Things no one has asked me to do: adjudicate the dispute on the existence of the fiscal multiplier (in plain English)
(I’ve tried to write this in plain English, because while it is at first glance a somewhat technical dispute in macroeconomics, it is also window into understanding the relationship between government and central bank policy which is so poorly understood by the vast majority of even informed observers. If even a few of the non-economists who I know read this end up getting it, then some good will have been done)
For those of you not in the know, Keynesian macroeconomics has this idea that government spending increases total spending at a multiple of the original amount, because for every (say) $100 spent, 90% (say) of that would then be spent by those who received that money from the government etc. The multiplier is equal to 1-’Marginal Propensity to Consume’, which in this case is that 90% figure. In this example, the $100 spend would increase total spending by $1000 (multiplier is 10, as MPC is 90% or 0.9).
Of course, this can only work to raise real GDP if there are unemployed resources. Because otherwise, that initial $100 spend represents the using of resources that would have otherwise been employed doing something else. Hence there is no ‘real’ fiscal multiplier in that circumstance, because that spend doesn’t represent an exogenous increase in economic activity, merely a transfer from the private to the public sector. The public sector may or may not be able to make better use of those resources, but the mere fact of spending more cannot make a difference and will either represent foregone private consumption or investment.
However, there are those such as economics blogger par excellence Scott Sumner who says that there cannot be a fiscal multiplier in any circumstance, or even a nominal fiscal multiplier, because it depends on the government doing something (increasing ‘Aggregate Demand’, or total spending on final goods and services) which the central bank actually controls. For example, in my native Britain the Bank of England is instructed by the government to manage Aggregate Demand in such a way as to produce a certain level of inflation. In my also-native US, the Federal Reserve has a somewhat less specific mandate to maintain low inflation and low unemployment. If the central bank is doing its job, then it is crystal clear (and I think not disputed on either side) the fiscal multiplier will be at or close to zero (unless the public sector employs resources more productively than the private sector, which is not the Keynesian contention). But if the central bank isn’t doing its job, and has let demand fall below a level consistent with its mandate, then the question gets a lot trickier to answer.
There are two ways you can attack the question – conceptually and empirically. Let’s start with the latter first. Because the impact on government spending isn’t geographically uniform, there is an approach to ‘measuring’ the multiplier by comparing across geographies. These studies (such as this one, which has shown a decent fiscal multiplier of 1.5) suffer from a pretty devastating objection, which is that they could show a fiscal multiplier even if it was actually zero (or, indeed, if it was negative!). This is because government spending really would shift economic activity towards these regions, as the distribution of the central bank’s offsetting mechanism will almost certainly not have the same geographic profile as the government spending. Therefore, the increase in government spending will make it really look like those places where the government is spending are doing better. Indeed, they will be doing better, relative to other parts of the country. But that won’t be because the pie is bigger, but rather because their piece of it is. Score 1 to Sumner on this point.
However, on the conceptual point I think Sumner has a weakness. He is of the view – and it is almost, almost true – that the level of Aggregate Demand in the economy is completely determined by the central bank. If the central bank is letting Aggregate Demand fall too low, it is because they ‘want’ it to. If the government did anything to change it, they would simply offset the change in AD to bring it back to their desired level. In Sumner’s model, the central bank has some implicit objective (inflation of x, unemployment of y, some mix of the two, some mix of GDP growth and inflation etc.) that it then tries to achieve by setting AD at the level consummate with that target. In short, it rules out the possibility that AD is not at the level the central bank wants it to be. It is not possible that the central bank could want higher AD, and be unwilling or unable to take action to achieve it. It can screw up, through either having a bad target or bad beliefs about how to achieve it. But it can’t be inconsistent.
There are thus two angles fiscal policy advocates can take against this – that there are circumstances where the central bank is unable to raise AD to its desired level, or circumstances where it is unwilling to. There is no evidence to believe the former, and every reason to believe that the central bank can increase AD if it wants to. I’m not going to discuss here the arguments for and against the existence of a so-called ‘liquidity trap’, as that would be the subject of a long post in and of itself (not because it’s a complicated case, but because so many bad arguments have been provided for its existence it would take time to refute them all). I believe the arguments for its existence are spurious, and the historical examples used to support it (e.g. Japan in the last 20 years) are spurious also. Let’s assume for the sake of the rest of the post that the central bank is operationally able to produce whatever level of AD it desires.
That then leads us to the final possibility, that the central bank desires more AD, is operationally able to provide more AD, and does not. It would, therefore, be the case that a fiscal expansion would not be offset by the central bank if it increases AD to the level the bank itself desires. It would at least appear at first glance to be irrationality of both the purest and most unforgivable kind. Could this happen? Has this, in fact, happened? I’m inclined to say, pace Sumner, yes it has. As Sumner himself has pointed out, the Fed is currently acting irrationally – quoting him quoting the Fed minutes last week,
[Fed] “Consistent with its statutory mandate, the Committee seeks to foster maximum employment and price stability. The Committee continues to expect a moderate pace of economic growth over coming quarters and consequently anticipates that the unemployment rate will decline only gradually toward levels that the Committee judges to be consistent with its dual mandate. Strains in global financial markets continue to pose significant downside risks to the economic outlook. The Committee also anticipates that inflation will settle, over coming quarters, at levels at or below those consistent with the Committee’s dual mandate. However, the Committee will continue to pay close attention to the evolution of inflation and inflation expectations.”
We expect to fail, but we’ll keep a close watch on things just to make sure. [Sumner]
Is it then so implausible that the Fed would, for want of a better term, tolerate government action that theoretically should bring inflation and unemployment closer to its mandate? I agree that the Fed could, if it wanted to, achieve this itself. And there is every reason it should want to. But it isn’t doing it. My understanding is that Bernanke has himself said that the government should do fiscal stimulus, despite the fact that it makes nonsense of the rest of his statements strenuously denying that the Fed is ‘out of ammunition’, and nonsense of his academic career prior to becoming Fed chairman.
Why might this be? How can we have a situation where the Fed wants x, can achieve x by doing y, and decides not to do y? I think it has something to do with the question as to why Prof. Sumner had to launch his seemingly tireless effort to persuade both his profession and the intelligent/informed observer that our beliefs about monetary policy are wrong. It has to do with the fact that basic, intuitive and widely-practiced ways of thinking about macroeconomics are hopelessly inconsistent. That is what is, I think, missing from the so-called ‘Market Monetarist’ model. It’s the question of why doesn’t everyone think this way*.
This matters because a central bank is, at the end of the day, a political institution. Appointees are political, and central banks feel political pressure whether or not there is a direct mechanism of accountability. Earlier in the post, I set up the problem in terms of what the central bank is ‘operationally able’ to do. My phrasing was deliberate. The Fed may be operationally able to do x, but may be unable in some other way. We don’t have a ‘public choice’ model of central banking – and I would be more than willing to supply one if I had any expertise or ability to do so. I believe it really is necessary in order for the Market Monetarist model to provide a satisfactory explanation of why central banks seem to be acting irrationally. The economist’s instinctive reaction to seeming irrationality is to ask ‘what incentive is missing from the picture?’, and it is a very healthy reaction to have. Let’s apply it to the catastrophic failures of central banking in the last three years.
Thinking about this clearly is important, because it is by no means an open and shut case that fiscal expansion would have worked in the US. The Fed has already done politically controversial things like Quantitative Easing, and it is a completely reasonable question to ask whether the Fed would have done that had there been more fiscal stimulus. I don’t know the answer to that question. But I have to agree with Noah that it is at least possible that fiscal stimulus could have helped, given the a plausible interpretation of the revealed preferences of central bankers. However, we simply do not have a sufficiently rich description of the central bank’s ‘reaction function’ to give a remotely satisfying answer.
In summary, I think it is ‘almost’ true that the fiscal multiplier is zero – but within that ‘almost’ lies some intellectual terra incognita that needs to be explored and mapped out. We need a model that takes into account the possibility of being wrong, and the effects on central bank incentives if a large portion of the political class are wrong. I don’t know how to go about doing that, but this humble correspondent is all ears.
*Even being the manager of the world’s largest bond fund clearly doesn’t make you immune to writing an op-ed on monetary policy that makes almost no sense whatsoever