Worthwhile Canadian Initiative

Just who is supposed to build the new infrastructure?

It would appear that the issue of whether or not the federal government should be increasing expenditures on infrastructure could determine who will form the government in the next few weeks.

But just who is supposed to build this new infrastructure? As I noted earlier, unemployment in the construction sector is at an all-time low. And in the comments in this post, Alex Plante makes an important contribution to this debate:

I'm a civil engineer in Montreal. I design road and municipal infrastructure projects for a large diversified engineering firm. Montreal is so far still in the midst of the construction boom that began in 2000. Several recent tenders our firm prepared for infrastructure projects went no bid or had very few bidders, which is something none of the most senior engineers had seen in over 30 years experience. Construction firms tell us they have too much work.

in an e-mail, he adds:

A problem that may or may not be unique to Quebec is that construction collapsed for 10 years in the 90's.  New civil engineering graduates could not find jobs, and by the late 90's enrollment in both university and junior college level programs shrunk drastically.
As a result, there is a missing generation of civil engineers, technicians and skilled construction workers in their 30's and early 40's.  I've noticed that the ages of my co-workers seem to cluster into 3 groups: a junior group in their mid to late 20's, a group in their late forties - early 50's (who hold all the management positions), and a small group in their early 60's.

I'd be surprised to learn that this situation is in fact unique to Quebec.

So if new infrastructure projects are promised, who will build them?

Does Canada need a Fiscal Stimulus? - Discuss

My own view, for what it's worth, is that Canada does not need a fiscal stimulus right now, but there's a possibility it might need one soon, so the government should get everything ready to introduce one as quickly as possible, but not pull the trigger quite yet. My preferred fiscal stimulus would be a temporary GST cut, and/or an increase in government investment spending, especially construction spending. (I would give a different answer for the US, UK, or Euroland, but that's not the question.)

Here is why I believe that.

On the timing, size and form of an eventual fiscal stimulus

The federal government released its economic update yesterday. Some things were silly: the nickel-and-dime budget cuts are at best pointless, and are much more likely to be counter-productive. One of the benefits of getting public finances under control was supposed to be that we wouldn't have to play these games. And the proposal to cut the subsidies to federal parties is clearly an up-and-in fastball designed to intimidate the opposition.

A noteworthy feature of the fiscal update was the lack of a proposal for an immediate fiscal stimulus. This has provoked shrieks of outrage and incomprehension (an example from the usually-sensible Alain Dubuc), but it's not at all clear why anyone would insist that Jim Flaherty should be starting an 11-figure spending spree right away.

This is going to be a long post, and the rest starts below the fold.

Monetary policy and durable goods deflation

In a recent post, Nick made an important point about the story we tell when we explain why deflation is a bad thing. If prices are falling, people will delay purchases, which reduces aggregate demand, which induces more rapid deflation, which reduces aggregate demand even more as the economy circles the drain. But to make this all work, you have to be talking about goods for which intertemporal substitution is possible: it's hard to see how that story would be applied to (say) the demand for milk.

So it's worth taking a closer look at the recent behaviour of the prices of consumer durables. In the US, I'm using the durable goods deflator from the monthly personal consumption expenditure series, and for Canada, I'm using the durable goods component of the CPI. (Canada doesn't have the monthly equivalent of the PCE, and I couldn't find a durable goods component for the US CPI.)

Ordinarily, the monetary authority is able to manipulate the real interest rate for durable goods to accelerate or delay expenditures on consumer durables. But the effectiveness of monetary policy becomes problematic when the rate of deflation exceeds the nominal interest rate. Even if the nominal rate is driven all the way to zero, the real interest rate will still remain positive - and there's nothing the central bank can do about it.

We're going to start with the US. Here is a graph of the effective federal funds rate and the rate of durable goods deflation:

The first thing to notice is that the prices of durable goods have been falling for more than 10 years. And according to the October data, the Federal Reserve is no longer able to bring the real rate of interest for durables below zero: deflation is now a real and pressing issue. But what's even more striking is what was going on in 2002-2003. There are any number of analysts (myself included) who think that the Fed kept interest rates too low for too long during this period. But there could be a credible case in its defense: the prices of durable goods were falling at an annual rate of 3-4%, so a tighter monetary policy might have tipped the economy into a deflationary spiral. In hindsight, of course, housing prices were not falling along with those of durable goods, and low interest rates resulted in a housing boom, bubble and bust.

Moving on to Canada, here is the overnight rate and the rate of durable goods deflation:

I'm not sure just what to make of the recent surge in durable goods deflation, or what its implications are for Canadian monetary policy. Much of it is certainly due to the rapid appreciation of the CAD, and the recent depreciation will likely bring it back down again over the next few months. To the extent that many (most?) of the durable goods purchased in Canada are imported, it's not clear just how serious of an issue durable goods deflation is for Canadian aggregate demand.

Will Deficit Spending in fact be Money-Financed?

It's easy to say that money-financed increases in government spending should be used as a weapon of last resort, if the central bank runs out of ammunition as nominal interest rates fall to zero. But would government deficits in fact be money-financed?

If the central bank targets the monetary base, the answer is trivial: if the bank wants to increase base money to pay for the deficit (printing money and giving it to the government to spend) then deficits will be money-financed; otherwise they won't. But the Bank of Canada (like others) is not targeting base money; it uses a nominal interest rate instrument to target inflation. So if the Bank of Canada keeps the same target rate of inflation, will increased government deficits in fact be money-financed?

I started thinking about this question after reading Willem Buiter on the subject, but my answer is a bit different http://blogs.ft.com/maverecon/2008/11/monetise-public-debt-and-deficits/

The answer matters for two reasons. First, if deficit spending were bond-financed rather than money-financed, future taxes would have to be higher (or future spending lower), to pay the extra interest on the higher future debt. Anticipating higher taxes in future, some people might cut consumption today in response. This offsets some of the effect of increased government spending (and in the extreme case of Ricardian Equivalence, offsets it completely, so an extra $1 of government spending crowds out $1 of private spending). Second, even ignoring any effect on current spending, a higher future government debt is a Bad Thing (bad for future taxpayers, bad for the government's credit rating, bad for its ability to fight future emergencies, etc.).

In simple models where the central bank uses an interest rate instrument, the stock of money is demand-determined, and the demand for money is determined by three things: the price level (positive relation); real income (positive relation); and the nominal rate of interest (negative relation). The stock of money is, well, a stock. And the flow of government spending is, well, a flow. So if a temporary increase in the flow of government spending has a cumulative cost of (say) $100, we can say that it is 100% money-financed if it causes a permanent increase of $100 in the stock of money.

In normal economic times, if the government increases spending, so aggregate demand increases, the central bank will need to offset that increased aggregate demand completely to keep inflation on target. In a closed economy, the central bank raises the rate of interest to do this. This increase in the nominal rate of interest reduces the demand for money, so the money stock falls. The central bank is forced to sell interest-paying bonds to buy back non-interest-paying money. So the government debt in public hands (i.e. not owned by the bank, because since the bank is owned by the government anyway that doesn't matter) actually increases by more than the $100 the government borrowed to finance its spending. The extent to which increased deficits are money-financed is actually temporarily negative in normal times. But when the temporary increase in government spending comes to an end, the interest rate returns to where it was before, so does the stock of money, and so the government spending was 0% money-financed, 100% bond-financed.

In an open economy it's a little more complicated, because the exchange rate appreciates, and so the central bank won't need to raise the interest rate as much to keep inflation on target, but the result is qualitatively the same. (Trust me, or work it out for yourself.)

In normal times, the only effect of government spending on the demand for money is via its effect on the rate of interest. The central bank makes sure it doesn't affect real income or inflation. But in abnormal times, when the central bank has lowered interest rate down to zero, and it's still not enough to prevent a recession, or prevent inflation falling below target, the result is very different.

To keep things simple, assume that a cumulative increased government expenditure of $100 will work very quickly to bring back normal times. And assume that the alternative scenario is to wait for a deus ex machina to bring the economy back to normal in 10 years. Let's focus on money demand after those 10 years have passed, so the economy is back to normal under either scenario.

The price level after 10 years will be higher if the government increased spending than if it waited for the deus ex machina. It's higher because inflation has been at the 2% target for those 10 years, rather than in a deflationary spiral. This will mean money demand is higher. (Note that if the central bank had been targeting the price level, rather than inflation, the price level by definition would be exactly the same under the two scenarios when the economy returned to normal.)

Real income after 10 years will be higher if the government increased spending than if it waited for the deus ex machina. A 10 year recession would have caused lower investment and hence a lower capital stock, so less output and income. This will also mean that money demand is higher. (Note that in an AK growth model real output would never catch up, and this effect would be permanent, but in a Solow growth model it would eventually disappear).

Nominal interest rates after 10 years would probably be about the same under either scenario, so this won't affect money demand. (This assumes the inflation target stays the same, and ignores any effect of lower investment during the recession on real interest rates when the economy gets back to normal.)

So with higher price level and real income, money demand and hence the stock of base money will be higher after 10 years if government spending increases than if we wait for the deus ex machina. So the increased government spending will be, at least in part, permanently money-financed. Could it be 100% money-financed, or even more than 100% money-financed?

Lets do some math. Suppose a cumulative expenditure of 10% of annual GDP is enough to rescue the economy. Suppose the money base is also 10% of GDP. That means the money base would have to double (comparing the two scenarios) to make the government expenditure 100% money-financed. That means money demand would have to double, which could only happen if nominal income were also double after 10 years (comparing the two scenarios). By the Rule of 70, that would require nominal income growth to be 7 percentage points higher for 10 years under the first scenario than under the second. Yep, that's in the right ballpark: 2% inflation vs (say) 3% deflation gives us 5% difference in inflation; and 2% real growth vs (say) 0% real growth gives us 2% difference in real income growth; 5% + 2% = 7% difference in nominal income growth, which would make it 100% money-financed.

So yes, if increased deficit spending were used as a policy of last resort to prevent a deflationary recession, a significant chunk of that spending, perhaps all of it, perhaps even more than all of it, would in fact be money-financed.

A Useful Definition of "Deflation" - the Principle of Charity

Some prices are falling, some are falling quickly, and some are still rising. Whether we've got deflation, and how much deflation we've got, depends on what you include in the index, with what weights, and over what time period. Do we include asset prices for example? It depends on how you define the word. Now you can define "deflation" to mean anything you want, but some definitions might be more useful than others.

We fear deflation because aggregate demand for newly-produced domestic goods and services depends on the real rate of interest. The real rate of interest is the nominal rate minus expected inflation, and since deflation is just negative inflation, the real rate of interest is the nominal rate plus expected deflation. We fear deflation because deflation causes expected deflation, which raises the real rate of interest, which lowers aggregate demand, which causes excess supply and a recession, which causes deflation to increase. If we are close to the lower bound on nominal interest rates, this is something we ought to fear.

So right now, the best definition of "deflation" is the definition which maximises the truth-value of the sentence: "Aggregate demand for newly-produced domestic goods and services depends on nominal interest rates plus expected deflation". In other words, we should choose a definition of deflation which would provide the best fit in an estimated aggregate demand equation. We should define "deflation" using economists' theory plus the philosophers' Principle of Charity

http://en.wikipedia.org/wiki/Principle_of_charity

Let's see how this meta-definition works.

If the price of antique furniture is falling, does that count as deflation? No, because even though people would postpone buying antique furniture if they expect prices to fall, this has no effect on current output and employment, because nobody produces antique furniture. (OK, the services of antique dealers might take a hit.)

If the price of houses is falling, does that count as deflation? Yes, if it's new house prices that are falling, because people will postpone buying a new house if they expect new house prices to fall, which will reduce the demand for newly-produced houses. But old house prices (pre-owned?) don't matter, for the same reason that antique furniture prices don't matter.

If the price of gasoline is falling, does that count as deflation? To the extent that people postpone buying gas if they expect falling gas prices, and to the extent that gasoline is domestically-produced, the answer is yes. But if the intertemporal elasticity of demand for gas is zero the answer is no. And if the gasoline is all imported (and if we hold the exchange rate constant), the answer is again no.

If the price of new cars is falling, does this count as deflation? Yes, if people postpone buying a domestically-produced new car. But perhaps we should also look at the cross elasticity of demand for mechanics, to keep the old cars running longer.

You get the picture. Anyone with an Euler equation, some elasticity estimates, and some estimates of the extent to which the good is currently produced domestically, can keep this list going.

The key point is this: the weights on various goods which make up an index which creates a useful definition of deflation in the current debate over monetary policy with a zero lower bound will be very different from the CPI weights, because the CPI definition is designed to be useful for a very different purpose. (The best CPI weights are those designed to maximise the truth value of the sentence "If nominal income rises at the same rate as inflation, then utility is the same".)

So, with that definition of deflation in mind, do we currently have deflation? Should falling share prices be included? (And I do mean "falling", not "low", because falling share prices will have different effects on consumption and investment demand than low share prices.) What's the time period? I don't have the answers, but I think I have a way to find the answers.

posted by Nick Rowe

Shooting Bears: Ammunition and Central Banks

[ Statistics Canada just announced the seasonally adjusted CPI fell by 0.5% from September to October. And it wasn’t just gas prices. Core was down too (not seasonally adjusted). That’s deflation (on a monthly basis).]

 Metaphors are like models, only with unclear responsibilities. Can the Bank of Canada (or the Fed, if you like) ever run out of ammunition? If so, should it sometimes hold its fire, to save some ammunition for later?

I will answer the second question first, because the answer is clear: if the enemy is deflation, and if there is any danger of running out of ammunition, the Bank of Canada should definitely not try to save ammunition for later. On the contrary, prudence requires it to fire more bullets than it thinks it needs to kill the bear, and to fire them as soon as possible. The answer to the second question is less clear, but the Bank of Canada can run out of ammunition. The only policy of last resort is money-financed fiscal policy.

Suppose, for the sake of argument, that each 25 basis point cut in the Bank of Canada’s overnight rate target is one bullet, so that if it fires nine of those bullets, to bring the target from the current 2.25% to zero, it runs out of ammunition. If it expects inflation to fall below its 2% target, the Bank of Canada will need to cut interest rates – to fire some bullets. If it gets the number of bullets right, and sets the actual interest rate equal to the “right” rate of interest, inflation stays at the 2% target. Suppose the Bank gets it wrong, and does not cut interest rates enough (does not fire enough bullets). With the actual rate of interest above the “right” rate of interest, the economy goes into recession, and inflation continues to fall. The recession causes households and firms to cut spending, which makes the recession worse. The fall in actual inflation causes a fall in expected inflation, which causes a rise in real interest rates (for a given nominal rate), which further cuts spending, and makes the recession worse still. The worsening recession means that the “right” rate of interest falls too, so a bigger cumulative cut in the actual rate of interest will be needed than if the Bank had got it right the first time. (Yep, this is just Wicksell’s cumulative process, plus a multiplier/accelerator and a feed-back loop from actual to expected inflation, and I have avoided the words “natural rate” to avoid needless definitional arguments).

The longer the Bank waits before firing the right number of bullets, the more bullets it needs to fire. The bear is getting bigger every minute it lives, and the number of bullets needed to kill the bear gets bigger as the bear gets bigger. If the hunter is uncertain about how many bullets is the right number, and has a limited number of bullets, he should fire more bullets than he thinks is the right number to kill the bear, and should do so as soon as possible. Ditching the metaphor, if A is the actual nominal rate of interest, and R is the “right” nominal rate of interest (defined as the one at which inflation would equal the target), then if the Bank has preferences defined over the deviation of inflation from target, and hence over the gap (R-A), has lagged and hence imperfect information on R, and if R changes over time proportionately to the gap (R-A) plus serially correlated noise, then any risk that the lower bound on A might become binding in future implies that the Bank should set A below E(R), even if the Bank’s preferences are symmetric. (I am sure that any mathematically-competent economist could prove this formally.)

So, if there is any danger that the Bank might run out of ammunition, it should cut interest rates by more than it thinks is enough, and should do so quickly. It should not save some ammunition for later; prudence dictates just the opposite.

That was the easy part. But can the Bank run out of ammunition?

James Hamilton http://www.econbrowser.com/archives/2008/10/deflation_risk.html says we can always use “quantitative easing” to increase the money supply even if the rate of interest falls to zero, so deflation can always be cured. Paul Krugman http://krugman.blogs.nytimes.com/2008/11/15/macro-policy-in-a-liquidity-trap-wonkish/

says that the money supply increase would need to be permanent to work, and it may be difficult to credibly commit to a permanent increase in the money supply, so deflation is a real danger. To put it another way, if the central bank runs out of interest rate bullets, or if these bullets turn out to be blanks, what else can it use?

First, the Bank of Canada could make a commitment to keep interest rates low in future. If credible, this would reduce long rates of interest, as well as short rates, and have a greater impact on demand for goods. This would help, but only a limited amount. Even a perfectly credible commitment to keep interest rates permanently at zero could at most reduce the nominal long interest rate on safe assets to zero, and this might not be enough. With expected deflation sufficiently high, the real rate of interest on long bonds could still be too high, in which case those expectations of deflation would be confirmed by subsequent experience.

Second, the Bank of Canada could engage in “quantitative easing” by buying safe nominal assets, like government bonds (or CMHC-insured mortgages, which are ultimately insured by the government). Again this would help, but again it might not be enough. It would circumvent the credibility problem of the first option above, but at best it could only bring the nominal long rate of interest down to zero.

Third, the Bank of Canada could engage in quantitative easing by buying risky nominal assets, like uninsured mortgages, commercial paper, or ‘troubled assets’. In principle, with a finite supply of risky nominal assets, the Bank of Canada could push the risk premium on risky assets down to zero, so that both safe and risky bonds would pay zero nominal interest. This would make the expected (or risk-adjusted) nominal rate of interest on risky bonds negative, which would certainly help stimulate risky investment financed by such bonds. But there are three problems: first the sheer magnitude of the open-market-operation would be mind-boggling; second the Bank of Canada would suffer losses; third the supply of risky nominal bonds would not remain finite. It would not take a financial genius to manufacture an infinite supply of lottery tickets, sell them to the Bank of Canada, hold the proceeds in cash, then keep the proceeds when the Bank of Canada loses.

Fourth, the Bank of Canada could engage in quantitative easing by buying real assets, like shares, houses, or farmland. The rent/price ratio on farmland is a real rate of interest, since nominal rents can be expected to rise and fall with inflation and deflation. By buying farmland (and renting it out), the Bank could push up the price of farmland, push down the rent/price ratio, and reduce the real rate of interest on farmland. This would give farmers an incentive to make new farmland, if the price of farmland exceeds the cost of clearing scrub (so Tobin’s Q exceeds one). The same is true if the Bank buys houses, or shares in Canadian firms. By buying real assets, the Bank pushes down the real rate of interest on those assets. This stimulates new investment spending. Yet even this form of monetary policy has limits. If the expected rate of deflation is (say) 5%, people can earn 5% real interest just by holding cash, and so the Bank of Canada would be unable to push the rent/price ratio below 5% without buying all the farmland in Canada. Farmers would offer it all to the Bank of Canada when prices approached 20x rents. The same is true of houses and shares. The Bank would eventually face a perfectly elastic supply of real assets, and might be unable to raise Tobin's Q sufficiently to create enough investment to prevent deflation. If the Bank could credibly  commit to buying all the farmland, and keep on bidding up the price thereafter (unless deflation stopped), that might do the trick. But how could such a commitment be made credible? (Maybe by appointing a Marxist-Leninist as Governor?)

Of course, even if buying real assets worked, exactly the same effect could be achieved by the Bank of Canada printing money and lending it to the government to finance new government investment, or any increase in government expenditure. Or a money-financed tax cut or transfer increase (which is just helicopter money). The difference between government and private farmers is that the government might not care if it would be more profitable to hold cash than to build more sewers.

The Bank of Canada could also buy foreign exchange in an attempt to force the exchange rate down and increase net exports. But in a global deflation, this beggar-my-neighbour policy would not be a good way to make friends. Plus it might not work, if the supply of foreign exchange from foreign central banks were perfectly elastic, as it would be if they resist.

Pure monetary policy does not stop when nominal interest rates hit zero, but the Bank of Canada can nevertheless run out of ammunition, if expected deflation is sufficiently high. But money-financed fiscal policy remains as the weapon of last resort.

And by the way, since the Bank of Canada can indeed run out of ammunition, it would be very unwise to reduce the target rate of inflation below the current 2%. Reducing the target to zero inflation would have meant eight fewer bullets in its magazine. It’s only got nine bullets left right now (plus a couple of weird ones), and will almost certainly need to use some of those nine.

Nick Rowe (until I can figure out how to do "author tags")

Nick Rowe joins Worthwhile Canadian Initiative

Over the past few weeks, Nick Rowe of Carleton University has been sending me some very fine pieces of analysis that I've posted here. This has increased the quality of this blog immensely, so I'm very pleased to announce that he has agreed to join me and post as an author in his own right.

He will start posting just as soon as he figures out how.

Are we seeing the end of a 15-year stock price bubble?

I spent much of the latter part of the 1990's looking at this graph:
 
During this period, Pascal St-Amour and I were trying to explain just why stock prices were so volatile; the fundamentals driving the standard C-CAPM were far too smooth to generate swings like this. Our story was based on a two-state model in which investors alternated between periods of low risk aversion (bull markets) and high risk aversion (bear markets). We made it operational with a latent Markov process and we estimated it using Bayesian MCMC techniques, which were then still relatively new to econometrics. We met with some success: our model could leverage fairly small movements in the atemporal coefficient of relative risk aversion (between 2.20 and 2.65) into swings of about 0.4 in log real asset prices.

(AER version, ungated version)

Every now and again, I've thought about updating this paper, but each time, I've decided against it. Here's why:

There was simply no way that our model could handle post-1995 data.

Which brings me to the question of whether or not there is any model that could. It's hard to imagine a model in which movements in discount rates and/or expectations about payoffs could generate those data.

Why aren't all econometricians Bayesians?

Andrew Gelman offers this wonderful graph:

If you have a simple model and lots of data, then it doesn't matter if you use classical or Bayesian methods. Otherwise, classical methods are more likely to generate more in the way of thorny, pointless digressions ("Hey, wait a minute! Are the data trend- or difference-stationary?") than actual answers.