Thursday, September 30, 2010
The Federal Reserve, the Bank of England, and the Bank of Japan are considering further quantitative easing. It's an explicit statement, as with the Federal Reserve and the Bank of England, or implied by the fact that the foreign exchange intervention will eventually be sterilized if the policy rule is not changed, as with the Bank of Japan. Why more easing?
In response to this question, BCA Research (article not available) presented a version of the quantity theory of money. They looked at the simple linear relationship between the average rate of money supply growth (M2) and nominal GDP growth (P*Y).
The chart is a reproduction of that in the BCA paper, but with a sample back to 1959 (they went back to the 1920's when M2 was not measured). The relationship illustrates the 5-yr compounded annual growth rate of money (M2) against that of nominal GDP, and has an R2 equal to 50% - okay, but not perfect.
Nevertheless, the implication is pretty simple: the current annual growth rate of M2, 2.8% in August 2010, corresponds to an average annual income growth just shy of 4%. Sitting beneath a behemoth pile of debt relative to income, 4% nominal GDP growth is unlikely provide sufficient nominal gains for households to deleverage quickly or "safely".
However, notice the 2000-2005 and 2005-2009 points, where the relationship between M2 and nominal GDP growth deviated away from the average "quantity theory" relationship. Would a broader measure of money account for the weak(ish) relationship in the chart above? Yes, partially. (Note: the relationship almost fully breaks down at an annual frequency.)
These days it's all about credit. I'm sitting in Cosi right now - bought a sandwich and charged the bill on my credit card. Actually, I prefer to use cards. But M2 doesn't account for this transaction as money if the balance is never paid in full. M2 is essentially currency, checking deposits, saving and small-denomination time deposits, and readily available retail money-market funds (see Federal Reserve release).
One can argue about the merits of including credit cards balances as "money", per se. However, the sharp reversal of revolving consumer credit, and likely through default (see the still growing chargeoff rates for credit card loans), would never be captured in M2. The hangover from the last decade of households using their homes as ATM's (i.e., home equity withdrawal) and running up credit card balances to serve as a medium of exchange is dragging nominal income growth via a sharp drop in aggregate demand.
The Federal Reserve discontinued its release of M3 in 2006, which among other things included bank repurchase agreements (repos). Including M3, rather than M2, in the estimation improves the the R2 over 30 percentage points (to 81%).
This is a very small sample, and removing the latest data point from the original estimation improves the R2 slightly to 64%; but clearly there's something going on here. I think that it's fair to say that we may be disappointed by the M2 implied average nominal GDP growth rate over the next 5 years (4%).
According to John Williams' Shadow Statistics website, M3 is still contracting at (roughly because I don't subscribe to the data) 4% over the year. The relationship in the second chart implies that nominal GDP will fall, on average, about 4.5% per year. Japan's nominal GDP never contracted more than 2.08% annually during its lost decade, but the implication is that "things" may not be as rosy as the M2 measure of money suggests.