Wednesday, September 16, 2015

The Menzie Chinn (rhymes with Pnin) test


The 1954 Fall Term had begun. ....

...the bestowal of a particularly generous grant was allowing the renowned Waindell psychiatrist, Dr. Rudolph Aura, to apply to ten thousand elementary school pupils the so-called Fingerbowl Test, in which the child is asked to dip his index in cups of colored fluids whereupon the proportion between length of digit and wetted part is measured and plotted in all kinds of fascinating graphs.
The University of Wisconsin's chief chartist is busy;
I have been stressing the international implications of a potential interest rate increase as a rationale for deferring monetary tightening.
Which--give him a few minutes;
This is actually a harder question to answer than one would think.
You don't know the half of it, guy.
A regression (in first differences) over the 199M03-2015M08 period yields a statistically insignificant negative coefficient on the interest differential, with zero adjusted R-squared. Augmenting with VIX leads to an increase of adjusted R-squared to 0.15, but no change in the coefficient on the interest differential (the explanatory power is provided by the VIX).

As I’ve pointed out before, the US dollar does seem to be correlated with the shadow Fed funds rate. Without a good measure of rest-of-world interest rates, I use the shadow rate for the euro area.
Why not use squirrel entrails mixed with;
Eye of newt, and toe of frog, 
Wool of bat, and tongue of dog,
Adder's fork, and blind-worm's sting,
Lizard's leg, and owlet's wing,
That would yield every bit as useful a conclusion as;
Δet = 0.0014 + 1.550Δ(it-it*) + 0.121ΔVIXt + 0.818Δ(it-1-it-1*) + 0.090ΔVIXt-1 + ut

Adj-R2 = 0.52, SER = 0.008, Nobs = 78, DW = 1.54. bold face denotes statistical significance at the 10% msl using HAC robust standard errors.

The cumulative impact of a one percentage point interest rate increase is 2.33.
You can take it to the bank.

1 comment:

  1. It is misleading to look at the past, pick out likely factors, and declare that a theory is correct.

    === ===
    http://www.johndcook.com/blog/2011/06/21/how-to-fit-an-elephant/

    Renowned mathematician John von Neumann: "With four parameters I can fit an elephant, and with five I can make him wiggle his trunk."

    By this he meant that one should not be impressed when a complex model fits a data set well. With enough parameters, you can fit any data set.

    It turns out you can literally fit an elephant with four parameters if you allow the parameters to be complex numbers.
    === ===

    ReplyDelete