Saturday, August 27, 2011

The Economist on GDP targeting

The Economist has a short article on the possibility of shifting to a target for nominal GDP. (Really, that is redundant, because GDP is nominal, and the estimate of real output and income is real GDP.) They even mentioned Scott Sumner!

I think they do a good job of describing the advantages of targeting nominal GDP. Then they discuss the disadvantages--

For all its theoretical merits, a switch to NGDP targeting would throw up some new problems—and old ones. The Fed has not exactly sat on its hands since the financial crisis began in 2007, so it is far from clear it could easily reach the new goal. Moreover the theory of how inflation targeting works is well-established: by anchoring the expectations of businesses and households, it encourages them to set wages and prices that will comply with the target. That anchoring can allow central banks to look beyond temporary surges in inflation such as those driven by commodity prices. One worry about NGDP targeting is that it would be harder to pin down people’s expectations, not least since nominal GDP may be a harder concept to grasp and is measured less often.

The biggest risk may be the loss of credibility. Whatever its shortcomings, inflation targeting has yielded a reasonably stable macroeconomic environment for the past two decades. Central banks in America and Europe have at least avoided following Japan into deflation, which raises debt in real terms and so makes deleveraging even more of an uphill struggle. Asking central banks to ditch inflation targeting and to pursue another goal could do more harm than good particularly if it left people less certain about the central bank’s ultimate commitment to prudence and stability. That is why a switch to NGDP targeting, whatever its virtues, should not be undertaken lightly.

What do these concerns amount to?

1. The Fed has been "loosening," but NGDP is well below its previous trend. Perhaps it is impossible for the Fed to make NGDP rise enough to reach some target value higher than its current value.

2. The theory behind inflation targeting is well understood--households and firms set prices and wages based on the central bank's inflation target. How can we theorize about how firms and households set prices if GDP is expected to be on a stable growth path. Or, perhaps, how will firms and households set their prices when all they know is that NGDP can be expected to be on a stable growth path in the future?

3. If inflation expectations are pinned down, central bankers can look beyond temporary surges inflation due to commodity prices Inflation targeting pins down those expectations. Perhaps NGDP targeting make it necessary for central bankers to respond to temporary surges in inflation.

4. NGDP is a harder concept to grasp than inflation. Perhaps households and firms won't understand what the central bank is trying to do.

5. NGDP is measured quarterly, and inflation is measured monthly. If NGDP has gotten off track, the central bank won't discover that for several months. If the CPI is getting off track, the central bank will find out in one month. Perhaps the lags in GDP reports will keep central banks from taking sufficiently prompt action to reverse mistakes.

6. Inflation targeting worked very well during the last 20 years. Perhaps NGDP targeting will work less well.

7. Inflation targeting suggests that central banks are responsible and prudent. Perhaps NGDP targeting would make people doubt the prudence and responsibility of central bank.

What are my responses?

1. This is the claim that the Fed is "out of ammunition."

Where to start?

The Fed can lower its interest rate target. It is currently at a range between .25% and 0. It can be reduced to something slightly below zero, but no more than the cost of storing hand-to-hand currency.

The Fed can stop paying interest on reserve balances, and further, make the interest rate on them negative, but no more than the cost of storing hand-to-hand currency (vault cash.) In other words, the Fed can charge banks for the privilege of holding an asset with zero interest rate and credit risk.

The Fed can undertake QE3. The Fed can purchase large amounts of government bonds. The experience of Sweden, which has purchased bonds equal to 25% of GDP (as David Beckworth has pointed out,) suggests that the needed amount may be large. In proportion, that would be a level of base money of $4.5 trillion.

And finally, if the Fed commits to doing all of this if necessary, it may need to do little, because the demand to hold money will fall. Firms and households will spend some of the money they are currently holding.

To the degree that real output is currently at capacity, then the increase in money expenditures will be inflationary. Firms and households will be motivated to spend now before prices increase.

To the degree that increases in spending and sales result in firms expanding production and employment, the increase in prices will be smaller. However, the expected increase in employment and capacity utilization will motivate firms to purchase capital goods to expand capacity and households to expand consumption because they are less worried about losing their jobs. And, of course, the newly employed people will earn income and consume more.

So, regardless of what combination of the increase in expected spending generates additional real output or higher prices, there is a motivate to reduce existing money balances (cash) and purchase output.

2. While the "theory" of targeting a growth path of GDP may be less well developed than the theory of inflation targeting, the rough outlines are not difficult to understand.

Generally, households and firms can set prices exactly as they do under inflation targeting. If the GDP growth path is 5% and we start on the target, and no supply shocks are expected, then with capacity growing 3%, the price level can be expected to increase 2% per year.

If there is an adverse supply shock, then the price level will rise to a higher growth path, and the inflation rate will rise above 2%. Once the price level reaches its new growth path, it will return to 2% inflation. To the degree such a supply shock is anticipated, the expected inflation rate will be higher.

However, if the central bank is targeting inflation, then this is exactly what happens when there is an unanticipated supply shock. The inflation rate rises and the price level moves to a higher growth path. The central bank then targets inflation, so prices continue to rise 2% from there.

If households and firms anticipate the supply shock, then they will expect higher inflation. Of course, if the central bank anticipates the supply shock and it contracts to offset it, then it is creating an excess demand for money to offset the supply shock.

This is exactly what the central bank would be required to do if it were targeting the growth path of the price level. This is why targeting the growth path of the price level is a bad idea. And if central bankers don't understand the difference between inflation and growth path targeting, then inflation targeting is not all that well understood. Worse, if firms and households are unsure how the central bank will respond to future supply shocks, then they don't understand what the central bank is doing.

Is there no difference between NGDP targeting and inflation targeting in terms of price setting? There is a difference. The difference involves unanticipated shocks to aggregate demand. If nominal GDP grows more quickly than the target, then real output and the inflation rate will rise. The price level moves to a higher growth path. With NGDP targeting, the central bank restricts money growth and slows money expenditures so that NGDP returns to target. Real output falls or at least grows more slowly so that it returns to the growth path of potential output. And the inflation rate slows so that the price level returns to its previous growth path.

With inflation targeting, the increase in the price level is allowed to persist. NGDP is on a permanently higher growth path. Inflation returns to 2% and real GDP growth slows, as it returns to potential. Presumably, this occurs because costs (including money wages) grow more quickly, move to a higher growth path, and so choke off the excess growth in real output.

One might argue that allowing a permanently higher growth path for prices is the least bad response to this sort of inflationary mistake. However, we are currently suffering the consequences of the other sort of mistake--NGDP has fallen off a cliff. Real GDP has fallen a tremendous amount, and the price level has grown more slowly and fallen to a slightly lower growth path. Inflation has been close to target, and real GDP is supposed to grow back to potential, presumably through costs (including wages) growing more slowly than the trend inflation.

It is the current situation that has led some to advocate price level targeting. Get the price level back up to its previous growth path. By my calculation, that would require about 5% inflation over the next year. But price level targeting has bad consequences when there is an adverse supply shock. And switching between inflation targeting and price level targeting on a case by case bases is no rule at all.

3. Read the above explanation again. Instead of central bankers being confused by the difference between inflation and price level targeting, and worrying about whether or not a surge in inflation is temporary, they just target the growth path of NGDP. This would require them to tolerate a surge of inflation due to supply side factors. It would require them to reverse an excessively expansionary monetary policy that was leading to more rapid growth in nominal expenditures that was impacting, in the first instance, the demand for commodities. One of the ways that excessively rapid growth in NGDP appears is more rapid growth in the prices of goods with flexible prices. Preventing this is what the target requires.

4. Perhaps NGDP is harder to grasp than inflation. NGDP targeting isn't that hard to grasp. Total spending in the economy grows at a stable rate. (My preferred version is easier to explain. The productive capacity of the U.S. economy usually grows 3% per year. NGDP targeting keeps total spending in the economy growing at that same rate. I will grant that having it grow at a 2% faster rate is harder to justify.)

How easy is inflation to grasp? Certainly households and firms have heard about inflation. Many of them may not have ever heard of GDP, and if they have, the significance of "nominal" is probably lost on nearly everyone.

However, how much help is it for households and firms to think they know what inflation means?

For example, the Fed's practice of focusing on core inflation rather than headline inflation causes tremendous confusion. Aren't gasoline and food an important part of the typical person's budget? Why is the Fed ignoring the high prices that are really hurting me? (But, but....)

The actual construction of price indices, and particularly consumer price indices, is aimed at measuring changes in real standards of living. The goal is to get a good measure of what is happening to per capita real income or real wages. Is it really the role of a central bank to stabilize the "cost of living" in that sense?

In particular, is monetary policy to change with every adjustment in our estimates of changes in quality?

Sure, households and firms think they understand inflation, but do they really understand it? And to what degree are they think it means--the things I buy are more expensive, so my standard of living is worse?

In my view, the proper standard to judge monetary institutions, including a central bank's policy, is whether or not the quantity of money is held in balance with the demand for hold money--avoiding monetary disequilibrium. Slow, stable growth in money expenditures on output is the least bad approach.

5. NGDP is currently measured quarterly and the CPI is measured monthly. While shifting to monthly measures of NGDP would be possible and may be desirable, for both approaches, the goal should be to stabilize the expected future value NGDP or the CPI.

It would seem likely that a mechanical feedback rule between past values of NGDP or the CPI, and some instrument of monetary policy, like the change in the quantity of base money or some interest rate, would work better with more frequent measures of the goal.

However, this is not a sensible approach. The current quantity of money should be determined so that the expected level of NGDP (or the CPI) will be on target. To the degree current values of the CPI provide information about what NGDP will be in the future, then they can be taken into account.

I have been proposing that the Fed target NGDP four quarters in the future--not one quarter in the future. Similarly, if I still favored targeting the price level, then I would advocate targeting the CPI one year in the future, not one month or one quarter in the future.

Years ago, Sumner argued for index futures targeting because trying to target the June CPI in June would be a mistake because it has already been partly determined in June. He argued for targeting the July CPI in June, and the August CPI in July.

When you are targeting the CPI one year from now or NGDP one year from now, the number of observations we get of GDP or the CPI between now and one year from now really isn't that important. NGDP targeting doesn't require that a central bank only look at the most recent reported value of NGDP in order to determine what it should do.

6. Inflation targeting did work well between 1984 and 2008. That is 24 years. During that 24 years, NGDP stayed very close to a 5% growth path. For the last 3 years, inflation targeting, particularly combined with periodic, judicious changes in the federal funds rate, has been a disaster. It is time for a change.

7. Keeping total spending growing at a slow stable rate is responsible. A policy of printing up money based upon what the government wants to spend is not consistent with slow, stable growth in NGDP. Creating inflation to bail out debtors, including the government, is not consistent with slow, steady growth in NGDP.

If the government benefits from money creation, it is true that creating less money than is consistent with slow, steady growth in NGDP will reduce how much money the government gets from money creation.

If NGDP is allowed to collapse, then the least bad result would be a lower price level, and that would increase the real value of debts, including those of the government. Avoiding such a collapse is better for debtors.

But these are hardly reasons to stabilize inflation rather than NGDP.


  1. Excellent commentary. Much enjoyed.

  2. 2.5% NGDP growth YoY in 2Q 2016. Did you really think you would ever see the day?

  3. It will be interesting to see the target for GDP but just got to have wise ideas going forward into it. I never need to do anything over the top with help of OctaFX broker since they take care of all with the sort of luxurious offers present here especially the rebate program where we can get up to 15 dollars per lot size trade which is even with the losing trade, so that’s why I find it so special and helps me working out nicely.