Do markets learn from each other?  For example, is the S&P market this year following a similar path to bonds last year, with every trepidatious move down being requited with a rise? Are such "learnings" graduated to the point of regularities. And is it a domino effect or a path of least resistance or consilience or convergent evolution or what have you? What do you think?  Can it be quantified? Should it be quantified?

Allen Gillespie answers:

Yes they do.  In the old days that could be because some pits closed before other pits, so you could have individuals walk between pits and trade both.  Today it would be the result of correlation trading desk and carry trades.  The correlations, however, do change depending on what the Street owns - it is simply a balance sheet effect. There is also the issue of relative levels v. absolute levels.  This is important because leveraged and professional traders may act on relative relationships more quickly and at higher price levels, where as, unleveraged traders need both an attractive relative and absolute relationships to act.  This is the problem with ZIRP.  Discount models become meaningless near zero i.r. as values become highly sensitive to small changes and price changes gain increasing amplitude.  It encourages leveraged carry trades regardless of the absolute levels sought by cash buyers. That's the issue now no? Inflation targeting is just the gold standard, and under the gold standard short term interest rates were highly volatile while long term ones were not because there was no long term inflation except for natural growth which caused periodic revaluations of the currency. So, the Fed views long bonds as the same as cash, however, this is because the Chairman does not believe that currency is supposed to serve the function of acting as a store of value.  If it cannot serve that function then it truly is worthless.  

So, stocks are following bonds because there's 600 bps of carry (relative to the Ten Year last August) or 800 bps of carry (relative to ZIRP) or 500 bps relative to corporates.  I call it the old bankers trade (3-6-3), borrow at 3, lend at 6, play golf at 3. Stocks then would give you an additional 280 bps on the 6. The issue now is 0,3,6 (borrow at 0, lend at three, get up at 6 because at $1360 on the S&P the earnings yield is 6.4 and this probably does not adequately reward investors in a world where the currency is worthless and where the value of a $1 perpetuity at 10 bps is $1000 and at 11 bps $909 and at a mere 15 bps back to the deal with the devil $666 that the S&P made in 2009.  Another way to consider it is to think you will get that 6.4 two thirds of the time or 4.25% which is still better than a bond, but is it adequate?  AA bonds have a 20 year cumulative default probability of 2.71% but an average loss severity of 36.5% or so at anything less than 2.3% they really have used your money for free for 20 years.  Didn't our president say something about spreading the wealth around? Well, using money for free would meet my definition of how to do it.  So, we are in the ABT market (anything but treasuries) since last August and I truly suspect current bond issues will someday be referred to as those Obamanations like the old not worth a Continental all made possible by Wall Street's deal with the devil at Fed.  The rub is the math works both ways.  

One way to quantify would be to take all market periods (say one year) and then run a correlation against the second market 1 year forward, with the hypothesis being that would be see consistently high correlation numbers.  Thus one would see not just the direction of the relationship but the consistency.





Speak your mind

2 Comments so far

  1. Craig on February 22, 2012 10:29 am

    The S&P and copper mostly move together but February isn’t similar. Maybe it’s the short-term relative strength of interest rates and oil prices compared to stocks. FedEx and the transports seem to not like the oil price and interest rate relative strength either.

    With growth rates of inflation and the economy adding up to near 5% and Fed Funds at 0%, nothing makes much sense. The Fed operations are kind of like playing golf with GOTCHAs. You get someone to give you 3 gotchas and shout 1 gotcha during the opponent’s swing early in the round. The key is to not use them up, so he spends the whole round worrying about when you might scream next during his swing. Maybe the Fed power is more like a GOTCHA than they know. None of the economists talk about the Fed’s lacking ability to influence a less capital intensive service economy. Nobody talks about the lack of elasticity on every level from national economy down to the states. If you look at 2-year growth rates (compares to a 4yr avg) of the Conf Board indexes, the lagging is tracking the coincident index higher in this recovery. The only time they even touched was in the declines of 1973-75 and 1981-82. We’ve never seen a recovery like this. Similar to how inflation’s lag can be used to estimate duration of an expansion, the lack of lag by inflation or lagging economic indicators can’t be a good thing. The economy shows annual growth rates with lagging indicators outpacing the economy since April. Maybe 2-year grates will eventually invert also but leading indicators like the stock market normally weaken before that happens.

  2. douglas roberts dimick on February 22, 2012 11:37 pm

    Unsystemic Valuation is Systemic

    Lowly to those masters of the universe here at daspec, I will not venture into trader or banker domains concerning bond and index trending (or treading). That said, as to the issue of what is the proper model for quantification – yes, it can be quantified – of market correlations, consilience (not convergent evolution, whatever that means) is defined by systemic ecology when modeling unsystemic risk.

    The architecture of electronic exchange market systematics requires models to define points of entry and exit. Therefore, indicator convergence/divergence first requires a rules-based declaration of protocol(s) for quantifying consilience of a given set of program indicators.

    Allen citing desk and carry trades provides an example , whereby (as he so notes) distinguishing between “relative and absolute levels” of leverage presents the problem that “the math works both ways.”

    Then there is Craig’s “nothing makes much sense” surmise concerning inflation and Fed Funds. Why would it? The rules of the game keep changing in Washington as well as around the rest of the world now that global economics are dominating national and regional markets.

    I just viewed a TED youtube… Paddy Ashdown: The global power shift…


    Ashdown speaks of how global power is shifting not only laterally but also vertically with globalization of economies as well as within the markets themselves. The Point relative to Victor’s query as to a model?

    The rules are changing. Moreover, with globalization of power, and not just nation-state power, those correlations that our masters of the universe have in the past so steadfastly banked upon now increasingly face unregulated space. So of course not much makes sense.

    Ten years ago, when I began researching how to approach markets for the purpose of constructing an investment strategy if not vehicle, trends in SEC and GAAP (and Basel) indicated that external (including fundamental or nonsystemic) dynamics of market correlations were increasingly disparate. Insider/outsider and churn-and-burn paradigms were no longer functioning in accordance with the old brown-shoe constructs.

    Case in point: Where is the consistency of valuations today both within and among asset classes?

    Having constructed on averaged more than one function-indicator design per day during this past year, my answer to our Chair’s query as to a “proper model” is that there is not one model. You can use Pi or sine or squareroot or a multiple-power series to correlate for purposes of quantifying consilience.

    What one cannot do, it appears to me, is batch securities or correlate markets – like-kind or otherwise – as has been done prior to 2008 without considering the physics of electronic exchange mechanisms. The apples-to-oranges regime of stochastic mappings for algorithmic correlations to define points of convergence/divergence, such as for benchmarking, increasingly invites such models being consumed by the vary vacuums of valuation that they are attempting to quantify.

    I could be wrong here. Is there an absolute operating in the markets?

    Fear and greed so one may cite. Sure, but market and investor psychologies present that constant set of variables that existed from the onset of free market exchange, well before the rise of electronic exchanges.

    My take, The Theory of Quantitative Relativity, posits that electronic change systematics offer the best if not only rules-based spatial array of price/time correlations. Victor has written a fair amount on nonrandom sequences; his approach is predominantly predicated on statistical indication. Consider, though, what we may term as condition(s) precedent… What are the assumptions that define price-action for patterning statistical domains that indicate market congruencies?

    Linear mapping of market topology, for instance… How do we define order-size domains relative to plotting (high-low) polarity of price action?

    Many consider Victor to be the father (or one of them) here. The gravity of his question, therefore, seems to take us to the core of market ecology.

    That said, one may find unsystemic risk now to be ever more so correlated to system risk. To the extent that one concurs for purposes of modeling, then rules-based configurations become determinative for finding indicator-function correlations.



Resources & Links