An NGDPLP Target vs a higher Inflation Target as insurance against the ZLB

This is not a new point (Scott Sumner has made it before), but it needs repeating.

The Zero Lower Bound is something we want to avoid. Here are two ways we can change monetary policy to reduce the risk of hitting the ZLB. Both work, but one comes at the cost of higher average inflation.

A permanently higher inflation target is like working at a second job just in case your house burns down and you need the extra income to buy a new one. An NGDP Level Path target is like buying fire insurance that gives you the extra income only if and when you need it.

1. Permanently raise the Inflation Target.

If the central bank had always been targeting 3% inflation rather than 2% inflation then it would have set a nominal interest rate path (approximately) 1% (one percentage point) higher than what it actually set. That will work to reduce the probability that it would want to set a negative nominal interest rate. And that's a good thing. But it comes at the cost of permanently higher inflation.

2. Switch to a Nominal GDP Level Path Target.

For the same average inflation rate, there are two reasons why NGDPLPT would have a lower risk of hitting the ZLB than would IT.

2a. First, assume that the central bank gets monetary policy right and always hits the target. Under an NGDP target, there will be times when Real GDP growth is higher than average so inflation is lower than average, and other times when RGDP growth is lower than average so inflation is higher than average. Theory tells us that equilibrium real interest rates will probably need to be lower than average at those times when RGDP growth is lower than average, because people want to consume less and save more when they are less optimistic about future income. Those are the times we would want a higher target rate of inflation to avoid the risk of hitting the ZLB, and an NGDP target delivers the higher inflation only at those times when it is most needed, rather than all the time.

2b. Central banks sometimes make mistakes and miss their targets. Suppose a central bank fails to respond to a negative shock and lets Aggregate Demand fall below target. That is a time at which the risk of hitting the ZLB is greatest. The beauty of a Level-Path target for NGDP is that people will expect a temporarily higher rate of inflation and/or growth rate in real GDP, and will increase desired spending as a result, at precisely those times when Aggregate Demand and NGDP have fallen below target.

3. It is true that a Price Level Path Target would also provide insurance against hitting the ZLB after a negative AD shock, by causing a temporarily expected higher rate of inflation after AD fell. This is a definite benefit of a PLPT over an IT. But an NGDPLT provides even better insurance coverage than a PLPT. That's because the Short Run Aggregate Supply Curve (or Short Run Phillips Curve) may be fairly flat, and the economy might be doubly unlucky to get hit by an adverse SRAS shock at the same time as an adverse AD shock (stuff happens). So the temporary increase in inflation needed to bring the price level back to target under PLPT might be small or even non-existent. The NGDPLT provides a more diversified insurance portfolio, with two assets in the portfolio. NGDPLPT says that there will be a mix of temporarily higher inflation and/or temporarily higher real GDP growth whenever AD falls and the risk of hitting the ZLB is greatest. One or the other should work. It's a safer insurance portfolio, because it's more diversified.

And a PLPT doesn't increase the inflation rate when there's a secular decline in potential RGDP growth, while the NGDPLPT does.

If the risk of hitting the ZLB is what scares you, but you don't like higher inflation, NGDPLT beats PLPT beats IT.

Comments

[Bill Woolsey]

Great post, Nick.

Also, when there is an adverse productivity shock, which reduces the natural interest rate and the expected growth rate of real output, the increase in the expected inflation rate offsets the effect of the decrease in the natural interest rate on the equilibirum nominal rate. This is true even if nominal GDP and expected nominal GDP stays on target. There is less chance that expectations of falling future productivity and a falling natural interest rate will require negative real interest rates even without monetary disequilibrium and a drop in aggregate demand.

P.S. Why aren't you at the Canandian Economics Association meetings?

[Andy Harless]

"Central banks sometimes make mistakes and miss their targets."

One point you don't address is that the target may be impossible to hit at the intended horizon, even by a central bank that reacts perfectly to perfect information, because, given current price and output levels, there may be no target-consistent combination of next period output and price level and current period real interest rate that is on both the SRAS curve and the IS curve. (Moreover, even if such a rational expectations equilibrium exists, there could be a Tinkerbell problem -- especially since agents may not know whether such an equilibrium exists even if it does, or they may not know that one another know it exists, or they may just believe one another to be too pessimistic even if they all know that they all know it exists, etc.)

Of course with inflation targeting the target may be impossible to hit at any horizon, which is even worse.

[Nick Rowe]

Bill: Thanks!

"P.S. Why aren't you at the Canandian Economics Association meetings?"

I feel guilty about not going. But I'm confident you did a great job carrying the MM flag at the PEF/CEA.

Ummmm, because I think unnecessary flying is immoral given AGW and didn't have time for the 4 day drive? No, I need a better excuse. Because I need a quiet life at the moment, and air travel and meetings would have taken too much out of me. They are a bit stressful. Blogging controversies are quite stressful enough (I'm taking next week off, starting tomorrow).

"Also, when there is an adverse productivity shock, which reduces the natural interest rate and the expected growth rate of real output,..."

I was thinking that adverse productivity shock would be the same as a secular potential RGDP slowdown. We NGDPers might need to spend more time thinking about whether real shocks cause expected future RGDP growth and real natural rates of interest to be positively correlated under all conditions or just most of the time.

Andy: "...there may be no target-consistent combination of next period output and price level and current period real interest rate that is on both the SRAS curve and the IS curve."

I'm trying to think of conditions where there wouldn't be (absent ZLB), and I can't. Unless there are discontinuities in SRAS and/or IS curves, which would cause problems of hitting any target that were any weighted average of P and Y.

[Sina]

Except for a one time shock that can be made negligible with enough warning communicated, what is fundamentally different about 4% inflation over 2% in terms of the real economy?

You're just accepting that 4% is worse, and I don't see why that must obviously be so.

[Andy Harless]

"conditions where there wouldn't be (absent ZLB)"

I think the conditions would only occur at the ZLB. My point is, if there is a shock large enough to hit the ZLB (or if you're already at the ZLB because agents just generally prefer future expenditure to current expenditure on the relevant margin), the central bank may not be able to avoid hitting it, and under these conditions the central bank may be forced to deviate temporarily from its target, even if it was previously on target, fully anticipates the shock, and accurately anticipates the effects of it actions.

[Nick Rowe]

Sina: "You're just accepting that 4% is worse, and I don't see why that must obviously be so."

Good. Because it isn't at all obvious that average 4% inflation is worse than average 2% inflation.

Here are the standard textbook arguments why it might be worse (none of which are really obvious):

1. "Shoe Leather". Inflation is a tax on holding currency that pays 0% interest. People will try to avoid that tax by running back and forth between the banks and the shops more frequently, so they hold less currency on average while still doing the same amount of shopping. Like (nearly) all taxes, the efforts people take to avoid paying it have real costs.

2. "Menu costs". Firms have costs of changing prices, and will want to change prices more often, so the costs per year will be higher.

3. "Relative price distortions". If firms don't change prices more frequently, and one firm changes prices every January and a second firm changes prices every July, their relative prices will go up and down seasonally for no good reason by a bigger amount.

4. "Bad memory for dates". It's hard to know if $3,000 is a good or bad deal for a canoe today if you can remember you paid $2,000 for the last canoe but you can't remember how long ago that was, so you can't adjust for inflation. So sometimes you pass on a good deal and sometimes you pay too much.

5. "Non-Indexation". It's hard to properly index the tax/benefit system for inflation (God knows why it's so hard, but that's what it seems).

6. "Confused accountants". Again, God knows why, but accountants just do not get inflation, and you just can't explain it to them (I've tried), so they make all sorts of mistakes that matter more when inflation is high.

7. "Confused people". It's not just accountants who get confused. Regular people think a dollar is a dollar, even at different dates.

Plus, if we have gotten used to 2% inflation, and expected 2% inflation, and a lot of people have made plans and promises (like debts) based on 2% inflation, it's a bit arbitrary to change it to 4%, and it may take a very long time (decades) for the short run effects of that change to die out.

(I've probably forgotten a couple).

On the plus side, we have:

1. Avoids the ZLB on nominal interest rates (as in this post).

2. Avoids the possible ZLB on nominal wage changes (absolute downward nominal wage rigidity). Workers who would accept a 1% nominal wage increase with 4% inflation (if they think that's all that's possible) will pointblank refuse a 1% nominal wage cut with 2% inflation (even though their real wages would fall by the same 3% in both cases) even if they lose their jobs. So maybe we get lower unemployment with higher inflation, at least up to a point.

[Nick Rowe]

Andy: OK, I get you now.

[Sina]

Nick:

Some of those are good points, but many of those my gut tells me will have very small effects. But, okay, fine, I get what you're saying. (And your plus-side #2 is my favourite part about a higher inflation target.)

But with NGDP targetting, don't we run the risk of stunting real growth in times when it is large (when/if the next technological revolution happens)? That's what I'm most worried about. If you can convince me that that won't happen, I'm all on board for NGDP targetting.

[Nick Rowe]

Sina: "Some of those are good points, but many of those my gut tells me will have very small effects."

Your gut is not alone (for small increases in inflation, when it's currently low). There is a literature where people try to quantity those effects, but I'm not up on it. Maybe, some of those effects might affect the long term growth rate, which would magnify the present value of those costs.

"But with NGDP targetting, don't we run the risk of stunting real growth in times when it is large (when/if the next technological revolution happens)?"

Maybe. That depends on the model of the SRAS curve. But suppose it does temporarily stunt growth when growth would otherwise be extra fast. By symmetry, it should also temporarily prevent growth slowing as much when growth would otherwise be extra slow. If so, it won't have much effect on growth on average. It won't be ideal, and Real Business Cycle theorists would say that we want growth to speed up or slow down when technological change is fast or slow. But I'm more worried about the Aggregate Demand side, because I think that's where the big danger lies for long run growth. It's hard to invest for the future when you don't know if the economy will be in a recession just after your new factory gets built, or you get your degree.

[Bill Woolsey]

Sina:

Think about it from a micro perspective.

Some firm develops a more efficient production process. From its perspective, resources prices are on the same growth path, but unit cost are lower. It cuts prices and sells more. What is the problem?

A firm develops a new product that period really like. Frims spend more on it. The firm makes a profit. Of course, there are other things they like less. Less is spent on them. This frees up resources to produce the new products that people like better.

Now, the actual prices people pay for things may stay the same on average. But, the goods are better. So given our current conventions we would say that the quality adjust price level is falling. (Or rising more slowly.)

How does this interfere with the motivation to introduce products people like better? I don't see how.

Personally, I favor zero percent trend inflation. And that means a NGDPLT equal to the trend growth rate of potential output. But if potential output speeds up or slows down, and deflation or inflation develops and persists, I don't really see it as a big problem.

It is a mistake to carry over intutions about the effects of deflation when there is a drop in spending on output (or even disinflation when spending slows) to a situation where productive capacity begins to grow extra fast due to increases in productivity--either new and better products or better production methods.

Now, changes in population growth is another matter.

[Peter N]

Scott has 2 proposals, NGDP targeting and NGDP futures. The second is by far the most interesting.

The most important reason is, it seems to me, the possibility of avoiding the paradox that DeLong refers to in the post cited recently here.

http://delong.typepad.com/sdj/2012/06/a-fragment-on-the-interaction-of-expansionary-monetary-and-fiscal-policy-at-the-zero-nominal-lower-bound-to-interest-rates.html.

You could invest in expectation of a given NGDP and hedge against the downside risk. A thorough analysis of this is beyond my abilities, but Scott seems to be neglecting what may be his most convincing idea. Mechanism and concrete steps can, occasionally, matter.

[Ralph Musgrave]

Scott Sumner should ask a more fundamental question, namely what’s the sense in adjusting interest rates at all in a recession? The implication of doing so is that investment spending suddenly becomes more worthwhile relative to consumption spending just because there is a recession: an obvious nonsense.

In fact in a recession (certainly at the start of a recession) there is MORE THAN THE USUSAL amount of capital equipment lying idle. To that extent, investment spending is exactly what is NOT NEEDED.

[Nick Rowe]

Ralph: in a recession, when aggregate demand falls, it creates unemployment. Unemployed workers, and unemployed machines, and unemployed land too. No real surprise there.

If that fall in AD was caused by an exogenous fall in desired consumption, i.e. an increase in desired saving, then people presumably wanted to save more now so they could consume more at some future date. If you were a mythical central planner, and learned that your subjects wanted to consume less today and consume more in the future, you would want to divert some of today's resources away from producing consumption goods towards producing investment goods, so there would be higher future output to make that higher future consumption possible.

[Min]

Nick Rowe: "If that fall in AD was caused by an exogenous fall in desired consumption, i.e. an increase in desired saving, then people presumably wanted to save more now so they could consume more at some future date."

Well, if I am facing starvation, then you might say that I eat something now so that I can eat more at some future date. But that gets the picture wrong. It does not imply that I would rather put aside wheat for planting than make bread with it now. (Not that some wheat should not be set aside, but getting the big picture wrong means that the allocation will be wrong.)

Now, as in the 1930s, there is a private debt overhang. Doesn't the increase in desired saving have more to do with staying afloat, rather than deferring consumption? Even if people hope to consume more in the future?

[David Beckworth]

Nick,

Great post. I come to conclusion that we need to avoid the ZLB for a practical reason: it is hard for most folks to think coherently about interest rates there (i.e. they think low rates mean easy policy). This lack of understanding of the stance of monetary policy at the ZLB creates political pressures that make it difficult for the Fed to respond appropriately. Many observers think the Fed has done too much or is causing financial repression and thus want the Fed to be reigned in when in fact it has been too tight. All of this confusion could be cleared up if more people understood the natural interest rate (maybe that is indictment of us economic educators). Since a NGDPLT would reduce the chances of us returning to a ZLB and the confusion it creates, it is another plus for it.

[dwb]

excellent post.

[Nick Rowe]

Dear all: thanks for your comments, and nice words.

I'm blogged out after 5 posts in 5 days (God only knows how people like Mark Thoma and Paul Krugman do it, or Scott too), so I need to clear my head and I'm taking a week off.

The others (not sure who all went, but I know Frances did) should be back from the CEAs tomorrow or Monday, with lots of exciting new things to report.

Bye for now!

[Ralph Musgrave]

Nick, In your 8.50pm comment just above you say that if “that fall in AD was caused by an exogenous fall in desired consumption, i.e. an increase in desired saving, then people presumably wanted to save more now so they could consume more at some future date.”

I think there is a questionable assumption lying behind that “presumably”. Another possibility (one that has actually been taking place for the last two years or so, as I understand it) is that people simply want more savings (in the form of money), and households might want to cling onto that relatively large stock of money for a decade or more. In that case there is no reason to have a bias towards investment spending in a recession.

As to actually finding out how much of a reduction in AD is attributable to the above two possible causes, this is near impossible, seems to me: do we do household surveys to determine this? I think that is unrealistic.

Also, even if the population does plan to consume less now, and more in a few years’ time, it’s true that that desire could be accommodated by “storing up” capital equipment. But that involves an inefficient use of capital equipment: the equipment lies idle for a few years, and then during the “increased consumption” years, a higher capital equipment to labour ratio is used than is optimum. That all sounds very “iffy” to me.

Re getting exhausted, I try to have one day a week without caffeine. That causes me to doze or sleep most of the day, then come evening, I’m alert and ready to go despite having no caffeine in my system.

[OGT]

Nick- None of the standard text book "costs" of inflation seem terribly compelling, or rigorous, for that matter. I think 5-7 are probably closest to the mark, but they're just categories of money illusion, which can't be spoken of in polite circles. However, if RE isn't reliable enough to base monetary policy on, and it isn't or we'd target zero percent inflation, then I wonder how well NGDPLT stacks up in an adaptive expectations model, especially with adverse shocks as you describe above.

[Julian Janssen]

Great post. I'm getting to appreciate the NGDP-targeting argument.

For any who are interested, here are some musings on the cause of the Great Depression:

http://socialmacro.blogspot.com/2012/06/balance-sheet-great-depression.html

Criticism is appreciated.