E-piphany

Before getting into today’s column, let me first describe my plan of attack for the month of December. I plan to have several comments this week and next week, culminating in my annual “Portfolio Projections” piece at the end of next week. Then, for the last two weeks of the month, I plan to ‘re-blog’ some of my best articles from the last four years (editing out the current events, which will no longer be topical of course). Included in that list is an article on long-run returns to equities, one on Yellen’s defense of large-scale asset purchases, an article on the Phillips Curve, one on why CPI isn’t a bogus construct of a vast governmental conspiracy, and so on. Because I don’t expect some of the places where this column is ‘syndicated’ to post the re-blogs, you should consider going to the source site to sign up for these posts.

With that housekeeping complete, I want to turn today to a scholarly article I recently stumbled on which is worth a read even once you have read my synopsis and comments. The article, written one year ago by Samuel Reynard of the Swiss National Bank, is entitled “Assessing Potential Inflation Consequences of QE after Financial Crises.” It appears to be unpublished except as a working paper, which perhaps shouldn’t be surprising since it is so decidedly clear-eyed and takes the consensus view of QE to task.

What I love about this article is that Reynard’s view is remarkably consonant with my own – the only example I can come up with of a reasonably-placed central banker espousing such commonsensical views (Daniel Thornton at the St. Louis Fed gets an honorable mention though), backed with quantitative data and clear reasoning. Here is the paper’s abstract:

“Financial crises have been followed by different inflation paths which are related to monetary policy and money creation by the banking sector during those crises. Accounting for equilibrium changes and non-linearity issues, the empirical relationship between money and subsequent inflation developments has remained stable and similar in crisis and normal times. This analysis can explain why the financial crisis in Argentina in the early 2000s was followed by increasing inflation, whereas Japan experienced deflation in the 1990s and 2000s despite quantitative easing. Current quantitative easing policies should lead to increasing and persistent inflation over the next years.”

In the introduction, the author directly tackles current central bank orthodoxy: “It is usually argued that it is sufficient to monitor inflation expectations, and that central banks can avoid accelerating inflation by quickly withdrawing reserves (or by increasing the interest rate payed on reserves) once inflation expectations start rising. The monetary analysis of this paper however shows that there has never been a situation of excess broad money (created by the banking system) which has not been followed by increasing inflation, and that the increase in inflation occurs after several years lags.”

Reynard starts with the quantity theory of money (MV≡PQ), which I have discussed at length in this column. Regular readers will know that I am careful to distinguish transactional money from base money – as does Reynard – and that the sole reason inflation has not accelerated is that money velocity has declined. This decline is not due to the financial crisis directly, but as I have shown before it is due to the decline in interest rates. This makes monetary policy problematic, since an increase in interest rates which in ordinary times (that is, when there isn’t a couple trillion of excess reserves) would cause M2 to decelerate and dampen inflation will also cause money velocity to rise – offsetting to some extent the effects of the rising interest rates on the money supply. (Among other things, this effect tends to help cause monetary policy to overshoot on both sides). Reynard’s insightful way around this problem is to “model equilibrium velocity as a function of interest rate to reflect changes in inflation environments.” That is, the monetary equation substitutes an interest rate variable, based on a long-run equilibrium relationship with velocity, for velocity itself. In Reynard’s words,

“Thus the observed money level is adjusted…by the interest rate times the estimated semi-elasticity of money demand to account for the fact that, for example in a long-lasting disinflationary environment when inflation and interest rate decrease, the corresponding increase in money demand reflecting the decline in opportunity cost is not inflationary: the price level does not increase with the money level given that equilibrium velocity decreases.”

This is exactly right, and it is exceedingly rare that a central banker has that sort of insight – which is one of the reasons we are in this mess with no obvious way out. Reynard then uses his model to examine several historical cases of post-crisis monetary and inflationary history: Switzerland, Japan, Argentina and the 1930s U.S. He finds that there are downward rigidities to the price level that cause inflation to resist turning negative (or to fall below about 1.5% in the U.S.), but that when there is excess liquidity the link between liquidity and inflation is very tight with a lag of a couple of years. Reynard’s opinion is that it is this non-linearity around price stability that has caused prior studies to conclude there is no important link between money and inflation. As Fama observed back in the early 1980s, and I observe pretty much daily to the point that it is now a prohibited topic at the dinner table, when inflation is very low there is a lot of noise in the money-inflation relationship that makes it difficult to find the signal. But the money-inflation connection at higher levels of inflation and money, and over longer periods of time, is irrefutable.

In the last section of the paper, the author assesses the effects of current QE (through November 2012) on future inflation in the U.S. His conclusion is that “Excess liquidity has always been followed by persistent increases in inflation. Current quantitative easing policies should lead to increasing and persistent inflation over the next years.” The chart accompanying this statement is reproduced below.

As you can see, the model suggests inflation of 3-4% in 2013 and 5% in late 2014. While clearly inflation in 2013 has been lower than suggested by the chart, this isn’t supposed to be a trading model. I suspect that if get 3-4% in 2014 and 5%+ in 2015 (our forecast is for 3.0%-3.6% on core inflation in 2014 and 3.3%-4.8% in 2015), the issue of whether Reynard was essentially correct will not be in question!