David Altig on NGDPLT

Saturos directed me (via Mark Thoma) to this post by David Altig:

In the post on NGDP targeting I was in fact thinking about level targeting, and Gregor Bush’s last sentence gets to””in fact is“”the heart of our disagreement. I am just not willing to concede that anchoring long-term inflation by saying something like “2 percent, 3 percent, whatever” is the path to sustaining central bank credibility. Over the longer term, inflation is the only thing that monetary policy can reliably deliver, as the Federal Open Market Committee (FOMC) has clearly articulated in its statement of longer-run goals and policy strategy:

I suppose I’ve also said that inflation is all the Fed can control in the long run.  But when an economist makes that claim, they implicitly mean “any nominal aggregate,” not inflation per se.  So I’m puzzled by the way Altig phrases this, as if it’s an argument against NGDPLT.

But let’s say I’m misreading him on that narrow point.  Suppose he does agree that NGDPLT is an alternative feasible target.  What about the claim that inflation credibility is important to a central bank?

I’d argue that some sort of nominal anchor is important, but it need not be inflation.  The real argument here is which of the following two is better:

1.  Unstable inflation and stable NGDP growth

2.  Stable inflation and unstable NGDP growth.

I’ve often argued that number one is better, for two reasons:

1.  Almost all the welfare problems allegedly caused by unstable inflation are, on closer inspection, actually more closely correlated with unstable NGDP.

2.  Almost all the welfare problems allegedly caused by changes in the trend rate of inflation are, on closer inspection, caused by changes in the trend rate of NGDP growth.

These include labor market disequilibrium, lender/borrower unfairness, excess taxes on savings and investment due to high inflation/nominal yields, etc.

One possible exception is menu costs of price changes.  But the differences there are slight and NGDP targeting might better stabilize input prices, especially wages.

NGDPLT also has the advantage of eliminating the perceived need for suboptimal fiscal stimulus/bailouts, etc.

It’s not about credibility.  It’s not about which variable can the Fed control in the long run.  It’s about whether the economy does better on a stable inflation (or price level) path, or a stable NGDP path.



24 Responses to “David Altig on NGDPLT”

  1. Gravatar of Saturos Saturos
    19. January 2013 at 08:41

    As I said, more reasons to abolish inflation. (I see you’ve given up on that crusade.)

    Whilst you’re paying attention to my comments, what do you think of this Miles Kimball tweet (and his other ones in the past few minutes): https://twitter.com/mileskimball/status/292671678176186368

  2. Gravatar of Saturos Saturos
    19. January 2013 at 09:04

    In other news – Ryan Avent does a post on Japan, posts what may be the Economist’s first ever NGDP graph: http://www.economist.com/blogs/freeexchange/2013/01/japanese-economy

  3. Gravatar of Steve Steve
    19. January 2013 at 09:08

    Krugman throws out some raw meat:

    “it’s not clear how much difference it would have made if the Fed had grasped the scale of the danger back in 2007. The big errors came later, after the depth of the crisis was apparent to all, and they came mainly in fiscal and housing policy, not monetary policy.”


  4. Gravatar of Andy Harless Andy Harless
    19. January 2013 at 09:39

    Another advantage of NGDP targeting is that it promotes financial stability, which is something that central banks allegedly care a lot about. Let’s take an extreme example where most of a nation’s capital stock is (unexpectedly) destroyed and potential GDP falls by 50%. Either one of two things (or some combination) has to happen: the price level doubles (at one extreme) or nominal GDP gets cut in half (at the other extreme). But nominal debt contracts made before the loss were promises made on the expectation that GDP would not fall. Not only are most individual debtors unable to pay, the whole economy couldn’t pay their debts, because the resources they have promised to pay do not exist. If you let the price level double, it largely fixes this problem: it’s like an automatic 50% write-down of all debts. Most creditors are probably better off than they would have been, having to do difficult negotiations or accept default, and debtors are obviously better off. If, on the other hand, you require nominal GDP to be cut in half, the financial system is screwed.

    This example is extreme and unlikely, but it is qualitatively similar in some ways to what happened in 2008. Little physical capital was actually lost, but, in effect, the perceived productivity of capital, including the housing stock and financial institutions, was severely diminished. If the Fed had determined to keep NGDP on target, allowing for the acceptable possibility of inflation, the collapse of the financial system might have been avoided, or at least it would have been less severe.

  5. Gravatar of Saturos Saturos
    19. January 2013 at 10:10

    Again, it would help if people saw NGDP not as some misbegotten amalgam of a real and nominal variable, but rather as the aggregate flow of money M * V, with the supply of that money fully under the control of the central bank.

  6. Gravatar of Saturos Saturos
    19. January 2013 at 10:12

    Steve, does Krugman still not realize that the Fed was hiking real rates in September 08? How is that not a (the) main big error?

  7. Gravatar of Saturos Saturos
    19. January 2013 at 10:20

    I think noise in the price system is more closely related to inflation instability, though.

  8. Gravatar of Sumner’s Reply to Altig | The Penn Ave Post Sumner’s Reply to Altig | The Penn Ave Post
    19. January 2013 at 11:41

    […] to Altig Posted at 3:47 on January 19, 2013 by Mark Thoma Scott Sumner replies to David Altig : David Altig on NGDPLT [See here for David's post.] […]

  9. Gravatar of Rajat Rajat
    19. January 2013 at 12:44

    Yes, how do investors know where to invest (ie trust price signals in their market) if they do not know whether observed price changes in a given market are relative? With 2% inflation, I know that if I observe a 3% price increase, the sector is experiencing excess demand and I should invest. With a 5% NGDP target, and the same observed price change, I don’t know whether my sector is experiencing excess demand or supply.

  10. Gravatar of Greg Ransom Greg Ransom
    19. January 2013 at 13:08

    A stable inflation rate is impossible for the reasons well identified by Hayek.

  11. Gravatar of ssumner ssumner
    19. January 2013 at 13:12

    Saturos, That’s incomplete, What does Kimball think should be a martingale? (I’ll probably agree with him, but more information is needed.

    The Economist’s first ever NGDP graph is wrong–Japan’s NGDP is lower than 20 years ago. I’m told they plan to correct the graph.

    Steve, And people keep telling me “Krugman agrees with you, he thinks monetary policy can be effective at the zero bound.” Ummm , no he doesn’t.

    Andy, Good point, I should have mentioned that.

    Rajat, No one should EVER invest based on relative price changes—they should look at the expected profitability of investments.

    But if you really think relative price changes are important, just subtract measured CPI inflation from the local price change, then make your invesment decision.

  12. Gravatar of Bill Woolsey Bill Woolsey
    19. January 2013 at 13:24


    You need to think about this more. First of all, with a 2% inflation target, a 3% increase in prices does not mean increasing demand, and investing might be a terrible idea. You have to look at costs to determine profitability.

    Suppose the price is rising because costs are rising faster in that sector? Buying resources and producting more would slow the price increase, but would bit up costs as well.

    Anyway, suppose the price of imported oil rises and the central bank restricts spending growth so that all domestic prices grow more slowly so that the price level on average grows 2%. Every domestic product supposedly has “excess supply.” Do you not invest in nothing? In this circumstance, the opportuntity cost of producting everying has fallen, and so costs (like wages) should grow more slowly. This would make it profitable to continue producing the domestic output.

    If wage growth is sticky, then the effort of the central bank to prevent inflation creates the false incentive to not invest in anything. Of course, this leads to higher unemployment and hopefully, wages will sooner or later grow slowly, shift down to a sufficiently lower growth path, and the lower costs will now signal that despite the lower prices of domestic output, it is wise to invest in just about everything.

    If nominal GDP were targeting, the price of oil just rises, the price level rises as an arithmetic consequence. No one gets the signal to invest less in domestic production, and there is no need to slow the growth of wages. Real incomes (including real wages) fall, but basically because oil products are more expensive to buy.

    If you assume that the only reason relative prices change is due to changes in relative demands (more movies and less icecream) then a stable inflation rate (or price level) provides the appropriate signal. But nominal GDP targeting has the same consequence as price level targeting.

  13. Gravatar of Bill Woolsey Bill Woolsey
    19. January 2013 at 13:37

    In my view:

    The orderging from least bad to worse is:

    Nominal GDP level targeting.

    Nominal GDP growth targeting

    Inflation targeting

    Price level targeting.

    Successful price level targeting is a disaster because of supply shocks.

    Inflation targeting is better than price level targeting for supply shocks, though it is worse for demand shocks.

    Nominal GDP growth rate targeting is marginally better than inflation targeting on both counts.

    And nominal GDP level targeting is much better than nominal GDP growth targeting for demand shocks and maybe marginally worse for supply shocks.

    The market monetarist view is that a credible commitment is a great idea–and that the credible committment should be to a nominal GDP growth path. At any future date, people should know what total spending on output will be.

    A monetary authority can control spending on output. That is how a monetary authority controls the price level and inflation.

    What the central bank cannot do is control real output, employment, or unemployment. Market monetarists never advocate controling those things.

    We only advocate controlling what the central bank can control.

    And again, we believe that a credible committment to a series of future levels of spending on output is better than a credible committment to have the price level rise a fixed percent from wherever it happens to be.

    With nominal GDP targeting, what inflation actually happens to be depends on what happens productivity and supply. Lower productivity and reduced supplies are inflationary. And enhanced productivity and increased supplies are disinflationary.

  14. Gravatar of RebelEconomist RebelEconomist
    19. January 2013 at 14:19

    @Andy Harless: “nominal debt contracts made before the loss were promises made on the expectation that GDP would not fall”

    Rubbish. Debt contracts have always been made on the understanding that, short of default, which may be treated more or less harshly under the law, the debtor benefits from any upside and loses from any downside. If the counterparties want the alternative, they can agree an equity contract instead. Given that there seems to be little enthusiasm for equity contracts in, for example, the housing market, I could imagine that, under NGDP targeting, the counterparties would simply agree an indexed contract instead.

  15. Gravatar of Bill Woolsey Bill Woolsey
    20. January 2013 at 05:29

    Rebel Economist:

    With an economy that has a single firm, then what you say is correct.

    With a multi-good economy, debt and equity divide and share risk for particular firms as the investors desire. This tells us nothing as to whether they prefer to share risks born by other firms in other industries in the same way.

    The monetary regime has implications for risk sharing. An entrepreneur holds equity, perhaps because he is willing to bear the risk that he will shirk. Or that the customers won’t like his product. Or that the production process will be ineffective. That doesn’t mean he is especially willing to bear a larger portion of the risk that there will be bad weather in the wheat market.

    And, of course, treating new entrants to the labor force as if they are equity investors is pretty much criminal.

    In order to make sure that creditors are shielded from the effect of higher prices due to a bad wheat harvest, we will make sure that other prices fall (or just rise more slowly,)maintaining their real interest income. That new entrants to the labor force end up with no employment and no jobs is just a cost of this protection?

    Of course, we could try to set up a scheme where new entrants to the labor force just get really, really low wages. So to make sure that creditors are shielded from a bad wheat harvest, the monetary regime results in new entrants to the labor force getting exceptionally low nominal and real wages when that happens.

    A gold standard, a quantity of money rule, (and nominal GDP level targeting) all have he consequence that a bad wheat harveset raises the price of wheat and the price level. Everyone receiving fixed money claims (creditors) recieves a somewhat smaller real income–they can buy the same amount of everything else, but less wheat. You know, because of the bad wheat harvet.

    If, instead, we have a regime that stablizes the price level (or inflation rate,) then action is taken to force other prices down (or to keep them growing more slowly.) Some people (creditors) are shielded from this risk by the monetary regime. But that just imposes more of the burden on everyone one else.

    Now, over in the wheat market, the debt and equity agreements work just fine. The farmers whose crops are lost pay still owe their creditors. The equity owners took the risk.

    Of course, with price level targeting, the creditors in the wheat market are shielded from the impact of the higher wheat prices. They are compensated by lower prices like everyone else. For the farmer, to the degree that a contractionary monetary policy dampens the increase in wheat prices, they suffer a bit more than otherwise. (Those farmers whose crops were not reduced benefit less.)

    And as I began, if wheat is the only product, then the fall out on the “rest of the economy” doesn’t exist. And that, of course, is why Rebeleconomists is makeing this error. Imagining a single good economy, where “supply shocks” impact the quantity of “output” and so shifts in the price level frustrate the contracts made by those with the “representative agents” (single firm) that made the contract.

  16. Gravatar of ssumner ssumner
    20. January 2013 at 07:15

    Bill , I think I slightly prefer PLT to inflation targeting.

  17. Gravatar of Suvy Suvy
    20. January 2013 at 07:43

    “I’d argue that some sort of nominal anchor is important, but it need not be inflation. The real argument here is which of the following two is better:

    1. Unstable inflation and stable NGDP growth

    2. Stable inflation and unstable NGDP growth.

    I’ve often argued that number one is better”

    I absolutely agree with you. I think this idea that an economy has a low inflation, low unemployment equilibrium that we can guide it to is absurd. The fluctuations between unemployment and inflation may very well be necessary and healthy. We should not be in the business of managing the cycles in an economy. We should be in the business of not letting an economy get too out of hand one way or another.

  18. Gravatar of RebelEconomist RebelEconomist
    20. January 2013 at 09:21

    I think I understand what you are saying Bill – that the debtor is effectively agreeing to insure the creditor against shocks that have nothing to do with the debtor’s business. That may be true, but it may be difficult to unravel the different types of shock, and I certainly don’t think that you can say as Andy Harless did that “nominal debt contracts made before the loss were promises made on the expectation that GDP would not fall”. People know that there are recessions. And efficient markets thinking (which Scott says he believes) would suggest that the debtors do get paid for taking this risk on, in the form of lower liability costs.

    This is part of what is bugging me about the growing popularity of NGDP targeting. I suspect that there is a strong lobby, especially in current account deficit countries like the US and UK, who took out debt contracts knowing the risks perfectly well, and who would have been crowing about their own prescience if those risks had come off, but are now casting around for a bailout. And people like Scott are providing the academic cover for the politicians to offer such a bailout. They don’t care less about the economic merits of NGDP targeting, but “any port in a storm” will do.

  19. Gravatar of Andy Harless Andy Harless
    20. January 2013 at 11:31


    I think debt contracts made between 1995 and 2007 were made on the reasonable (or so it seemed at the time) expectation that there would not be a dramatic drop in (real or nominal) GDP.

    My view of moral hazard is rather the opposite of yours. I think that anyone who, over the past 5 years, has bet on low NGDP growth, needs to be punished. The precedent should be that, when we have a dramatic drop in NGDP and hit the ZLB, people learn to be scared of money and T-bills and not think of them as safe assets (unless they think of them as safe assets with a very negative real return). To set up the incentives otherwise creates a different sort of moral hazard, in which central banks are providing free insurance to the holders of cash, encouraging them to engage in behavior that is severely suboptimal for society as a whole.

  20. Gravatar of RebelEconomist RebelEconomist
    20. January 2013 at 13:39

    Andy Harless, they obviously were not following the Fed’s justifications for their easing in response to incipient crises in those years, such as LTCM (“seizing up of financial markets”) or the Y2K panic. Either monetary policy should have been tighter in the pre-crisis years, and debtors would have paid more interest then, or there really was something to worry about.

    You do not seem to understand the term “moral hazard”. Do you honestly believe that the holders of money have been rewarded for their investment strategy over the last five years? Or has it been the holders of risky assets. BoE research suggests the latter: http://www.bbc.co.uk/news/business-19356665

  21. Gravatar of Andy Harless Andy Harless
    21. January 2013 at 13:37


    Do you honestly believe that the holders of money have been rewarded for their investment strategy over the last five years?

    Yes, I do. They’ve lost less than 2%/year in real terms, which is much better treatment than they deserve, and possibly better than they expected. They have taken unfair advantage of nominal rigidity and central bank policy, the combination of which prevented them from bidding up the value of money initially and thereby allowed them to escape with only small subsequent losses. It was also an unfair one-way bet, because losing 2%/year was basically their worst case scenario, whereas they could reasonably hope that the Fed would lose control of the economy and send us into deflation. Holders of risky assets did well, but that’s because they drew a lucky outcome: we didn’t have another Great Depression; if we had, they would be screwed.

    I don’t buy the logic of your first paragraph. There are various bad things that might have happened if the Fed didn’t prevent them from happening. But the Fed did prevent them. The Fed exists, and people know that it exists, so people made debt contracts on the assumption that, should another crisis occur, the Fed would be able and willing to prevent it from developing into a severe recession, as it had been able and willing in the past.

    Even assuming the Fed was wrong to worry about such things (and obviously, in retrospect of Lehman, it wasn’t wrong), why should monetary policy, on average, have been tighter in that case? We were not experiencing problems with inflation; if anything, the trend in the inflation rate was downward. The response to LTCM and Y2K were minor blips in a policy that, on average, was right on the mark.

  22. Gravatar of RebelEconomist RebelEconomist
    21. January 2013 at 14:32

    I am afraid I just cannot follow your logic, Andy. You seem to be saying that it is reasonable for investors to expect the Fed to limit asset busts (which I would dispute anyway, because that seems to me to create a massive moral hazard problem, but leave that aside) but that post-crisis holders of risky assets just “got lucky”, which entitles you to claim that holders of money got “better treatment than they deserve”.

    Also, you need to go back and look at the Fed’s record on inflation. Based on remarks by Alan Greenspan, the Fed’s informal inflation target was considered to be 2% (and that based on the pre-Boskin measure). Yet, even with the supply shock of cheap manufactured goods from Asia, US inflation between 1995 and 2006 inclusive averaged 2.6%, with no significant downward trend. Not really bad, but nevertheless suggesting that monetary policy in those years was a bit too easy.

  23. Gravatar of Andy Harless Andy Harless
    23. January 2013 at 08:09

    It is reasonable for investors to expect the Fed to limit asset busts to the extent that it’s reasonable and feasible for the Fed to do so. Prior to September 2008, it seemed, based on experience, that it was reasonable and feasible for the Fed to avoid what turned out to happen in the subsequent months. After September 2008, agents revised their estimate of what was reasonable and feasible, such that there was a non-trivial chance that the Fed would not be able to avoid another Great Depression.

    As to the Fed’s inflation target, it was all rather vague prior to 2000, but around that point it became clear that the informal target was 2% on the personal consumption deflator, and, according to most interpretations, it was the core personal consumption deflator that was considered relevant, not the headline deflator. It was widely observed that the CPI grew faster than the consumption deflator, and it was also the case that the price of oil rose during 2004-2006 in what can reasonably be regarded as a one-time increase that the Fed’s inflation targeting policy could not reasonably be expected to offset (especially if it was targeting the core index). From December 1994 to December 2006, the core personal consumption deflator rose at an average annual rate of 1.7%, so by that measure, the Fed was too tight, not too easy.

  24. Gravatar of Andy Harless Andy Harless
    23. January 2013 at 08:16

    I should add, one of the virtues of NGDP targeting is that it clarifies precisely what is considered reasonable and (presumably) feasible. There would be no need for the Fed to limit asset busts directly, unless this was considered the cheapest way to achieve its NGDP target (taking into account any associated moral hazard as one of the costs). There might be questions about short-run feasibility of NGDP targets, but as long as the target is a level path, it should be achievable in the long run. In general, the NGDP targeting policy itself would tend to limit asset busts in a way that doesn’t increase moral hazard (and, as Scott argues, would have avoided most of the 2008-2009 financial crisis).

Leave a Reply