It’s useful to think about the precise difference between the “New Keynesian” (Woodford) and “New Monetarist” (Williamson) effects of monetary policy. In New Keynesian models, monetary policy is important primarily because it affects the real interest rate, shaping patterns of consumption and investment across time. If interest rates are high, it’s expensive to buy a car or build a factory, and even ordinary consumption becomes less attractive compared to the return from saving. If interest rates are too high, consumers spend less than the economy can produce, and we see a recession. A good example of this type of model is in Eggertsson and Woodford’s classic piece on the liquidity trap.
In New Monetarist models, monetary policy is important for a completely different reason. Certain kinds of “decentralized” consumption require you to hold money balances; by manipulating the relative rate of return on money, the Fed makes this consumption more or less attractive. When money is expensive, buyers consume less in “decentralized” markets, and we see a downturn. A good example of this type of model is in Williamson’s forthcoming AER article, which builds upon the canonical piece by Lagos and Wright.
It’s silly to deny the existence of either effect: in a qualitative sense, they both hold. But it’s also important to know the magnitudes, and that’s where my skepticism comes in.
Consider the traditional instrument of monetary policy: the federal funds rate. Suppose that the Fed announces a surprise rate hike of 1%, to last exactly one year. What happens?
First of all, even though the Fed only controls the nominal rate, we can expect to see a real rate increase of at least 1% as well. Why? Since the rate increase will last only one year, the Fed’s long-term policy rule remains the same, and long-term inflation expectations stay anchored at roughly the same level. The only way that we can get a real rate increase of less than 1%, then, is if there’s a temporary uptick in inflation—and in virtually any model, that’s associated with an expansion, not a contraction. Certainly this won’t result from an increase in the policy rate. At best, inflation will stay roughly constant; at worst, it’ll fall.
A one year real rate increase of 1% means that all consumption or investment today becomes 1% more expensive, relative to consumption or investment a year from now (which is pinned down by the same fundamentals that existed before the shock). The effect won’t be even across the economy: durable goods will fall much more than, say, food. But to summarize the situation, we can say that all $15 trillion of GDP become 1% more expensive relative to next year. This is the New Keynesian effect.
The New Monetarist effect, on the other hand, is driven by the nominal interest rate. A 1% increase in the federal funds rate pushes down the annual yield of non interest-paying base money, relative to other forms of liquidity, by 1%. Right now, paper currency is the only type of base money that doesn’t pay interest, and there’s a total of $1 trillion in circulation. Taking the midpoint of various estimates, let’s say that roughly half of that is held in the United States. Furthermore, of that $500 billion, let’s be exceedingly generous and say that all of it is being used to facilitate legitimate economic activity. Then a 1% increase in the federal funds rate increases the implicit cost of holding this money by $5 billion. This is the New Monetarist effect.
You don’t need to be an economist to see that in our calculations, the New Keynesian effect is vastly larger than the New Monetarist effect: ($15 trillion)*1% = $150 billion versus ($500 billion)*1% = $5 billion. That’s a factor of 30!
Granted, these back-of-the-envelope calculations don’t explicitly tell us what the economic effect will be. That’s a much more complicated calculation, and it depends on the precise mechanics of the model. But they certainly are suggestive, and unless the New Monetarists have some trick up their sleeves, it’s awfully hard to see how the New Monetarist effect can be nearly as important as the New Keynesian effect at business-cycle frequencies.
In fact, many obvious modifications of the model only widen the spread between New Keynesian and New Monetarist effects. Some sectors of the economy are much more responsive to costs (including interest rates) than others. When the Fed raises rates, we expect a large decline in the demand for cars, but a much smaller change (if any) in the demand for food at home. That’s because cars are durable goods that provide value over time—their exact date of production isn’t so important, and it can easily be shifted around in response to the cost of capital. Consumers’ ability to “intertemporally substitute” basic sustenance, on the other hand, is virtually nil.
But where is New Monetarist effect least relevant? In precisely the cases where spending is most flexible: fixed investment, large durable goods purchases, and other transactions that are virtually never made with cash. Our comparison above, therefore, actually exaggerates the extent to which the New Monetarist effect makes a difference.
The gap also becomes wider if we alter our thought experiment. Let’s say that the 1% rate hike lasts for two years rather than one. Then the New Keynesian effect becomes almost twice as large: all else equal, the economy two years from now remains pinned down by fundamentals, and consumption and investment today is now 2% more expensive relative to then. The New Monetarist effect, on the other hand, remains the same: it depends on the current cost of holding money, not the full path of interest rates.
Of course, a few caveats are in order. My analysis above is concerned with the Fed’s impact on the business cycle: the short term, not the long term. As we stretch our time horizon, the Fed’s ability to impact the economy via the New Keynesian effect diminishes, while the New Monetarist effect remains roughly the same. If we’re talking about the federal funds rate, I still don’t think that the New Monetarist effect is very important, for essentially the same reasons that I’ve dismissed the Friedman rule. Nevertheless, the comparison in this post doesn’t apply.
More importantly, I’ve limited the analysis in this post to conventional monetary policy—changes in the federal funds rate. As we’ve seen over the past several years, this is not the only form of monetary policy. Moreover, a great deal of liquidity creation takes place outside the Fed: at banks, or the Treasury. Modelling this activity is potentially very important, and I think that the New Monetarist program is quite promising when viewed in the right light.
But when it’s applied to changes in the central bank’s interest rate—which is still the channel for monetary policy in the vast majority of countries, the vast majority of the time—New Monetarism simply doesn’t offer much. As even the most casual back-of-the-envelope calculation will convincingly demonstrate, the New Monetarist effect is tiny compared to the New Keynesian effect, and there’s no reason to think that it’s essential for understanding the implications of monetary policy.