High-powered money (100 years of stability)
By Scott Sumner
When I studied economics, the term ‘high-powered money’ was often used synonymously with the monetary base, which consists of currency plus bank deposits at the Fed. This asset was called “high-powered” for three reasons:
1. It is determined exogenously by the monetary authority.
2. It is non-interest-bearing.
3. It is the medium of account.
As a result of these three factors, high-powered money is a sort of “hot potato”. When the Fed injects new high-powered money into the economy for reasons other than responding to an increased demand for liquidity, the public tries to get rid of excess cash balances by spending them. Prices and NGDP rise until the public is again content to hold the newly enlarged supply of high-powered money.
Today, bank deposits at the Fed earn interest, and thus are no longer high-powered money. Only currency remains high-powered.
The stock of high-powered money has risen by roughly 265 times over the past 100 years, from about $6.45 billion in early 1919 to just over $1.7 trillion today. NGDP has risen by almost the same proportion, leaving the high-powered money to NGDP ratio at roughly 8.2%, even after 100 years! People seem to want to hold about the same fraction of income in the form of high-powered money as they did 100 years a go. As a result, you might say that the Fed caused NGDP to grow 265-fold by increasing the stock of high-powered money 265-fold.
Actually, it’s a bit more complicated than that. The demand for high-powered money had fallen to 5.6% of GDP right before the Great Recession, and the fortuitous coincidence of almost no change in the ratio over 100 years is actually due to two offsetting factors. Technological progress reduced the demand for high-powered money over time, and very low interest rates plus increased foreign demand for US currency recently boosted the demand for high-powered money.
Even so, the Fed determines the long run path of NGDP via adjustments in the stock of high-powered money. (RGDP is up about 16.7 fold over the past 100 years, the rest is inflation.)
I’m hesitant to talk about MMT theory, because its proponents always insist that critics get it wrong. But unless I’m mistaken, MMTers seem to assume that money creation is an important source of funding for federal spending. George Selgin has a post criticizing MMT, and cites this passage from Stephanie Kelton, Andres Bernal, and Greg Carlock, who are MMT proponents:
As a monopoly supplier of U.S. currency with full financial sovereignty, the federal government is not like a household or even a business. When Congress authorizes spending, it sets off a sequence of actions. Federal agencies, such as the Department of Defense or Department of Energy, enter into contracts and begin spending. As the checks go out, the government’s bank ― the Federal Reserve ― clears the payments by crediting the seller’s bank account with digital dollars. In other words, Congress can pass any budget it chooses, and our government already pays for everything by creating new money.
This is very misleading. Nick Rowe has a post where he points out that money creation provides very little revenue for advanced economies, citing a rough estimate of 0.25% of GDP for Canada. If we assume 4% trend growth in NGDP in the US, then an 8% currency ratio will lead to steady-state revenue of roughly 0.32% of GDP, ignoring costs of producing money. That’s trivial compared to total federal spending, which is over 20% of GDP. Studies of the “Laffer Curve” for seignorage tend to produce maximum inflation tax revenue estimates on the order of a few percent of GDP, assuming a revenue maximizing inflation rate of several hundred percent per year.
Obviously those sorts of inflation rates are politically infeasible in the US, and hence as a first approximation it’s best to assume that money creation is not a significant source of funds to pay for government spending. Programs need to be funded with either taxes or debt, and of course debt represents future tax liabilities. The only possible exception is if the interest rate on public debt will stay persistently below the economic growth rate (as has been the case in recent years), in which case a government could earn a one-time revenue windfall by supplying the public with the extra Treasury debt it seems to crave. But it would be foolish to rely on that one-time gain to pay for expensive permanent programs such as “Medicare for all”.
In any case, whatever one’s views on the “dynamic inefficiency” argument for more federal debt, it’s misleading to claim that money creation provides a significant source of funds for government spending. Perhaps the confusion comes from the huge QE programs of the past decade, which looked to many people like a case of “monetizing the debt”. While bank reserves used to be high-powered money, today they are not (except for vault cash, of course.) In fact, QE largely exchanged one form of federal debt (T-bonds) for another (bank reserves.) QE does not pay for a significant fraction of government spending; for the most part it simply changes the maturity of the public debt.
PS. The Nick Rowe post I linked to is the best explanation of MMT that I have encountered. I also recommend the George Selgin post linked to above, as well as Paul Krugman’s critique:
[T]here are limits to the amount of real resources that you can extract through seigniorage. When people expect inflation, they become reluctant to hold cash, which drive prices up and means that the government has to print more money to extract a given amount of real resources, which means higher inflation, etc.. Do the math, and it becomes clear that any attempt to extract too much from seigniorage — more than a few percent of GDP, probably — leads to an infinite upward spiral in inflation. In effect, the currency is destroyed. This would not happen, even with the same deficit, if the government can still sell bonds.
The point is that under normal, non-liquidity-trap conditions, the direct effects of the deficit on aggregate demand are by no means the whole story; it matters whether the government can issue bonds or has to rely on the printing press. And while it may literally be true that a government with its own currency can’t go bankrupt, it can destroy that currency if it loses fiscal credibility.
Now, I am not predicting hyperinflation for the US . . . But the MMT people are just wrong in believing that the only question you need to ask about the budget deficit is whether it supplies the right amount of aggregate demand; financeability matters too, even with fiat money.
PPS. My early NGDP data comes from Balke and Gordon’s data set (actually GNP). My high-powered money data is from FRED, and from Friedman and Schwartz.
HT: Pat Horan