Why did (Canadian) inflation-targeting work in 1996 but fail in 2008?

In both 1996 and 2008 the Canadian economy was hit with a big shock. The 1996 shock was the change in fiscal policy, turning a large deficit into a large surplus. The 2008 shock was the global financial crisis. (Canada didn't really have much of a financial crisis; no banks failed, as usual.)

In both 1996 and 2008, the Bank of Canada kept inflation (more or less) at the 2% target (in 2008 total inflation fluctuated above and below target, but core inflation stayed close to trend).

In 1996 there was no recession. In 2008 there was a recession. Why? (I don't have an answer to that question, sorry.)

In my view, the main thing we are looking for when we choose a monetary policy target is that it prevents recessions. Inflation targeting passed that test in 1996, but failed that test in 2008. I used to be a supporter of inflation targeting, because it seemed to work well in practice. When it became clear to me that inflation targeting had failed in 2008, I stopped supporting inflation targeting, and switched to supporting NGDP level-path targeting.

But I would feel a lot more confident in my support of NGDP targeting if I understood why inflation targeting worked in 1996 but failed in 2008.

Comments

[Chris J]

Nick, in the post 2008 austerity debates some people used Canada 1995 as an example of austerity working. The response I remember reading was an export boom lead by exports to the fast growing USA made Canada a success in 1995. Is this interpretation evidence based?

[Michael Margolis]

A disaster for a trade partner is partly a supply-side shock. You used to have this great technology for making cosmetics -- send woodpulp across the Southern border, and it turns into lipstick. Then in 2008 the machine breaks down. Half the Yanks who used to broker the thousands of deals that made it work can't get the loans against receivables that used to be routine. It's like you forgot a technology -- a negative Real Business Cycle style shock. The central bank can neutralize the shock to net export demand, but it can't neutralize the loss of real gains from trade.

The fiscal change in 1996 was no such shock.

[Nick Rowe]

Chris: In my view, it's not so much an example of austerity "working", but an example of "austerity" not causing a recession. (Which shows that calling it "austerity" is a misnomer, because national income did not fall. We should simply call it by the more neutral "fiscal tightening" instead.)

There was an increase in net exports. I think that increase in net exports was not just a lucky coincidence, because the (real) exchange rate fell at the same time. We can get into a semantic debate about whether the Bank of Canada *decreased them* (active) or *allowed them to fall* (passive), but the interest rate and exchange rate did fall, which would provide the offsetting increases in investment and net exports.

If the increase in net exports had been just a lucky coincidence, caused by a fast-growing US economy, we would expect to see the exchange rate appreciate. Though you could legitimately argue that the fast US growth meant the exchange rate didn't need to fall as much as it would otherwise have done.

[Nick Rowe]

Michael: (I had to fish your comment out of the spam filter, don't know why. But if future comments don't appear, that's what happened. I will check from time to time, and so train Typepad to recognise you.)

Take an RBC model. A reduction in the fiscal deficit is a real shock that has real consequences. One of those real consequences will be a fall in the real exchange rate, just like when there's a fall in foreign demand. In a sense, a reduction in demand from Government (and from Canadian households cutting consumption demand if taxes increase) is like a (small) disaster for a trade partner. Canadian producers saw a reduction in demand from their traditional trade partners (Canadian government and households). Plus, at the more micro-level, production has to reallocate resources when there's any shift in relative demand.

Nevertheless, there may be something to what you are saying. I don't rule it out, but I don't feel clear on it yet.

[JKH]

The behavior of the stock market and the "wealth effect" was massively different.

Wouldn't that be a good place to start thinking about this?

[Market Fiscalist]

If the only thing that varied was demand for money then inflation targeting and NGDP targeting would be equal. But if relative demand fluctuations then inflation targetting may lead to too low an inflation rate for relative price adjustments to take effect if prices/wages are sufficiently sticky.

Simple example: The demand for labor drops as pessimism increases at the start of a recession. Wages need to fall relative to other prices but don't because of stickiness. An inflation target may be successfully hit, but if this doesn't allow the required relative fall in wages to be fully accomplished RGDP may still fall.

[Roger Sparks]

Nick,

Try thinking from a slightly different frame of reference. Any worker 'making goods for export' is working for an external economy no matter where the worker lives.

Using this frame of reference, in 1995 Canadian workers would have lost jobs due to the fiscal change from debt to surplus. Because there was no recession, the workers must have switched to export jobs (making things for export).

In 2008, the exports to many economic sectors were reduced. Canadian export driven workers would have job reductions.

Again using this frame of reference but shifting focus to the secondary support level of workers (those workers who support the primary workers who produce salable products), we would expect to see that in 1995, the secondary level saw relatively little changed in demand. In 2008 however, the secondary support level could expect to see a considerable change in demand as primary workers lost jobs.

In common to both years (1995 and 2008), the Canadian economy had been partly driven by deficit spending. The difference between 1995 and 2008 was that in 1995 the source of the deficit shifted from internally sourced to externally sourced.

I have not tested this proposed frame of reference with actual data. Of course you would want to test it before making a shift away from NGDP support.

[Majromax]

Perhaps the difference between a "supply-shock" and "demand-shock" for international trade is movement in the exchange rate?

In 1996, we could send wood-pulp across the border for cosmetics. The terms of trade weakened such that wood-pulp would not buy as many cosmetics, but we could still send over as much wood-pulp as we wanted.

In 2008, on the other hand, the exchange rate did not move that much (at least after the acute crisis was mitigated). We could exchange wood-pulp for cosmetics at roughly the same fixed rate, but the US simply would refuse to buy as much wood-pulp as we wanted.

1996 then looks like a supply shock, with imports becoming more expensive. 2008 looks like a demand shock, where imports stay the same price but the market doesn't clear.

In turn, this may have happened because of ZLB shenanigans. The situation of 1996 allowed the Bank of Canada, in the course of ordinary inflation-targeting, to create an interest-rate differential that supported a weakened exchange rate. In 2008, where both Canada and the US dropped rates, this happened to a lesser extent, and the US's AD shortfall spilled over into a shortfall in import demand as well.

[Kenneth Duda]

Sorry if this is uninformed and/or obvious, but did the Canadian CB hit the ZLB in 1996?

The reason that IT failed in 2008 seems kind of obvious, at least in the US, where the Fed was more or less following a Taylor rule. First, the Fed brought its interest rate target to zero. Second, inflation failed to materialize. Third --- and this is important --- the Fed failed to act concerned about missing its PCE inflation target. It basically said, "Well, we never really wanted inflation anyway, and our 2% target is really a ceiling, let's see if we can get this economy moving without any inflation, we're going to take some risks with QE but trust us we will never ever let inflation above 2%." And the markets said, "that's interesting. There's no way you're going to get the markets moving if you're going to be that timid about it." And we lost six years just like that.

If the Fed had announced a PLT (i..e, a commitment to run inflation above 2% as long as needed to hit its PLT, or equivalently, to bring average inflation to target in the medium term), that would have surely worked better. Of course, NGDPLT is even better than that, because the Fed would have reacted much more forcefully to the NGDP downturn in 2008-2009, with a credible commitment to permanent monetary expansion.

I feel like I'm partly just echoing things I've read on this very blog, so I'm sure none of this is news to you. Maybe my real question is, did monetary policy go rudderless in Canada in 1996 and/or 2008 like it did in the US in 2008? Because if the answers are "no" and "yes", then I think we understand why Canada had a recession in 2008 but not 1996.

-Ken

Kenneth Duda
Menlo Park, CA

[Michael Margolis]

In an RBC model, we think of a shock as a permanent shift to the production function. A negative shock is represented as a fall the A_t term that multiplies the Cobb-Douglas function of labor and capital. The loss of a foreign trade partner is certainly such a shift (although in 2008 it probably came with an expectation things would be restored). I can see how reduction in government activity could be such a shift, but not at all that it must be. Canada's government, unlike a foreign trade partner, is using the same factors of production desired by private industry, so if it is not using them well the retrenchment would cause a positive supply shock, not negative. The demand shock, of course, would still be negative.

I think the need to reallocate is, in this framework, just where working capital happens to go for a year or two -- it is investment, no problem. I don't wholly buy that in general, but I think it is about right in those cases where it is fairly easy to know which parts of the private sector are ripe for expansion. Perhaps that was so.

[Nick Rowe]

JKH: The stock market did indeed behave very differently, with a steady rise from 1993 to 1998.

https://ca.finance.yahoo.com/echarts?s=^GSPTSE#symbol=^GSPTSE;range=my

That is probably an important fact for this question. But it also raises the question why the stock market responded so differently in the two cases, and whether the different response is simply a consequence, rather than a cause, of the fact there was a recession in one case and none in the other. (Plus, the absence of recession in the 1987 stock market crash suggests the two don't always go together.)

MF: I think you are right, in that IT and NGDPLT will (generally lead to real shocks (that require relative price changes) having different effects. But both 1996 and 2008 could be seen as real shocks. So I wonder what the difference is between those two real shocks that lead to IT having very different effects?

Majro: suppose, counterfactually, the Bank of Canada had held the exchange rate fixed in the mid-1990s. The same fiscal shock would then presumably have caused both a recession and inflation to fall below target. But if 2008 was purely a demand shock, that the Bank of Canada was either unable or unwilling to offset, why didn't inflation (especially core inflation) fall (relative to trend)?

Ken: A big difference between Canada and the US in 2008 was that Canadian monetary policy was not rudderless. The 2% inflation target had been in place since 1994, and the Bank of Canada kept repeating its commitment to that 2% target, and (by and large) hit that target, so it was a reasonably credible target. Now headline inflation did fluctuate a lot around 2008, but it was as much above as below target, and core inflation kept close to trend. Graphs here.

[Nick Rowe]

Now Typepad is putting *me* in spam, as well as Michael! The cheek of it!

[Kenneth Duda]

Nick, thanks for the response. I had missed your 2013 post, and I had not realized Canadian inflation was on track all through the 2008 NGDP plunge. Fascinating.

Does the Canadian equivalent of the CBO estimate an output gap for Canada? Do you think there was an output gap in 2008? If there was, then the puzzle is what can cause prices to rise in an environment where NGDP is falling and there is lots of spare capacity. I always assumed the mechanism of CPI inflation was an excess of money chasing a scarcity of supply, buying up inventories of finished goods. If there was excess aggregate supply, then inventories would remain high and prices couldn't rise.

On the other hand, if there wasn't an output gap, then you'd have to believe there was a negative supply shock in 2008 that more-or-less exactly offset the negative demand shock. Seems far fetched. But I'm curious what you think. Is that possible?

-Ken

P.S. Sorry for my crude AS/AD model. I'm not enough of an economist to express these thoughts in terms of the A_t multiplier on the Cobb-Douglas function or whatever.

[Nick Rowe]

Ken: employment and RGDP fell, and the unemployment rate went up by about 2.5%. So by any normal measure, it was a standard recession, with above-normal spare capacity, though not an especially severe recession.

http://www5.statcan.gc.ca/cansim/a47

But if you define an output gap as "a difference between actual output and the level of output at which inflation remains on target", then there wasn't really an output gap. The Bank of Canada talks about "output gaps", but its real-time measures are prone to big revisions. It can come down to just trying to fit a smoothed curve through the data. It's a horribly arbitrary exercise. Output gaps are in the eye of the beholder. But I would say there definitely was one, despite the failure of inflation to fall.

[Kenneth Duda]

Maybe the difference is that in 1996, the shock was local to Canada, so foreign demand replaced government demand, whereas the 2008 shock was a global AD shock so trading partners were no help in maintaining AD?

This would be an easy hypothesis to check against the data. Did net exports increase in 1996 to offset the fall in G?

If so, at least the difference in NGDP response to the demand shock is understood. But I agree, how the Canadian economy managed to maintain steady 2% CPI inflation in the face of plunging NGDP does seem like a very important mystery.

Any chance the CPI data is misleading? Have you looked at breaking it out into components? In the US, it seems like even the pathetically low inflation we have is mostly due to supply constraints in housing, if you believe Kevin Erdmann (http://idiosyncraticwhisk.blogspot.com/2015/01/the-complicated-role-of-homeowners-in.html --- see third graph in particular). If an effect like this was going on in Canada, except larger, it could explain everything --- perhaps there was, in fact, a simultaneous supply shock (leading some prices to increase) and demand shock (causing spending to plunge), and, stir in a little downward nominal price rigidity, and you get 2% inflation and a recession.

-Ken

[Tom Brown]

Nick, I have nothing intelligent to say about inflation targeting vs NGDPLT. However, I notice something you wrote: you said you used to support inflation targeting because it seemed to work in practice (respectable application of inductive logic there I'd say). And you abandoned it when it failed (again, this seems reasonable). However, can you say "it seems to work in practice" for your support of NGDPLT? For all I know NGDPLT is going to be the best thing since the wheel, but at this point in history is it really anything beyond an educated hypothesis? I don't know the answer.

[Tom Brown]

"Now Typepad is putting *me* in spam, as well as Michael! The cheek of it!"

Typepad sounds like a moron.

[Majromax]

> Majro: suppose, counterfactually, the Bank of Canada had held the exchange rate fixed in the mid-1990s. The same fiscal shock would then presumably have caused both a recession and inflation to fall below target. But if 2008 was purely a demand shock, that the Bank of Canada was either unable or unwilling to offset, why didn't inflation (especially core inflation) fall (relative to trend)?

Maybe our language of supply and demand shocks isn't flexible enough.

In both cases, we envision a shift to one or the other curve, such that there is a well-defined (new) market-clearing equilibrium that we would move towards with greater or lesser ease.

However, since the US was suffering from an internal aggregate demand shortfall, the magic export market wasn't clearing. The terms of trade remained fixed, but we simply couldn't sell enough wood pulp.

The better comparison would not be a shift in demand, but in an imposed quota. In this case, demand for exports collapses but demand for imports remains unsatisfied. The inside-Canada residual looks like a preference-composition shift rather than an unadulterated shock. (That is, the Canadian economy suddenly shifted from demanding 1 apple and 1 banana of its producers to 1.5 apples and 0.5 bananas.)

I think, from my untrained point of view, that can match the observation. With this kind of change in preference (without any time-shifting, note), demand for money remains the same to leading order, the effect on prices is ambiguous (with increases in some sectors and decreases in others), and output falls.

[John Handley]

Perhaps Canada was subject to a different set of frictions during each shock. If, for example, sticky prices were the primary friction in 1996, then inflation targeting would have (in theory) worked perfectly as it apparently did. Maybe Canada in 2008 was substantially different or the shock in 2008 had more to do with a different kind of friction in the Canadian economy where a constant rate of inflation did not mean a perpetual output gap of zero. I guess this means that NK models lack the kind of friction and/or shock that affected (or is it effected here?) Canada in 2008. I just wonder what frictions (besides other prices being stickier than the targeted price index) would make an NGDPLT optimal. Why not a PLT? Why not stick with inflation targeting in the confidence that global financial crises do not happen exceedingly often? It seems clear that economists need to find a model that explains the financial crisis before they switch to a new target, but I'm relatively new here so maybe that's not how it works.

[Majromax]

> It seems clear that economists need to find a model that explains the financial crisis before they switch to a new target

The market monetarist view of the financial crisis is that (for the US) the Fed inappropriately reacted to the early stages of the housing collapse by tightening monetary policy, both by increasing its headline rate and by "passive tightening" of doing nothing while aggregate demand expectations fell. That caused a downturn in one industry (housing) into a liquidity crisis relevant for the full financial sector.

[Chris J]

Nick, thanks. Agree about the "austerity" "working".

[Nick Rowe]

Ken: "Any chance the CPI data is misleading? Have you looked at breaking it out into components?"

There are umpteen different ways to calculate inflation. No doubt there are some measures that fell below trend. But unless we have some particular theoretical justification for choosing one measure, it's just ex post cherry-picking.

Tom: "For all I know NGDPLT is going to be the best thing since the wheel, but at this point in history is it really anything beyond an educated hypothesis?"

"Educated hypotheses" is usually about the best economists can do, when it comes to giving policy advice. I'm just trying to get us a tiny bit more educated, because if we can figure out maybe why IT failed in one case but not in another, that might help us understand whether NGDPLT might be a bit more robust.

John: " Why not a PLT?"

If you look at the CPI data during the IT period, you cannot distinguish it from PLT. (Well, actually the data are more consistent with PLT than IT !) So it looks like (de facto) PLT failed in 2008 too.

[Marko]

Shouldn't we be asking the same question about the U.S. ? :

https://research.stlouisfed.org/fred2/graph/?g=1EuW

Canada had a bit of a jolt in 95-96 , but overall things were quite similar , fiscally speaking , in the U.S. at the time.

I think you can chalk most of it up to the dotcom boom / peace dividend in the U.S.

Ah , yes , those were the good ol' days.

[Lorenzo from Oz]

I don't have an answer as such, but isn't the question: "why did the 1996 fiscal tightening not adversely affect income expectations but the 2008 shock did?". Which comes down to "why did the 1996 fiscal tightening not adversely affect income expectations?". Perhaps because it did not increase demand to hold money, so monetary policy focused on inflation targeting did not have perverse consequences?

[Nick Rowe]

Chris: maybe because I'm an old Brit, the word "austerity" to me refers to e.g. wartime (WW2) austerity, which really was austerity, but was accompanied by a very large budget deficit (the government borrowed to pay for military spending, of course). Austerity continued after the war ended, but the deficit shrunk massively.

The mid-90's change in fiscal policy succeeded in its goal of reducing debt/GDP ratio. (And you could easily argue that the low and falling debt/GDP ratio in 2008 was one of the reasons the Canadian government, unlike others around the world, was able to loosen fiscal policy in 2009.)

Marko: Hmm. Fair point. The change in Canada's fiscal policy was a little bit bigger, and steeper, (and more permanent), than the US, but yes, they are more similar than I had thought. But note that Canadian interest rates fell below US interest rates, plus the Canada/US exchange rate fell, which suggests a bigger IS shock in Canada than the US, and so greater need for monetary offset.

Lorenzo: Suppose some shock causes the demand for money to increase. If the Bank of Canada fails to increase the supply of money by the same amount, we would normally expect both a fall in NGDP and a fall in inflation. In 2008/9, we got the fall in NGDP, but not the fall in inflation.

[Mark]

I think IT targeting failed in 2008 b/c the BoC failed to achieve its inflation target. In 2009 inflation was 0.3%, way outside its 'operational guide'.

The overnight fell from 3% in August 2008 to 1.5% in December 2008. They could have cut interest rates faster and provided more stimulus and economic support.

Additionally, they failed on the wrong side of the target. If anything, the BoC should take some chances and try to overshoot inflation during a recession, just in case. In this case, the BoC was no different than most other CBs. Of course, that's why nGDP targeting is more helpful.

Side note, when talking about 1996 and 2008 I wouldn't feel comfortable pointing to a chart showing the BoC is on target most of the time.

[Mark]

"in 2008 total inflation fluctuated above and below target, but core inflation stayed close to trend".

You're right, in 2008 and 2009 core grew around 1.7%-1.8%. But (afaik) overall CPI is the target and the big miss on overall meant core should have risen by much more than it actually did to support growth. If enough stimulus had taken place to bring core up to 2.5%ish I think overall would have been within the operation guide. Would have probably needed less core growth cause that stimulus would have cause more import price increases imo.

In 1996-97, there was far less difference between core and overall.

[Nick Rowe]

Mark: in the longer term, 2% total inflation is the target. But the Bank calls core inflation its "operational guide", which I interpret to mean it tries to keep core growing smoothly at a trend compatible with 2% total (which, recently, has meant a little less than 2%). Deviations between core and total are mostly due to gas prices (and changes in the GST (VAT)). The Bank's view is that trying to stabilise total inflation would mean causing a big recession whenever the relative price of gas shoots up, to force other prices down.

[baconbacon]

Nick-

It seems you are getting close to repudiating the long run neutrality of money. If inflation targeting didn't prevent a recession but NGDP targeting would have you have only the option of NGDP targeting impacting the real economy in the short term. However, because the BOC was explicitly targeting inflation, not NGDP, you are functionally removing the expectations channel for inflation targeting having a real impact on short term RGDP. With no real short term impact from inflation expectations how do you get the long term impact needed to make money neutral in the long term?

*you can revive it by having the private sector WAY WAY WAY ahead of the economics profession and having them already understand that NGDP targeting is far superior and try to backdoor expectations again, but that is not something that is easily supportable.

[Nick Rowe]

bacon: ?

I *think* that in an alternate world, where the Bank of Canada had been targeting NGDP, and people *expected* the Bank to target NGDP, things would have been better than in the actual world.

I am certainly denying the short-run neutrality of money. As we look across possible worlds with different monetary policies, the variance of M, and the covariance of M wrt various shocks, has real effects on the variances and covariances (and probably means) of real variables. But the level of M has no real effects.

[baconbacon]

Nick- I am trying to reword here what I mean more carefully, without rambling, hopefully i will get back to that.

Can you answer this hypothetical- What if the BoC had secretly adopted NGDP targeting shortly before the recession. Would the outcome have been better or worse than what happened under inflation targeting?

[Mark]

Nick-
Point taken on core as being the operational guide. Nonetheless, overall CPI (according to their site) is their mandate and the point stands the Bank failed to keep inflation on, or close to, target in 2009 despite conventional capacity to add stimulus.

My (limited) sense is that the Bank has history of reacting to above-target overall inflation despite contained core (2000, 2003 auto premiums...) but treating below-target overall as transitory (2012-14). Sure, on average, over the last 25 years they've done a great job of keeping it around 2% on average but during the first big downturn since IT the Bank failed b/c they didn't properly react. To do so would have required negative interest rates by December 2008, probably QE soon after, and faster hikes after. My ultimate takeaway though is that IT fails b/c policy makers can't act aggressively enough, and probably our financial system couldn't handle that volatility...

[Kenneth Duda]

Nick, I was not arguing that CPI is the wrong inflation metric. I was arguing that it's possible for IT to fail miserably because whatever inflation metric you choose is right at 2% target, in the midst of plunging NGDP (AD shortfall), where the CPI is at 2% because of a combination of one component showing supply constraints that are so severe that they are binding even in the demand shortfall, whereas other prices are flat because of downward nominal rigidity. The conclusion is that IT is very harmful. It ignores plunging NGDP if the plunge occurs along with a supply constraint. What you want is NGDPLT, that reacts swiftly to plunging NGDP by supporting aggregate demand, which will cause the relative price adjustments your economy needs to reach full productivity, which you can't get without the AD support because downward nominal rigidity leaves prices flat (unadjusted in relative terms), with key markets not clearing (e.g., labor market).

Here's Kevin Erdmann's latest for the US:

http://idiosyncraticwhisk.blogspot.com/2015/08/july-2015-inflation-and-our-very-low.html

Note how inflation is close to target. But this is highly deceptive. Supply constraints in housing are giving us way over target shelter inflation, whereas inflation in everything else is basically zero, which is really scary, because along with downward nominal rigidity, it means relative prices aren't able to adjust.

I think this is probably the answer to your mystery. Why was Canadian inflation at 2% in the midst of the biggest AD collapse since the great depression? I bet it's because certain prices were rising due to supply constraints, and others weren't falling due to downward nominal rigidity. This is why we need NGDPLT so badly.

-Ken

[Tom Brown]

Thanks Nick.

""Educated hypotheses" is usually about the best economists can do, when it comes to giving policy advice."

When I think of macro economic science policy advice vs say medical science policy advice, I have a vision in my head of 13th century Europe... the king is deathly sick and his ministers, in a panic, send for Christendom's best physicians. After a careful examination, one recommends blood letting, another ear candling, and a third cupping.

Is that a good analogy for the "state of the art" today in macro or am I too pessimistic by a few centuries? If I'm wrong, why is there so much disagreement? Too many damn self deluded blood letters and ear candlers?

[Nick Rowe]

bacon: "What if the BoC had secretly adopted NGDP targeting shortly before the recession. Would the outcome have been better or worse than what happened under inflation targeting?"

My conjecture: probably a bit better. But people might have become very confused, and the inflation target might have lost credibility, which might possibly have made things worse.

Mark: OK. So IT works provided headline and core inflation stay together, and IT fails when they move differently. Interesting hypothesis. Needs a bit of theory to back it up. Why?

Ken: OK, so IT works provided relative prices don't change much, and fails when they do change much. And there is some sort of theory to back that up, like non-linear (asymmetric) price flexibility. Or the theory is that the prices that do change are not representative of the prices that want to change but can't. This gets back to my old posts about the assumed random flight of the Calvo fairy rigging the New Keynesian model in favour of IT. Or this old post.

[Nick Rowe]

Tom: I think we're a bit better than that, but not much.

Controlled macro experiments wouldn't get past the Research Ethics Board, and would cost trillions. Give me control of 100 countries' central banks for 100 years, plus a coin to flip to decide what monetary policy they follow, and I could make a much more educated guess between IT and NGDPLT.

[Bob Murphy]

Nick you're killing me man. Michael Margolis told you upfront the answer (or at least, a huge step toward it), and you're fighting like a kid trying not to go to bed.

OK, you can think of the Canadian federal government as a foreign trading partner. But, when this particular trading partner buys exports from the rest-of-Canada, it pays for them by borrowing from the banks in the-rest-of-Canada, and then paying these loans back by taking the necessary funds from the-rest-of-Canada at gunpoint in the future.

Surely this is an important part of the story? If you want to dress it up with sticky wages and whatnot, OK, but you seem really unsure as to whether this is even relevant?

[Nick Rowe]

Bob: so spell it out for me. Is this the rest of your story: So the reduction in G caused a reduction in expected future distortionary taxes, which increased investment demand to offset the fall in G? But then why did the interest rate and exchange rate need to fall a lot?

[Patrick]

Couple of thoughts from out here on the concrete steppes:

When you work for a firm with operations and sales in the US and Canada, a Canadian worker can easily loose their job if the US side of business contracts. Lots of work done in Canada results in 'exports' through e.g. emails, shared servers sitting in the US, etc. but that isn't captured by the export numbers because they aren't physical widgets passing through customs. If the US side of the business is decimated, it can show-up as job loses in Canada.

Along the same lines: If one works for a very large multi-national firm, that little corner of the business likely gets funding for from a big pool allocated throughout the business. If the pool shrinks because of an economic disaster somewhere other than Canada, the executives could easily decide to shrink the Canadian business in response because they are strategically allocating resources to e.g. preserve operations they feel have more growth potential in the long run.

[LK Beland]

Ken Duda may have a pont. The GDP deflator tells a slightly different story than CPI inflation.

The 2002-2008 trend was 3% annual deflator growth. It slightly accelerated late 2007 (maybe 2% above trend).
It cratered in 2008 (falling by almost 5 %; thus 8% below the 2002-2008 trend).
The 2009-2012 trend was about 2.5%.
The 2012-2014 trend was about 1.5%.

This mismatch between CPI and the deflator is probably related to the reason CPI-IT failed in 2008.

[Bob Murphy]

Nick,

OK this is the kind of thing where we must be talking past each other, because what I'm saying is really simple, I know that you must get what I'm saying, but from your answers it's not clear that you do. So I'm guessing you think there is a "signature" that is missing from the evidence that my story would produce, if it were correct. Yet I'm failing to see what it is.

So, suppose for years the Canadian government has been running a balanced budget, and it uses $1 billion of its tax revenues each year to buy wheat, which it then sends abroad to impoverished countries who otherwise would not have bought it.

Then, the government changes course, and cuts that line item to $0, and instead retires $1 billion worth of government debt.

We see the price of wheat drop a lot, the price of agricultural land drop a little, and the price of government bonds increase (i.e. interest rates fall).

But there is no recession. Farmers plant other crops to replace the acreage that previously went to wheat.

Would this be a mystery requiring explanation? Do we have to say that the Bank of Canada targeted soybean prices to offset the shock to wheat demand?

[Nick Rowe]

Bob: OK, now tell exactly the same story, only this time it is the foreign demand for Canadian wheat that drops (2008 rather than 1996). Why shouldn't the results be exactly the same? Canadian farmers plant other crops (for domestic consumption) to replace the acreage that previously went to wheat.

[Bob Murphy]

Nick, to exaggerate it, suppose that originally foreigners sent $1 billion worth of ball bearings that (for whatever reason) can't be made in Canada. They are essential for all sorts of machinery to operate.

Now there is an earthquake and all the foreign factories making those ball bearings are destroyed. So the foreign demand for Canadian wheat drops by $1 billion, and now Canadian producers can't run a bunch of their machinery.

You see how in these two exaggerated examples, things might play out differently? In my first scenario, the government originally took $1 billion of wheat ultimately at gunpoint, and removed it from the country with no offsetting benefit. Then it stopped.

In scenario two, foreigners originally added a bunch of critical ball bearings to the Canadian economy in exchange for some of its wheat. Then they stopped.

Scenario A was a good change, Scenario B was a bad change.

[Nick Rowe]

Bob: Maybe this will help stop us talking at cross purposes:

Assume initially a perfectly coordinated economy (Robinson Crusoe writ large). There are good shocks that raise utility, and bad shocks that lower utility. 2008 was a bad shock for Canada, in that sense. (You could argue that 1996 was either good or bad, though probably good.)

But just because a shock is "bad", in that sense, doesn't mean it should necessarily cause a coordination failure. (And just because a shock is "good", in that sense, doesn't mean it should necessarily not cause a coordination failure.)

In 1996 we saw no (obvious) coordination failure. In 2008 we saw what looked (to me) like a coordination failure. If the rest of the world (your customers and suppliers) sank beneath the waves, your utility will be lower, even if you respond perfectly. But a coordinated response would quite possibly see an increase in employment. "We are poorer dammit, so we had better work harder!". We wouldn't see an increase in unemployment, and excess supplies cropping up all over the place.

The 2008 shock caused monetary disequilibrium, and the 1996 shock did not cause monetary disequilibrium. There must be something about the Bank of Canada's monetary policy (IT) that created that different response to the two shocks. But what was it?

[Nick Rowe]

In 1996, keeping inflation on target was sufficient (to allow the interest rate and exchange rate to fall by a sufficient amount) to prevent an excess supply of goods and labour and prevent a coordination problem. In 2008, keeping inflation on target was not sufficient to prevent an excess supply of goods and labour or prevent a coordination problem. Why?

[Majromax]

> In 2008, keeping inflation on target was not sufficient to prevent an excess supply of goods and labour or prevent a coordination problem. Why?

I think it's international trade.

Look at the balance of payments information (CANSIM 376-0101). Between 1991 and 2000, current account receipts increased at a steady rate. Current account payments fluctuated such that sometimes there was a current account surplus and sometimes a deficit, but on the balance there was an approximate surplus.

Between 2008 and 2009, however, Canada's current account receipts fell by $136bn, from $648bn to $512bn. This is the difference between 2008 and 1996.

A year-over-year decline in current account receipts for Canada seems to be diagnostic of recession. This happened in 1990-1991 and again over 2000-2003 (gradual), in addition to 2008-2009.

[Frank Restly]

Nick / Bob,

"to allow the interest rate and exchange rate to fall by a sufficient amount"

If exchange rates are primarily a function of interest rate differentials (not absolute interest rate), then can a single central bank prevent a coordination failure? Also, does a zero bound problem for one central bank create an interest rate differential (and exchange rate) problem for other central banks? What did the interest rate differential between the U. S. and Canada look like in 1996? What did it look like in 2008?

All questions, no answers, but perhaps a path to finding some answers.

[Nick Rowe]

Bob (and others): I decided to expand my response to Bob into a post. Let's switch this discussion there.

[Patrick]

"In 2008, keeping inflation on target was not sufficient to prevent an excess supply of goods and labour or prevent a coordination problem. Why?"

NAFTA? In 1996, it had only been in force for 2 years.

[John Handley]

Majro: "The market monetarist view of the financial crisis is that (for the US) the Fed inappropriately reacted to the early stages of the housing collapse by tightening monetary policy, both by increasing its headline rate and by "passive tightening" of doing nothing while aggregate demand expectations fell. That caused a downturn in one industry (housing) into a liquidity crisis relevant for the full financial sector."

I know what the market monetarist view of the financial crisis is. That does not mean they have a model. If anyone know of a market monetarist model that's not just a slightly modified version of a new keynesian model with an extra sticky input price stuck somewhere, then I think it could be reasonable to support an NGDPLT.

[John Handley]

Nick:

I didn't really mean to ask specifically why not a PLT, I was just trying to point out that I think NGDPLT seems to be arbitrary without having a specific model behind it. IT and PLT have NK theory, the Friedman rule has frictionless models where there is an opportunity cost to holding money, what model or class of models says that an NGDPLT is optimal and/or significantly better than IT or PLT?

[Nick Rowe]

John: "...what model or class of models says that an NGDPLT is optimal and/or significantly better than IT or PLT? "

Here is my tentative sketch of an answer

But NGDPLT will never be exactly optimal, unless you rig all the parameter values to make it exactly optimal.

[Nathan W]

In 1996 there was no recession because global growth (especially USA) was strong.

In 2008 there was a recession because global growth (especially USA) was weak/negative.