5048.62: All judgments are provisional

It’s the 14th anniversary of the dotcom bubble peak.  Here are some closing prices of NASDAQ, the tech-oriented stock index:

March 10, 2000:  5048.62

October 9, 2002:  1114.11

Last Friday:  4336.22

I recall the general view in October 2002 was that the 5048 closing in March 2000 was utterly insane, an overvaluation of historic proportions.  In contrast, the closing value of 1114 was seen as; “Ah, that’s more like it.  I always knew the dotcoms were a bubble.”

I don’t wish to argue in favor of the 2000 market peak, it still looks absurdly high to me.  Rather I’d like to suggest that all such judgments are provisional.  If investors in 2000 had known what we know now (current NASDAQ level), the peak would have probably been closer to 3000.  If investors in 2002 had known what we know now, the low point would have been slightly higher than in 2000, say around 3200.

(These are ballpark guesstimates based on the notion that in equilibrium stock indices might be expected to rise at faster than the rate of inflation, but slower than NGDP growth.  The latter point is due to the fact that stock indices do not account for new firms being created, and the total value of all stocks should rise at about the rate of NGDP, in the long run.  This may be wrong, but I don’t think anything in my post hinges on these guesstimates being exactly right.  Also, I realize that if investors knew then what we know today stocks would have been risk free assets over the next 14 years.  Please ignore that fact and assume a normal return for an asset with that level of risk.)

As of today, it looks to me like investors were somewhat more insane in October 2002 than March 2000.  In absolute terms they might have been roughly equally wrong.  The values of 3000 and 3200 are roughly half way between the 2000 highs and 2002 lows.  But it’s percentage differences that matter in investment.  If you can’t see that, consider this example.  You are looking out the window at a new Mercedes parked at the curb.  You comment that the car probably cost $80,000 new, and your friend says he’d estimate it was about $1000 new.  If the actual value was $40,000, who would have made the more reasonable guess?  And yet who was closer in absolute terms?

I believe humans tend to notice absurdly over-priced assets more readily than absurdly underpriced assets.  People are more likely to talk about the fools who bought stocks when the NASDAQ was 5000, than those who foolishly sold when it was 1100 a few years later.  Or the fools that bought Vegas property in 2006, not those who sold London property in 1994.  Or those who bought Bitcoins at $1000, not those who sold Bitcoins when they were $5.  You might say that sometimes people might have to sell, but they never have to buy a particular asset.  That’s true.  But even people who sell and then reinvest in something else seem to get a pass.  It’s the buyers who pay too much who are mocked.

Of course all the judgments in this post are provisional.  If NASDAQ goes to 8000 in the next few years I’ll no longer view March 2000 prices as crazy.  And yes, 4336 may not be the “correct value” today, but it’s the best estimate that we have.  If Richard Rorty had been a finance professor he would have said the true value of a stock is that value which investors regard as true.  And he would have also said that if we later find out that that value “was not actually the true value,” all that would mean is that later on people regarded a different value as “true.”   All truth statements are provisional, (excluding the optimality of NGDPLT.)

PS.  I don’t ever recall Robert Shiller saying stocks were underpriced and that people should buy, but I recall him calling stocks overpriced on numerous occasions.  Can someone confirm this impression?  If true, why doesn’t he recommend that people buy stocks during periods like March 2009, when the S&P 500 was at 670, barely a third of its current value?

PPS.  Off topic, but I highly recommend this post by Evan Soltas.  His posts (and those of Yichuan Wang) are amazing thoughtful for such young bloggers.  Indeed for bloggers of any age.  I doubt I could have made as persuasive an argument for NGDP targeting as Evan just did.



40 Responses to “5048.62: All judgments are provisional”

  1. Gravatar of Nick Rowe Nick Rowe
    10. March 2014 at 06:48

    Scott: “The latter point is due to the fact that stock indices do not account for new firms being created, and the total value of all stocks should rise at about the rate of NGDP, in the long run.”

    This doesn’t affect the point of your post, but it’s not new firms being created (that just changes the weights in the index, I think) but whether or nor earnings are reinvested that matters. If all earnings are paid out as dividends, and nothing else changes, we should expect to see stock prices constant. If there are no dividends, only reinvested earnings (and share buybacks??) we should expect to see stock prices rise at the E/P ratio.

    The total return index gives you a better measure of what you are looking for.

    Again, none of this affects you point in any material way.

  2. Gravatar of Nick Rowe Nick Rowe
    10. March 2014 at 06:54

    BTW, this Canadian, Garth Turner, http://www.greaterfool.ca/ , who has been saying for years that Canadian house prices are overvalued, has also been saying for years (since the crash) that US house prices are undervalued. And advising people to sell Canadian houses and buy US houses. I think he’s probably half-right (not sure which half!)

  3. Gravatar of Tommy Dorsett Tommy Dorsett
    10. March 2014 at 07:04

    I don’t recall Shiller ever regarding stocks as cheap in 2009 or advising investors to buy them then. Indeed, the entire raft of perma bears like John Hussman we’re arguing that stocks were OVERVALUED at SPX 666 in March 2009.

    If you’re model said stocks were a sell in early 2009, something’s wrong with your model. If you’re still using the same approach as gospel today, five years into one of the most powerful bull market’s in history, something’s wrong with you.

  4. Gravatar of Ironman Ironman
    10. March 2014 at 07:46

    SSumner writes:

    “I don’t ever recall Robert Shiller saying stocks were underpriced and that people should buy, but I recall him calling stocks overpriced on numerous occasions. Can someone confirm this impression? If true, why doesn’t he recommend that people buy stocks during periods like March 2009, when the S&P 500 was at 670, barely a third of its current value?”

    Your perception is correct. The reason why is because the relative valuation of stock prices according to his Cyclically-Adjusted Price-to-Earnings (CAPE) ratio (where reported earnings per share are adjusted for inflation over a 10-year period and averaged), has consistently placed the ratio over its long term average (since 1871) since the early 1990s. (See chart).

    That’s partly attributable to the formation of the Dot-Com Bubble in the late 1990s, but the real question is why the CAPE ratio remained elevated after the bubble peaked in mid-2000 and entered its deflation phase, which extended into mid-2003.

    Much of the reason why turns out to due to an accounting change in how companies report their earnings, which was implemented as part of GAAP in 2001. As a result of the change, there’s an apples and oranges aspect to earnings reported before the change and afterward.

    Since Shiller never adapted his CAPE ratio valuation metric to account for the effects of the accounting change upon it, he consequently always sees the market as being overvalued with respect to the long term average of his CAPE ratio, which explains what you’ve observed and hopefully answers your question.

  5. Gravatar of Morgan Warstler Morgan Warstler
    10. March 2014 at 08:10

    “You comment that the car probably cost $80,000 new, and your friend says he’d estimate it was about $1000 new. If the actual value was $40,000, who would have made the more reasonable guess? And yet who was closer in absolute terms?”

    This is the problem I have with the way labor economists talk about MW.

    The research is always around “marginal” changes.

    When cons say rhetorically “well why not make it MW $50?” The idea is waved away as non-serious because it is non-serious.

    An increase in MW by 7x is non-serious.


    But GI/CYB makes a decrease in MW of 7X!

    We go from $7 to $1.

    And it is deadly serious. It’s something everyone can see and taste and smell, nothing about it seems crazy. Liberal and conservative econos have poked at it. Miles Kimball asks outright if anybody has a better idea…. (crickets).

    We can have a non-marginal change to MW, and there’s nothing I can see in economic literature that says a massive price shift can’t clear the market.

    Even NASDAQ stays within a factor of 5!

    Also, more edits to GI/CYB plan:


  6. Gravatar of Patrick R. Sullivan Patrick R. Sullivan
    10. March 2014 at 08:34

    ‘And advising people to sell Canadian houses and buy US houses.’

    Well, I don’t know about Canadian houses, but buying US houses for investment purposes, is big right now;


    ‘B2R Finance, America’s Buy to Rent Lender, is the leading provider of residential buy to rent mortgages for property investors. The company offers cost effective and innovative lending solutions dedicated to residential rental property investors. We operate in partnership with the nation’s leading commercial and residential real estate experts to ensure that our loan products and customer service are in the forefront of the growing residential rental investment market.’

  7. Gravatar of Major_Freedom Major_Freedom
    10. March 2014 at 08:54

    1. Re: Rorty

    Rorty’s philosophy self-contradicts.

    Rorty claims that no objective, common ground foundation for truth exists. He believes Rationalism to be an illusion; an artificial constraint. Truth, to Rorty, is relative. He advanced “Hermenautics”. He wrote:

    “Hermeneutics sees the relations between various discourses as those of strands in a possible conversation, a conversation which presupposes no disciplinary matrix which unites the speakers, but where the hope of agreement is never lost so long as the conversation lasts. This hope is not a hope for the discovery of antecedently existing common ground, but simply hope for agreement, or, at least, exciting and fruitful disagreement. Epistemology sees the hope of agreement as a token of the existence of common ground which, perhaps unbeknown to the speakers, unites them in common rationality. For hermeneutics, to be rational is to be willing to refrain from epistemology””from thinking that there is a special set of terms in which all contributions to the conversation should be put””and to be willing to pick up the jargon of the interlocutor rather than translating it into one’s own. For epistemology, to be rational is to find the proper set of terms into which all contributions should be translated if agreement is to become possible. For epistemology, conversation is implicit inquiry. For hermeneutics, inquiry is routine conversation” – Rorty, Philosophy and the Mirror of Nature, pg 318.

    After reading this, is it not reasonable to then ask: “OK, what then of Rorty’s arguments?”

    If Rorty is correct, that there is no such thing as truth based on common, objective ground, then all of his statements concerning the efficacy of the human mond cannot be presented as objectively true either. Yet it seems Rorty is seeking to claim for himself, as all relativists do, an objective foundation for his own pronouncements, while at the same time denying that any such foundation exists for others.

    In short, if Rorty is right about epistemology, then his own thesis is not what he makes it out to be.

    Notwithstanding the above critique, does anyone else just get the impression that there is an air of craziness surrounding Rorty’s beliefs? That truth is only what we say it is? Really? That if someone regards the Earth as flat, then the truth is that the Earth is flat? Anyone who seriously accepts this philosophy would be prone to making mistakes, and advancing relativism. Speaking of relativism:

    2. Re: Sumner’s contradictory beliefs:

    “All truth statements are provisional, (excluding the optimality of NGDPLT.)”

    See? Other non-NGDP based theories of truth are Rortian. Their claims are necessarily relative, based on merely regarding something as true. But just like Rorty, Sumner is claiming a non-relative, non-provisional foundation for his own beliefs. Marxists did this. Nazis did this. It is the kind of polylogism that denies any counter-arguments as necessarily false because they don’t agree. Objective truth has nothijg to do with it. This is really just a lazy attempt to shut down debate.

    What if someone else said “All theories are provisional, except Austrian theory”? How would anyone refute this, without also refuting what Sumner just did?

    This blog post is evidence the author is going mad.

    3. Re: Soltas

    Soltas failed to consider the real side distortions that are caused by inflation, including NGDPLT. There are at least two types of distortion.

    A. In a world market, country level NGDP targeting would have the same type of distortion as would firm level or city level NGDP targeting. In a world market, a country with forced NGDP stability will not release the necessary resources to the rest of the world, as would NGDP that fluctuates according to market forces, should those resources be better utilized in another country. Changes in the foreign exchange rates in a country level NGDPLT framework are insufficient, because continued profitability won’t change management and ownership anywhere near as effectively.

    2. There would also be distortions within the country, between industries. Despite the fact that the theory of NGDLT only considers aggregate factors, it nevertheless has relative effects due to the transmission mechanism of the credit expansion -> inflation system. If there are market driven nominal spending deflationary or inflationary forces, then the central bank forcing NGDP through inflation or deflation will not affect all prices and all spending equally. There is a persistent heterogeneous effect in inflation. The Fed does not increase everyone’s money balances equally. It affects the balances of the primary dealers (on record, black ops off the record) first. These funds then get spent and respent throughout the population in a sequential manner. This objective truth is why it usually takes 12-18 months for an expansionary Fed policy to “work.” It takes time for the new money to increase different people’s incomes. While NGDP itself is not significantly affected by this transmission mechanism, if at all, it is nevertheless true that relative incomes are being affected. Relative income changes lead to relative real side changes. But because these relative real side issues are purely a result of the central bank, and not market forces, the economy moves away from the optimal path, and the capital structure becomes more and more distressed, and the deflationary forces build and build in counter-force to the persistent NGDPLT. The end result is that the money supply growth rate will have to reach infinity in order to prevent the capital structure from correcting which requires relative deflation from the past. This of course is impossible, which is to say NGDPLT cannot be permanent, and will have disasterous consequences if continually imposed exogenously. Price inflation targeting fails not because it wasn’t NGDPLT, but because it is socialist.

  8. Gravatar of bill bill
    10. March 2014 at 09:02

    Shiller started giving a speech about stocks being over priced in 1995. By 1999 or so, he turned that into a book. He wasn’t wrong in 1995 – just early. 😉

    This was a great post by the way.

  9. Gravatar of Jerry Brown Jerry Brown
    10. March 2014 at 09:45

    I don’t understand the Mercedes example. If my friend and I both make an offer to buy the car based on our estimates of its value, I’m the one who is going to get hurt. It doesn’t matter that my estimate is more reasonable. Its even worse with a stock purchase. At least with a car purchase, it is possible that I would be able to use it if I couldn’t sell it. There is no reason I can think of to buy stocks except to make money, so any loss is always a bad thing.

    Or do you mean that there is a “for sale” sign on the car asking a certain price? I can see that my friend would be losing an opportunity to make money at any price between $1000 and $40,000 whereas I would still benefit even though my original value estimate was way too high?

  10. Gravatar of Brian Donohue Brian Donohue
    10. March 2014 at 09:49

    Pessimism = gravitas.
    Optimism = flightiness.

  11. Gravatar of Kevin Erdmann Kevin Erdmann
    10. March 2014 at 10:27

    Great post!

    This is a fundamental reason why one of the persistent anomalies in financial returns is the return to value and to smallness. Lacking wisdom, which pretty much describes all of us, cheap cynicism serves as a substitute. Nothing hurts a reputation more than seeming naïve. It’s very difficult to increase assets under management by explicitly expressing faith and optimism in corporations.

  12. Gravatar of Doug M Doug M
    10. March 2014 at 10:34

    Stock prices and stock returns can a do grow faster than NGDP. And yet, the market cap of total issuance as a percent of GDP does not grow all that fast. This is because equity net-issuance is generally negative.

    So, what was the “rational” price for CCMP (Nasdaq Composite) in 2002 in light of the information that CCMP would be at 4300 12 years later?

    If you bought at the bottom, you would have experienced ~ 12% annualized total return, which is not far off of the long run total return of equities. So, you could say that 1114 was pretty darn close to the “rational price” or you could say that equity valuations have been “irrational” for most of last 150 years.

    Whereas, if you bought at 3200, you would have made 2.5% per year, which would probably be below your expectations.

    Is 12% fundamentally too high of a long-run expected return for equities? Is there an equity risk premium? To get this 12% annualized return, you have to accept 30% annualized volatility. This leads to a Sharpe ratio of ~ 0.3 for stocks.
    0.3 is a much lower Sharpe-Ratio than many other asset classes. Junk bonds have a Sharpe ratio of about 0.5 and Mortgage backed Securities have a Sharpe Ratio near 1. This means that if you could borrow at the risk free rate, you could buy bonds and have the same returns as equities for significantly less risk. And if you are a bank or an insurance company, you do. But since you are not a bank, you can’t and you don’t and you buy the equities that are giving you the relatively crappy return of 12%.

    Equities are also tax-advantaged over bonds, making them a more attractive option for most individuals.

  13. Gravatar of ssumner ssumner
    10. March 2014 at 10:44

    Everyone, Lots of good comments, I mostly agree.

    Jerry, You forget that an investor at the lower price buys many more shares than at the higher price. So even though their absolute loss per share is smaller, the total loss from shorting stocks that rise from 1 to 40 is much greater than going long on stocks that fall from 80 to 40.

    Doug, Ignore my previous comment. I accept your point, but 12% still seems far too high to me. The long run stock returns have been fairly high, but they almost certainly exceeded expected returns by a wide margin. At 12% almost anyone could easily become rich.

  14. Gravatar of Doug M Doug M
    10. March 2014 at 10:46


    Good post and thanks for the link.

    In the chart you link to, equities look to be dirt-cheap in the 1970s. But in real terms this is the beginning of a 10 year bear market (stock prices move sideways while the price of everything else rose).

    I think the PE ratio is upside-down way of looking at the market. “Earnings yield” is a much more intuitively satisfying number. And then we can compare the earnings yield to the long-bond yield, to decide if equities are cheap to bonds. At which point, equities would have looked far less cheap in the 70’s and less rich today.

  15. Gravatar of Tom Brown Tom Brown
    10. March 2014 at 10:52


    “Doug, Ignore my previous comment.”

    What comment?

    @Major_Freedom, check it out:

    “Everyone, Lots of good comments, I mostly agree.” – ssumner

    “Everyone” … that includes you. You can declare victory now!

  16. Gravatar of Doug M Doug M
    10. March 2014 at 10:55

    “Doug, I am talking about stock values, not total returns. Stocks values (in aggregate grow with NGDP in the long run, which means that a subset (a stock index) grows at less that NGDP.”

    Stock market capitalization grows at approximately the rate of NGDP. I agree with you there. But, the index level is not the market capitalization. Index levels are discounted by the number of shares outstanding. Index values reflect the net price changes of the constituents. That is, if the number of shares is declining, the market cap stays constant, yet the index level continues to rise.

    When you look at the levels of the Nasdaq composite, you are quite literally “reasoning from a price change.”

  17. Gravatar of Tom Brown Tom Brown
    10. March 2014 at 11:32

    MF, re: “Everyone, Lots of good comments” … I think that means you’re finally getting through to him. Keep up the good work!

  18. Gravatar of ed ed
    10. March 2014 at 12:16

    Doug, you’re theory is right but I think it is too strong to say that “equity net-issuance is generally negative.” Net issuance has varied over time, but since 2000 it has been net positive rather than negative, and in any case the effects over time haven’t been particularly large.

    An easy way to see this is to look at the S&P 500 index divisor, available at the S&P website if you download “index earnings.” Here are the year end values for the last few years. Increases indicate net issuance, decreases indicate net repurchases.

    2013 8924.03
    2012 8934.61
    2011 9052.93
    2010 9088.35
    2009 8902.83
    2008 8692.85
    2007 8763.44
    2006 8974.73
    2005 9015.96
    2004 9314.65
    2003 9250.51
    2002 9214.85
    2001 9113.82
    2000 8872.78
    1999 8381.82

    Overall I think Sumner’s original point stands pretty well.

    (I don’t know where to get the Nasdaq index divisor, but I am going to guess that net stock issuance over has been even higher there, due to lots of options and stock-based compensation in the tech sector.)

  19. Gravatar of Major_Freedom Major_Freedom
    10. March 2014 at 13:59


    I don’t think so. The game is called pretend my commwnts don’t exist, and if you don’t play, you’re in Sumner’s bad books. Whatever you do, don’t ever agree with me or in any way intellectually flatter me. Then he blows a gasket.

  20. Gravatar of Tom Brown Tom Brown
    10. March 2014 at 14:06

    MF, OK, thanks for the tip 😉

  21. Gravatar of Doug M Doug M
    10. March 2014 at 14:13

    ed, I don’t think the S&P divisor is sufficient.

    If a company splits its shares, there is no new issuance but the divisor increases.

    If company A acquires company B for cash and company C is added to the S&P 500 to bring the list back up to 500 names, net issuance is negative, but the divisor increases.

    According to the Fed Flow of funds data.

    Non-Financial corporate issuance
    2009 -62.1
    2010 -277.4
    2011 -472.2
    2012 -399.5
    2013 -383.7

    The reason I am looking an “Non-financial” is that the financials data includes issuance of ETFs which isn’t really new issuance. If we included financials but excluded funds, the numbers would be more negative. Even with the ETFs the numbers are negative, just not as negative.

    “At 12% almost anyone could easily become rich.”

    That is what every financial consultant will tell you. However, that 12% is nominal and it is before taxes. And, that 12% number is endpoint sensitive. And, it comes with significant risk. While 12% might not be the ideal estimate to use in your financial planning, it is in the right ball park.

  22. Gravatar of ed ed
    10. March 2014 at 15:23


    According to the S&P index methodology document, a stock split does not change the divisor.

    But thinking some more, I now agree with you that some other things besides share issuance/repurchase do in fact affect the divisor, chiefly when a new firm is added, either replacing a firm that has become less valuable or because two firms in the index merge.

    The numerator in the index is the total market value of the publicly traded shares of the top 500 firms. I think we all agree that in the long run this will probably not change too much as a fraction of GDP, so cannot grow any faster or slower than GDP in the long run. Mergers/acquisitions/spin-offs change this fraction in the short run, but don’t change the long run fact.

    So looking at the denominator (AKA divisor) seems to me like a good way to account for all the factors that Sumner needs to account for in order to relate GDP growth to growth in the index level. Dilution from new firms and new shares makes the denominator grow, and indeed that has been the general pattern in past decades. But lately the denominator has been more or less constant or even shrinking, due mainly I would suppose to share repurchases.

    So while I still think the divisor is very useful for this debate, it ultimately supports your point that the index might grow as fast (or faster) than GDP.

  23. Gravatar of RyGuy Sanchez RyGuy Sanchez
    10. March 2014 at 15:49


    Their was a flaw in the earnings data set that was in used in your link. Essentially it was a spliced earnings and operating earnings series that making equities look more attractive than they really are:

    “A particularly thoughtful and well-argued essay that makes this case surfaced in recent weeks. Normally, we address research debates by allowing historical evidence to speak for itself without referring to specific analysts, but we have received a number of notes asking us about our view on this particular ‘The CAPE Ratio is Broken’ argument. In this case, the analyst suggests that the earnings per share of the S&P 500 should be adjusted to use a less volatile series in calculating the 10-year average of earnings. The adjustment suggested in this particular piece was to replace S&P’s reported earnings with an EPS series that is available on the Bloomberg terminal. This is where the error crept into the debate, so we will first focus on this EPS series from Bloomberg.

    We can’t call the Bloomberg EPS data series an operating EPS or a reported EPS because – spoiler alert – it’s actually a combination of the two. The series was created by combining an operating earnings series with a reported earnings series. This makes any comparisons of recent EPS data in this series incompatible with earlier data in the same series. In short, it would suggest that recent valuation levels are more attractive than they really are.”


  24. Gravatar of RyGuy Sanchez RyGuy Sanchez
    10. March 2014 at 15:53

    Anyway Shiller P/E is bunk, it creates a inflation adjusted long-term P/E number that is then used to predict the nominal returns over the next 7-10 years. If it were truly a good indicator it would show good predictive power of REAL equity returns over 7-10 years. It doesn’t UNLESS you break it down into macro environments (postwar boom, great inflation, great moderation), then its predictive ability is much better.

  25. Gravatar of rwperu34 rwperu34
    10. March 2014 at 17:03

    The thing that jumps out at me from the article Ironman linked, the author identifies the 70s as high inflation/tight money (in the cart). Then he says this;

    “The 1980s bull market began in 1982, when Paul Volcker finally conquered inflation (or so the narrative goes), making it possible for the Fed to shift to a looser monetary policy.”

    Seems to me like this blog would think the opposite, that the 70s high interest rates and inflation were due to loose money and things have been policy from 1982 forward was tighter than the previous decade.

  26. Gravatar of Al Al
    10. March 2014 at 17:55

    It seems like there should be some limit to provisionality, if only for the sake of organizing opinions. If price action in 2014 can justify prices in 2000, then why shouldn’t 2016 invalidate such prices, then 2020 revalidate, and so on? In the meantime, the business models wrapped in legal shells may have adjusted so that subsequent price action offers minimal testimony to the quality of investor reasoning in 2000.

    It’s like if someone asks you to travel back in time to kill Hitler. You would correctly note that you have no idea whether such an action will produce worse outcomes in 2014. The respondant would correctly note that you don’t know whether it produces better outcomes in 2100. And yet you must make a decision. Actually, I’ve talked myself into agreeing with Sumner. All judgements are provisional, including this one, and judgements on judgements to the infinite power.

  27. Gravatar of Doug M Doug M
    10. March 2014 at 20:08

    Ed, you are right that a stock split does not change the divisor of the S&P, but it does change the divisor of the Dow Jones.. I got myself turned around there.

  28. Gravatar of Saturos Saturos
    10. March 2014 at 23:20

    C.f. Richard Rorty, there’s a reason why philosophers aren’t that rich.

    Otherwise, the post seems very convincing.

  29. Gravatar of ssumner ssumner
    11. March 2014 at 08:14

    Doug, You have a couple each making $70,000 a year. The put $30k into a 401k at age 30 and live on $110,000. Very plausible. I did something similar (except I married slightly later.)

    By age 66 the money has grown to $4 million. Recall taxes don’t slow the growth of a 401k. Even with 2% inflation that’s a crazy expected return.

    And if that’s just from a single year’s saving!! They can also save money at age 31, and age 32, and age 33. They could have many $10s of millions of dollars. From the sort of income earned by nurses and teachers and cops. A 12% expected return is utterly implausible.

  30. Gravatar of Doug M Doug M
    11. March 2014 at 10:31

    You are correct in principle but your math is a little bit off.

    36 years at 12% is 60x or 30,000 turns into 1.8 million. And again that is nominal and not real… So at 2% inflation your 1.8 million is only worth 900 thousand of today’s dollars. It is tax deferred and not tax free. You will still have to pay your taxes when you take that money out.

    So, is 12% a reasonable expectation? Over the last 36 years the S&P has returned…wait for it… 12.4% annualized.

    That is 36 years that included the 2008 credit crisis, and the dot com implosion and the 87 crash. And, while I don’t have the numbers in front of me, if you took the 36 years before that, the total return was 12% plus or minus 2%.

    Now, you can certainly argue that the next 36 years will not look like the last 72.

  31. Gravatar of RyGuy Sanchez RyGuy Sanchez
    11. March 2014 at 11:25

    Doug M

    30 year treasuries are currently at 3.7%, is there really 8% of free money per year lying around, i find that hard to believe.

  32. Gravatar of dlr dlr
    11. March 2014 at 13:15

    Scott, I can’t believe only Doug M has partially called you out on this. Your assumption about stock indices growing below NGDP growth kills half the point of this post and the parenthetical doesn’t save it. Real stock returns over a lot of years and many countries have averaged 6-7% and reflect risk premiums. If you think real yields should be lower because you blind risk takers should be able to get rich, you probably should also markets should trade at (say) 25X normal, forward earnings — since most assets are real — which they almost never do. Unless your post is meant to be an addition to the equity premium puzzle literature, I think you have to accept the (provisional!) equity risk spreads the world has actually experienced.

    While 12% nominal is a bit high, a better guess for the fair but not risk-free real return from a slightly higher beta segment of the market like the Nasdaq is ~7-8% real plus inflation (inflation has averaged 2-2.5% since 2000). That would mean an investor in 2000 would want 10% total nominal returns (7-8% real) from the Nasdaq in 2000, and thus would pay about 1300 while collecting a hypothetical starting dividend yield of ~60bp (the actual dividend yield of the Nasdaq in 2000 was 20bp).

    Your are exactly right that all judgments are provisional, but you are not right that today’s provisional pretense of knowledge renders 2002 worse than 2000. The 2000 Nasdaq is still almost 400% of the hindsight Nasdaq, while the 2002 low is about 72% of the hindsight Nasdaq (2000 Fair value * 1.09^2). You’re going to have to wait until tomorrow for your provisional reversal on this one. For the Nikkei, you might have to wait a bit longer.

  33. Gravatar of Major_Freedom Major_Freedom
    11. March 2014 at 14:17


    “Your are exactly right that all judgments are provisional”

    What about the statement that all statements are provisional? Is that provisional as well? If so, then it cannot be apodictically true, in which case anyone who believes that is obligated to consider the argument that some statements are not provisional, is true.

  34. Gravatar of marcus nunes marcus nunes
    11. March 2014 at 15:06

    “I don’t ever recall Robert Shiller saying stocks were underpriced and that people should buy, but I recall him calling stocks overpriced on numerous occasions. Can someone confirm this impression?”
    In a Board presentation on November 1996, Shiller and Campbell were so bearish on the stockmarket that in a speech on Dec 5 Greenspan coined the “Irrational Exuberance” expression.
    That day the Dow had closed at 6437 points. A little over three years later it peaked at almost 11500 points!
    No, Shiller never said stocks were “cheap”!

  35. Gravatar of ssumner ssumner
    11. March 2014 at 15:20

    Doug, Ask people who put $30,000 into their 401k whether they expect that one year contribution to turn into $900,000 in purchasing power when they retire. remember, we are talking about a single year worth of contribution. That would be a spectacular success.

    The past 36 years are simply wildly atypical. In addition, you are looking at total returns, and in my post I made no claims about total returns, i was discussing stock indices. The Dow has been measured since the early 1900s and has risen at less than the rate of NGDP growth. Thus I would expect future increases to be less than NGDP growth.

    dlr. You said;

    “Your assumption about stock indices growing below NGDP growth kills half the point of this post and the parenthetical doesn’t save it. Real stock returns over a lot of years and many countries have averaged 6-7% and reflect risk premiums.”

    That’s a complete nonsequitor. I am talking about the fact that stock indices tend to rise at less than the rate of NGDP growth over the long term, which is true, and you respond with total returns. Did I say anything about total returns in my post?

    I don’t know of any model of risk that would justify the sort of EXPECTED real returns you suggest. And as your Japan example clearly shows the US is an outlier. Many markets have suffered collapses at one point or another over the past century. The US economy has been THE success story of the past 100 years. But no one knew that 100 years ago, and hence I very much doubt the actual returns were expected. Would you estimate expected return for internet stocks by looking at actual returns of Google? No one should study risk premiums by looking at US market data. If I knew NASDAQ averaged real returns of 7-8% going forward I’d put all my wealth there without batting an eye. Yes, it’s done well recently, but past performance is no guarantee of future success, nor is it a good way to estimate expected returns.

  36. Gravatar of Major_Freedom Major_Freedom
    11. March 2014 at 17:00

    “I am talking about the fact that stock indices tend to rise at less than the rate of NGDP growth over the long term”


    I would say that stocks grew faster than NGDP up until 2000. This is reflected by the fact that the 2nd derivative appears to be quite larger for stocks than for NGDP.

    For the S&P, if we take the 2000 peak to the current peak of the S&P, then NGDP has been growing faster than stocks.

    The growths have been the same if we consider 2002 intersection of S&P and NGDP, to the present.

    Since 2009, it’s not even close. Stocks by a mile.

    Over the long run then, I think it’s more reasonable to argue that stocks increase faster than NGDP, on average.

  37. Gravatar of Kevin Erdmann Kevin Erdmann
    11. March 2014 at 19:35

    I, for one, am tickled by the irony that in the comment section of this excellent post about the pessimistic bias, Scott has been pulled into an argument where he is saying that a 4% equity risk premium that persisted for a century is the result of a lucky break – that a cumulative excess return of some 5000% over that time doesn’t reflect an expected return premium for equities. On average, less than every 20 years, equity portfolios have doubled government bond portfolios. Scott says, if he thought he could expect those returns, he’d put his money in equities “without batting an eye”, but since he’s not a Pollyanna like these naïve commenters, he’s not biting.

    And, he says, indignantly, that if you told a young couple that $30,000 invested today could be worth $1.8 million by retirement if it was invested in a volatile equity index, they would rightly scoff at you. And, the fact that commenters are using historical experienced returns to make this claim is no defense!

    Scott, re-read your own post, please! 😉

  38. Gravatar of Kevin Erdmann Kevin Erdmann
    11. March 2014 at 23:10

    Using Kenneth French’s data:

    Using annually constructed equity portfolios (each portfolio was funded just once, at initiation), by stock market cap quintile, for portfolios funded from 1927 to 1980, the portfolios reached 50x their original nominal value, on average:
    25, 30, 31, 33, and 38 years after funding.

    The lowest quintile never took more than 34 years to provide 50x total returns and the middle 3 quintiles achieved 50x the original nominal value in 40 years or less in all but 3 portfolios (out of 162 portfolios).

    A young couple using a 1 time tax-free savings of $30,000 had a million & a half by retirement in pretty much every historical period of the past 85 years, if they invested in equities.

  39. Gravatar of dlr dlr
    12. March 2014 at 05:26


    The reason it’s important to talk about total return and not index levels is because the Nasdaq’s payout ratio was unusually low in 2000, and would have been unusually low (far below the average payout ratio for a world equity index) even at 1/3 the price.

    And no, the US is not Google. Real stock returns for the US from 1900-2000 (when investors are displaying their risk-preference/expectations) were 6.7%. Here are the others:

    Belgium 2.5%, Italy 2.7%, Germany 3.6%, Spain 3.6%, France 3.8%, Japan 4.5%, Denmark 4.6%, Ireland 4.8%, Switzerland 5.0%, UK 5.8%, Netherlands 5.8%, Canada 6.4%, South Africa 6.8%, Austria 7.5%, Sweden 7.5%. The average is 5% and the size weighted average is around 5.5%. If we use a 5% real return plus 1% for the Nasdaq’s excess correlated volatility you get a 6% real return plus 2.4% inflation. That’s about 8% nominal after accounting for the low adjusted dividend yield. That still gets you 1500 provision FV on the Nasdaq in 2000.

    Here’s the kicker. The real yield available on ten year inflation protected securities in early 2000 when the Nasdaq peaked was 4.5%. Are you going to tell me that expected returns for the Nasdaq should have been below that?? You can argue that the 20th Century countries on the list are also essentially akin to Google, but it seems like a silly argument relative to comparable lowest risk available returns.

  40. Gravatar of ssumner ssumner
    12. March 2014 at 05:53

    Kevin, You are making the same mistake as other commenters. You need to look at the ratio of stock indices to NGDP, not total returns. That was my claim.

    Let’s take the 1927-80 period you cite. I can’t find the exact data, but I believe the Dow rose from about 200 to 1000, a 5 fold increase. In contrast, NGDP rose from about 90 to 2850, a more than 30 fold increase. The gap is so large that it would presumably hold for other indices as well. So even in the time period you select my claim is strongly supported.

    Regarding the “century of data” argument, sorry, but that’s a misuse of statistical significance.

    As far as my own personal investments, I’ve been in stocks for all but one of the last few decades, my comment was on investing in NASDAQ specifically. I’ve had huge returns from my investment, but they were also far greater than anticipated by me.

Leave a Reply