There is no bubble in the US stock market

Before you start reading this post note that I am not an equity market analyst and this is not investment advice. Rather it is an attempt to discuss the impact of monetary easing on the US stock market and to what extent the Fed’s actions have created a stock market bubble.

It is quite often said these days that the recovery we have seen in the US stock markets since early 2009 in some way is “phony” or “fake”, and that it has been driven by “easy money”. Even some policy makers, both in the US and other places, seem to think that there is a bubble in the global stock markets, which is a result of overly easy monetary policy.

To test these views in a simple way, I have estimated a model for the S&P500 going back to 1960.  It is a simple OLS regression. You can do something more fancy, but that is not the point here. It is all indicative. If you have a better model, I would love to see it.

Take a look at the graph below. The blue is the actual performance of S&P500, while the green line is the model “prediction”. Please note that I have estimated the model until 2007 to avoid the results being influenced by the monetary policy shift over the past five years (it doesn’t change the result in any substantial way, however, to estimate the model until today).

SP500 model

As you see, the model fits the actual performance of S&P500 over the decades quite well. The model is quite simple. I used only three explanatory variables – A corporate Aaa-rate long-term bond yield to capture funding costs and/or an alternative investment to stocks. I used the nominal Personal Consumption Expenditure to capture demand in the US economy. It is also a proxy for earnings growth and finally I used the ISM New Orders index as a proxy for growth expectations. All variables have the expected signs and are statistically significant.

Yes, it is a simple model, but it seems to work quite well in terms of fitting the actual level on the S&P500 over the years.

If anything stocks are still cheap (and monetary policy too tight)

As mentioned, I estimated the model with data until 2007, but I have used the model to “predict” how the stock market should have performed according to the model from 2008 until today.

The results are quite clear: Since 2008, stock prices have consistently been lower than what the model predicts. Only recently have stock prices approached the level predicted by the model. Based on this, it is quite hard to argue that stock prices in the US are overvalued. In fact if anything stocks remain “cheap” relative to the model predictions.

I don’t want to argue this too strongly and I am certainly not giving any advice on whether to buy or sell the US stock market at these levels – all kind of things tend to move the market up and down. What I am arguing is that the view that there is a bubble in the US stock market is pretty hard to justify based on my simple model. If you can come up with a better model, which can show that there is a bubble, I am all ears.

Therefore, it is very hard to argue in my view that overly easy monetary policy has distorted the pricing of US stocks. What has happened rather is that stocks became extremely cheap relative to “fundamentals” in early 2009, and what we have seen in the past five years is a closing of this “valuation gap”. Has monetary policy helped close the “gap”? Yes, that is likely, but that is not the same as saying that there is a bubble.

Concluding, there is no empirical reason in my view to claim that US monetary policy has unduly inflated US asset prices. And hence the performance of the US stock market over the past five years is not an argument for monetary tightening. It anything it is an argument that monetary policy has remained too tight.

PS if you are interested in the model output see below (it is not rocket science):

  • Model 1

    S&P500, rebase 01-01-1960 = 100.0

    • Observations 576
      Degrees of freedom 572
      R2 0,9597105111
      F 4541,7504443086
      Prob-value(F) 0
      Sum of squared errors 16279,404325032
      Standard error of regression 5,3348380549
      Durbin-Watson 0,0489587816
      AIC 6,1933143441
      HQ 6,2051117912
      Schwarz 6,2235650918
    • Coefficients Standard error t Prob-value
      Intercept -627,703293948 7,0591041987 -88,9210976745 0
      x1 -3,4836423233 0,096018312 -36,2810202582 0
      x2 28,8240799392 0,2485205896 115,9826635779 0
      x3 -0,0228854283 0,0303242775 -0,7546899778 0,4507456199
    • Legend
      x1 Corporate Benchmarks, Moody’s Aaa-Rated Long-Term, Yield, Average, USD
      x2 PCE
      x3 ISM PMI, Manufacturing Sector, New orders, SA
    • Covariance matrix
      Intercept x1 x2 x3
      Intercept 49,8309520883 0,0191128279 -1,6803093367 -0,067856727
      x1 0,0191128279 0,0092195162 -0,0048925278 0,0007896205
      x2 -1,6803093367 -0,0048925278 0,0617624835 0,0003876941
      x3 -0,067856727 0,0007896205 0,0003876941 0,0009195618
Leave a comment


  1. These stock market strategists are of the same opinion. But they think monetary policy is lax and this, coupled with economic growth, will push stock prices higher therefore this is not a bubble:

  2. Agreed. Whining about monetary policy boosting stocks is like complaining that a ship captain is ‘artificially’ keeping the hull away from icebergs. In an economy with an upward and roughly stable NGDP trend, stocks should regularly make new highs.

  3. cacramer01

     /  December 11, 2013

    Reblogged this on cacramer01’s Blog.

  4. Benjamin Cole

     /  December 11, 2013

    Excellent blogging, andy that is a great model. I am a fan of simple, rugged models, and this is a great one. The more complicated the model, the more fragile and difficult to access they become.

    My guess is that this model is right, and stocks are not overvalued, or if they are, it is not by a lot, ala the 1990s.

    Except for a few markets neither is USA real estate overvalued.

    The QE bubble hunt may well turn out like the QE inflation-hunt, which turned out like a snipe hunt.

  5. Jordan

     /  December 11, 2013

    I assume you looked at stationary in the residuals? What about a time trend. I want to know the relationship here is real and not just an artifact of regressing a couple of series with an upward trend on each other. Thanks.

  6. Diego Espinosa

     /  December 11, 2013

    Please note, the model did not say there was a bubble in 2007 either.

    You can define a bubble as a mispricing. Stocks were therefore not “mispriced” in 2007. Or maybe they were mispriced contingent on the vast credit mispricing ending (this is in fact what happened). Which begs the question: is there a ‘contingent’ or ‘conditional’ mispricing of stocks today? One answer is that they are mispriced conditional on margin mean regression. Another is they are mispriced conditional on the Treasury bond mispricing ending. Yet a third says they are mispriced conditional on those two things happening at the same time.

    • Diego, was there a bubble in 2007? I don’t think so. There was monetary policy failure in 2008.

      • Diego Espinosa

         /  December 12, 2013

        Assume the crash was mostly a policy error. If agents had some knowledge such an error could occur, they would have hedged against it, thus lessening its impact. For instance, banks and households could have carried more liquidity on their balance sheets, relied less on short term funding, reduced leverage, etc.

        Why didn’t agents anticipate the potential impact of a Fed policy error? If it was so easy to diagnose ex post, surely agents would have been able to discount the possibility of it occurring? The only explanation I can think of was that agents believed the Fed was infallible. If this was the case, then there you have it: the bubble was in confidence about the Fed’s ability to stabilize the economy without generating huge unintended consequences. This is a plausible explanation, and it may explain our current circumstances as well.

  7. Viktor

     /  December 24, 2013

    “All variables have the expected signs and are statistically significant.”
    Really? T-ratio by new orders (variable x3) is -0,75 which doesn’t seem to be much significant.

  8. Dan S

     /  May 22, 2014


    Have you updated this at all (today is May 22 ’14)? Just curious what it has to say now.


  9. I use Aswath Damodaran’s estimate* of the equity risk premium to determine over/under valuation. The ERP has been elevated since the Crash because of the low bond yield. The current ERP is 6% which is very high by pre crash standards. If bond yields rise, that will normalize. But I do not expect bond yields to rise: I expect stock prices to rise.

  1. There is no bubble in the US stock market | Fifth Estate
  2. There is no bond market bubble | The Market Monetarist

Leave a Reply to ViktorCancel reply