The lag in the effect of monetary policy on output employment and inflation may or may not be a problem. It depends on the nature of the shock: whether the shock has a shorter or longer lag than monetary policy. It should not have been a problem in the current recession, where the shock was in financial markets.
In the CPAC roundtable on the $50 billion deficit a few days ago, I said that I wished that monetary policy had been more aggressive a few months ago, so that we might not have needed quite so large a fiscal deficit. Glen Hodgson replied that monetary policy had a long lag in its effect on output and employment, so that a temporary fiscal stimulus was needed. I countered, rather lamely, that fiscal policy had long lags too.
What I said about the lag in fiscal policy was correct, but it wasn't what I should have said. I should have said that an exogenous shock to monetary policy might have a long lag in its effect on output and employment, but a change in monetary policy that is a response to a financial market shock might have an immediate impact on financial markets. It could have prevented the lagged effect of the financial market shock on output and employment. It could have been timely.
I remember at the time thinking about Scott Sumner's post disputing the alleged long and variable lags in monetary policy. But I didn't understand his point well enough to articulate it. Maybe I still don't fully understand the point he was trying to make. But that doesn't really matter. I'm going to use what I've learned from thinking about Scott's post to make my own point.
The Bank of Canada has been targeting 2% inflation. Even though inflation has been very close to 2% on average, it has also fluctuated around that target from year to year.
The lags in monetary policy are crucial in explaining why inflation has fluctuated around the target. If there were no lags, the Bank of Canada could easily have kept inflation exactly at 2%. As soon as inflation started to edge up above 2%, the Bank could tighten monetary policy, and bring it back down again. As soon as inflation started to edge down below 2%, the Bank could loosen monetary policy, and bring it back up again. No forecasting ability or macroeconomic knowledge required.
You don't need a crystal ball or a degree in mechanical engineering to drive a car at the speed limit. You just need to watch the speedometer and remember which way to move the gas pedal. A simple cruise control mechanism can do it. That's because the lags are trivial. Add a 1-minute lag to the gas pedal, and you would need to look at the road ahead and be able to calculate how much extra gas you need to offset the approaching hill.
The Bank of Canada implicitly recognises the lag in monetary policy when it says that its objective is to return the inflation rate to the 2% target over the "medium term", by which it normally means 6 to 8 quarters.
Because there is a lag in the effect of monetary policy on inflation, and the Bank of Canada does not have a crystal ball to forecast shocks, and does not have perfect knowledge of how the economy responds to shocks and to monetary policy, the Bank inevitably makes "mistakes" that are only apparent with hindsight.
When it is responding to its own past "mistakes", the lag in monetary policy matters. That's the Bank of Canada operating like a cruise control mechanism, which can only "see" the speedometer.
But sometimes the Bank of Canada is responding to other shocks, not its own past "mistakes". And exogenous shocks also have lags in their effects on output, employment, and inflation.
If the exogenous shock has a shorter lag than monetary policy, that's a problem for the Bank. Even if the Bank sees the shock, and correctly calculates its impact, and how big a monetary policy response will be needed, the effects of the monetary medicine will arrive too late. Changes in sales taxes, or commodity prices, might be examples of shocks that have a very short lag.
But what if the exogenous shock has the same length of lag as monetary policy? Then the lag in monetary policy is not a problem. If the Bank sees the shock, and responds immediately, the effects of the medicine will arrive at exactly the right time — at the same time as the effects of the shock.
Shocks to financial markets seem to match that description. Monetary policy has a near immediate impact on financial markets, and a lagged effect on output, employment, and inflation. Shocks to financial markets also have a near-immediate impact on financial markets, and a lagged effect on output, employment, and inflation. Why wouldn't the two lag structures approximately match?
An immediate monetary policy response to financial market shocks should cause an immediate countervailing shock to financial markets. If the magnitude of the monetary policy response is correct, the two shocks should cancel out, and have no net impact on investment and consumption plans, and no net lagged effect on output, employment, and inflation.
In the current recession, the main shocks to financial markets were seen in Fall 2008. The Ted Spread peaked in mid October. That was when monetary policy needed to be at its most expansionary. The world's central banks were lowering interest rates in response to financial market shocks. But falling interest rates are not the same as low interest rates. And low interest rates can be seen as a sign of overly tight past and expected future monetary policy as much as loose current monetary policy.
What we are seeing now are the lagged effects on output employment and inflation, of the lagged response in monetary policy to the shock to financial markets.
Worthwhile Canadian Initiative 
This seems plausible as long as the central bank can identify the nature (or degree) of the financial markets shock that has occurred. The bank presumably can’t observe the shock itself but only its effects, and no doubt the signals are imperfect… a specific rise or fall in certain market interest rates might indicate a variety of different underlying shocks, and the bank has to estimate which is its most plausible model of what has happened and what the appropriate response is.
So each cut in Bank of Canada interest rates might have been a correct response to whatever underlying shock the Bank thought had occurred at that time.
Perhaps the bank’s goal should be to exactly cancel out the effect on financial markets – if some market interest rate rises by 0.5%, the central bank should change policy until that same rate goes back where it was before. But this doesn’t necessarily mean it has successfully counteracted the underlying shock.
In the car analogy, a given speed reduction might mean that the car’s on a hill, or it might mean there is a puncture – the appropriate response might not be the same in each case. Even if you could get back up to the previous speed by pushing the gas pedal, the medium-term effects would be very different.
Leigh: I agree. It still leaves open the question of getting the magnitude of the monetary response correct, even if time lags are not an issue. At this point Scott would invoke his NGDP futures market proposal, but I didn’t want to get into that here.
For the Bank to be able to operate in “cruise control” mode, where the only indicator they need to look at is the target variable itself, you need very quick feedback from the instrument to the target variable.
Somewhat related is Paul Krugman’s lecture at the LSE (Part 1 & 2 are up, part 3 is tomorrow):
http://www.lse.ac.uk/collections/LSEPublicLecturesAndEvents/live/LSELive_previous.htm
(BTW, thanks to Nick and Adam I think I actually understood the IS/LM stuff he talks about. Thanks guys.)
“Shock”.
Is this really the right picture?
Is this really the right language?
This was a slow, slow build with an inevitable outcome identified in advance by many Hayekian macroeconomists.
There is no “shock” anywhere to be found in all this.
Are we stuck with a false picture and a false language given us by the macroeconomists who didn’t see what was happening?
Greg: Scott has an interesting take on this. He considers the ‘shock’ not to be the underlying cause (or chain of causes), but the turning point where enough people in the market realise what is happening and react to it.
In this argument, the slow build might be true but doesn’t count as a shock until it results in a change in monetary behaviour.
Although this approach simplifies the analysis, it feels a little circular: redefining the shock to be simultaneous with its effects distracts us from examining the causes.
Thanks Patrick.
Greg: I am working within the standard framework which sees exogenous “shocks” hitting the AD curve, and monetary policy responding to those shocks to try to keep AD growing smoothly. You can say that’s a “Keynesian” perspective, but only in the sense that “we are all Keynesians now” (or at least, most of us are, and many of us were Keynesians pre-Keynes). Yep. It’s not the Austrian framework.
One of these days I will do a post on “non-super-neutralities of money”, which will be an open invitation to you especially to tell me where the standard framework is wrong. Maybe I should do that soon.
Leigh: “Although this approach simplifies the analysis, it feels a little circular: redefining the shock to be simultaneous with its effects distracts us from examining the causes.”
I’ve been thinking vaguely similar thoughts myself. And I worried over them in writing this post. How do we define “shock”, and “a change in monetary policy”? I can see the advantages to Scott’s approach, where a change in monetary policy is defined as a change in the path of MV (which = NGDP). But I can also see the “tautology” (circularity) aspects of this definition. I ducked the question here, and left “shocks” and “monetary policy” open to more traditional interpretations. Because my main point was about the lags, and wanted to avoid muddying the waters by bringing up other questions, interesting as they are.
Nick, If you don’t want to cite a monetary crank like me, use the following argument:
Lars Svensson has said that money policy should target the forecast. He prefers internal Fed forecasts, BTW. Normally that’s what the Fed does. But last October it stopped doing that. How do we know? Because it started saying that the likely trends in inflation and real growth were unwelcome, and called for assistance from fiscal policymakers. You don’t even have to mention NGDP futures, just talk about Fed forecasts of inflation and real growth. Now here is what is so wonderful about targeting the forecast–there is no lag at all. Forecasts respond immediately to changes in instrument settings.
Even without NGDP targeting, there are lots of proxies for RGDP expectations, and of course there is the TIPS spread for inflation, so the Fed can get a rough idea of whether they were falling short. And they did notice they were falling short, but simply decided to play it safe and not put the pedal to the metal. And that’s why they failed. And that’s also why the ECB and BOJ failed.
This is an argument that cites a Princeton economist, not a Bentley economist, and doesn’t require any weird NGDP futures markets. You are better off not even mentioning my name in polite society.
BTW, What I say here does overlap somewhat with your comment about financial markets, and also Leigh’s comment. So I don’t mean to imply that you guys missed this point entirely, I just think there is an even stronger argument here than most people recognize.
BTW#2, Monetary policy lags are not long and variable, it’s just that they are hard to identify in post war data. Prior to WWII they are easy to identify, and the lags are short.
Scott: I do want to cite you. Because I got the idea for this post from you. Monetary crank or no ;).
Slightly off-topic: why does everyone assume that Lars Svensson invented inflation forecast targeting? (Or do I misunderstand what everyone assumes?) As far as I know, when inflation targeting was first introduced in NZ and Canada, it was always inflation forecast targeting. They adjusted the instrument so that the Banks’ internal forecast of future inflation (at a 1 or 2 year horizon), was at the target. Maybe Lars Svensson articulated what they were doing, or justified it, but they were already doing it, as far as I know. I have just sent an email to my colleague Chuck Freedman, asking his views on this (he was deputy governor at the BoC and in there right at the beginning in setting up the inflation targeting framework.
I’ve just been skimming through the Bank of Canada’s press releases that are issued when it sets the overnight rate, 8 times a year. It nearly always repeats the assertion that its current and expected future actions are consistent with inflation returning to the 2% target over “the medium term”. Sometimes it is more explicit about the date, and when it is so, the “medium term” has been stretched a little, to just over 2 years. This seems a bit different from the Fed.
Nick, Yes, many of us have been talking about targeting the forecast well before Svensson. I use him because he is a big name and an articulate proponent of the idea. It adds prestige to my approach.
Correct me if I am wrong, but I’m not sure the BOC is doing what you claim. Couldn’t you interpret that statement as meaning “we hope to get back to a 2% inflation rate over the next two years” not “we expect the price level to be 4% higher 24 months from now.” Forecasts targeters like Svensson and I prefer the latter, and I don’t think this is what the BOC is doing. If they are, then there is really no AD problem at all that need to be solved. Does anyone seriously believe that the BOC looks with pleasure on the expected growth in NGDP and P over the next 24 months?
Even if I am wrong, it suggests the need for “level” targeting, to keep central banks honest. otherwise you could have a situation like 1929-33, where prices kept falling, but at each point central banks suggested that the wrost was over and prices would level off from this point forward.
What the BoC means is that they expect, and intend, that the year over year inflation rate, 2 years from now, will be 2%. In other words, they expect that the price level will rise by 2% between June 2010 and June 2011.
Yep, they are inflation targeting, not price level path targeting. I am now leaning towards price level path targeting (and the BoC may switch to PLP targeting in a couple of years). But they have made a commitment not to change the target midstream, but to review it every few years, so any change in the target would be pre-announced. And I think they are right to do that. They are following a rule, and keeping their commitments (or, at least trying to, roughly).
The strange thing is, that even though they say they have been inflation targeting rather than PLP targeting, if you look back over the last 2 decades, it looks more like PLP targeting. Mistakes on the upside tend to follow mistakes on the downside, and vice versa.
Nick, Thanks for clarifying that. My view is that they should target inflation between now and mid-2010, not between mid-2010 and mid-2011. So I wouldn’t call their policy “targeting the forecast.” Even a backward-looking central bank will insist that it hopes inflation will get back on target at some point in the future.
That comment about the price level is interesting, but of course level targeting is most stabilizing if it is expected, not if it happens because mistakes happen to cancel out. So if they are going to do that, they should let the markets know.