

Jul 07 02:27 PM

[Fred Voetsch:]

"The yellow highlighted circles represent periods of extreme risk premiums, that is to say a “peak” in risk premium. Since 1881 there have been 10 major peaks in risk premium; 9 out of the 10 peaks resulted in a major market rally at the 3, 6 and 12 month time horizon."


This is perhaps the worst, most misleading data I have seen in quite awhile. Brian, there were many more than "10 major peaks in risk premium", you just decided to circle 10. I could easily double...maybe triple the number of circles on that chart.

The data is useful; how you have chosen to present it is not.

Jul 07 02:35 PM

[Alex Trias:]

Great article, although I agree with those commentators who point out (correctly, in my view) that a "peak" cannot be discerned until after the fact.
What is the current risk premium on the S&P? Using Shiller's data (which he posts for free on his web page), I'd say the average earnings on the S&P 500 for the past decade is probably around $55. Let's assume the law of averages applies, so the current earnings on the S&P 500 will grind a bit higher. On those assumptions, then at today's prices, the earnings yeild on the S&P 500 is around 6%. - about 175% higher than the yeild on a ten year Treasury. Not a bad premium, sure, but at the bottom of previous bear markets, we've seen a premium of 300% or more. That suggests to me that a "peak" in the earnings premium may not be in place at this point.
Upshot: stocks may or may not be cheap right now - I tend to suspect they are, and I also think that earnings could trend higher in the next six months as the global economy recovers. That by no means implies equities are likely to rally from this point on, or that downside risk is limited. In fact, at the bottom of an epic twelve year bear market, you'd really expect stocks to be wildly cheap, irrationally cheap, universally unloved by individual and profesional investors alike.

Jul 07 02:57 PM

[fleeing omega:]

i agree with those above who point out that these sorts of relationships and equilibria assume that the underlying financial system is intact and functioning normally and that the people have confidence in the same (regardless of how any given instrument is performing).

and so it would seem that this is in fact holding ... at the moment.

it seems to me that the govenment and fed basically need the market to cycle in its current apparent range so that neither the financial system nor the treasury gets crushed outright (by falling equities on the one hand or falling treasury prices with ramped up interest rates on the other). so as long as the PPT and Fed can tweak their respective venues, the whole "fix" can slog along with no one the wiser (but don't bet yer a$$ on a real V here).

yet if confidence in the governments solvency (california anyone??) or the fiat currency (chinas take on the long bond?) fall into doubt at the global level, well jeepers, the seesaw back and forth of domestic equities and treasury prices could end rather abruptly with both going in the same direction at the same time. i think you know which direction that would be. in which case the march lows would look like the top of mt everest.

i am not saying (or hoping) that this happens, but it appears to be closer to happening than it has before. in other words, if one were ever to ponder the consequences of such an outcome, now would be that time. not too surprising that most would prefer not to even consider the possibility.

disclosure: i am NOT long PMs, yet, simply because i am hoping the near term deflation will provide a better entry point.

Jul 07 03:01 PM

[roblogic:]

This is a HORRIBLE misuse of the Fed Model. A rising equity market increases the P in the E/P ratio, so all other things equal, an increasing stock market will lower the Risk Premium. Of course a declining Risk Premium correlates with an increasing equity market; the former is dependent on the latter.

One may want to argue that the Risk Premium is higher than historical levels, and that a return to "normal" Risk Premia will drive the equity market higher. But I see very little evidence of an abnormally high Risk Premium from the chart above.

And shame on the editor for making this an "Editor's Pick". I want my 15 minutes back.

Jul 07 03:03 PM

[manuel:]

may be the author means a peak after a series of lows, but there would still be more


On Jul 07 02:35 PM Fred Voetsch wrote:

> "The yellow highlighted circles represent periods of extreme risk
> premiums, that is to say a “peak” in risk premium. Since 1881 there
> have been 10 major peaks in risk premium; 9 out of the 10 peaks resulted
> in a major market rally at the 3, 6 and 12 month time horizon."<br/>
>
>
> This is perhaps the worst, most misleading data I have seen in quite
> awhile. Brian, there were many more than "10 major peaks in risk
> premium", you just decided to circle 10. I could easily double...maybe
> triple the number of circles on that chart.
>
> The data is useful; how you have chosen to present it is not.

Jul 07 03:58 PM

[KawKaw:]

I feel sorry for the analyst that stayed up all night doing your work. This stuff is not complicated. Stop focusing on the weeds in front of you and take a look from the 30,000 foot level. The only analysis and/or perspective that should be trusted is that of a person who has no investments and does not make money from advising people.

The prices reflected in the S&P 500 are ridiculously high...why do I say that? Because I carry a short position on equities. Maybe if I was long, I would buy into the chart analysis. Our economy is a lot more complex than it was 30 years ago, let alone 100. The implications of the total debt crisis facing the U.S.A. are so grave that no one wants to even talk about it...let's just roll it forward.

Green Shoots!!!

We all know that any given data set can be compiled and used to present a particular point of view. So just stop it already with this nonsense. If the black box operations of the computer traders and the bellowing voices of CNBC had not propelled the markets falsely upwards after the March crash, we would have actually seen a true clean up by now. You can't possibly think that equity prices can fall to a low and stay there for a couple of weeks, and then miraculously we recover.

Until we start to pay off our debts and start to live within our means, nothing will change.

Jul 07 04:14 PM

[manuel:]

if you were long you should at least wait till it hits the march lows before you buy...... the article just tells you that probably socks will go down but not below march lows, so at least wait to see if they get there before you buy


On Jul 07 04:14 PM KawKaw wrote:

> I feel sorry for the analyst that stayed up all night doing your
> work. This stuff is not complicated. Stop focusing on the weeds
> in front of you and take a look from the 30,000 foot level. The
> only analysis and/or perspective that should be trusted is that of
> a person who has no investments and does not make money from advising
> people.
>
> The prices reflected in the S&amp;P 500 are ridiculously high...why
> do I say that? Because I carry a short position on equities. Maybe
> if I was long, I would buy into the chart analysis. Our economy
> is a lot more complex than it was 30 years ago, let alone 100. The
> implications of the total debt crisis facing the U.S.A. are so grave
> that no one wants to even talk about it...let's just roll it forward.
>
>
> Green Shoots!!!
>
> We all know that any given data set can be compiled and used to present
> a particular point of view. So just stop it already with this nonsense.
> If the black box operations of the computer traders and the bellowing
> voices of CNBC had not propelled the markets falsely upwards after
> the March crash, we would have actually seen a true clean up by now.
> You can't possibly think that equity prices can fall to a low and
> stay there for a couple of weeks, and then miraculously we recover.
>
>
> Until we start to pay off our debts and start to live within our
> means, nothing will change.

Jul 07 04:39 PM

[irondoor91:]

"Earnings" are what the accountants want to say they are. The real cash-in-pocket spendable, investable yield comes from dividends. Not a day goes by without another dividend-cut announcement. Falling prices have been/will be chasing falling dividends in a race to the bottom. Until price declines get ahead of dividend cuts, the market will continue to go down.

The dividend yield on the Dow at the end of the crash in '29 was 16%. This bear is of a greater-degree than that one due to the massive deleveraging that is going to continue to take place. The Dow dividend yield will exceed 16% at the bottom in around 2014.

On the other hand, if all the Dow stocks stop paying dividends, the yield (and probably the P/E ratio) will be infinite.

Jul 08 12:32 AM

[scheinine:]

Yields on Treasuries may increase dramatically. For the 3-9 months after the other peaks in the Risk Premium did the government try to auction as much debt as we expect to occur this time (as an order of magnitude comparison)?

Jul 11 02:02 PM

[Brian Kelly:]

Not only the one time gains from banks, but the elimination of GM as a component in the S&P 500 has had a major impact on aggregate earnings.

Of course, there is never one indicator that can paint the whole picture, but I'll take any help I can get!


On Jul 07 06:11 AM Moon Kil Woong wrote:

> Interesting, but do you think the massive one time accounting gains
> the banks and financial institutions reported are real and should
> be factored into such a study? Personally, I don't think so. Of course
> reversing them out is such a pain. That's why most chartists tend
> to revert to oversimplification destroying the basis for their charting
> all together.
>
> There are probably tons of inconsistancies that should be taken into
> account with such studies and I have yet to see very many charts
> where the analyst has attempted to rectify the varagries between
> different time periods. They tend to make the same argument all the
> time, "It all cancels each other out." Somehow I don't think that
> passes muster in class, so why should it get a pass in the real world?

Jul 23 06:04 PM