Tuesday, November 8, 2011

Supply Shocks and Nominal GDP Targeting

Adam P is a bit irritated with all the attention being given to nominal GDP targeting and has been tossing some "volatility critique" and "optimality critique" grenades our way.  Fortunately, the volatility critique grenade was a dud, though I am still am holding and sizing up the optimality one.  Hopefully, it doesn't blow up. Where is the love Adam P?

His latest bombardment on the nominal GDP camp comes in an attempt to critique what I think is one of the biggest benefits of nominal GDP targeting: how it deals with supply shocks.  This latest bombardment, however, is not a dud but ironically ends up blowing apart inflation targeting rather nominal GDP targeting.

Here is why.  Inflation is the result or symptom of underlying shocks to aggregate demand (AD) and aggregate supply (AS).  Monetary policy, however, can only meaningfully influence AD so that is where its focus should be.  This cannot happen with strict inflation targeting because it requires the central bank to respond to any change in inflation, regardless of whether it is caused by AD or AS shocks.  In other words, inflation targeting causes the central bank to respond to AS shocks when it should only be responding to AD shocks.  A nominal GDP target acknowledges this distinction and appropriately focuses monetary policy on the cause (AD shock) not the symptom (inflation).

Adam P., therefore, is therefore right to claim that monetary policy cannot "fix" a supply shock, but he has it completely backward when he claims that this does not happen with inflation targeting:
[T]he thing about supply shocks is that there really isn't anything monetary policy can do to "fix" the problem, one of the reasons that inflation or price level targeting is better [than nominal GDP targeting] is exactly because there is no attempt to fix a problem that is not amenable to a monetary solution.
On the contrary, a strict inflation-targeting central bank is forced to respond to AS shocks when they occur.   For example, assume a new technology makes computers significantly faster.  All else equal, this productivity-enhancing AS shock would create disinflation and put upward pressure on the natural (i.e. equilibrium) interest rate. A central bank adhering to a strict inflation target would be forced to respond to the disinflation by lowering its target interest rate.  This response, however, would push the target interest rate down just as the natural interest rate was increasing, a destabilizing development.  Stated differently, this response would add unwarranted monetary stimulus to an existing boom.  A nominal GDP target, on the other hand, would allow the inflation rate to fall and the target interest rate to rise with the natural interest rate. 

To make this example concrete, assume a nominal-GDP growth-rate target of 5 percent. This technology shock might temporarily result in 5 percent real economic growth and 0 percent inflation under this regime. In contrast, a rigid inflation target of say 2 percent in conjunction with the 5 percent real economic growth would require 7 percent nominal-GDP growth, or a potentially destabilizing surge in spending. Better to ignore the supply shock and allow the temporary disinflation than to have an unsustainable boom in spending. 

Now consider a negative AS shock caused by a super-virus that temporarily shuts down most computer systems. This negative shock would decrease productivity and increase prices. This might, for example, result in 0 percent real economic growth and 5 percent inflation. Here, a 2 percent inflation target would require a tightening of monetary policy that would further constrict an already weakened economy. A Federal Reserve that was targeting nominal GDP would not face this dilemma. It would simply keep total current-dollar spending stable at 5 percent growth and allow the supply shock to work itself out.  Yes, there would still be a recession and rise in unemployment, but nowhere near as pronounced as a central banking choosing to further strangle an economy just to maintain an inflation target.

Though not explicitly arguing for nominal GDP targeting, Lawrence Christiano, Roberto Motto, and Massimo Rostagno in this NBER paper raise the same criticism of inflation targeting.  They formally show that focusing "too narrowly on inflation may inadvertently contribute to welfare-reducing boom-bust cycles in real and financial variables." The authors show that if (1) there are positive productivity innovations and (2) monetary policy follows a standard Taylor rule that responds to deviations of inflation from its target then boom bust cycles can be generated. The authors explain that in
the equilibrium with the Taylor rule, the real wage falls, while efficiency dictates that it rise [following a productivity shock]. In effect, in the Taylor rule equilibrium the markets receive a signal that the cost of labor is low, and this is part of the reason that the economy expands so strongly. The ‘correct’ signal would be sent by a high real wage, and this could be accomplished by allowing the price level to fall. However, in the monetary policy regime governed by our Taylor rule this fall in the price level is not permitted to occur: any threatened fall in the price level is met by a proactive expansion in monetary policy.
In other words, these authors are arguing that by forcing monetary authorities to respond to changes in inflation that come from AS shocks, inflation targeting becomes destabilizing.  Now these authors say nothing about nominal GDP targeting, but their point above that the price level should be allowed to fall in response to a positive AS shock implies nominal GDP targeting--which allows for this very thing--would be better.  

This flaw with inflation targeting--treating all changes in inflation the same--is a big reason why the Fed added too much stimulus in the early-to-mid 2000s.  It misread the disinflationary pressures then as indicating weak AD rather than rapid productivity gains. This flaw also explains why the Fed failed to add monetary stimulus at its September, 2008 FOMC even when all signs where indicating a sharp collapse in AD.  As I said before, it is better to target the cause than to target the symptom.  


  1. If there were never any supply shocks would inflation targeting and NGDP targeting drive the same policies ?

  2. Hmmm. I think my response to the same question is a little bit different from yours. And maybe both are different again from Bill's.

    I think you need to distinguish between the LRAS and the SRAS curves. (Or the long and short run Phillips curves). Different "supply" shocks can cause the two curves to move different amounts. I can rig up a theory in which IT is optimal for supply shocks. I can rig up another theory in which NGDP targeting is optimal. It depends. In general, neither is optimal, and it's a question of which is less suboptimal in practice. I think that's more likely to be NGDP.

    Ah well, we never really pretended to all think exactly alike.

  3. Rob: The simple answer is "yes", but a more correct answer is "no". Sorry. Because central banks will make mistakes under either policy, so there will still be demand shocks. And the legacy of past demand shocks will look like supply shocks, to which IT and NGDPLPT will respond differently.

  4. You have no idea what your up against. The FED's technical staff is literally criminal. I would include Volcker, Greenspan, & Bernanke being criminal. Justice, knowledge & right doesn't triumph. Been there, done that.

  5. Flow 5,

    I am not sure I understand your point. Can you elaborate?

  6. But what central banks are actually supposed to do is FLEXIBLE inflation targeting, which allows them to judgementally allow for supply shocks.

    The recent problem has been, however, that some central bankers have not been doing what they are supposed to do. They have tended to adjust less for positive supply shocks than negative supply shocks. For example, the Bank of England has never undershot its acceptable range for inflation even in the early 2000s but has overshot it, even before the financial crisis, by up to 2%+. Similarly, the Fed did not need to keep the Fed funds rate as low as it did in the early 2000s, when (even Boskinised) CPI inflation almost never fell below 2%.

    The problem has been the central bankers, I suspect out of concern for their own popularity (remember "Maestro"?), rather than the policy framework. While it is clearly not perfect, I would expect that straightforward inflation targeting is better at keeping central bankers honest than a framework that includes real activity. Let the elected government worry about real activity.

  7. BTW, as for:
    "I am still am holding and sizing up the optimality one. Hopefully, it doesn't blow up."
    I think I defused it here.

  8. "To make this example concrete, assume a nominal-GDP growth-rate target of 5 percent. This technology shock might temporarily result in 5 percent real economic growth and 0 percent inflation under this regime."

    What happens if you get 3% or lower real GDP and 0% inflation?

  9. Rob: Like Nick said, they should be the same.

    Nick: I think this issue goes back to our discussion on what happens to AD when there is an productivity shock.

    RebelEconomist: Yes, flexible inflation targeting is different, but the critique of NGDP was not about the flexible version. I agree that central bankers haven't consistently used their flexible inflation policies and that is why I want them to commit to a NGDP level target.

    Anonymous: If real GDP is 3% and inflation is 0% then the Fed (1) either the Fed is not targeting NGDP in the first place (or) if it is and some shock caused it then a NGDP level target would correct for this miss next period. Knowing that the Fed would make up past mistakes, would tend to make such misses less likely in the first place.

  10. About the 3% real GDP and 0% inflation, I believe I'm asking how is the 5% real GDP going to be enforced. For example, what if some people don't want/need more stuff and/or some people can't afford more stuff?

  11. David,

    I've been having a look at the Christiano et all paper and thus far I think I agree with your interpretation. It does appear that stabilizing NGDP would do better in this example.

    In fact I think you can make this an even stronger example of NGDP targeting out performing inflation targeting by considering the effect of false positives, that is signals for future productivity increases that subsequently don't occur.

    The whole thing works through the expected monetary easing casusing things to start happening in advance. If the productivity increase then fails to materialize you get an even worse reversal. This is I think a better description of the boom/bust of the last decade.

    Of course, that doesn't quite imply NGDP targeting is universally better. I still think the 1970s type episodes are examples where stabilizing inflation is clearly preferred but this example appears to work as you've advertised.

  12. Hey Adam P,

    Thanks for stopping by. I think we all agree that just having an explicit target would be an improvement for the Fed.