« Keynes just left Canada | Main | Old Keynesian vs New Keynesian fiscal policy »


Feed You can follow this conversation by subscribing to the comment feed for this post.

Depends which assets, no? This little model assumes fixed-income securities valued at the risk-free rate, so there's no read-through from NGDP to expected cashflows, and no varying risk or liquidity premia to the discount rate.
Let's say housing is the bubble. If you guide down NGDP growth, you are likely to slow the rate of growth in house prices. Same for equities. Both assets have cash flows that correlate positively to NGDP, so the change in the nominal discount rate is not the whole answer.
We also need a conception of what a bubble is to answer this properly.

louis: I think you are right. It depends. But if anyone is going to advocate doing something like this, we should at least be reasonably sure we've got the sign right.

Don't target NGDP. Target financial stability and NGDP will take care of itself.

srini: I think it's more plausible to argue the other way around. A big unexpected drop in NGDP is likely to cause a lot of defaults, whether it comes from declining real GDP or debt-deflation. The Canadian 1982 recession is a good example of this. The unemployed were especially likely to default on their mortgages, unsurprisingly.

In the simplest form of the NK model, I think this corresponds to a lower NGDP target (assuming we're trying to lower output).

The NKPC under certainty is:

pi(t) = a*x(t) + beta*pi(t+1)

where pi is inflation and x is the output gap.

In steady state, pi(t+1) = pi(t), and assuming beta < 1 this implies that ss inflation is increasing in ss output.

Thus to lower ss output, we need to lower ss inflation, and since ss NGDP growth is just equal to inflation, this means a lower NGDP target.

I'm assuming a one-time surprise permanent change in the NGDP target, not announced in advance, and successfully implemented effective immediately. This means that we immediately jump to the new ss on the announcement. The result will be an immediate fall in output and inflation to the new ss. The real interest rate will be unchanged, but the nominal rate will fall because of lower expected inflation.

There aren't assets in the model besides a one-period bond, but if there were I'm assuming their prices would fall because of the lower output. (For instance, if there's a fixed stock of capital, and households trade claims on this capital income, then the price of these assets falls.)

Other assumptions (like sluggish adjustment of inflation, etc) may change things. Of course, this model doesn't have anything that would give rise to financial stability concerns, or that would make us prefer NGDP targeting in the first place.

jonathan: but isn't the beta < 1 a bit of an embarrassing artefact of the Calvo Phillips curve, that NK macroeconomists would prefer to get rid of? It's a very minor and model-specific non-superneutrality, and we wouldn't want to rest financial stability on such a weak foundation.

Nick: Is it? Beta < 1 simply means that firms discount the future when setting prices, and so higher expected future inflation raises current inflation less than 1-for-1. Seems reasonable to me.

(Incidentally, if beta = 1, then there would be *no* change in NGDP target necessary to effect a change in ss output in either direction.)

But I agree that a permanent change in policy target might not be the right way to think about this. If instead we wanted to temporarily lowering output, then this would require a period of *higher* output growth than usual. Of course, this also involves lower inflation, so the resulting path of NGDP growth is ambiguous. (In the simplest model, it probably rises over time).

(I'm taking it for granted that we're trying to improve financial stability by lowering output.)

jonathan: some NK macroeconomists force Beta = 1 by assuming that prices rise at expected/target inflation rate for firms not visited by the Calvo fairy. Like the sticky information models, IIRC. Plus, given real interest rates this low, Beta = one is a good approximation anyway.

"But I agree that a permanent change in policy target might not be the right way to think about this."

I was (somewhat vaguely) thinking about a temporary change, with temporarily sticky inflation.

"(I'm taking it for granted that we're trying to improve financial stability by lowering output.)"

I'm thinking that's problematic. We might be *willing* to (temporarily) lower output, if that's the price of increasing financial stability, but it's not obvious to that that is a sensible cure. A temporarily higher expected inflation and growth rate of output might deflate the bubble, if it causes higher real and nominal interest rates. Which is not, I think, what most people are assuming.

I would announce greater loan scrutiny and leave NGDP alone, but that's just me.

Lord: me too. But you are not playing this fun game!

The dotcom bubble vs the housing bubble in the U.S. might provide a useful contrast in trying to answer this question.

Domestic nonfinancial debt/gdp in red , left scale , as the financial stability indicator. Net worth/gdp in blue , right scale , as the bubble indicator. This was the Great Moderation , remember , so ngdp growth was rock steady throughout the period in question , until , well , you know :


IMO , Greenspan was correct in disregarding the late '90s dotcom bubble and in maintaining his ngdp target , since nonfinancial debt/gdp was stable. It would probably have been useful to employ more specific measures , like raising margin requirements , to limit the extent of the bubble , however.

The response to the collapsing dotcom bubble was more problematic. Starting in mid-2002 or so , it should have become apparent that debt/gdp was on a strong upward trend. Housing price increases started soon thereafter. That's when the ngdp target should have been lowered. You could argue that specific measures aimed at curbing excessive mortgage lending should have been employed instead , but mortgage debt was the driving force behind ngdp growth , so I suspect the ngdp target would have been missed had it been maintained while these measures were employed.

As to the 2015 bubble , my reasoning above might lead to the conclusion that I think current ngdp targets can be maintained , since the financial stability indicator , debt/gdp , is stable. However , if you look at the sum ofmonetary base + nonfinancial debt as a ratio to gdp , it's again on a rising trend , so I can understand why the Fed is getting antsy.

Marko: "Housing price increases started soon thereafter. That's when the ngdp target should have been lowered."

Why do you think that lowering the NGDP (growth rate) target would have reduced rather than increased the house price bubble (assuming it was a bubble)? Shouldn't a lower expected NGDP growth rate cause *lower* real and nominal interest rates, which would push house prices up even more?

(I'm starting to think that maybe people aren't getting my point in this post.)

I get your point , but I think the point is silly. Do you think the average person shopping for a mortgage has a clue about "expected NGDP growth rates" ?

Would bankers and brokers , who were making a killing pushing mortgages out the door , react to any "expectations" chatter from the Fed , whether it's chatting up or chatting down , unless it was accompanied by strong regulatory action or direct influence on rates ?

I suppose the Fed could say " We're raising rates , because a higher NGDP target will cause rates to rise anyway , and we're raising our NGDP target ". They could say that , but they'd sound pretty dumb doing it.

I'm talking about the real world , not a word game.

This is a nice post.

If bubbles are the result of mispricing assets, then there is no reason why the risk free interest rate should move to address them. I think the best way to fight against bubbles is via regulation -- e.g. increase downpayment requirements for home purchases in case of a housing bubble, or increase requirements for purchasing stocks on margins, etc. It's not like the CB only targets inflation -- most central banks play an important oversight/regulatory role of their nation's financial sector, and this is the proper regime in which you address systemic mispricing errors.

The alternative would be to move the whole term structure of rates because of (temporary) pricing errors, which is akin to a store constantly changing the price it charges for its products to offset a software bug in the checkout machine.

I agree that monetary policy is the wrong instrument to use against bubbles, but I think Nick and RSJ's 5:15 pm post are giving the counter argument short shrift.
One way of looking at speculative bubbles is a market where a large portion of demand for the asset is due to expected future appreciation in the asset's price, and perhaps an extrapolation of recent rates of appreciation into the future. Momentum investing, "greater fool" dynamics, and all that. Bubbles tend to pop dramatically because once the momentum breaks, buyers lose their appetite for the asset (i.e. demand is positively correlated with price).
Bubbles are often fueled by borrowing, so tightening liquidity or access to credit can weaken the upward price momentum and trigger a pop. They are also vulnerable to market uncertainty, so greater perceptions of economic risk can also stop the bubble cycle.
Lowering the target for near term NGDP growth can definitely pop a bubble in risky assets, whatever the medium-term effect on risk free interest rates.
I'm not even sure what a bubble in risk-free assets would look like.

rsj: Thanks!

"If bubbles are the result of mispricing assets, then there is no reason why the risk free interest rate should move to address them."

I think that's important. At the very least, it's not at all obvious how and why it would work. Changing the risk free rate might simply change the fundamental price by the same sign and magnitude as the market price, leaving the bubble component unchanged.

louis: hmmm. If the only way to burst bubbles is to create greater NGDP uncertainty, it's not obvious it's worth it. Unless you have a sort of "lots of small forest fires rather than one big one" argument.

Well, I do believe the CB can pop the bubble of any asset by hiking rates enough. E.g. if everyone needs to pay cash for a house up front, then the price/rent multiple of housing will be very low.

But can the CB raise the price of an asset that is depressed by lowering rates? Here, I'm not sure. When something is too cheap, there is some credit constraint at work and its not clear that lowering the risk free rate will remove this constraint. It would be better to try to remove the constraint directly -- e.g. make it easier for more people to borrow against the asset.

At the same time, by intervening this way, you are going to be screwing with all prices, asymmetrically: it's more likely that all prices become less correct if you make everyone more credit constrained than less credit constrained.

What a mess.

Nick - what I was thinking was not so much uncertainty as to NGDP growth, but uncertainty as to the effects of a lower NGDP path on specific asset cash flows and risk premia.
Still, I agree on substance that monkeying with monetary policy to deal with financial bubbles is misguided. Only looking to argue that for most real world bubble "lower NGDP" is the more likely answer to the cheeky question in this post.

I think Nick has the causation reversed here. If a central bank wants to raise interest rates, it just raises interest rates (constricts the supply of base money, if you like, or raises the interest rate on reserves). The question is not how to effect the interest rate increase but how to rationalize it. Most central banks, I expect, would think that an interest rate increase today is more consistent with a lower NGDP target, because the relevant causation goes from interest rates to NGDP, and in standard New Keynesian (i.e. neo-Wicksellian) models the sign is unambiguous. (Neo-Fisherites would say otherwise, but central banks, thankfully, don't use their models very much.) A central bank might be concerned that the Fisher effect would cause longer-term interest rates to go down when it raises short-term rates, but I would guess that, even with this concern, it would judge that raising short-term rates today (and lowering the NGDP target to make it consistent with this move) is likely to have a net effect in the direction of bubble-popping. (It's not clear the effect would be large enough to succeed in popping the bubble, though, unless it were a "shock and awe" policy change, in which case it would surely not be worth the tradeoff for NGDP stability.)

Andy: I think you are getting to the heart of it. But I think we need to distinguish between on-equilibrium path and (threats of) off-equilibrium path play.

What keeps NGDP on (or inflation) target is the central bank's willingness to raise r if NGDP starts to rise above target. But if the target itself is raised (and everyone knows it has been raised, because this is the central bank's new rule, and it is transparent), then the equilibrium r would also need to be raised (to prevent NGDP rising even further and rising above target).


Doesn't the central bank's response depend on how much leverage is used to buy an asset? If home prices bubbled up, but all homes were purchased from retained income, would the central bank give two hoots?

You might say the same thing about the central bank trying to hit an NGDP target. Suppose one day it takes 10% more debt to generate 1% more NGDP growth, then the next day it takes 30% more debt to generate 1% more NGDP growth. Should the the central bank be concerned about the efficiency of it's actions and how should that efficiency be measured?

Really, that is what defines a credit bubble - additional credit raises market prices but results in few new goods.

Not sure if this is going to work, but:

This is a graph comparing total credit market debt outstanding to nominal GDP. Notice the downward trend. From about 1960 to 1975, the "credit multiplier" was routinely between 50% and 90%. After 1975, credit was used less efficiently (in terms of $ changes in GDP per $ of changes in debt). The continuous downward trend began in earnest in 1992.

IF I was using monetary policy to pop the bubble, I’d raise short rates.

That will boost the short term real rate immediately – and by more than the nominal rate increase due to marginally lower inflation expectations.

It will cause the yield curve to flatten compared to where it was. Short rates would be higher and long rates would move by less or might even drop slightly.

At the same time, IF I was using NDGP targeting, I’d expect/announce a temporary slowing of NGDP and a temporary drop in the NGDP target.

IF you believe in the bubble-pop effect, you can’t do the announcement without raising short rates and still have the expected effect.

Later on, I’d drop rates when I thought the effect on the bubble was sufficient and revert to the original NGDP target.

Chuck Norris needs a weapon (although bare hands will suffice).

@louis: Depends which assets, no?

Letting the monetary effects aside: if Treasuries are the bubble, would it be the most natural thing to just short them? Meaning the central bank should sell them (, assuming it has some)? If some other asset class is bubbling maybe the valuation sensitivity with assets the central bank can sell/short is an important factor?

>> And suppose that actual NGDP was on target, and was expected to remain on target in future. And suppose you were Governor, and one of your advisors gives you some important news. Financial markets are in a bubble, so the prices of financial assets are above their fundamental values. And you know your advisor is always right about this sort of thing (just suppose).

You've built a contradiction in the premise here.

If NGDP is "expected to remain on target in the future," and that is of paramount relevance, then we're dealing with an efficient market. On the other hand, the clairvoyant advisor represents private knowledge, giving the Governor of the BoC information about fair pricing that is not reflected in the financial market.

So, you have the markets being right on NGDP and wrong on a financial bubble, and the absolute certainty of both is necessary to this story.

If we did live in such a world, then the proper action would be for the Bank of Canada to announce the existence of a bubble. The markets would then react to correct bubble pricing, NGDP expectations for a given path of BoC intervention would fall, and consistency with NGDP-targeting would then cause the markets to expect (and receive) more BoC intervention.

The comments to this entry are closed.

Search this site

  • Google

Blog powered by Typepad