Mark Carney, Governor of the Bank of Canada, spoke yesterday about short run "speed limits" to economic recovery. This is a concept you don't hear very often, so I thought I would briefly discuss it. It's easily confused with the much more familiar long run "speed limits" to economic growth. And those long run "speed limits" aren't really speed limits; they are level limits.
Actually, I think the Governor's explanation slightly confuses those long and short run speed limits.
"One of the mistakes one can make in these situations is to overestimate the speed limit of the economy in a severe recession," the bank's governor Mark Carney told the House of Commons finance committee.
"The reality is -- it's an unfortunate reality -- but in this adjustment process capacity is lost, investment is delayed, productivity is slower than it otherwise would be and so those inflationary pressures could come back sooner than otherwise (expected),"
The idea of a long run speed limit is familiar to anyone who understands the idea of a vertical long run Phillips Curve. If you try to use monetary policy to increase aggregate demand and keep the unemployment rate permanently below some natural rate, the result will be ever-accelerating inflation.
Associated with that natural rate of unemployment is a natural level of output, and that natural rate of output should be growing over time with a growing population, capital stock, and technology, so there is also a long run natural rate of growth in output. That long run growth rate in the natural rate of output, or the speed at which the Long Run Aggregate Supply curve is moving to the right over time, can be thought of as a long run "speed limit". If you use monetary policy to try to exceed that long run speed limit, sooner or later you will try to get unemployment below the natural rate, and the result is ever-accelerating inflation.
Now you can think of the long run limit on monetary policy as a speed limit, but it is not helpful to do so. Because it's really a limit on the level of output, not a limit on the speed of output growth. You cannot keep the level of output permanently higher than LRAS, even if the growth rate of output is the same as the growth rate of LRAS.
The short run speed limit however, really is a limit on speed. There is an extra term in the Phillips Curve. The idea is that inflation (relative to expected inflation) depends not just on the deviation of unemployment from the natural rate (or the deviation of output from its natural rate), but also on the rate of change of unemployment (or output). If we are in recession, so unemployment is above the natural rate, that will put downward pressure on inflation. But reducing unemployment, even if you are still in a recession, will put upward pressure on inflation.
If there really is a short run speed limit, too quick a recovery in aggregate demand will cause inflation to rise, even while the economy is still in recession.
What could cause a short run speed limit? I see the short run speed limit as due to costs of adjusting output quickly. Even if unemployed workers are available, it takes time to hire them. Even if a firm has unused capacity, it may take time to increase production, and get the extra output to the buyer. If AD increases faster than firms can adjust output, the result will be inflation, even if firms could adjust output to meet demand without raising prices if given longer to do so.
My explanation is different from the Governor's explanation. The things he is talking about -- the lost investment -- would mean the LRAS curve stopped growing at its previous rate, so the natural rate of output is lower than it would otherwise have been. That's a form of hysterisis. We cannot just extrapolate the past output trend to estimate future potential output. That's a level effect, not a rate of change effect. And lowered productivity would mean lowered output for given employment, but would not prevent the bank getting unemployment back to the same natural rate quickly.
So if we see inflation start to rise when the economy begins to recover, even when unemployment remains higher than normal, that doesn't mean we have hit the limit on the level to which unemployment can be reduced. But it might mean we have hit the limit on the speed at which unemployment can be reduced.
test
Posted by: Nick Rowe | May 01, 2009 at 12:39 PM
Hi Nick,
Along these lines, frictions and stuff you might be interested in this guy:
http://elsa.berkeley.edu/~chetty/
A couple nice papers about consumption also being subject adjustment frictions, that is, being sticky.
Posted by: Adam P | May 01, 2009 at 03:06 PM
Maybe another limit to growth at the micro-level is what is going to happen with CAD/USD exchange rates and Canadian interest rates. The flight to quality has the perverse effect of making USD stronger when they are in the crapper. Normally with a strong CAD you might want to encourage consumption of foreign goods and services like tourism and finance sector taking over USA banks, encouraging productivity increasing capital investments (from USA manufacturers). And with weak dollar obviously manufacturing gets a boost.
Now, I can't imagine businesses and public entities have any idea what the exchange rates and interest rates will be over the short to medium term (over long-term, debt must still rule?).
Posted by: Phillip Huggan | May 01, 2009 at 03:23 PM
Why would the speed limit matter? Do central bankers still think it is their job to target real GDP? If so, then I'd think we are in real trouble. If they target inflation at 2% or NGDP at 5% or any other nominal variable, then there is no need to worry about speed limits. Or am I missing something? Are those nominal aggregates inappropriate targets for a mid-sized open economy like Canada?
Posted by: Scott Sumner | May 01, 2009 at 11:04 PM
Adam: Good link (as always, you find them)! I like Chetty's view that "habit persistence" in consumption can be explained in terms of transactions costs and technology of the goods themselves, rather than just a primitive psychological propensity.
I have the same sort of idea at the back of my mind to explain "output persistence" -- transactions costs of hiring new inputs quickly. Or maybe just the time it takes for goods to flow through the supply chain from raw materials to finished products in the stores. We always ((almost?) model y as adjusting instantly to AD, (while allowing that P takes longer), but it is unlikely it does.
Phillip: you are right. I want to generalise your point. More generally (because it is not just the mix between net exports and domestic absorption that might change), when AD expands (whether due to monetary, fiscal policies, or whatever) it is unlikely that the mix of increased demand between different goods will be exactly the same as the mix of existing excess supplies. So resources will need to switch between sectors. And this may take time.
Scott: deeply ingrained at the back of my mind is the unspoken assumption that the central bank targets 2% CPI inflation! (And that fact alone is some measure of the Bank of Canada's success in anchoring 2% inflation into our habits of thought as a quasi-constitutional guarantee.) Given that, they do need to worry about the short-run speed limit, in interpreting the data.
But more generally, I'm just saying that we shouldn't be surprised if inflation starts to pick up even if unemployment has not yet returned to normal. It won't mean we have already hit the natural rate.
Posted by: Nick Rowe | May 02, 2009 at 07:34 AM
Scott,
speed limits matter in calibrating the policy response that is needed to return the economy to the inflation target.
Posted by: 123 | May 03, 2009 at 05:32 PM