Jeff Hummel on David Andolfatto
By David Henderson
David Andolfatto, an economist and Vice-President of the St. Louis Fed (and apparently, going by his CV, a fellow Canadian) wrote last year:
The fact that bonds become close substitutes for money when their yields are similar explains how the supply and demand for bonds can influence the inflation rate. Normally, we think of an increase in the demand for bonds as lowering bond yields. This is correct. But what happens when those yields approach the corresponding yield on interest-bearing money? (In the old days, when interest on reserves was zero, this limit was called the zero-lower-bound). An increase in the demand for bonds in this case must manifest itself in other ways. One way is for the price-level to fall. That is, a market-mechanism for expanding the real supply of nominal bonds is for the price-level to fall. One way this manifests itself is as China selling its goods for less [sic] USDs [U.S. dollars] to acquire the USTs [U.S. Treasuries] it so desperately wants. (italics in original)
This is from David Andolfatto, “A monetary-fiscal theory of inflation,” August 6, 2017.
A macroeconomics professor whom I taught as a Masters student recently asked San Jose State University economics professor Jeff Hummel, to comment. Jeff shared his answer with me and I thought it was so good that it should receive more attention. Here’s Jeff’s answer, with only slight editing by me:
1. I agree entirely with Andolfatto’s critique of the New Keynesian ass-backwards approach to the Phillips Curve. You can see the germ of this New Keynesian view in Mankiw’s macro text, where he presents the following formula for the Phillips Curve: inflation = expected inflation – Beta (observed unemployment – natural rate of unemployment) + supply shock. While formally correct as far as it goes, it implies that inflation is the dependent variable, which reverses the causation. When Friedman introduced the natural-rate hypothesis, inflation was an exogenous variable (determined by monetary policy), with the difference between actual and expected inflation causing cyclical unemployment. New Keynesians now think in terms of cyclical unemployment causing inflation, which of course indirectly and surreptitiously resurrects the old fallacy of cost-push, sustained inflation.
2. Andolfatto and I also agree about the similarity between interest-paying reserves and low-yield Treasuries. But we look at the similarity in opposite ways. I think of interest-paying reserves as having ceased to become genuine outside money and instead becoming another form of government debt. He thinks of low-yield Treasuries as becoming part of the broader money supply. Both approaches offer complementary explanations of why Quantitative Easing (QE) and its large-scale asset purchases (LASP) failed to generate any significant inflation. My approach suggests QE didn’t involve any significant increase in the true monetary base. Andolfatto’s approach suggests QE simply involved the Fed swapping one form of money for another, without any significant increase in some broad measure of the money stock. (Btw, notice the parallels between his approach and what the Bank of Japan is now doing, as discussed in my comment on the link Fred sent me.)
3. The most intriguing and trickiest part of Andolfatto’s argument is the part you quoted. The clearest explanation of his reasoning is his response to JP Koning in the comments: “When the yield on bonds falls to [the] yield on money, they are essentially perfect substitutes. What is relevant for the price level now is the total money supply, not its composition between money and bonds. Hence, an increase in the demand for bonds is like an increase in the demand for money, and is disinflationary.” In other words, he is treating low-yield Treasuries as part of the money stock and then correctly stating that an increase in money demand, ceteris paribus, pushes the price level down. But I find this a misleading way to analyze what is going on.
4. His conclusion that an increased demand for Treasuries can affect the price level has some similarity to Cochrane’s Fiscal Theory of the Price Level (FTPL). Cochrane, however, includes all government debt at any yield in his analysis and derives its impact on the price level through expectations about future government primary surpluses. [DRH note: a government achieves a primary budget surplus by having government revenues exceed government spending that doesn’t include interest on the government debt.] If future primary surpluses are expected to decline, this will cause the price level to rise (reducing the real value of the debt), and the reverse if future primary surpluses are expected to increase. This feature of the FTPL does not seem to play a role in Andolfatto’s reasoning.
5. Instead, Andolfatto seems to be implicitly assuming some kind of significant market segmentation between Treasuries and other debt securities. How otherwise could the demand for Treasuries be a unique factor holding inflation down, given that a decrease in the price level raises the real value of all debt securities, private and public? After all, critical to his argument is what precisely drives the increased demand for Treasures. Obviously a shift in demand from other forms of money to low-yield Treasuries won’t do, given his assumption they are perfect substitutes. As he clearly states, what he has in mind is “a portfolio substitution effect where savers redirect resources away from private capital spending (including expenditure on recruiting activities) toward money and bonds.”
6. But such a “portfolio substitution” should show up as a significant gap between Treasury yields and yields on private debt. But the empirical evidence fails to support such segmentation. As Cochrane points out in his “Eight Heresies of Monetary Policy“, the risk premium spreads are currently not that high. They may be slightly higher than in other recovery periods, but nowhere near as high as during the financial crisis. In short, Andolfatto is mis-identifying as “portfolio substitution” what in reality are general factors, whatever they may be, responsible for the now widely accepted decline in the natural (or neutral) real rate of interest.
7. I heartily applaud Andolfatto’s effort to create an open-economy model that includes a “growing foreign appetite for U.S. money/bonds over the past decade.” As you know, this is a major factor that I think has been overlooked by others. I agree that it is still contributing to low interest rates worldwide. But his analysis confuses an encompassing, global change in the demand for and supply of savings with a mere portfolio substitution.
8. Andolfatto’s also exhibits an exaggerated view of the Fed’s impact on market interest rates. Yes, if the Fed raises the interest rate on reserves, this will lower its remittances to the Treasury. But this by itself will not significantly affect market rates. And if market rates aren’t independently rising, a rise in the interest rate on reserves should cause banks to raise their reserve ratios even higher, tightening on broader monetary growth still further. Any resulting increase in government debt is not likely to offset this disinflationary pressure, justifying the neo-Fisherian conclusion, unless admittedly the growing government debt leads people to expect the Fed to reverse course. Yet what is going on in Japan with its huge government debt makes such an impact on expectations appear unlikely.