There is no such thing as fiscal policy

It can be rather traumatic for children to see their parents fight. I feel a bit like that when I see two of my heroes Scott Sumner and David Glasner discuss fiscal policy. The whole thing started with Scott picking a fight with a couple of Keynesians. To be frank that discussion really didn’t turn me on and even though I read most of what Scott writes this was not a discussion that I was particularly interested in. And it is certainly not my plan to address what the discussion really is about – let me just say I think Scott makes it unusually complicated – even though I think he is right (I guess…). Instead I will try to explain my view of fiscal policy or rather to explain why I think there really is no such thing as fiscal policy – at least not in the sense that Keynesians talk about it.

In an earlier post – “How I would like to teach Econ 101” – I have explained that there seems to be a disconnect between how economists think about microeconomics and macroeconomics. I think this disconnect basically also creates the misunderstanding among Keynesians about what fiscal policy is and what it can do.

The way we normally think of microeconomics is an Arrow-Debreu world with no money. Hence, we have a barter economy. As there is no money we can not talk about sticky prices and wages. In a barter economy you have to produce to consume. Hence, there is no such thing as recessions in a barter economy and hence no excess capacity and no unemployment. Therefore there is no need for Keynesian style fiscal policy to “boost” demand. Furthermore, it is not possible, as public expenditures in barter economy basically have to be funded by “forced labour”. “Taxes” will be goods that somebody is asked to “pay” to government and government “spend” these “revenues” by giving away these goods to other people. Hence, in a barter economy fiscal policy is a purely redistributional exercise, but it will have no impact on “aggregate demand”.

Therefore for fiscal policy to influence aggregate demand we need to introduce money and sticky prices and wages in our model. This in my view demonstrates the first problem with the Keynesian thinking about fiscal policy. Keynesians do often not realise that money is completely key to how they make fiscal policy have an impact on aggregate demand.

Lets start out with the standard Keynesian stuff:

(1) Y=C+I+G+X-M

Where Y is GDP (nominal GDP!), C is private consumption, I is investment, G is government expenditure, X is imports and M is imports. There is nothing wrong with this equation. It is an identity so that is no up for discussion.

Lets make it a little simpler. We assume that we have a closed economy so X=M=0. Furthermore lets assume that we define “private demand” as D=C+I and lets write nominal GDP as P*Y (Keynesians assume the P=1). Then we get the following equation:

(1)’ P*Y=D+G

In the barter economy P*Y is basically fixed hence it must follow that an increase in G must lead to a similar decrease in D. There is full crowding out.

So lets introduce money with another identity – the equation of exchange:

(2) M*V=P*Y

Combining (1)’ and (2) with the following:

(3) M*V=D+G

This basically explains why Scott keeps on talking about monetary policy rules when he discusses fiscal policy. (By the way this is a very simple IS-LM model).

Hence, the impact of fiscal policy on nominal GDP/aggregate demand crucially dependents on what happens to M and V when we increase G.

Under NGDP level targeting M*V will be fixed or grow at a fixed rate. That means that we is basically back in the Arrow-Debreu world and any increase in G must lead to a similar drop in D as M*V is fixed.

However, lets say that the central bank is just an agent for the government and that any increase in G is fully funded by an increase in the money supply (M). Then an increase in G will lead to a similar increase in nominal income M*V. With this monetary policy reaction function “fiscal policy” is highly efficient. There is, however, just one problem. This is not really fiscal policy as the increase in nominal GDP is caused by the increase in M. The impact on nominal income would have been exactly the same if M had been increased and G had been kept constant – then the entire adjustment on the right hand side of (3) would then just have increased D.

There, however, is one more possibility and it is that a change in G in someway impacts money-velocity (V). This is what happens in the traditional IS-LM model. Here an increase in G increases “the” interest rate. As the interest rate increases the demand for “bonds” increases and the demand for money drops. This is the same as an increase in V. This model in my view is rather ridicules for a whole lot of reason and I really should not waste people’s time on it, but lets just say that the whole argument breaks down if we introduce more than the two assets – money and “bonds” (Google Brunner and Meltzer…). Furthermore, lets say that we are in a small open economy where the interest rate is given from abroad then changes in G will not influence the interest rate (as least not directly) and hence fail to impact V. If the interest rate is determined by inflation expectations (or NGDP expectations) then the model also breaks down.

But anyway lets assume that this is how the world works. But lets also assume that the central bank has a NGDP level target. Then the increase in G will lead to a drop in money demand via a higher interest rate and thereby to an increase in V. However, as the central bank targets a fixed level for NGDP (M*V) an increase in V will have to be counteracted by an “automatic” drop in M. So again the monetary policy reaction function is crucial. In that sense it is also rather tragic that we had a long debate during the 1970s between old style Monetarists and Keynesians about the size of the interest rate elasticity of investments and money demand with out having any discussion about how this was influenced by the monetary policy rules. (This is not entire true, but bare with me).

One can of course play around with these things as much as one wants, but to me the key lesson is that fiscal policy only have an impact on aggregate demand if the central bank plays along. Hence, fiscal policy does not really exist in the sense Keynesians (normally tend to) claim. “Fiscal policy” needs to be monetary policy to be able to impact aggregate demand.

That said, fiscal policy of course can have an impact of the supply side of the economy and that is ultimately much more important – especially as the ill (lack of aggregate demand) the Keynesians would like to cure cease to exist if the central bank targets the NGDP level.


PS don’t expect me to write a lot more about fiscal policy. The idea that fiscal policy can be used to “stimulate” aggregate demand is just too 1970s for me. Even New Keynesians had given up on that idea during the “Great Moderation”, but some of them now seem to think it is a great idea. It is not. And no this is not some “Calvinist” idea I have – I just don’t think it will work.

UPDATE: Scott continues the fiscal debate (lets stop it Scott, we won long ago…)

UPDATE 2: Sorry for the typos…trying to write fast – sitting in Warsaw airport waiting to board on my flight back home to Copenhagen.

UPDATE 3: Back home with my fantastic family in Denmark – while my family is now asleep I see that Scott has yet another fiscal policy comment.

UPDATE 4: Nick Rowe sketch a very similar model to mine. Apparently Danish and Canadian monetarists think alike.

%d bloggers like this: