The preponed government spending multiplier may exceed one

From very casual observation, I get the sense that a lot of extra government spending in the last couple of years was to build stuff that would have been built anyway. They just built it a bit earlier. It was preponed government spending. Am I right?

Anyway, I'm going to assume I'm right, because it makes a really neat macroeconomic thought-experiment, as well as being possibly relevant to Canadian fiscal policy.

I'm going to make two more assumptions to keep the thought-experiment as clean as possible.

1. The real rate of interest on government bonds is zero for the interval between the preponed expenditure and when it would have been spent otherwise. Roughly plausible.

2. The stuff the government builds is mothballed at zero cost until the time it would have been built otherwise. They dig the hole, but don't do anything with it until they need it. They repair the bridge a couple of years earlier than it needed repairing. They build the bridge, but it doesn't go anywhere until they connect it up a couple of years later. Maybe a bit unrealistic, but it makes the thought-experiment cleaner.

What's so neat about this thought-experiment is that we can totally ignore the present value of the future tax liabilities. They are exactly the same whether or not the government expenditure is preponed. It doesn't matter whether Ricardian Equivalence is true or not. What's also neat is that we can totally ignore whether the government expenditure is useful and productive or not. It's going to be done anyway, and the only question is when -- now or later. And preponing the government expenditure is useless by assumption, because it's mothballed during the preponement years. So it can't be a substitute or complement to private consumption or investment during the preponement years.

OK, what are the macroeconomic consequences of preponing the government expenditure? Would aggregate demand be shifted forward in exactly the same way by preponing the same amount of government transfer payments?

If the economy has unemployed resources due to deficient demand, both this year and next, then there is no real benefit from preponing the government spending by one year. We get higher demand, output and employment this year, and lower demand, output and employment next year. Adding the two years together the multiplier is zero.

But if the economy has unemployed resources due to deficient demand this year, but not next year, there are benefits to preponing demand. We get an extra bridge (or whatever) almost for free. The resources that would have been used to build this bridge next year are now freed up to build something else instead. Like a second bridge. And the resources used to build the bridge this year weren't doing much anyway. If the government had spent the same amount on preponing transfer payments instead of preponing the bridge, everything else would be the same, except we wouldn't have whatever it is the freed-up resources would be building next year instead of the bridge. That's a multiplier of one. (The national income accountants can decide among themselves whether to add it to this year's or next year's GDP.)

Actually, it's maybe even better than that. If we prepone the bridge, government expenditure will be lower next year. That means private consumption will be higher next year. (Because government spending crowds out private spending under "full employment"). Assume people know this. The Euler equation for intertemporal optimisation tells us that an increase in expected future consumption will cause an increase in desired current consumption. So preponing the bridge also increases current consumption (and this effect is over and above any effect that preponing an equal amount of transfer payments would have).

Let's do a back of the envelope calculation. Assume future government spending crowds out future consumption spending dollar for dollar. (That's an upper bound, because private investment and net exports will be crowded out too). Assume the Euler equation tells us that desired current consumption increases dollar for dollar with expected future consumption. (That's an upper bound too, with imperfect foresight and borrowing constraints). But, given those assumptions, a $1 preponement of government spending would increase current consumption by $1, and future consumption by $1, giving a multiplier of 2. (We get the bridge either way, only it's this year instead of next year).

Of course, if monetary policy were quick enough and aggressive enough, this would all be academic. But then, I am an academic.

Not sure if any of this is original. Probably someone else has thought about this before. But I think it's right.

Comments

[Trevor]

Doesn't this depend on whether the rate at which assets built by the preponed spending depreciate slower than people's discount rates? I'm willing to believe this to be the case, but am I right?

Also, we might also consider preponed government spending to be of a lower quality. Something about the nature of government expenditure planning might mean a rushed-job costs more today than a careful one next year.

[david]

This looks like inverse RBC, where instead of recessions caused by shifting intertemporal preference toward the future on the part of individuals, we have recession antidotes caused by shifting intertemporal preference toward the present on the part of government. ;)

Provided that the recessionary shock is principally a shift in tastes ("confidence"?), the government can just substitute its own intertemporal plans in the opposite direction, by preferring the present. This being RBC, there is no role for monetary policy to play. We just have entities swapping intertemporal consumption patterns, and it can even be wholly rational on the part of the government (in view of lower costs).

[Adam P]

Nick, this is a really, really good point!

A quibble, you say "if monetary policy were quick enough and aggressive enough". Excuse me?

One clear implication of your story is that the preponement is not inflationary at all, might even be disinflationary, so why do we think monetary policy should change? (Yet another reason not to target NGDP growth perhaps?)

[Determinant]

On the downside you only lose what you spent, and you wanted to spend that anyway because you were in a slump. On the upside if the economy needs that extra capacity and the bridge should be hooked up ahead of schedule, you just have to spend the final 5% to do that and you get a few extra years of productive use out of your asset.

[Ken Houghton]

You're so close it hurts.

In any environment where "crowding out" is a serious issue, the tendency will be to defer G. As such, preponing (great phrase, plan to steal it) will not happen except in cases of major shifts. Instead, I/S projects will be deferred (minor adverse selection, though not enough to impact significantly from "equilibrium") until resources are available.

The result is the same, but the delay in public projects probably shortens the business.cycle as well, due to making new business marginally more difficult to start.

[Frances Woolley]

Nick, great post, especially happy to see one of my favourite words (preponed) on the blog...

[Nick Rowe]

Trevor: you may be right on depreciation. I think I am implicitly assuming zero depreciation while they are mothballed. I'm not sure why or whether people's discount rates would matter. It might matter a bit when I get to the bit about the Euler equation? But the real interest rate is assumed to be zero for the year or two when they are preponed.

Yep on the rush jobs. I'm implicitly assuming there's no change in quality. These are all "shovel-ready" projects.

david: it does look more like RBC. "Build bridges when the rain falls, because the private sector will stop making hay". I expect I'm just putting that on top of a Keynesian deficient demand outcome. Reminds me also of Keynes, telling housewives to stock up on bed sheets during the recession, IIRC!

Adam: Thanks!! I hoped you might like this one. I'm still thinking through your point about whether it's inflationary of deflationary. I expect it depends on the effect of preponement on future monetary policy?

Determinant: my assumption that the bridge is mothballed until next year biases the case against preponement. Because it means we ignore the benefits of one year's use of the bridge. My assumption of zero depreciation while mothballing (and zero costs of mothballing) biases the case the other way. This is how I think of it: there is some optimal time to build the bridge under normal circumstances (ignoring the recession). Small preponements or postpoonements around that optimal time will only have second order of smalls costs.

Ken: Not sure if I understand you there. Are you saying that optimal government policy would be to both postpone projects until there's a recession, and prepone them when there's a recession? Agreed. It should be symmetric timewise.

Frances; Thanks! I think I was reading about Indian English somewhere a few days ago. Then, when I was thinking about this post, no other words would work anywhere near as well. So, the Indians took my ancestors' language, and improved it. Well done Indians!

[Dave]

So it has come to this. It is novel and insightful to demonstrate that counter-cyclical government capital spending is a good idea?

I realise that macroeconomics can be counter-intuitive. But can we no longer trust our instincts on the veracity of the bleedin' obvious?

[RSJ]

Now, I am confused as to whether you are talking about government expenditure, or government consumption of output.

If you are talking about government deficit spending, which is what is relevant for present value of future tax liabilities, then the same argument works by having the government make a transfer payment.

More generally, there cannot be any crowding out due to pulling spending forward during busts and pushing it back during booms, as long as the long run NPV is constant as a result of these shifts. Any shift that leaves long run NPV unchanged would not cause any excess saving behavior.

But the NPV will always be constant as long as the policy is constant, so that you cannot argue that countercyclical deficit spending -- if part of a long-announced policy -- is crowded out when the spending occurs.

And the same argument would apply to cyclical shifts in deficits to a reduced revenues, either due to a reduction in revenues, or due to a policy to reduce tax rates during busts and increase them during booms.

The argument would apply to any policy-based spending of any kind. You are left with the conclusion that fiscal policy is effective at mitigating recessions, but with a cost of possibly dampening booms, provided that it is policy based. Basically a policy version of the Keynesian story, no?

[Nick Rowe]

Dave: you could say it's come full circle. Or you could interpret my result as saying that a cut in expected future government spending will increase demand today. Which certainly wasn't "bleedin' obvious" when I first learned Keynesian macro, in 1971. No, we can't trust our instincts. Our instincts of the veracity of the bleedin' obvious are probably based on the writings of some defunct academic scribbler, as someone once said.

RSJ: you lost me there.

[RSJ]

OK, Suppose there is a boom every even year (we start at zero), and a bust every odd year. The economy is growing at some fixed trend. Wages are also growing at the same trend.

During the bust, 10% of the people lose all income, while those who remain employed keep the trend wage. During the boom, 100% of the people are employed at the trend wage.

What is the rational savings behavior? With log utility, everyone will want to save too much, as they are risk averse. They will save due to fear of future income loss, not future taxes.

Now suppose the government announces an unemployment insurance policy, in which it will pay the going wage to those who are not working during the bust, and it will tax everyone a fixed fraction their (post-benefit) income every year. The fraction is chosen to make the budget balanced (or a constant debt to GDP ratio -- it's not important) over the entire cycle (e.g. both even and odd years).

In this case, the rational savings behavior is to just pay your taxes and not save any of your disposable income out of fear of future tax hikes. Even though you know that the government will run surpluses during booms (as it will be collecting taxes but not paying the unemployment insurance).

In fact, without the policy, people would save more out of fear of future income loss, and the recessions would be much deeper.

[K]

I love this thread. Da gummint is ufishunt!

But seriously... Fantastic post, Nick. And good point, RSJ: consumption smoothing adds additional net positive utility.

[Determinant]

I thought that was one of the long-run features of Keynesian economics. In fact the US, UK and Canada all ran budget surpluses in the late 1940's and early 1950's under this very theory, to dampen the post-war boom to reduce inflation and keep things sustainable, the bust being a very recent memory.

However aside from that short-lived period there have been few instances of other long-run government-drag operations during booms, Canada perhaps having come closest during the Chretien years.

[Mike]

Broken Window Fallacy. The bridge that was prematurely rebuild had its carrying value DESTROYED. The nation is worse off by that amount. So the preponement is not merely shifting resource utilization frm the future to the present, it's a redistribution from all users of the public good to the suppliers of the inputs.

The recession was CAUSED by preponement of durable goods. Resources which would have been used for housing and durable goods were employed in advance of their natural utilization rate. This causes a resource debt that must be burned off by time, unless there is some supernatural growth in demand such as a population or migration boom.

Government policies which bring resources forward are counterproductive. If we knew the end date of a recession, then this could be done without much cost - interest on the debt. But bringing resources to the present reduces future demand for them, particularly with respect to long lived assets. These expensive assets are also usually paid for with debt, so we increase leverage.

The problem is that resource costs increase with utilization, debt costs compound, and you're hoping for a miracle to arrive to solve the problem while you kick the can down the road.

The CAUSE of the recession is not exogenous, so you're not just biding time until it goes away. The cause is ENDOGENOUS. Our governments borrowed too much, creating too much foreign exchange reserves which came back to fuel real estate bubbles. Our citizens and businesses borrowed too much to consume.

Why do egg heads always try to make problems more complicated than they really are? We borrowed resources from the future, and that resource debt must be repaid by leaving them idle until they are actually needed again.

Monetary and fiscal policies are a hair of the dog policy, and they don't work.

[Jim Rootham]

And your evidence for this is?

In particular, compare the Great Depression with the Great Recession and show me how monetary and fiscal policies don't work.

[Nick Rowe]

About to teach my class, but:

Any Canadians out there who can say whether or not they agree with my casual observation/sense/hunch/guess that a lot of government spending over the last couple of years was stuff they would have done anyway -- they just did it a bit earlier? Because I don't know this for a fact, and your sense will be at least as good as mine on this question. Anecdotes wanted, in other words.

[Frances Woolley]

Nick, by spending do you mean spending or do you mean tax cuts e.g. pension income splitting? Would the home renovation tax credit fit the preponed expenditures story? I remember we talked about this credit at the time.

This page http://www.fin.gc.ca/access/budinfo-eng.asp has links to last year's and previous year's budgets if you want to take a look at the spending initiatives. The 2009 budget was the one with all of the infrastructure stuff: http://www.budget.gc.ca/2009/pdf/budget-planbugetaire-eng.pdf.

The 2009 infrastructure spending really fits the preponed expenditure story well, and that came in at $12 billion (projected) in the 2009 budget. The 2011-12 total federal expenditures (projected) are $278.7 billion.

[Phil Koop]

Great post, Nick.

@Dave, Ricardian Equivalence is supposed to be the bleedin' obvious reason why fiscal intervention can't work. The sharp end of Nick's point is that even if Ricardian Equivalence is granted, that conclusion does not necessarily follow. Not so bleedin' obvious.

@Ken Houghton, "prepone" is not a very new neologism. The OED dates it to the '70s: http://oxforddictionaries.com/view/entry/m_en_gb0658300#m_en_gb0658300.

[edeast]

Here is a map of the projects, I had looked through it a while ago, to write a comment on your other ricardian post, because I fundamentally disagreed with your pointless hole digging and filling example. I thought it would be really hard for municipal gov's to come up with pointless work. In my town there was a push on to get the application in to build the new community center/hockey rink which they were going to do anyway. It was preponed.

[Nick Rowe]

RSJ: OK. On average, with government insurance against unemployment, people's desired *stock* of savings will be lower (and in a growing economy the *flow* of savings would be smaller too). The equilibrium real rate of interest would be higher, and investment would be lower.

K: Thanks! Whether governments *in fact* do this though is an open question. They might, in practice, do the exact opposite, and spend more in good times and less in bad times. And this whole post assumes that monetary policy is not or cannot do the job. This post is me in Keynesian mode. Making the best case possible for countercyclical fiscal policy. (Or, maybe it's an ideologically motivated post with an election looming to say "Hey, the Conservatives did a really great job! ;-) ) (Actually, I wrote it because it seemed like a really neat thought-experiment and helped me get my head clearer about fiscal policy. Trying to get one's head clear on neat ideas in economics trumps everything else).

Determinant: it was *supposed* to be one of the main features of Keynesian fiscal policy. Small G in booms, big G in recessions. Didn't always work like that though in practice. Plus, the old Keynesian theory talked about how the increase in current G increased current Y, but there was no explanation of how a cut in expected future G would increase current Y. Plus, there was all the stuff it ignored, like expected future taxes, and whether G would be useful or not.

Mike: If the real rate of interest is zero, and if the bridge can be mothballed at zero cost with zero depreciation, there is no real cost to bringing it forward. (And there can be a benefit, as I have shown). Sure, not all government spending will fit those assumptions.

Frances: I mean spending, not tax cuts. (But a cut in the marginal tax rate on private spending, like home renovation, is somewhere in between). Thanks for the links.

Phil: thanks! Yep, fiscal policy works here even under full REP.

edeast: lovely map. I checked through a few local ones, and they seem like they might fit my assumption.

[RSJ]

Nick,

The point is Barro assumed that incomes were constant, and tax rates were volatile. That is the rationale for saving today in anticipation of higher tax rates tomorrow. He couched it in language such as "concern for future generations", but the key assumption is not concern for the future, but constancy of incomes across time.

But what we see is income volatility, in which case the crowding out argument is significantly undermined. To the degree that tax rates are stable -- and they are much more stable than deficits -- then there is no rationale for saving deficit spending. Only to the degree that they are volatile would there be a desire to save, and even then, you would not save the full amount if the future tax hikes were proportional to the future increase in income, which is also what happens.

For all the effort at including intergenerational transfers, he assumed away the business cycle in its entirety, and somehow this argument is meant to undermine the efficacy of countercyclical spending.

[Dave]

"Ricardian Equivalence is supposed to be the bleedin' obvious reason why fiscal intervention can't work"

OK, I take your point. Nick has done something remarkable by demonstrating the bleedin' obvious in spite of a bleedin' ridiculous assumption like Ricardian equivalence.

I suppose my point really was that countercyclical capital spending is obviously the right thing to do from a microeconomic perspective of maximising the efficiency of government expenditure. However, I can see that this does not necessarily make it right for macroeconomic reasons.

[RSJ]

"OK. On average, with government insurance against unemployment, people's desired *stock* of savings will be lower (and in a growing economy the *flow* of savings would be smaller too). The equilibrium real rate of interest would be higher, and investment would be lower."

Yes, with less risk you lower the demand for savings, so the risk-free rate needs to increase to convince savers to save the same amount. However you also have less risk, so the actual rates charged to investors could be higher or lower, with more or less investment. The risk-free rate rises, but neither savings nor investment needs to decrease. It depends on which effect dominates.

[Mike]

Nick, I have no idea what you mean by the bridge being "mothballed."

You were talking about replacing an existing bridge that has remaining useful life with a new bridge to bring forward some output.

You're not "mothballing" anything - you're destroying a perfectly good asset to justify building another one. It should be obvious that the destruction of an asset does not in any way make society better off. It's a classic example of the Broken Window Fallacy.

The problem is that the way Classical and Keynesian economists account for GDP would never pass muster with GAAP or IFRS. The reason we have accounting rules is to make financial reporting reflect economic reality.

The destruction of a perfectly good bridge reduces the assets of society. Since neither government nor economists maintain a societal balance sheet, we recognize the income from building a new bridge and marvel at the new asset, but don't subtract the loss of the previous asset. It's as silly as claiming we should start a world wa to end a recession, as if the flurry of economic activity justifies the destruction of resources and misuse of resources.

Producing any durable good before it is economically justified will eventually idle at least as many resources as was used to produce it. It's only a matter of time.

You would need miraculous gains in productivity to ever get ahead in that game, because you would then idle fewer resources than you pulled forward. The problem with that argument is that your inefficient consumption and investment will not produce those productivity gains. In fact, you're likely to reduce productivity.

Every iron worker who you keep employed building that unnecessary bridge has a higher valued use somewhere else (as opposed to unemployment). You rob that industry of labor and you rob that worker of the opportunity to gain worthwhile human capital.

You simply cannot get ahead with a continuous stream of inefficient uses of resources any more than you can craft a winning strategy in craps from a combination of wagers, all with a negative expectation.

if the miraculous, exogenous, deus ex machina of productivty enhancement fell from the sky, you would still have been better off by NOT constructing a public good which, on net, is a resource destroyer.

simply put, all these planning schemes are wealth transfers to the owners of the labor and capital for the selected projects at the expense of everyone else. But because the benefits are concentrated and the costs diffuse, there are few complaints. No one steps up and claims ownership of the useful bridge that is being torn down...

except me.

[Nick Rowe]

RSJ: "The point is Barro assumed that incomes were constant, and tax rates were volatile."

The Barror-Ricardo Equivalence Proposition assumes many things. But it does not assume that income is constant. Take any expected path of current and future income and any expected path of current and future taxes. That expected path of taxes will have an expected Present Value. What REP says is that the PV of taxes will not be affected by the choice of whether you financed G by current taxes or future taxes. It follows from the long run government budget constraint PV(G)+current debt=PV(taxes).

Mike: think of it this way. Your window is not broken, but you know it will break tomorrow (the seals are going). You need to work tomorrow, but you don't have any work or anything else you want to do today. So you replace your window today, even though it doesn't need replacing till tomorrow.

Yes, there's a cost to doing this: 1. you have to pay one extra day's interest, and 2. Your new windows will break one day earlier 20 years from now than if you had waited a day to replace them. But if those costs are small, it makes sense to replace your windows one day before they break. My assumptions 1 and 2 assume those costs are zero. "Mothballing" (in this example) means preserving the new windows so they don't suffer any wear and tear during the one day, and so last the extra day 20 years from now.

In my bridge example, there is no existing bridge, but you know you will need a bridge next year, but you build it this year when the resources to build it have a low opportunity cost. And "mothballing" means you don't use the bridge until next year, and the bridge doesn't suffer any depreciation if it's not being used. "Mothballing" is the right metaphor. It's what you do to clothes you are not wearing to stop them depreciating while they are sitting unused in your closet.

[Adam P]

Mike: "you're destroying a perfectly good asset to justify building another one".

No, this doesn't even have a toehold on what Nick is saying. Where do you get the idea anyone is suggesting tearing down a perfectly good bridge to replace it with another?

Nick is entirely clear that he's talking about a project, in this case replacing a bridge, that was anyway going to be done within a couple of years. Bridges generally last something like 100 years I'd imagine so if the governement foresaw a need to replace it within a couple of years then it would clearly not be "perfectly good".

On the other hand, since only a completely irresponsible government waits until the bridge literally falls down before replacing it you can imagine that they have a window of several years in which to replace the bridge where the government is roughly indifferent as to which year work actually takes place in.

Nick's point is that, given this flexibility you can have a positive effect by choosing well the time when you actually do the work (and fund it, recessions tend to mean low governemtn borrowing rates).

[Nick Rowe]

Dave: "I suppose my point really was that countercyclical capital spending is obviously the right thing to do from a microeconomic perspective of maximising the efficiency of government expenditure. However, I can see that this does not necessarily make it right for macroeconomic reasons."

Yep. One of the things I am still trying to get my head clear on is the precise difference (if any) between the microeconomic and macroeconomic Cost Benefit Analysis of preponing capital spending. They might give the same answer to the question of whether or not to prepone the bridge. But I think they would give a different answer on the question on how big the benefits would be if preponing is the right decision. For example, it might be that Macro NPV=k times Macro NPV, where K is the multiplier (and K will be greater than one)?

[Nick Rowe]

Aha! Adam and I were posting at the same time, and we have said the same thing.

Adam: "...you can imagine that they have a[n interval] of several years in which to replace the bridge where the government is roughly indifferent as to which year work actually takes place in."

(I've edited Adam's quote because his using "window" instead of "interval" threw me on first reading, because I thought he was still talking about windows. It's early here).

That's the key assumption. Assume, from the micro perspective, there are slowly increasing marginal costs and slowly decreasing marginal benefits of postponing the date at which you replace the bridge. There is an optimal time t* where the MB curve cuts the MC curve. But small deviations of t from t* (replacing the bridge a little bit early or a little bit late) have second order of smalls microeconomic costs. The gap between MB and MC is small close to t*, and only gets big when we move a long way away from t*. So the macroeconomic benefits of preponing the bridge by a small amount could easily outweigh the microeconomic costs of doing so.

[Nick Rowe]

Draw a graph with a smooth curved hill. The optimal time to change the oil in my car's engine is at the top of the hill, where the net benefits are maximised. I think t* in this case is 10,000 kms. But there's very little net loss in changing it at 9k instead of 10k. There's a bigger loss in changing it at 8k instead of 9k, and an even bigger loss in changing it at 7k instead of 8k. And the same thing for 11k vs 10k, etc., so it's roughly symmetric around 10k. So I don't change it at precisely 10k (t*). I usually change it a bit before or after 10k, on a warm day when I'm not doing anything else.

(This is the same intuition behind the small menu cost argument. A firm that faces a downward-sloping demand curve has a profit function (profit as a function of price) that looks like a smooth rounded hill. The profit-maximising price P* is the price that puts it at the top of the hill. But it is close to indifferent to anywhere near the top of the hill. So small costs of changing prices mean the firm won't move to the top of the hill, if it's not too far away from P*.)

[RSJ]

"The Barror-Ricardo Equivalence Proposition assumes many things. But it does not assume that income is constant."

OK, in Barro's proof, in period n, generation n-1 and generation n are alive, and they split the income so that g(n-1) gets rK(n) - x(n), and g(n) gets x(n) + w(n). x(n) is the intergenerational transfer occurring within the same time period. Whether the income is wage or capital income does not matter. The point is that generation n gives some income to generation n-1 -- in Barro's language, we are at an "interior point".

Now Barro (incorrectly) models taxes as lump sum payments levied on single generation at a single future time rather than as being levied on all generations that are currently alive on a percent of income basis, where all income is taxed, or in a consumption tax model, all consumption is taxed, regardless of the age of the person receiving income or making expenditures.

This unrealistic description of future tax obligations becomes material when income is variable -- when there is a business cycle that occurs within the life of each generation -- although it is not material when the business cycle is smoothed out.

Barro then argues that if generation n had wanted to decrease generation n+1's income, then they would not have made the intergenerational transfer to begin with, so that they will increase their bequests to "undo" the effects of the future tax obligation. That is the basic crowding out argument -- it's amazing how little math is in this paper.

But suppose every even year there is a boom and every odd year there is a bust. And suppose that taxes are paid as a percentage of all income by everyone receiving income who is alive.

Given that both generations pay more in taxes during boom years, and both generations pay less in taxes during busts, and given that each generation lives for 1 boom period and 1 bust period, then each generation pays the same amount of taxes, and yet the government can deficit spend during busts and run surpluses during booms.

Such a policy does not undo any intergenerational transfers -- it has nothing to do with intergenerational transfers at all, but rather inter-temporal transfers that leave total disposable income unchanged for each generation.

The inter-temporal shift will not only improve utility, but it will not affect intergenerational bequests (within a single period): If generation n cares about generation n+1, then they will care about generation n+1 in both periods in which it is alive, not just in the first period in which it is alive.

Generation n will not mind if generation n+1 pays more taxes during the (boom) period and pays less in taxes during the (bust) period, or vice-versa -- just as it has paid more during booms and less during busts. So if generation n happens to overlap with generation n-1 during a boom, then they will not decrease their bequests because the government is running a primary surplus, and if generation n happens to overlap with a generation n-1 during a bust, then they will not increase their bequests because the government is running a primary deficit. Because they know that the situation will be reversed in the next period, so that the countercyclical deficit does not subtract from their bequest anymore than the surplus adds to their bequest. Their bequest has the same effect on the subsequent generation's welfare in either case.

What would cause an increase in bequests would be an expected hike in future tax rates -- Barro's argument may be applicable to structural deficits, but not to countercyclical deficit spending.

Which, I thought, was the monetary interpretation of your "preponed" post.

[Mike]

I'm sorry I assumed you were replacing an existing bridge when that was not part of your story. But it might as well be, and it doesn't have to be, for my argument to hold.

Adam, a bridge is a long-lived public good which provides roughly the same benefit every year of its existence. So it is "perfectly good" its entire useful life.

Yes, it would be irresponsible to wait until the bridge fell down. I didn't say or imply that. The bridge has a useful life of X years and it should be replaced at that time. If it is replaced before , resources are destroyed.

But lets talk about the nonexistent bridge (or other project).

There are many projects which have benefits and costs. The net present value already takes into account the financing costs, so as rates decline, more projects become admissible. Also, because benefits may be higher in the future (say, because population is increasing), a lower interest rate will make projects admissible NOW that would only have been admissible in the future when benefits became higher.

This is basic C-B analysis, hardly requiring a PhD to invent a new word for it or construct a new model. Lower rates means more admissible projects, all else equal.

But our budget is finite and our ability to service our debt is finite. So not ALL admissible projects will or should be undertaken.

Nick, you're waving away crucial realities with assumptions. You CANNOT assume away expenses for depreciation, maintenance, operation, and accretion of retirement costs. You can't assume away external costs. Those are the micro problems.

The macro problem is that you are still treating the end of the output gap as:

1. Set to expire at a certain time
2. Exogenous to our decisions and policies.

Our low interest rates are a matter of POLICY, not exogenous fact. We will pay a high cost for that later.

Even if interest rates are low from low aggregate demand, there is a REASON for that. Businesses currently have extremely low financing costs and tax incentives to invest, and they still are not doing so. Don't you think their reasons might also apply to the public sector?

You are implicitly including a macroeconomic social objective into what should be a cold, hard project planning process. Latin American countries did that with their SOEs and it ruined their economies.

Bringing resource utilization forward in time through artificial means is precisely what caused this recession. Doing more of the same will not solve it. That will only make it worse.

The costs you are ignoring are enormous, and exacerbate the misallocation of resources. Stop picking at the scab and let it heal.

[david]

Am I wrong in reading Mike's disagreement as springing from a wholly different set of microfoundations here? Recessions in Rowe's vision are implicitly not caused by a microeconomic misallocation of resources, which Mike is assuming is the case.

[Mike]

You are reading correctly David. I am not "assuming" recessions are caused by misallocation of resources, I'm asserting that. Residential investment and durable goods manufacturing have disproportionate impacts on changes in GDP because of the abundance of resources they use, the durability of the output, and the debt which is used to purchase them. When there is any kind of subsidy, incentive, or expectation which results in overproduction in these markets, the resources used to make them and the financing to support them are idled for years thereafter. That's how we get business cycles.

People who view recessions resulting from exogenous shocks are doing so for the sake of convenience, and they are conveniently led to the wrong conclusion about how to "fix" the problem. The assumption of exogenous shocks means we can't stop them - we can only respond to them. This presents enormous problems because the very things which purport to "fix" the recession are exactly what caused them to begin with:

- Using too much debt
- Employing resources where they WERE employed because hey, a "construction worker" is a "construction worker", and wood and metal don't have anything better to do than build houses and cars.

There's absolutely nothing wrong with building a bridge when we NEED a bridge, and it would be fortuitous if we actually needed a bridge during a recession. But that only temporarily employs resources, and the distant shore of recovery moves further away from us.

No consequence of human decisions can ever be considered exogenous. Even an unlikely and uncontrollable event, say, an earthquake, isn't exogenous either - because we can and should make resource allocation decisions based upon the probability of the event and the loss given the event. If something is truly unforeseeable and the costs inestimable, then by all means model it as exogenous.

If interest rates are low, then we'll naturally decide to build bridges (and other things) sooner than we had anticipated, not because we WANT to restore full employment but because it is now a CORRECT decision to do so in the cost-benefit analysis. I reject a "full employment" motive as a legitimate benefit in C-B analysis. To do so is nothing but a wealth transfer.

There is only one correct way to fix a problem of overproduction: Slow production, let the excess burn off, and resume at a normal pace as the economy grows.

The incorrect ways are to open the floodgates of immigration and to subsidize consumption/investment on things we think we might need. That simply causes more market distortions.

A marvelous expansion of technology could also do the trick, but these miracles don't fall from heaven very often, and they usually have to be already deep in the pipeline of development.

[Antipodean lurker]

I think a comparison with Australia's "Building the Education revolution" program to build infrastructure in every school (public or private) during the 2008-09 recession would be interesting. Projects tended not to be actually "shovel-ready", and therefore there were a lot of hastily put together projects that attracted a high "quick-start" premium, then were subcontracted out with a lot of cream-skimming.

The projects took longer to start and complete than expected. However, they turned out to come online in the medium term (18-24 months vs the promised 12-18 months) and were actually very effective in combination with cash fiscal stimulus to benefit-receiving households, in avoiding 2 consecutive quarters of negative growth in Australia.