Thursday, May 3, 2012

David Andolfatto Can Feel More Confident About NGDP Targeting

David Andolfatto recently asked the following question about nominal GDP (NGDP) level targeting:
What is the theoretical underpinning for NGDP targeting? And what is the empirical evidence that leads them to believe that an NGDP target right now is a cure for whatever ails us right now?
He wants to know why proponents of NGDP level targeting are so confident that it would help the sluggish economy. Nick Rowe, Scott Sumner, David Glasner, Bill WoolseyMark Thoma, and Jason Raves replied to Andolfatto's first question by providing theoretical justifications for being confident.  Here I want to focus on his second question, but before doing that let me quickly note two important creditor-debtor problems that a NGDP level target would overcome in the current crisis.

The first problem is restoring the expected relationship between creditors and debtors that prevailed prior to the economic crisis.  This is the 'risk sharing' problem recognized by David Andolfatto that a price level or strict inflation target cannot address.   A NGDP level target would solve this problem by restoring nominal incomes to their expected pre-crisis paths when debtors signed their nominal debt contracts.    It is possible for nominal debt contracts to be renegotiated, but as Steve Randy Waldman notes this is very difficult in practice. 

The second problem is that there is a massive coordination failure among creditors now.  Creditors could increase their spending to offset the debtor's drop in spending as the latter deleverages.  The reason creditors have not--non-bank creditors are 'sitting' on money assets while bank creditors are destroying them as they are acquired from the deleveraging debtors--is because they are uncertain about future economic activity. These actions by creditors create an excess demand for money or, equivalently, a shortage of safe assets. Now imagine that something simultaneously catalyzed creditors' expectations such that it got them to start using and creating money assets again. The excess money demand problem would disappear and there would be a recovery.  Enter a NGDP level target and its ability to shape expectations.  If credible, this target would raise expected aggregate nominal spending and, in turn, raise expected aggregate nominal income.  It would also probably lead to a temporary surge in inflation.  All would reduce excess money demand by creditors and spark a nominal spending recovery.    

Okay, so what is the empirical evidence that a higher level of NGDP would make a difference now?   The most obvious answer is that those advanced economies currently doing the best are the ones where aggregate nominal spending has remained on or near its pre-crisis trend.  Case in point is Germany.  Its NGDP has remained very close to its pre-crisis trend and consequently, its economy is doing relatively well.  The German unemployment rate has actually fallen since the crisis and consumer confidence is holding up (though slightly down in recent months).  The rest of the Eurozone has not remained close to its pre-crisis trend and we all know how well things are going there.  From this perspective, the Eurozone crisis is nothing more than a NGDP crisis.  (Yes, the Eurozone is a structurally flawed currency union and some countries like Greece are serially defaulters, but Spain and  France? This figure says it all.)

The table below shows the NGDP gap (% deviation from pre-crisis trend) and the unemployment rate for a number of advanced economies.  The data comes from Fred and where possible harmonized unemployment rates are used.  Look closely and you should see a relationship between the two series.

In case the relationship is not clear, here are the two series placed in a scatterplot.  The figure shows that those countries with NGDP closer to their pre-crisis trend typically have lower unemployment rates.   

The fitted line in this scatterplot indicates that a 1% decline in the NGDP gap will reduce the unemployment rate by 0.325%.  That implies the United States could lower its unemployment rate by almost 2% if it closed the NGDP Gap.

Another bit of evidence that suggests a higher level of NGDP would help the current economy is the figure below.  It shows the relationship between the NGDP gap and the risk premium as measured by the Moody's BAA yield - 20 year treasury yield spread.1   This figure indicates that a higher level of NGDP would lower the risk premium.  In other words, the shock to commitment that makes it so hard to privately produce safe assets now would be mitigated by a higher level of NGDP.  This too would reduce the excess money demand problem and spur economic recovery.

A final but important piece of evidence is FDR's very own QE program in 1933.  He had publicly called for the price level to return to its pre-crisis trend and then backed up the rhetoric with a devaluation of the dollar (relative to gold).  As Gautti Eggertson shows, this policy dramatically altered expectations and sparked a robust recovery in 1933.  This implicit price level target of FDRs was  no different than a NGDP level target in this case.  It required radically altering expectations and then being willing to act upon it.  A NGDP level target would do the same today.  It would commit the Fed to buying up as many assets as needed to restore aggregate nominal spending to some pre-crisis trend.  Just the expectation of the Fed doing that may itself cause the market to do much of the heavy lifting.  Portfolios would automatically adjust from lower-yielding safe assets to higher-yielding riskier assets. This portfolio adjustment by itself would raise asset prices and reinforce the recovery.

And what is the downside?  What if adopting a NGDP level target does not improve the economy?  The upshot here is that with a level target there is still a strong nominal anchor in place.  The NGDP level target allows rapid catch-up growth (or contraction) in nominal spending without causing long-run inflation expectations to become unmoored.  Thus, there is far less risk in trying NGDP level targeting than many observers assume.  

1 The 20 year treasury yield is not available for part of the 1980s and 1990s.  This period's yield was estimated using the 10-year treasury yield.


  1. what about if we factor in U6 unemployment figures? the factoring out of workers who have either stopped looking for jobs and have effectively left the labor market and those who are underemployed given their credentials puts the total level of unemployment here in the US at just over 20%. is it worth mentioning? not sure... not necessarily meant as a criticism either because what you have here makes alot of sense. im just wondering if incorporating U6 paints a better picture.

  2. Great blogging--but crickey, is this a serious discussion?

    A central bank can print lots of money and it will not lead to growth and then inflation?

    Really, if QE will not lead to growth and inflation, then let's do about $5 trillion worth of QE buying of Treasury bonds, and wipe out one-half of our national debt.

    The Fed can keep it on their books for a next 200 years.

    Surely, the develeraging alone would be beneficial, and, say critics, cannot cause inflation.

  3. Good post David.

    "The fitted line in this scatterplot indicates that a 1% decline in the NGDP gap will reduce the unemployment rate by 0.325%."

    That immediately reminded me of Okun's Law. About the same value for the parameter, and a better fit by the looks of it.

  4. Great post.

    However, it looks like a positive NGDP gap would reduce unemployment even further. How does that work?

  5. How on earth do you get a +7.8% gap for Australia? Our NGDP tanked in '09 and there hasn't been any catchup since.

  6. great post.

    a different way but equivalent way to think about it here in the US is loan delinquencies, which are strongly positively correlated (90%) with unemployment (or equivalently the output gap) when homes prices have declined. As long as UE stays high, the delinquency-foreclosure - home price decline cycle will continue, putting pressure on the contruction industry.

  7. "The first problem is restoring the expected relationship between creditors and debtors that prevailed prior to the economic crisis."

    You are confusing relationships with outcomes, David. A cornerstone of finance is that, for the investor, fixed income debt is safer than equity, and vice versa for the borrower, and the relative returns are comensurate with the relative risk. The relationship did not change; it just so happened that when the dice fell recently, the investors have done better than the borrowers - unless the borrowers default of course. If you introduce a policy regime like NGDP targeting that changes this relationship ex-ante, then you can expect debt interest rates to be higher in future. And if you intervene to change outcomes ex-post, you introduce another form of uncertainty which will act as a drag on future economic activity.

  8. Rebel Economist:

    A gold standard and money growth rule have the same implications as nominal GDP level targeting for supply shocks for creditors and debtors.

    The impact of "flexible inflation targeting" is a bit difficult to describe. Flexibility is whatever the central bank wants.

    Introducing a price level target regime does shield creditors for "rest of the economy" supply shocks, in a way that historical regimes do not.

    Because of what you take as a "cornerstone of finance," this peculiar regime seems natural to you.

    Your error is to confuse the micro and the macro. Of course, the reality is that the finanice-thinking is micro. Lenders to a particular business are shielded from the risks peculiar to that business by the owner's capital.

    That the owners are using their capital to shielf the borrowers from risk due to supply shocks in unrelated industries just doesn't come into play.

    Of course, if you imagine a one good economy, then the macro is the micro.

  9. @Rebel economist

    fixed income debt is safer than equity, and vice versa for the borrower, and the relative returns are comensurate with the relative risk. The relationship did not change; it just so happened that when the dice fell recently, the investors have done better than the borrowers

    no, fixed income "safety" is a function of credit risk, and delinquencies are strongly correlated with unemployment or the output gap (see below, out put gap on inverted scale).

    So, debt contracts riskiness is a function of the future ngdp path, which did change. changing the ngdp path increases debt charge offs.

  10. Nick: interesting comparison to Okun's Law. It would be good for someone to do a more thorough analysis along these lines.

    Bill: Maybe that positive NGDP gap is reducing the cyclical component of unemployment. It is possible for NGDP to overshoot after all.

    dwb, that makes sense to me.

    Rebel Economist What Bill and dwb said.

  11. Declan,

    I used 1990-2006 as the basis for the pre-crisis trend in Australia. Look at this picture of the natural log of NGDP for Australia and you will see the 2009 dip has been corrected:

  12. Declan,

    I went back and estimated the trend through 2007. The NGDP gap falls to 5.9%. If it is done through 2008 the gap is 3.2%. But going that far is problematic because NGDP starts growing above trend at those times. That is why I stopped at 2006.

  13. Thanks for replying to my comment Bill, but I am afraid that like last time we discussed this (on moneyillusion), I don't understand your argument. Hopefully you will see this response and we can get this important point - which occurred to David Andolfatto too - sorted out.

    To me, it is as simple as this: Assume that an investor has a choice of a fixed income or equity claim and cares about real returns. Assume that the real return on equity is loosely linked with real GDP but that variations of real return are not so extreme that default is a possibility. If real GDP is unexpectedly low for any reason, and the investor's real return is guaranteed by inflation targeting, he does better with a fixed income claim. In an NGDP targeting regime, on the other hand, in which shortfalls in RGDP are made up by inflation, the real value of a fixed income claim is devalued.

    In short, NGDP targeting makes fixed (nominal) income claims more like equity claims.

    What is wrong with that reasoning?

  14. dwb,

    You are right that the "safety" of a fixed income contract is less when economic conditions are poor, but I would guess that nevertheless, the effect of credit losses on fixed income returns is small compared with the variability of equity returns.

  15. You are right that the "safety" of a fixed income contract is less when economic conditions are poor, but I would guess that nevertheless, the effect of credit losses on fixed income returns is small compared with the variability of equity returns.

    First, the ngdp relationship DID change, thats #1: ngdp is about ~6% below trend or full employment, and home prices are ~35% below their peak. Also, *both*
    both inflation and credit losses can hurt.

    However, thats not the right way to think about it. When the ngdp path goes down, assets lose money, and both equity and debt lose money.

    The correct question is not "how much do investors lose," the question is "what is the reservation price that the parties are willing to pay to cancel the contract when the ngdp relationship changes unexpectedly."

    With equity, you never have to repay. So, when the ngdp path changes, you can re-gear output costlessly (to the extent there are no other fixed costs). its like a sunk cost. Debt, conversely, is more like a fixed cost.

    Suppose you are in one of the 25% of underwater homeowners.

    Unexpectedly, the relationship changes so you would like to cancel your mortgage and renegotiate it. You can't, its very costly.

    In fact, its so costly that 70% of homeowners with and LTV between 115 and 150% are current. they wont default *as long as they can pay* because it will damage their credit, they will lose the house, its costly to move, etc etc.

    Most estimates I've seen peg the cost of default in the neighborhood of 30-60% of the value of the house.

    If it were a 30 year fixed price wage or materials contract with a 40% NPV cancellation fee you would say "god da## thats a sticky price."

    As i said, underwater homeowners will continue to default when they lose their ability to pay - which means delinquencies will remain high while UE is high.

    So I strongly disagree with your statement that ngdp targeting would lead to higher credit spreads ("debt interest rates to be higher in future"). Quite the opposite: by reducing the uncertainty surrounding the debt contract (e.g. the borrowers inability to repay due to unexpected ngdp path declines) credit spreads should decline.

    Also, I've mainly argued the ngdp-decline path... but the same logic applies to the unexpected ngdp-increase path (which is inflationary) stabilizing ngdp reduces the inflation uncertainty premium.

  16. ^^ poor editing. "Suppose you are in one of the 25% of underwater homeowners.

    Unexpectedly, the relationship changes "

    --->Suppose the ngdp relationship unexpectedly changes and you become one of the 25% of homeowners under water.

  17. "What is wrong with that reasoning?"

    Excessive aggregation.

    Smith can invest in a 50% share of Farmer Jones' profit from raising corn for $100 or he can lend $100 to Jones at 5% interest.

    He chooses to make the loan rather than buy the share. Farmer Jones makes $20 before interest and gets to keep $15. Jones gets $5. If he had chosen to be a partner, Jones would have had $10 and Smith would have had $10.

    Now, there is a fire in a computer factory. The resulting decrease in the supply of computers raises their prices. If all the other prices in the economy remained the same, then Smith's $5 and Jones $15 would purchase less.

    You don't really care about Smith, but Jones' has a slightly lower real return.

    The central bank contracts the quantity of money enough to force down other prices in the economy so that it pefectly offsets the effect of the higher computer prices.

    It so happens that there is no effect on the price of corn.

    But the sales of cars falls enough that car manufactuers cut car prices a tiny bit. Pay hikes are postponed and some workers are laid off.

    Because of the policy response to the computer fire, forcing other prices down enough to offset the increase in the price of the computers, Smith's $5 purchases the same. Sure, less computers but more cars. Of course, in this example, Farmer Jones is the same.

    Your claim is that if the central bank failed to impose this cost on the car industry, and the result was simply that Farmer Jone's $15 profit and Smith's $5 interst income purchases slightly fewer computers (and the same amount of other stuff,) then this makes the relationship between Smith and Jones into an equity relationships.

    No it doesn't. And even if the central bank's monetary contraction hurt corn sales too, so that it also hurts Farmer Jone's profit, along with the car companies and workers into an equity relationship.

    I understand perfectly well that price level stabilizing is a government intervention that shifts all of the risk away from debt holders and puts it on workers and equitiy holders.

    But it is not an appropriate baseline. This does not happen with gold or silver. It doesn't happen with a fixed quantity of money. It only happens if the government manipulates the quantity of money enough to make it happen.

    It isn't that nominal GDP targeting causes inflation in order to offset lower real GDP. It is rather than when the price of one good rises, it doesn't use a monetary contraction to force other prices down enough to offset the increase.

    If the price of a good rises because of a decrease in supply, then the price rises, and the price level rises as a matter of artithetic and money claims buy less of the good with the higher price and the reduced supply.

    If there is only one good, then the price level is the price of the good. All debt contracts are loans to those producing the good. Any decrease in supply is a decrease in the quantity of that particular good.

    But when there are many good, then yes, there is no surprise that people make like to have someone else bear risk for them.

    I have no problem with this. People can lend using an price level index, if they can find willng borrowers.

    You know, if gas prices rise, your interst rate goes up.

  18. I find your examples a little complicated Bill, but here are some remarks.

    I agree with you that production shocks that do not affect the output of the borrower do not affect the relationship between the investor and borrower. In practice, however, I think that real variations in GDP affect the fortunes of many if not the vast majority of borrowers.

    Your characterisation of how NGDP targeting would work seems insufficiently general to me. Even if a production shortfall did not raise the relative price of the product involved, the authorities would supply enough money so that a higher price level across all products kept NGDP on track.

    I am not claiming that NGDP targeting turns a debt relationship INTO an equity relationship; merely that it makes it MORE LIKE an equity relationship in that creditors are expected to bear some of the risk of hard times. I am not sure that you are disagreeing with this view.

    You make an interesting point about indexing of debt. If investors and borrowers value the characteristics of debt contracts, then perhaps they would recreate them in an NGDP targeting world by indexing contracts.

  19. dwb, the relationship to which I refer is the nature of the investor/borrower contract. That remains the same whether NGDP goes up 6% or down 6%, or house prices go up 35% (which they did and I do not recall borrowers offering to pay more interest at the time in the light of their unexpectedly good fortune) or down 35%. Typically, such contracts do not include clauses saying "except if things are that bad", although of course both sides accept the possibility of default.

  20. David,

    Well, of course it is a subjective choice how long you average the trend over, but to me going back to 1990 is far, far too long. If people like Scott Sumner now think that going back to the pre-2007 5% trend in the US is excessive, how can you justify going back two decades?

    I look at that graph and I see a level shift down followed by a resumption of growth at about the previous rate. I mean, would you seriously look at that path of NGDP and predict there was no recession in '09, without knowledge of the role commodity prices played?

  21. Typically, such contracts do not include clauses saying "except if things are that bad", although of course both sides accept the possibility of default.

    of course they do. the nature of the investor/borrower contract is that the law provides a mechanism for abrogating debt contracts (forclosure/default/bankruptcy). The borrower has that right under law. If you are a finance geek you can think of it as an option with a high exericse cost (in for example the merton model). Its not a free option, thats the cost of credit risk.

    that law is importnant! if income goes down significantly, and mortgages are way underwater, millions will excercise my options! Now, it has some cost (lawyers, credit report stain, and so on).

    Ccompanies ruthlessly exercise their bankruptcy option all the time (check out a case study on general growth properties, its interesting, because they emerged from bankruptcy without equity holders getting wiped out because they they went through bankruptcy even though assets>debt).

    If you are a manager at a real estate company and have not been through a bankruptcy you are considered inexperienced.

    "which they did and I do not recall borrowers offering to pay more interest at the time in the light of their unexpectedly good fortune"

    the argument is symmetric. unexpected inflation is costly too.

  22. dwb,

    I wanted to avoid raising the bankruptcy option, because I imagine that it is a relatively unusual end to a debt contract, but since you mention it, I think that bankruptcy is a much better way of dealing with unexpectedly hard times than engineering higher inflation. Right from the start of the financial crisis, I have said that, instead of bailouts, governments should have spent money on facilitating transition, including an expansion of bankruptcy courts if necessary. The good thing about bankruptcy is that its influence is confined to contracts that have clearly failed, whereas inflation affects all holders of nominal claims, including those who made wise choices or even were actively trying to steer clear of macroeconomic risk.