NGDP targetting? really?

Posted by

There’s a bit of a fuss about a GS report that supports NGDP targeting. Free Exchange shows some of the GS charts, which purport to demonstrate that outcomes would be better under NGDP targeting.

I am not so excited about NGDP myself.  It’s worth remembering that we have been here before (well, sort of).

The Fed Greenbook inflation forecasts were of the GDP deflator (the difference between nominal and real GDP) until 1989; and the dual mandate was well established by that time. Additionally, the arguments that justify real GDP targeting for inflation targeting central banks — the (Okun) relationship between growth and unemployment, and the (Phillips curve) relationship between unemployment and future inflation — were old (and flaky) by this time.

So, why did the Fed abandon the GDP deflator for CPI? Because targeting the GDP deflator didn’t work.

Consider a drop in the value of the USD (which increases import prices), or an increase in oil price. In both cases, the price of imports would increase, and CPI and the GDP deflator would move in opposite directions (CPI rises, the GDP deflator falls). Say an NGDP targeting central bank eased monetary policy in this case. I would expect a further decline in the value of their currency, a further increase in import prices, and hence a further decline in NDGP.

An increase in export prices, or an appreciation of the domestic currency would have the opposite effect — it would boost NGDP. No one serious thinks a central bank should tighten monetary policy if the currency appreciates by 25%

Among other things, the 1970s taught central bankers that CPI was a better guide for policy.

I doubt anyone except Chicago Fed President Evans is arguing for an NDGP target. If Bernanke cannot get an explicit inflation target (something most professionals are convinced is clearly beneficial), I cannot see a move to targeting NDGP.

Any move to target the GDP deflator would be a retrograde step. The unemployment part of the mandate justifies further action, and the Fed is taking action.

They are not pointing to inflation and saying ‘at target – we have done well’ – which is behavior that would potentially motivate congress to change the Fed’s Mandate.


  1. You seem to be saying that (one version of) NGDP targeting wouldn’t add anything to current policy formulation and could give misleading signals. I don’t think Sumner, for example, is hung up on the GDP deflator and often talks about CPI but focuses primarily on expectations including TIPS markets. I suppose one litmus test for whether you think this adds anything is whether you agree with Sumner that monetary policy was too tight from mid-2008, both before and after Lehman. Sumner points out that at the September 2008 meeting (2 days after Lehman), the Fed’s concern was equally divided between recession and high inflation, whereas TIPS markets were only expecting 1.2% inflation over the next 5 years and asset markets were plunging. Clearly, the Fed was missing something, but was it the idea that monetary policy had been too tight and suddenly got a lot tighter or was it that they underestimated systemic risks (or fear of these risks) in the financial system? Not sure if you have blogged about these issues before. Another litmus test is whether you think the Fed should be doing more than they are now given the recent fall in asset prices.

    1. I totally disagree. Neither question tells us if NGDP targets are a good operational guide.

      FWIW i think money was tight in 2007/8 – most FCIs tell you that, especially those that include quantity variables, due to the callapse of ABS issuance. It is clear that money is tight again right now.

      The LM curve is not the central bank’s curve, they just mmove their cash rate to try and influence it. So, of course the fed can be at 0bps, and LM can be too high.

      Sent from my iPad

      1. I think Rajat’s point is that Sumner doesn’t care about GDP deflators or CPI or whatever price index is used. In fact he’s expressed a high level of disdain for the idea of inflation measurement. What he cares about is *expected* NGDP (and therefore the implied expected inflation).

        He doesn’t like idea of model based forecasts (neither do I – I know how these things work – ie. badly). Rather he argues market based forecasts of NGDP / inflation are what should drive monetary policy. See his paper “Let a Thousand Models Bloom”. Given there is no NGDP futures market, the best market indicator of expected NGDP at present is combo of TIPS and asset prices. On TIPS even the regional Fed research suggests, that after adjusting for liquidity biases and other frictions, that TIPs are more or less reasonable guide for expected inflation. And I think asset price reactions to Fed policy statement *tend* to reflect market expectations of impact of monetary policy decisions. They ain’t perfect (that’s not EMH after all) – but they are best guide we have.

        As to the scenarios you suggested. Helpful sensitivity analysis – sure. But the point of Sumner’s NGDP level targeting is that the market finds the optimal levels of monetary accommodation. That means the non-linear effects of the scenarios you point out and the endogenous inter-linkages are resolved – on average – optimally by market. Another thing to point out is that oil or currency or other types of external shocks could be addressed by the “level” part of NGDP level targeting. And you can’t ignore the feedback that might come from introducing ngdp level targeting. The oil market will notice such a monetary policy response. It will then price in monetary policy accordingly. You can’t stop supply shocks and their effects; what you can affect is quickly the oil market responds to growth expectations. If the oil market knows that the Fed is going to smooth pull levers to get level growth of 5% p.a., then I argue it wouldnt respond so violently (absent supply shock of course).

        As to whether anyone aside from Evans is backing ngdp targetting, I think you should have a look at Hall and Mankiw’s paper “Nominal Income Targeting”. Recently Mankiw advocated level inflation targetting in his regular op-ed. It’s only a short step to NGDP level targeting – indeed I think he has expressed support for Sumner’s ideas in the past.

        1. I will check the Hall and Mankiw paper and revert.

          I suppose my point is that the CPI inflation targetting formula worked pretty well – great moderation and all that. If it is busted, it is on the macro-prudential side.

          Sent from my iPad

      1. Thats ok. I just did a summary of this Sumner post. Some of it relevant:

        • Price index that should be stabilised in the sticky one – i.e. some nominal wage index. however, nominal wage index data is poor. therefore go with ngdp. it’s a close proxy
        • Low inflation does not cause liquidity traps; low ngdp growth does. based on china vs japan in 1980s
        • Diff between nominal rates and inflation not as important as diff between nominal rates and nominal gdp
        • Inflation metrics bad because they are inconsistent with NK models. BLS uses rental equivalent; NK models would prefer new purchase prices (i.e. Case shiller). Case shiller down 30%, BLS data up.
        • Supply shocks won’t affect NGDP; NGDP is unambiguously demand side
        • Price level theory is ignorant of quality and utility. This is why nominal wage should be relevant index. What really matters is how much more of nominal wage is required each year to maintain equivalent utility. But nominal wage data isnt great. Therefore use ngdp
        • “PS. Let’s also recall the long and distinguished intellectual tradition of NGDP targeting proposals: Hawtrey, Hayek, McCallum, Taylor, Mankiw, Selgin, and many others.”

  2. Sorry, fifth summary point should read:
    • Effect of supply shocks on monetary policy can be dealt with endogenously. If you target inflation, suppply shock drives up inflation and so monetary response is to contract. I.e. you end up targeting oil prices. But NGDP targets purely the demand side.

Comments are closed.