A critique of Andolfatto and Spewak's Debt Monetization: Then and Now

Debt Monetization: Then and Now by David Andolfatto and Andrew Spewak at On the Economy. Andolfatto and Spewak (A&S) show this graph of the monetization rate:

Graph #1, from Andolfatto and Spewak

Their graph shows "the percent of [Federal] debt held by the Federal Reserve—the monetization rate—against personal consumption expenditures (PCE) inflation." In a note they add:

We excluded intragovernmental holdings from our definition of debt. However, the conclusions of our analysis would not change significantly if we included these holdings.

To start, I want to duplicate the graph. From the note, I know they are looking at debt "held by the public" and not the "gross" Federal debt. So I know where to begin.

The Debt Held By the Federal Reserve, there are two series for that: an older one and a newer one. I need both. A search of FRED for federal debt held by the public turned up only one series that goes back as far as 1953. But it is called "Gross Federal Debt Held by the Public" which seems a contradiction in terms. To me, at least, "gross" means all of it, and "held by the public" means only part of it. Anyway, that series is annual. The A&S graph is quarterly. So I knew my first attempt would not look right:

Graph #2: Old (blue) and New (red) Federal Reserve Holdings as a Percent of the Federal Debt
(You'd have to ask the Fed how they got "Holdings" wrong and why they had to change it.)

The annual data is too smooth. Plus, my numbers are too high. I peak above 22 .5% in the mid-1970s. They peak near 20%. So I switched out the denominator for the blue line. I went with the Federal government component of TCMDO instead:

Graph #3

The numbers came down some, and are quarterly. Not sure why the two measures of debt are so different. But now my blue line looks like what A&S show. So I changed the red line also, and added PCE Inflation for the hell of it:

Graph #4: Yeah, that looks like what they show

I used the first "PCE Inflation" I found that had quarterly data going back at least to 1953.

Then, since Adolfatto and Spewak say

From 1953 to 1974, the monetization rate increased hand-in-hand with inflation, with both peaking near the end of 1974.

I put the 1953-1974 data into the Data Analysis "Regression" form in Excel. First time I used that. I got an R square value of 0.60. So I'm thinkin' that point six oh (or more) means "hand-in-hand".


I'm dropping inflation from my graph. I just want to look at Federal Reserve holdings of Federal debt. I want to compare the A&S version, Fed Holdings as a percent of the Federal Debt, to my version, Fed Holdings as a percent of the rest of the debt: The "non-Federal" debt, I call it. The part of TCMDO that A&S ignore.

Graph #5: Fed Holdings of Federal Debt as a Percent of the Federal Debt (blue & red)
and as a Percent of Debt Other than Federal (green & purple)

Relative to the Federal debt, Fed holdings are high and variable. Relative to debt other than Federal, Fed holdings are low, and were falling until the crisis.

Just before the crisis, Fed holdings relative to Federal debt were near the middle of their historical range. That may have been high enough to cause inflation (A&S don't say) but it is not likely that Fed holdings in that range would have caused the crisis.

By way of contrast, just before the crisis Fed holdings relative to non-Federal debt were the lowest they had ever been. This could have caused the crisis. I'm not saying it did; not today, anyway. But I am saying the "relative to non-Federal debt" measure is at least as important as the "relative to Federal debt" measure.

And that means non-Federal debt is at least as important as the Federal debt when we're talking about the economy. I hope you'll keep that in mind when you are looking at debt, and when you see other people looking at debt.


Andolfatto and Spewak offer an explanation of inflation. In particular, they offer an explanation for why the 1953-1974 increase in Fed holdings of Federal debt was associated with inflation but the more rapid increase in the 2009-2017 period was not.

One can look at their graph and ponder a similar question about the increase of Fed holdings in the 1992-2003 period. As I indicated above, though, that increase is not discussed in their article.

The explanation of inflation that they offer has to do with banks' incentive to lend. A&S measure this incentive as a difference of interest rates: the interest rate on a one-year T-bill minus the interest rate paid on reserves.

Here they explain the inflation of the 1953-1974 period:

As the Fed ramped up its debt monetization during this span, the average rate of interest on a one-year Treasury bill was about 5 percent, and the interest rate paid on reserves was literally zero.

With the spread between interest rates so large, banks had more incentive to use their newfound reserves to make loans than to hold onto them. These loans would then create money, which would boost the money supply and have inflationary effects. While several factors led to the Great Inflation of the 1970s, the gradual rise in the monetization rate was likely a significant one.

A&S compare two periods and find greater incentive to lend in the earlier period because the spread was greater. Because of this greater incentive, then, they see the "rise in the monetization rate" as a significant contributor to the Great Inflation of the 1970s.

It makes good sense to say that the greater "spread" increased banks' incentive to lend. But I think the low level of private debt and the economy's "golden age" vigor in the 1953-1974 period also increased banks' incentive to lend. And the high level of private debt and the residual trauma from "financial crisis" in the 2009-2017 period reduced banks' willingness to lend. Andolfatto and Spewak mention neither the vigor nor the lingering effects of the financial crisis. The vigor of one period, the trauma of the other, for A&S these somehow fall under ceteris paribus -- "all else equal".

Oddly, too, they emphasize the increase in monetization of debt in the early period, but de-emphasize the rise of interest rates that occurred concurrently. I expected Andolfatto and Spewak to give more weight to interest rates -- particularly in this case, as they note that "The larger the spread, the more profitable it is for a bank to lend". It is not the rise in monetization but the rising level of interest rates which supports their "greater incentive" argument.

During this time, however, with the interest rate on one-year Treasurys rising from one or two percent (in 1953) to eight or nine percent (in 1974), A&S refer only to "the average rate of interest on a one-year Treasury bill".

Emphasis added.


Let me regroup. Andolfatto and Spewak say that the increasing monetization rate, which increased reserves and the lending capacity of banks, was more likely to lead to inflation at a time when the incentive to lend (measured as an interest rate spread) was high. Lending was more likely to happen at the higher constant ("average") rate spread, as it would be more profitable.

Sure. But the rate of interest was not constant. It was rising. The interest rate spread was increasing. If the spread was increasing, surely the anticipated increase in bank profits would be more than if the spread was constant. Surely, the increase of interest rates was as significant as the increase in the monetization rate. Maybe more significant: Monetization provided an opportunity, but it was interest rates that provided the incentive.

Why then do A&S make the rising-monetization-rate argument, but smother the rising-interest-rate argument? I can't say. But even though they emphasize the monetization, their argument is that the interest rate spread led to the lending that led to the Great Inflation.

A&S argue that higher interest rates caused inflation by increasing the incentive to lend. I like this argument a lot; it fits my way of thinking. But somebody should point out that monetary policy uses higher interest rates as a way to fight inflation.


Andolfatto and Spewak argue that the increase in Federal Reserve holdings of Federal debt led to inflation in the 1950s and '60s and '70s but not after 2008 because the interest rate spread was different in the two periods. In the earlier period the spread was large, and in the latter it was small. In the earlier period the incentive to lend was large; in the latter it was small.

A&S consider the supply side of bank lending, but ignore the demand side. They consider the incentive to lend, but not the incentive to borrow. They present half a picture.

They ignore borrowers' interests. In essence they are saying that banks' incentive to lend was great enough in the early period that it caused the lending to happen. As if the borrowers were left saying to themselves: Well dammit, we borrowed all this money. Now what are we gonna do?