“We should throw out all self-contradictory propositions, unmeasurable quantities and indefinable concepts” said Joan Robinson. I’m not sure if this is universally true. But there’s one place I would like to start – with the output gap. The idea here is that there is a difference between actual and unobservable potential GDP, such that inflation falls when actual GDP is below its potential but rises when it is above it. My first beef with this is an empirical one. As Eric has said, “the striking property of inflation of the last 20 years has been its complete invariance to shocks and policy changes.” UK inflation hasn’t varied much since the early 90s – compared to its variance in the 70s and 80s - despite significant cyclical swings. And insofar as it has moved, it has been largely
chris considers the following as important:
This could be interesting, too:
chris writes Against retail politics
chris writes Beliefs and interests
chris writes On class difference
chris writes The need for class politics
“We should throw out all self-contradictory propositions, unmeasurable quantities and indefinable concepts” said Joan Robinson. I’m not sure if this is universally true. But there’s one place I would like to start – with the output gap.
My first beef with this is an empirical one. As Eric has said, “the striking property of inflation of the last 20 years has been its complete invariance to shocks and policy changes.” UK inflation hasn’t varied much since the early 90s – compared to its variance in the 70s and 80s - despite significant cyclical swings. And insofar as it has moved, it has been largely due to moves in oil prices and sterling. Since 1996 (when the current vintage of retail sales data begins) there has been a significant negative correlation – minus 0.65 – between CPI inflation and annual changes in retail sales. That’s hard to reconcile with the idea that inflation rises when demand hits a “potential” level, but easy to reconcile with the notion of inflation being driven by shocks to supply, sterling or commodity prices*.
It’s not just in the UK that the idea of the output gap has been unhelpful. The same has been true in the euro zone. Zsolt Darvas has pointed out that the ECB has over-predicted inflation for years. It’s belief that a closing of the output gap would lead to rising inflation has been systematically wrong. For example, Mario Draghi said 12 months ago that “the economy has been growing consistently above current estimates of potential growth.” And yet core inflation today is exactly the same as it was then – 1.2% (pdf).
Granted Mr Draghi went on to say that “the responsiveness of inflation to slack has weakened in recent years” and that “estimates of the size of the output gap have to be made with caution.” But these look like what Karl Popper called “immunizing strategies” – tricks to defend a theory from any attempt to falsify it.
So, what’s wrong with the output gap in theory? I’d suggest three things.
- Potential output, or capacity, is not fixed. A study (pdf) of a US steel mill by Igal Hendel and Yossi Spiegel has shown this. They show that the mill doubled production over 12 years even with the same plant and workforce because every time the mill seemed to be at “full capacity” its managers found ways of tweaking production methods to eke out more output. “Capacity is not well defined,” they conclude. This suggests that a closing of the “output gap” leads not to rising prices but to higher productivity. Which means that if monetary policy tightens as the “output gap” closes policy-makers will deprive us of productivity gains.
- Companies don’t necessarily respond to higher demand by raising prices. Back in 1986 Julio Rotemberg and Garth Saloner showed (pdf) that they might well cut prices in a boom simply because there are more customers around then and so firms will fight to win them in the hope of winning more return business in future. (This should be more likely to happen when real interest rates are low or negative, because they increase the present value of future revenues.)
- The notion of potential output doesn’t apply to the intangible economy. It might make sense to speak of a restaurant as having a level of full capacity: it has only a limited number of tables**. But what is Google’s capacity? Or Spotify’s? Or that of the computer games industry? As Jonathan Haskel and Stian Westlake say, one feature of the intangibles economy is its scalability. But scalability renders the notion of potential output redundant.
I suspect, then, that the idea of the output gap is a bad one. But is it dangerous?
In one sense, no. In fact, it serves a useful purpose. A central bank that targets inflation can use the idea of an output gap to conduct counter-cyclical policy. In a downturn it can say “there’s a large output gap which will reduce future inflation, so we’re cutting interest rates.” Even if inflation doesn’t depend upon the output gap, this is the right thing to do.
In another sense, though, the idea is dangerous. If you believe the output gap is zero then you will infer that all government borrowing is “structural”. There will therefore be pressure upon governments to conduct unnecessary austerity.
And here lies a big and under-appreciated distinction. It is one thing to use unmeasurable quantities in one’s own personal thinking where the costs of error are small. It is quite another to use them for policy-making where bad ideas do real damage and even destroy lives. The bar for what represents a good idea should therefore be high in policy-making. And I don’t think the idea of an output gap gets over it.
* Yes, I know the omitted variables bias. Feel free to point out what these are in this case, and show their empirical importance.
** In truth, of course, the restaurant business has had excess capacity in recent years, despite the Bank of England’s belief that the aggregate output gap is small.