Does Y determine MV or is it MV that determines P?

Scott Sumner a couple of days ago wrote a post on the what he believes is a Great Stagnation story for the US. I don’t agree with Scott about his pessimism about long-term US growth and I don’t think he does a particularly good job arguing his case.

I hope to be able to write something on that in the coming days, but this Sunday I will instead focus on another matter Scott (indirectly) brought up in his Great Stagnation post – the question of causality between nominal and real shocks.

This is Scott:

“I’ve been arguing that 1.2% RGDP and 3.0% NGDP growth is the new normal.  The RGDP growth is of course an arbitrary figure, reflecting the whims of statisticians at the BEA.  But the NGDP slowdown is real (pardon the pun.)”

The point Scott really is making here (other than the productivity story) is that it is real GDP that determines nominal GDP (“NGDP slowdown is real”). That doesn’t sound very (market) monetarist does it?

Is this because because Scott – the founding father of market monetarism – suddenly has become a Keynesian that basically just thinks of nominal GDP as a “residual”?

No, Scott has certainly not become a Keynesian, but rather Scott fully well knows that the causality between nominal and real shocks – whether RGDP determines NGDP or it is the other way around – is critically dependent on the monetary policy regime – a fact that most economists tend to forget or even fail to understand.

Let me explain – I have earlier argued that we should think of the monetary policy rule as the “missing equation” in the our model of the world. The equation which “closes” the model.

It is all very easy to understand by looking at the equation of exchange:


The equation of exchange says that the money supply/base (M) times the velocity of money (V) equals the price level (P) times real GDP (Y).

The central bank controls M and sets M to hit a given nominal target. Market Monetarists of course have argued that central banks should set M so to hit an nominal GDP target. This essentially means that the central bank should set M so to hit a given target for P*Y.

We know that in the long run real GDP is determined by supply side factors rather than by monetary factors. So if we have a NGDP target then the central bank basically pegs M*V, which means that if the growth rate in Y drops (the Great Stagnation story) then the growth rate of P (inflation) will increase.

So we see that under an NGDP targeting regime the causality runs from M*V (and Y) to P. Inflation is so to speak the residual in the economy.

But this is not what Scott indicates in the quote above.

This is because he assumes that the Fed is targeting around 2% (in fact 1.8%) inflation. Therefore, IF the Fed in fact targets inflation – rather than NGDP – then in the equation of exchange the Fed “pegs” P (or rather the growth rate of P).

Therefore, under inflation targeting the Fed will have to reduce the growth rate of M (for a given V) by exactly as much has the slowdown in (long-term) growth rate of Y to keep inflation (growth P) on track.

This means that under inflation targeting shocks to Y (supply shocks) determines both M and P*Y, which of course also means that “NGDP slowdown is real” (as Scott argues) if we combine a slowdown in long-term Y growth and an inflation targeting regime.

Scott won – so he is wrong about causality

Scott since 2009 forcefully has argued that the Federal Reserve should target nominal GDP rather than inflation. I on the other hand believe that Scott has been even more succesfull than he believes and that the Federal Reserve already de facto has switched to an NGDP targeting regime (targeting 4% NGDP growth). Furthermore, I believe that the financial markets more or less realise this, which means that money demand (and therefore money-velocity) tend to move to reflect this regime.

This also means that if Scott won the argument over NGDP targeting (in the US) then he is wrong assuming that that real shocks will become nominal (that Y determines M*V).

The problem of course is that we are not entirely sure what the Fed really is targeting – and neither is most officials. As a consequence we should not think that the monetary-real causality in anyway is stable. This by the way is exactly why we can both have long and variable leads and lags in monetary policy.

For further discussion of these topics see these earlier posts of mine:

The monetary transmission mechanism – causality and monetary policy rule

Expectations and the transmission mechanism – why didn’t anybody think of that before?

How (un)stable is velocity?

The missing equation

The inverse relationship between central banks’ credibility and the credibility of monetarism


Long and variable leads and lags

Scott Sumner yesterday posted a excellent overview of some key Market Monetarist positions. I initially thought I would also write a comment on what I think is the main positions of Market Monetarism but then realised that I already done that in my Working Paper on Market Monetarism from last year – “Market  Monetarism – The  Second  Monetarist  Counter-­revolution”

My fundamental view is that I personally do not mind being called an monetarist rather than a Market Monetarist even though I certainly think that Market Monetarism have some qualities that we do not find in traditional monetarism, but I fundamentally think Market Monetarism is a modern restatement of Monetarism rather than something fundamentally new.

I think the most important development in Market Monetarism is exactly that we as Market Monetarists stress the importance of expectations and how expectations of monetary policy can be read directly from market pricing. At the core of traditional monetarism is the assumption of adaptive expectations. However, today all economists acknowledge that economic agents (at least to some extent) are forward-looking and personally I have no problem in expressing that in the form of rational expectations – a view that Scott agrees with as do New Keynesians. However, unlike New Keynesian we stress that we can read these expectations directly from financial market pricing – stock prices, bond yields, commodity prices and exchange rates. Hence, by looking at changes in market pricing we can see whether monetary policy is becoming tighter or looser. This also has to do with our more nuanced view of the monetary transmission mechanism than is found among mainstream economists – including New Keynesians. As Scott express it:

Like monetarists, we assume many different transmission channels, not just interest rates.  Money affects all sorts of asset prices.  One slight difference from traditional monetarism is that we put more weight on the expected future level of NGDP, and hence the expected future hot potato effect.  Higher expected future NGDP tends to increase current AD, and current NGDP.

This is basically also the reason why Scott has stressed that monetary policy works with long and variable leads rather than with long and variable lags as traditionally expressed by Milton Friedman. In my view there is however really no conflict between the two positions and both are possible dependent on the institutional set-up in a given country at a given time.

Imagine the typical monetary policy set-up during the 1960s or 1970s when Friedman was doing research on monetary matters. During this period monetary policy clearly was missing a nominal anchor. Hence, there was no nominal target for monetary policy. Monetary policy was highly discretionary. In this environment it was very hard for market participants to forecast what policies to expect from for example the Federal Reserve. In fact in the 1960s and 1970s the Fed would not even bother to announce to market participant that it had changed monetary policy – it would simply just change the policy – for example interest rates. Furthermore, as the Fed was basically not communicating directly with the markets market participant would have to guess why a certain policy change had been implemented. As a result in such an institutional set-up market participants basically by default would have backward-looking expectations and would only gradually learn about what the Fed was trying to achieve. In such a set-up monetary policy nearly by definition would work with long and variable lags.

Contrary to this is the kind of set-up we had during the Great Moderation. Even though the Federal Reserve had not clearly formulated its policy target (it still hasn’t) market participants had a pretty good idea that the Fed probably was targeting the nominal GDP level or followed a kind of Taylor rule and market participants rarely got surprised by policy changes. Hence, market participants could reasonably deduct from economic and financial developments how policy would be change in the future. During this period monetary policy basically became endogenous. If NGDP was above trend then market participant would expect that monetary policy would be tightened. That would increase money demand and push down money-velocity and push up short-term interest rates. Often the Fed would even hint in what direction monetary policy was headed which would move stock prices, commodity prices, the exchange rates and bond yields in advance for any actual policy change. A good example of this dynamics is what we saw during early 2001. As a market participant I remember that the US stock market would rally on days when weak US macroeconomic data were released as market participants priced in future monetary easing. Hence, during this period monetary policy clear worked with long and variable leads.

In fact if we lived in a world of perfectly credible NGDP level targeting monetary policy would be fully automatic and probably monetary easing and tightening would happen through changes in money demand rather than through changes in the money base. In such a world the lead in monetary policy would be extremely short. This is the Market Monetarist dream world. In fact we could say that not only is “long and variable leads” a description of how the world is, but a normative position of how it should be.

Concluding there is no conflict between whether monetary policy works with long and variable leads or lags, but rather this is strictly dependent on the monetary policy regime and how monetary policy is implemented. A key problem in both the ECB’s and the Fed’s present policies today is that both central banks are far from clear about what nominal targets they have and how to achieve it – in some ways we are back to the pre-Great Moderation days of policy uncertainty. As a consequence market participants will only gradually learn about what the central bank’s real policy objectives are and therefore there is clearly an element of long and variable lags in monetary policy. However, if the Fed tomorrow announced that it would aim to increase NGDP by 15% by the end of 2013 and it would try to achieve that by buying unlimited amounts of foreign currency I am pretty sure we would swiftly move to a world of instantaneously working monetary policy – hence we would move from a quasi-Friedmanian world to a Sumnerian world.

Without rules we live in Friedmanian world – with clear nominal targets we live live in Sumnerian world.

PS Today is a Sumnerian day – hints from both the Fed and the ECB about possible monetary tightening is leading to monetary policy tightening today. Just take a look at US stock markets…(Ok, Greek worries is also playing apart, but that is passive monetary tightening as dollar demand increases)

%d bloggers like this: