Daily Speculations

The Web Site of Victor Niederhoffer & Laurel Kenner

Dedicated to the scientific method, free markets, deflating ballyhoo, creating value, and laughter;  a forum for us to use our meager abilities to make the world of specinvestments a better place.

 

Home

Write to us at: (address is not clickable)

12/15/04
Pause for Cause, by Kim Zussman

From: Gregg van Kipnis: "If there are lag structures in a relationships (that usually argues for causality, unless you think time runs backwards)..."

Recently had the pleasure of attending Junto as a guest, which featured speakers Larry Kudlow and Steve Kagann, NY state chief economist. Dr. Kagann made the point that "blue" (Dem) states had the highest unemployment and taxes and worse economies than "red' states. The message was less government is better which makes sense. However there was the causality question, which I did not phrase very well but will try now:

Does voting Democrat cause states to have worse economies, or do worse economies tend to vote democrat? (there is a related hypothesis about taxes and economies). Addressing such (and all related) questions by introducing a lag seems like a logical leg up on causality:

Define fraction Democrat voting as (Dem votes)/(Rep+Dem votes), and use as independent variable following Presidential elections to evaluate effect on dependent state economy variable (such as state per capita GDP) in 4-years following election. Repeat regression over many Presidential cycles, and look for significant relationship.

OTOH, it is possible that states with poor economies tend to vote Dem. For this one could reverse the variables and the lag such that prior 4-year's state economy per capita is the independent and the election vote (fraction Dem) the dependent.

In medical research there are often many confounding variables. This week's Time features a timely article on sleep biology, and refers to the old saw about needing >8hr/nite is false since prior research showed people who sleep>7.5 die sooner. Story fails to mention that one symptom of depression is somnolence, and depression shortens lifespan for many reasons. Other known relationships which are not causative include male seated height and prostate CA risk, male pattern baldness and coronary disease, earlobe fold and coronary disease, left-handedness and shortened lifespan (possibly flawed study).

In speculation we look for lagged dependence with statistical significance, especially recently. Is this equivalent to causality, and should we care? If a certain move in bonds presages (by back-testing, with significance) a move in stocks, does one cause the other, and if so should we be more confident in the relationship? What if the bond move effects a multitude of factors, a cascade, which ultimately moves stocks. Does one need to know the mechanism? Perhaps economics would approach the deterministic veracity of physical sciences upon elucidation of mechanistic cascades were it not for the profit incentive of non-disclosure.