The fiscal stimulus may well have been ineffective. But that doesn't mean that it wasn't a good idea.

A flurry of punditry has been provoked by a Fraser Institute thingy claiming that the federal government's fiscal stimulus program has had essentially no effect on the Canadian economy. (Question: why do they insist so much on the fact that they use numbers from Statistics Canada? Where do they usually get their national accounts data? The previous week's Lotto 6/49?)

This claim may well be correct. But it will take a lot more work than what I've seen there or in the reactions to that piece before I'll be convinced either way. And even if the fiscal stimulus was ineffective, it's not clear to me that it was a mistake.

Here is an incomplete list of reasons why we would expect fiscal policy to be less-than-very-effective:

1. Crowding-out. Increased government spending/borrowing pushes up interest rates, which reduces private investment expenditures.
2. Import leakages. The increase in expenditures may go to foreign suppliers, not domestic producers.
3. Timing. By the time some projects start spending money, the recession will be over.
4. Political interference. 'Nuff said.

Now, let's get in the wayback machine and head back to March 2009. The most recent data from February had shown that employment had fallen by 2% in the space of four months - a pace we had never seen in previous recessions. The Bank of Canada's overnight target was at 0.50%, and was set to go to its current level at the next announcement in April. The prospect of hitting the liquidity trap was very, very real.

Entering the liquidity trap would have removed the first item from our list of objections to fiscal policy. In a liquidity trap, interest rates are stuck at zero, and so there is no crowding-out.

Import leakage was still a potential problem, which explained all those meetings of various governments in which everyone promised to participate in a coordinated program of fiscal stimulus. It could well be that Canada could have said "We're all right, Jack" and refused to join in, secure in the knowledge that we would benefit from the fiscal stimulus leakages from other countries. But that sounds to me like a very risky strategy. We may, at some point in our future, benefit from a coordinated response to an economic crisis.

In March of 2009, the case for fiscal stimulus seemed strong enough to me that I was willing to set aside my reservations associated with points 3) and 4). And I was even open to arguments based on using fiscal policy as an instrument for convincing Canadians that we should not panic, that we had the fiscal resources to deal with the recession, and that we should keep calm and carry on.

One year later, it seems at least plausible that the fiscal stimulus wasn't necessary. Although we came very close to the liquidity trap, we didn't actually fall in: the recovery has been driven by interest-sensitive sectors.

Of course, that's not really how we should be evaluating the fiscal stimulus program. In order to do so, you have to construct a structural model that corresponds with what we observed, and then run it in an alternate scenario in which the fiscal stimulus package didn't occur. I'd be particularly interested in how the whole defying-the-rest-of-the-world-and-refusing-to-join-in scenario would have played out.

I would not be surprised if such an exercise shows that the fiscal stimulus was ineffective. But that doesn't mean it wasn't a good idea at the time.

Comments

[Christopher Hylarides]

I think we (Canada) were in no danger of a liquidity trap. Virtually every shock to the Canadian economy during this recession is/was due to external factors. Some of the pre-recession domestic expansion may have been based a bit on the american bubble, but in most of the non-export oriented sectors of the economy it was almost as if the recession didn't happen. And now as inventories get low companies are hiring again. This is even with the high Canadian dollar.

I remain sceptical of stimulus' effect. I feel that either way time would have healed the economy, so long as there wasn't a massive monetary contraction.

[Nick Rowe]

I feel roughly the same way. We didn't know how bad things were going to get. They could have turned out worse. Fiscal policy as an insurance seemed a reasonable argument.

I haven't gathered up the strength to wade into the Fraser Institute argument. But there's something I don't understand: how the hell can anyone tell whether or not Canadian fiscal policy worked by looking at Canadian data? There's only one data point? Zero degrees of freedom.

If we had 2 identical Canadas, and applied fiscal policy to one, and not to the second, then we could tell if fiscal policy made a difference.

If we had 100 non-identical countries, tossed a coin to decide which 50 got fiscal policy and which 50 didn't, then we could tell if fiscal policy made a difference. (Or maybe 150 countries, and sacrifice a goat in the third group as a placebo.)

But what the hell can anyone tell from looking at one country?

(That's why I did my "Canada, Australia, New Zealand" post. Though, of course, a sample of three is not a lot better than a sample of one.)

[Nick Rowe]

Using a structural model could answer the question, as you say. But then a structural model is really just imposing your priors on the data. The structural model itself will tell you whether fiscal policy works. It pre-judges the whole question.

Arnold Kling said something similar recently, in an important but very nihilistic and depressing post, that got ignored.

[Patrick]

It's sorta like a sky diver deploying her parachute and, upon discovering that she would have landed in nice soft snow covered old growth pine trees, cursing the uselessness of the parachute.

[Just visiting from Maclean]

Q: In a structural model, how would you treat infrastructure spending? Would you assume it was simply a time shift (ie gets spent in years 2&3 instead of say yrs 4&5) or is it incremental spending that would otherwise not be made if not for the recession?

If it was the first situation, would it not necessarily still be better overall assuming the recovery is more tepid in yrs 2 &3 relative to yrs 4&5?

[brendon]

Rule number one is always "consider the source". I have a problem believing most of the research that comes from the Fraser Institute.

It is also important to remember that the infrastructure portion of the stimulus hasn't fully hit yet. What Fraser is measuring is the impact of tax breaks and industry support. We know the latter is probably a bad idea but may have prevented some significant job losses.

It might be interesting to conduct a similar VAR analysis for Canada as what Romer did for the estimates of stimulus impact on GDP and employment in the US.

[Kosta]

Maybe I should look into their methodology, but I find their numbers hard to believe. Quoting their press release government stimulus packages, including the federal government’s $47.2 billion Economic Action Plan, contributed little to Canada’s economic turnaround in 2009 ... The study found that government spending and infrastructure investment contributed only 0.2 percentage points to the change in GDP growth between the second and third quarter of 2009 and nothing between the third and fourth quarter.

Looking at their numbers, and assuming that only 25% of the stimulus was spent in the two quarters they report number for, you get a multiplier for gov't spending of around 0.1 (or is it 0.025?). What, did the gov't just bury the money and then forget to tell everyone where they buried the loot? The numbers just don't make sense.

[Mike Moffatt]

"In order to do so, you have to construct a structural model that corresponds with what we observed, and then run it in an alternate scenario in which the fiscal stimulus package didn't occur."

If you're arguing point 3 (timing), you are essentially arguing that there was no fiscal stimulus package to speak of. How do you compare two virtually identical scenarios? (One where there was no fiscal stimulus at all vs. one where there was no fiscal stimulus during the recession). Unless you are modeling what the effect of money spent in late 2010 would have on the economy in 1st quarter 2009.

[Nick Rowe]

Mike: " Unless you are modeling what the effect of money spent in late 2010 would have on the economy in 1st quarter 2009."

I'm not sure if you were being serious there, but I think it could very well have had an effect. People and firms spend on C and I (or don't cut their spending as much as they would have) in anticipation of future demand. Lenders don't foreclose on people and firms in anticipation of future jobs and orders. Stock prices rise in anticipation of future demand. Real interest rates fall in anticipation of higher expected future prices. Etc.

[Just visiting from Macleans]

(One where there was no fiscal stimulus at all vs. one where there was no fiscal stimulus during the recession).

Back in the day when I was preparing AFEs (authorization for expenditures) depending upon the nature of a project - you could budget anywhere from say 5%-20% on upfront engineering costs (depending upon the complexity of the project and its scope) on capital projects in the resource sector. Infrastructure projects are probably similar (low end for a relatively simple paving job, on up for say replacement of a bridge or whatever). The AFE preparation and approval process preceeded going to bid, and followed the preparation of a feasibility study/business case preparation and approval.

So, when the bottom falls out of a commodity market (or a general downturn in the economy) short term reacting management (in the private sector at least) often cut back on the upfront work first. The SNC Lavalins and other EPCM (engineering construction and project management) firms are used to booms and busts, and staff up/down as required.

So, accelarating an infrastructure spending timeframe (if that is what the stimulus program was doing) has benefits beyond the technical recession timeframe where only a portion of the upfront funds will be spent. It keeps engineering/technical teams intact (a key consideration for productivity), and due to the lag time required for "shovel ready" projects, helps to smooth out the boom/bust nature of the construction industry. And can take advantage of lower costs in low demand periods.

[Mike Moffatt]

"I'm not sure if you were being serious there, but I think it could very well have had an effect."

I was being serious. I agree it might have had an effect. But how do you model it and empirically test it?

[Mike Moffatt]

RE: "I'm not sure if you were being serious there, but I think it could very well have had an effect. People and firms spend on C and I (or don't cut their spending as much as they would have) in anticipation of future demand. Lenders don't foreclose on people and firms in anticipation of future jobs and orders. Stock prices rise in anticipation of future demand. Real interest rates fall in anticipation of higher expected future prices. Etc."

People spend/invest today because they believe things will get better in the near future thanks to the stimulus. That makes some sense to me - the stimulus is acting as a placebo effect. In which case, we should try to find a far cheaper placebo.