Arnold Kling  

Scott Sumner vs. Milton Friedman

Life Expectancy Statistics... Masonomics Watch...

Tyler Cowen writes,

Professor Sumner's views differ from the monetarism of Milton Friedman by emphasizing expectations rather than any particular measure of the money supply.

Allow me to expand on the history of monetary thought embedded in that statement.

In Capitalism and Freedom, Milton Friedman wrote,

The rule that has most frequently been a price level rule; namely, a legislative directive to the monetary authorities that they maintain a stable price level. I think this is the wrong kind of a rule because it is in terms of objectives that the monetary authorities do not have the clear and direct power to achieve by their own actions.

Friedman's view was that changes in monetary growth rates affect nominal GDP with a "long and variable lag." As a result, if the Fed tries to target nominal GDP (or a similar target, such as the price level), it will proceed to overshoot and undershoot, in the words of James Tobin, "like an amateur shower tuner."

This brings up the topic that Benjamin Friedman (no relation) referred to as that of targets, instruments, and indicators. Nominal GDP or the price level are potential targets of monetary policy. However, the Fed only has certain instruments--open market operations, the discount rate, and reserve requirements being the traditional three. Finally, the Fed can use indicators, which could be any economic data, to decide how well it is doing at hitting the target.

To change from Tobin's metaphor, consider a hockey metaphor. Businessmen are fond of citing Wayne Gretsky to the effect that a great hockey player does not skate to where the puck is now but instead skates to where it will be soon. A hockey player who does nothing but skate in the direction of the puck will simply waste energy skating all over while the puck gets passed around the rink. Similarly, in Friedman's view, if the Fed tries to hit a price target, the economic environment will keep changing so rapidly that the Fed may wind up destabilizing. Right now, for example, you might say that the economy is weak and we need more money growth. However, if today's money growth affects the economy two or three years from now, by then the problem may be excess inflation.

In Sumner's model of the economy, monetary policy takes effect more quickly because monetary growth targets affect expectations. I do not buy that model, for reasons I have occasionally suggested, although I probably owe people a fuller explanation. Not today.

What I find most troubling right now is that fiscal policy is subject to long and variable lags. I am willing to bet that the majority of the fiscal stimulus enacted earlier this year will actually hit the economy after positive economic growth has been restored. Furthermore, the multiplier effects of fiscal policy tend to hit with a lag, so we will be getting even more stimulus late in the game.

It is true that employment growth will be sluggish (if my guess is right), so it might not seem as though stimulus in 2011 is so bad, but the sluggish employment growth will reflect structural problems (mismatch between growing sectors and the employee skill base), not cyclical problems. The bottom line is that I think that in 2011 and 2012 we may be experiencing increased inflation while unemployment is high. This would be true whether we tried using monetary policy to stimulate the economy or using the unfortunate delayed-action fiscal policy to stimulate the economy.

Incidentally, in the chapter on fiscal policy in Capitalism and Freedom, written in 1963, Friedman bemoans the fact that the push for fiscal stimulus always leads to permanently higher levels of government spending. It reads as if it were written yesterday.

Comments and Sharing

COMMENTS (9 to date)
E. Barandiaran writes:

In addition to the "technical" problems associated with lags, you have the fundamental problems associated with the credibility of any authority (rules never eliminate all discretion). So let me copy my comment on Tyler's post:

If you knew the world history of central banking, you'd agree that any central bank is as credible as this guy

pete writes:

I would like to know more in depth why you don't buy Sumner's expectations view of the economy.

Niccolo writes:


I agree with your analysis about the delayed effects and general inaccuracy of monetary policy. But my question is, what do you think can be done to overcome this? Would adding market forces be enough? How would market forces change the dynamics of money stability?

It is a question I have often wondered about, and I'd be appreciative of any insights to possible solutions.

Richard A. writes:

The Friedman quote you came up with refers to the P in GDPn=PxGDPr. Changes in GDPn act faster than changes in P in response to changes in M. What Friedman favored was a smooth growing M. What Sumner favors is a smooth growing GDPn.

M x V = GDPn
In the above equation, I suppose you could call Friedman's approach open-loop monetarism.

By determining M based upon what GDPn does, you've got a feedback loop. I suppose you could call this closed-loop monetarism.

Scott Sumner writes:

Arnold, Nearly 30 years later there had been a lot of progress in forward-looking monetary rules, and Milton Friedman came around to my way of looking at things. Check out this post:

Don't worry, you don't have to read the whole thing!

Here is a key quotation:

“..Hetzel has suggested…that the Federal Reserve be instructed by the Congress to keep the difference between a nominal and an indexed bond yield below some number. [This] is the first nominal anchor that has been suggested that seems to me to have real advantages over the nominal money supply. Clearly it is far better than a price level anchor which is always backward looking.”

Bill Woolsey writes:


You wrote:

"In Sumner's model of the economy, monetary policy takes effect more quickly because monetary growth targets affect expectations."

There is no monetary growth target in Sumner's approach, so they cannot take effect more quickly because they affect expectations.

Sumner's proposed instrument is open market operations (though not limited to T-bills.)

Sumner's proposed target is expected nominal GDP one year in the future.

Sumner's proposed indicator is an index futures contract on nominal GDP one year in the future. Rather than watching the price of those contracts and using a feedback rule for open market operations, the Fed is to trade the contracts at the target price and undertake ordinary open market operations based upon its net position on the contract.

Because the Fed isn't trading these contracts now, Sumner spends a good bit of time discussing what he considers inferior indicators like the inflation expectations shown by the TIPs market as well as the performance of stock markets and organized commodity exchanges.

The issue is always how are open market operations today impacting expected nominal income in the future.

So, is there any truth to your interpretation? Yes.

Open market operations impact the monetary base. Sumner believes that the monetary base today impacts expected nominal GDP on year from now.

So, it isn't exactly that a target for monetary growth is taking effect more quickly because the target effects expectations. The change in base money resulting from the open market operations are impacting expected nominal income one year in the future.

The other element of Sumner's thought is that changes in expected nominal income one year from today will impact nominal income between now and one year from now.

_There is no feedback rule from past values of nominal GDP to the current growth rate or level of the monetary base._ Further, the size of changes in base money are not determined by historical relationships between changes in base money and nominal GDP.

RD writes:

So then you agree that fiscal policy will in fact stimulate the economy, and even have multiplier effects?

fundamentalist writes:

I don't think Sumner's plan will work. Regardless of expectations, econometric analyses of actual changes in the money supply and actual changes in ngdp show a lag of about 4.5 years to max effect.

Nevertheless, it would be nice to have the feds try Sumner's theory. It certainly couldn't be worse than what the feds have been doing for the past two decades.

Expectations aren't everything, but they're pretty important. When it looked like sub-prime would remain largely contained, the economy was doing ok ... and commodity prices were even rising in early/mid 2008. When things "suddenly" fell apart in October, surely GDP itself didn't fall as fast as the stock market. But people's expectations fell quickly ... and then they acted on those expectations, which in turn reduced GDP.

If the Fed met the increased demand for money -- and was totally committed to maintaining NGDP, then individuals (and, in aggregate, markets) would be surrounded by conditions that tended to persuade them to continue with the activities that are a part of GDP.

Final note: everyone should read Sumner's critique of long and variable lags. May or may not be correct, but worth pondering.

Comments for this entry have been closed
Return to top