Flaws in Warren Buffett's reasoning re: Treasuries "bubble"

Discussion in 'Economics' started by Cutten, Mar 1, 2009.

  1. Cutten


    Buffett recently described US Treasuries as a bubble - and not just that, he said it was arguably bigger than the prior housing and dot.com bubbles that preceeded it. I examined his comments from the latest shareholder report, and would like to present a calm, factual, logically reasoned explanation of why he is totally wrong - not only that, why he is making extremely basic noob errors of macroeconomic analysis. Note that this is not a market prediction, or a personal opinion - everything I will say in this post is demonstrable fact, derived from commonly accepted or provable data, and drawing only those conclusions which, using rigorous logic, necessarily derive from that data and from accepted theories of what that data means and what can be extrapolated from it. I also happen to think Buffett is wrong from a personal/trader perspective, but that is not germane to this post.

    The primary subject under consideration here is whether US Treasuries are in a historic bubble, comparable or greater than the dot.com and housing bubbles that preceeded it. I will argue, and prove indisputably, that under no circumstances could a rational person consider them to be a historic bubble at current price levels (3% yield on the 10 year). All the data shows is that, at worst, they are a bit pricey - and one can make a plausible argument that they are fair value, and maybe even a little cheap (if the economy gets really bad in the next 5-10 years). However, I am not in this post trying to say they are cheap or fair value. I'm simply trying to prove that they are NOT, under any possible view, a historic bubble comparable to dot.com, housing, Japan 1989, gold 1980, 1929 and so on.

    Another secondary goal is to demonstrate that Buffett either lacks sound macroeconomic analytical and/or forecasting skills, or has for some reason not exercised them in this case (which implies that one should not rely on his views in future). Buffett is a shibboleth at present and I will argue that the world's collective opinion of Buffett's analytical skills (in the macro area only) is arguably a far greater "bubble" and collective delusion then the price of US Treasuries. This second claim is far more controversial, and sure to annoy a lot of people, so I would prefer if discussion is confined mainly to the topic at hand - are US bonds in a historic bubble. Bashing Buffett merely amuses me in passing (I generally rate him highly), solely because so many people admire him brainlessly, like sheep.

    I will continue in the next post.
  2. Cutten


    To answer the question, first we need to ask what is a bubble? If we can't agree on that, then we'll just have endless semantic debates, and end up talking past each other. So let's agree on a definition - here's mine:

    Bubble (noun): a situation where the price of an asset is so expensive that it is way beyond any reasonable valuation that could possibly be inferred by the fundamentals.

    Note that under this definition, merely being expensive is not a bubble. Manhattan has been expensive for 100 years, for example. The fact that the price later crashes does not mean it was previously a bubble - Detroit real estate was never a bubble, but it crashed due to deteriorating fundamentals. Something is only a bubble if the price is *so high*, that even an optimistic reading of the fundamentals cannot justify a valuation anywhere near that.

    From this definition, it's clear that there are several steps needed to be able to identify bubbles:

    1 - you must have a sound valuation model. If you cannot determine at least a range of reasonable values, there is no way you can say something is overvalued or not.

    2. You must be able to make at least a range of plausible predictions for future fundamentals. If you do not know if the US will be around tomorrow, or will go into 50% per annum deflation, or hyperinflation by next month, then you cannot value any assets in it. On a more limited basis, if you have no clue what the S&P earnings for 2009, 2010, 2011, 2012 will be, it's hard to judge value because you have no clue on future earnings.

    N.B. You don't need to be able to predict things precisely (which is impossible), but you do need to be able to come up with the 3 or 4 most likely possible outcomes on a fundamental basis. For example right now, you could say either the US has a fast steep recession and recovers, a lengthy recession, or a long depression - and maybe in the former case you get high inflation returning too. So that's 4 different but possible scenarios - reasonable people can disagree on which will occur, but no reasonable person can say any of those forecasts is ridiculously implausible.

    3. You must know what kind of valuations are "normal", and what kinds are absurdly aberrant. For example if fair value on the S&P is 15, but the S&P frequently ranged from a PE of 1 to as high as 100 every few years, then you could not say it was a bubble just because it hit 50. Whereas if a PE of 50 is only attained once or twice per century, then it is much more likely to be truly aberrant and hence quite likely to be a bubble.

    So, to summarize:

    A bubble is when an asset becomes insanely overvalued such that even the most optimistic assessment of the fundamentals gives us a a valuation way lower than the current price.

    To be able to judge if something is a bubble, you need to know i) how to value it ii) what are the main plausible scenarios that could arise in the future iii) what is the historically normal range for valuations, and what is the bubble range.

    If we apply this to US Treasuries, we should be able to answer whether Buffett is right or not.
  3. the question you need to answer is if you loan the us government money for 30 years for a 3% return will you make or lose money after inflation at maturity in 30 years. most people would say no.
  4. Cutten:

    You made similar arguments in another thread where I called a short term top when TLT was in the 115 area. TLT is now at the 101 area as of this Friday.

    You viewed at the time that shorting TLT in the 115 area was a "crazy" thing to do. I wrote at the time that you were wrong, and that the bonds would fall. The bonds have fallen, and TLT is back to 101 area.

    The market proved your reasoning to be wrong. If TLT goes back up to the 115 area, if you had longed the bonds at that level, and if I were you, I would accept the fact and get out that losing position.

    Market has spoken. Your reasoning and/or facts are wrong. Sorry buddy.

    PS: Anyone who shorted TLT at 115, or bought the inverse of it, take your profits and get out, with TLT at 101. Bank those profits, and you can always go back if shorts bring it back up while covering.


  5. fhl


    I think that going into the kind of debt we're talking about here while at the same time using the tax code and new regulations to discourage investment, and thus the means to pay it back, makes it look to some that doing it without printing buckets of money is increasingly implausible.
  6. Cutten


    So, how to value US Treasuries? The main valuation components are inflation expectations, and the risk premium. Inflation expectations should be easy to understand. The risk premium is made up of two components - default risk (for foreign investors, it also has currency risk), and the general risk premium (which applies to all investable assets including stocks, corporate bonds, munis, real estate etc).

    Treasuries are easier to value than most bonds, because the default risk is very low. The US is less likely to default than pretty much any country, and certainly at worst it is no more likely to default than other developed countries. So I think it's reasonable to assume a negligible default risk premium - and Buffett indeed does not even mention default risk as a reason for his view.

    So, that leaves us with two other components - inflation expectations, and the general risk premium. Thanks to the existence of TIPS, which have almost identical credit risk, and which pay out based on actual future inflation, we can isolate these factors. Current TIPS real yields are 2.06%, 10 year Treasuries yield 3.01%. Therefore inflation expectations are 0.95% for the next 10 years, and the risk premium for Treasuries is 2.06% (actually it may be slightly different, since they could have a liquidity premium over the less-liquid TIPS, but this effect should have a negligible impact on prices).

    To say that Treasuries are overvalued (let alone in a historic bubble), we only have these two factors to go on - either the risk premium is too low, or the inflation expectations are too low. If neither is too low, then the bonds are either fair value, or actually undervalued. For them to not just be overvalued, but a historic bubble, then the risk premium and/or inflation expectations must be so low that NO plausible forecast of future fundamentals could justify such low yields. Let's see if that's the case.

    I will start with the real risk premium, since that is easy to dismiss. Here is the real yield of TIPS over the last 6 years:


    As you can see, the 10 year real yield ranged from 1% to 3% during this period, a relatively narrow range. The current yield is 2.06% - right in the middle of the historical range. For a relatively safe asset such as Treasuries a 2% real yield is pretty normal. Thus, there is no way one can say the real yield is abnormally low - if anything, it is a bit on the high side (i.e. cheap) because of the safety features of government debt during a financial crisis.

    Since the real yield cannot be called low at all, let alone abnormally "bubble" low, that leaves us only with the inflation expectations. Given our 3 valuation components, and the total normality of current valuations in 2 of those, only inflation is left.

    Now, remember our previous definition. Being expensive is not a bubble. Prices subsequently falling is not a bubble. A bubble is where there price is so high that no reasonable fundamentals could remotely justify the valuation. Even under the most bullish likely scenario, the price is still too high even for that.

    This leaves us with one last question - are there any plausible scenarios that justify a 1% inflation forecast, or at least something close to it, for the next 10 years? If so, then Treasuries cannot be in a bubble. To call them a bubble means that you think no reasonable individual could rationally expect inflation to stay at an average of 1% or lower until 2019, under any plausible circumstances.

    Is that a reasonable claim? Let's examine the evidence in my next post...
  7. Cutten


    No, that is not the question. The question you posed there is simply "are 30 year treasury bonds yielding 3% going to outperform inflation?". I am not, in this post, saying whether they will or won't outperform inflation. I am asking "are they a bubble, similar in magnitude to the dotcom and housing bubbles, as Warren Buffett is claiming?". My contention is that they are not.

    Your criteria for judgement - what "most people would say" - is also flawed. Most people in 1997 would not have said that investing in US stocks for the next 12 years would underperform inflation, but that is exactly what happened. Most people in 2007 would not have said house prices would crash by the most in 75 years, or that the S&P would fall 50% in just over a year, but that is exactly what happened. Most people don't have the slightest clue about how to forecast financial markets or investments. That is why I am not relying on the judgement of "most people".

    Remember, an investment asset being a bit overvalued, or losing some money, does not mean it was in a bubble. The Nikkei is lower than it was in 1998, but no one in 1998 thought it was a bubble and no one in 2009 thinks that the Nikkei circa 1998 was a bubble. A bubble has strict criteria - INSANE overvaluation, and the certain loss of huge amounts of capital. Not mild overvaluation and *possible* (not certain) small loss of capital.

    For what it's worth, I will actually put forth some arguments for why the conventional wisdom might be wrong i.e. that a 10 year at 3% might actually be not only fair value, but possibly *undervalued*. But you don't need to agree with that to think Buffett is wrong in his bubble claim. Even people who would not touch bonds at current prices can accept my reasoning, and just say they are too expensive, or have too much risk at current levels.

    Your post does raise the question what is a US treasury. I am focusing on the 10 year, and I do agree that the 30 year seems worse value and more risk. However, I do not think even the latter is a bone fide bubble, let alone comparable to the dotcom era or housing bubble.

    So, back on topic - are US Treasuries a bubble? Please can anyone posting here only address the points that I make in this thread relating to bond market valuation, NOT spurious, extraneous, or arbitrary and meaningless assessments such as "most people say" or "Buffett says" or "the trend is down" or whatever. I am not arguing based on those factors, I am arguing based on universally accepted valuation metrics for bonds, and for bubbles. So please limit your criticism or agreement to those metrics and the conclusions I draw from them.
  8. Cutten


    Again, that's not the point of my thread. I am not trying to make a bull case for bonds here. I am simply saying that there are no logical grounds for calling them a bubble of historic magnitude at current prices.
  9. janvir19


    Agree with Cutten. Treasuries are one of the few assets that are undervalued right now. When the yield on the 30 year hits 1.5%, then it's a potential short. Right now the output gap is massive and inflation is years away...
  10. Cutten


    Back on track...

    "This leaves us with one last question - are there any plausible scenarios that justify a 1% inflation forecast, or at least something close to it, for the next 10 years? If so, then Treasuries cannot be in a bubble. To call them a bubble means that you think no reasonable individual could rationally expect inflation to stay at an average of 1% or lower until 2019, under any plausible circumstances.

    Is that a reasonable claim? Let's examine the evidence"

    To do this, we need to look at what are the plausible range of forecasts for inflation over the next 10 years, which requires us to forecast the possible range of economic outcomes.

    Let us then ask - what forecasts, if they were to come true, would result in inflation of 1% or lower? I can see a few:

    1) Great Depression rerun - trade wars, banking collapse continues, restrictive regulation and legislation crushing enterprise, large government borrowings crowding out productive private sector investment, and so on. What will that do to inflation? It won't be a matter of what the inflation rate will be, it will be a matter of how much *deflation* we have. Inflation could easily be -1% or -2% for the next 10 years under this scenario.

    2) Long-term stagnation - think Japan 1990s. Less extreme than the first scenario, this would see very low growth over the decade, and inflation could be perhaps in a -1% to +1% range.

    3) Eventual recovery with anaemic growth - here would could postulate a low growth economic with restrictive credit and conservative bank lending, resulting in fairly low inflation in the 0-2% range.

    Which forecasts would disprove it?

    4) Normal recovery with normal growth. Here inflation would come back in 2010/2011 and increase to 2-4% for the rest of the decade as in a typical growing economy. The banking system would operate just as it did in the 1990s or 2000s, once the crash and recession were over.

    5) Major inflation >4%. The recovery occurs, and then for some reason the Fed pursues a deliberate policy of creating inflation, e.g. to help Congress pay off its large debts.

    Now, scenarios 1 to 3 would all be consistent with inflation at 1% or less on average for the next 10 years. Thus if ANY of scenarios 1 to 3 are remotely plausible, we cannot be in a bubble for Treasuries. Remember, a bubble requires valuations so stratospherically high that no plausible scenario for fundamentals and valuations could justify such high prices. If even one scenario is plausible and would justify it on a strict valuation basis, there's no bubble.

    All that remains then is to examine the plausibility of scenarios 1, 2, and 3. If any are plausible, then there is no bubble, and Buffett is totally wrong.
    #10     Mar 1, 2009