Friday, October 5, 2012

The velocity of money and inflation

Before I get in to this to any extent, let me first say that I have not drawn any conclusions nor do I have any profound insights on following. If you have been following the debate on inflation, you undoubtedly have heard one pundit or another talking about how there cannot be any inflation in the system because the velocity of money is very low and is declining.

Without getting too much into the background here, this argument stems from the equation MV=PQ, where M is the money supply and PQ is essentially productive output (Q) and price levels (P). Working a little algebraic magic yields the velocity of money (V) or V=PQ/M. As the theory goes, V and P are related and generally move in tandem. So are these pundits correct?

If this was any year prior to 1990, I would say most definitively yes. However, the apparent relationship between the velocity of money and inflation (here defined as the year-over-year change in CPI) has broken down since then. See the chart below. Prior to 1990, the change in monetary velocity and inflation appears directional the same.




Since 1990, There does not appear to be any relationship whatsoever or at the very least a tentative relationship. In fact, the following is the graph of the 10-year rolling correlation of the 1-year change in both monetary velocity and inflation, and it shows this breakdown in the relationship, beginning in or around the mid-1990's.



What does mean or signal? I am not sure. Could this be due to productivity? Potentially. The correlation again appears to have shifted in years, specifically increasing in the years beginning in between 2006 and late 2010. At the very least, I think the graphical analysis suggests the relationship between monetary velocity and inflation is more complex than what is implied.

All that glitters 10/5/12- still on the sidelines with gold shares

I remain on the sidelines as it pertains to the precious metal shares. The three time-variable models I run continue to suggest that the future performance on the Phily Gold/Silver Index (ticker XAU) will underperform the average results.

The 1-year model is currently running at a 1. Reminder- a positive number indicates negative future results. 


This is up from -0.2 a month ago and -1.4 in the beginning of July. Since the beginning of 2000, the performance of XAU following periods when the indicator was 1 or above was 60 basis points below the buy-and-hold average on a 1-week forward basis. On a forward 3-month basis, the performance of the XAU was 295 basis points below the average when the timing indicator was 1 or above.

As for the 6-month model, the 6-month timing model is currently 2.1.


Again, this indicator is up from a 0.97 a month ago and -1.3 three months ago. Since the beginning of 2000, the performance of XAU following periods when the 6-month indicator was 2 or above was 174 basis points below the buy-and-hold average on a 1-week forward basis and 512 basis points on a forward 3-month basis.

Lastly, the most volatile model, the 3-month model, is in an 'underperform' range with an indicator level of 1.5.



The 3-month indicator has improved from a 2.2 a month ago, but is above the -0.9 seen three months ago. The underperformance trends in the three model are similar to both the 1-year and 6-month models. On a forward 1-week basis, the XAU has fallen behind the average performance by 143 basis following an indicator of 1.5 or greater and 530 basis points on a 3-month basis.

One last note, I ran a scenario analysis on the historical results and calculated the 3-month performance when all three models were at or above presently seen levels. This situation has occurred 67 times in all the weeks since the beginning of 2000, As for the performance, the XAU has underperformed it's own average by 529 basis on three month basis and has been up only 43% of the 3 month periods since 2000. I would get more constructive on precious metal shares if money supply increase significantly, the price of gold pulls back, or provided the time decay present in the timing model.