« Growth and Development: The Very Long Run | Main | Oil, art, and silver plate »

Comments

Feed You can follow this conversation by subscribing to the comment feed for this post.

I guess this is a silly question, but what does the word "fiscal" mean? It can't just refer to government spending and taxes, since the word fiscal can apply to consumption and investment. I guess what I want to ask is.... from the perspective of "all markets are money markets" what is the difference between a monetary and fiscal change? No one ever defines these terms, they just assert their existence and point to IS/LM.

Is the answer that "monetary" refers to money as a stock, and fiscal refers to money as a flow, thereby splitting them between the LM and IS curves, respectively?

For a long time I thought that "fiscal" changes simply refereed to changes in money demand, but that doesn't seem right.

JoeMac: "Fiscal policy" has a fairly well agreed-on meaning in economics. Any change in government spending on newly-produced goods and services (G) and/or taxes (net of transfer payments, which are negative taxes) (T) is fiscal policy. It's the meaning of "monetary policy" which is less agreed on. Traditionally, it was changes in the stock of base money. New Keynesians (for example) may mean changes in the central bank's interest rate target.

If a government increases spending, but pays for it by increasing the growth rate of base money (rather than by issuing bonds, or increasing taxes), that would be seen as both fiscal plus monetary policy.

[MMT guys: no, let's NOT get into an argument about this, because I know you have a different definition, but arguing about that here is off-topic.]

But the IS curve doesn't only refer to government finances, because it includes non-government consumption and investment. In that case, fiscal must mean something else. Let's pretend there is no government and assume that M is frozen. What does it mean to say that there is a "fiscal" change in consumption and investment.

I don't know how to put this into words. Here's my best attempt... where is the word "fiscal" in MV = PQ?

Talking about bad shocks, what about a natural disaster? That makes us poorer, on average. Yet we increase gov't spending, don't we? Why should we not react similarly to a man-made disaster?

Min: suppose my car falls apart. That makes me poorer. But it also makes me buy a new (to me) car. I spend more on cars and less on other things, because the marginal benefit of spending on cars has now increases (because I have one less car) relative to the marginal benefit of spending on other things. Plus, the car is a durable good. It's sensible to have an increase in debt if it is used to finance investment in durables.

It depends on the disaster. Does it increase the marginal benefits of government spending vs private spending? Does it increase the marginal benefits of durables vs non-durables?

JoeMac: "Let's pretend there is no government and assume that M is frozen. What does it mean to say that there is a "fiscal" change in consumption and investment."

I don't think we would talk about "fiscal policy" at all, if there were no government. We could still talk about shocks to private consumption and investment causing the IS curve to shift. Fiscal policy is only one of the things that causes IS to shift.

Possibly-naive question from a non-economist: would it be way off base (and missing the point of the post) to suggest that, in the real world, politicians pursue pro-cyclical fiscal policies because they, and the voters, think of the government like an individual household? When times are good, households spend; when times are tough, they tighten their belts.

If that's right, I leave it to others to judge whether or not to call that "incompetence, ignorance, or failure of the political system".

Jeremy: the theory you have just put forward is, roughly, one of the, if not the, received theories about why fiscal policy is procyclical!

Nick,

Suppose that real GDP follows a random walk with drift. Suppose also that government spending follows a random walk with drift. If the two series are co-integrated that would tend to suggest that fiscal policy is pro-cyclical. In other words, if they are cointegrated then they are driven by the same stochastic trend. Then, conditional on the two variables being cointegrated, we would need to show that there is a positive relationship between the two variables to suggest that there is pro-cyclical fiscal policy.

However, suppose that they are not cointegrated. It is still possible to estimate a positive correlation between the two variables even if they are completely unrelated. (This is a variation of the spurious regression problem that results from estimation when the variables have unit roots.)

Does this seem right?

Josh: that *seems* right. But I never have properly got my head around the spurious regression problem you mention, so I can't say for sure.

But the statistical/econometric insight I thought might be more relevant is the idea that, if you fit a trend line to data series generated by a random walk with drift, it will always look like a cycle around a trend. (I *think* that's right).

Nick, I interpret the gist of your argument to be:

Shock to GDP => permanent effect on future GDP => change in government fiscal policy to account for new expectations of future income

Basically, upon receiving a shock (maybe a technology boom or an adverse recessionary shock), the government needs to re-optimize its fiscal strategy. If the government does this optimally, then positive shocks imply permanent increases in spending and vice versa.

In order for the story to hold, it appears we need Ricardian equivalence and the government has to actually make its decisions in this "optimizing" manner. I am skeptical that these assumptions hold. Some things to add to the story about how governments may actually make spending decisions include:

-political economy considerations (spend more in an election year, spend more when left of centre parties in power?)
-prior commitments (i.e. in Canada we provide healthcare - demography would imply the fiscal burden will go up as the baby boomers get old)
-international considerations (i.e. spend money on the military to go to Afghanistan to make the Americans happy)

I guess my point is, I don't think the way that we model government spending mimics how governments actually decide on spending. As well, the permanent income hypothesis type story requires Ricardian equivalence (I haven't thought about it in a while but I believe so?). That is a whole other can of worms!

Government as a complementary good. More construction needs more roads and utilities. More commerce and transportation needs more bridges and ports. More children needs more schools and teachers (after a five year lag). This imparts a co trend but the pro cyclical aspect would be cyclical revenues and reducing debt not being sexy.

Countercyclical fiscal policy addresses unemployment and labor dislocation. Most of the automatic stabilizers, such as unemployment insurance, address income loss by displaced labor, investment in retraining, etc. GDP is NOT the only factor that matters. Income distribution matters. Employment matters.

The upward spiral is a too high demand, wage-price spiral. Wages and prices are very sticky so the reverse spiral is not a low demand negative wage-price spiral, it is a negative demand-employment spiral.

These self reinforcing spirals will always spiral up or down in response to external and internal shocks unless there is moderation of these spirals with monetary or fiscal policy. Monetary policy is supposed to fine tune the spiral to target 2 percent wage inflation (current supposed target). Regulatory policy is supposed to reduce the big shocks such as a financial collapse. Fiscal policy pulls its weight when monetary policy is ineffective or not selective enough. Monetary policy is impotent for stimulating employment at the zero lower bound. This requires fiscal policy to address too little demand for goods, services and labor. In an economy with a spiral moving slowly up at 2 percent, fiscal policy can address market failures in a single sector without monetary policy having to hammer the whole economy to address a single bad apple.

I am not sure the pro-cylical policy would be optimal if you add a drift to the random walk. Economies grow over time and even if the detrended GDP looks like a random walk, I think that there is strong evidence suggesting a long run trend (since the Industrial Revolution).

If the government knows about the long run trend, I wonder if optimal G could be countercyclical depending on other details of the model - productivity of private versus public investment, optimal welfare mechanisms (I remember you suggesting that fiscal policy should take care of distributional/insurance issues etc.).

Please ignore/delete the last 4 words forming an incomplete sentence at the end of my previous comment.
[Done. NR]

If there is a drift term, then over the long run the drift term dominates and the series becomes, for all practical purposes, just a trend.

Regarding the possibility of a cointegrating relationship, it might be better to think in terms of one nonstationary series on the private sector side (i.e., C + I) following one random walk while the "G" time series would follow what would APPEAR to be a random walk, but is in fact a reaction. If GDP is the sum of C + I + G, then the cointegraging relationship would be stationary. This gets back to the old image of a dog dragging his drunk owner home from the bar. The drunk follows a random walk. As the drunk veers to the left the dog pulls to the right. The dog's counter-drunk actions would APPEAR to be a random walk as well. But together the two random walks follow a stationary path home. On the other hand, this whole thing seems a bit "out there."

Kevin: "As well, the permanent income hypothesis type story requires Ricardian equivalence (I haven't thought about it in a while but I believe so?)."

If a household is borrowing constrained, then the permanent income hypothesis won't hold, and nor will Ricardian Equivalence (well, RE might still hold, depending on the nature of the borrowing constraint, but it might not). But governments are not normally borrowing constrained (some, e.g. Eurozone, are, but that would put a whole new light on why governments do procyclical fiscal policy).

Lord: Hmmm. That idea might be worth thinking about.

jonny: your comment doesn't really address the post. I suddenly have an image of an old lady dressed all in black, who sees a heretical thought, and furiously Keynesian crosses herself repeating the Keynesian catchecism to try to push it out of her mind.

primed: If growth in GDP were constant, I would expect G and T to rise (roughly) in proportion. So I don't think adding a drift to the random walk would change things much.

2slugbaits. Neat image; it's a new one for me. But isn't the dog doing countercyclical policy?

Nick, I think that you just restated the old "structural (supply side) vs. demand side" argument of the nature of recessions. If "the cycle" means real shocks, then we are in RBC world and pro-cyclical policy is good. If we have demand induced recession, then procyclical policy is bad.

If we have both shocks at once (is this what you implied by the article?), then I still think that procyclical policy is bad. This is something that Scott likes to say all the time: we need to fix the demand problem first, because it is the easier one. Once we solve it, we are back in the RBC world and we may start to solve our supply problems even employing standard "procyclical" policy.

So to say it differently, the heart of the problem you presented here in this post is a definition of recession. It is not just a decline in RGDP or RGDP/capita or whatever. You yourself presented several very good conditions for when to call that we are in recession, like this one: "When it is harder then usual to sell things for money".

Nick: Welcome to Real-Time economics!

"Take an extreme (and unrealistic) case. Suppose that (real) GDP followed a random walk."

Why do you think this is unrealistic? Think about it; the predictability in GDP movements disappears after a few quarters; at the one and two year horizons, most models, forecasts, surveys, etc. have no detectable predictive ability. But with an annual budgeting process, the fiscal policy maker works at planning horizons of one year and up.

But there's also an active theoretical and empirical literature on pro-cyclical fiscal policy. Alesina and his co-authors offer models based on political explanations for pro-cyclical policy (which I think is what Frankel argues in his post.) That leads people to argue that we need more fiscal "discipline" (think balanced-budget rules) to compensate for bad political incentives.

More recently, evidence from Europe argues that the problem is closer to what you describe. (Cimadomo has a great survey: http://ideas.repec.org/p/fip/fedpwp/11-25.html) In particular, in a world with important unit roots in GDP, output gaps get revised a lot ex post. That means that when we judge whether fiscal policy is procyclical or counter-cyclical, we could get different answers depending on whether we use what people thought the gap was at the time, or what we now think the gap was. Alesina et al. used ex post measures of the gap. Cimadomo's survey finds that fiscal policy looks much less procyclical when we use contemporary measures of the gap. That suggests Alesina's political story is much weaker than we thought.

2slugbaits:

"If there is a drift term, then over the long run the drift term dominates and the series becomes, for all practical purposes, just a trend."

Yes, the drift term dominates....but Nick's story about misperceiving the unit root is unaffected by a constant drift term. (To convince yourself of this, try the following.
1) Estimate the drift rate of your favourite real GDP series.
2) Simulate a random walk with that drift rate.
3) Detrend your simulated series with your favourite detrending method (HP? Band-Pass? Linear Trend? etc.)
4) Compare the estimates that you get in (3) with those that you get if you detrend recursively (i.e. rolling through the sample, adding one observation at a time.)

So instead of twisting our minds with thoughts of cointegration and trends and drifts rates, it might be easier to understand things with some transformations.

We've been talking in terms of a level relationship Y = C + I + G ( + X - M, remember? )
Let's divide everything by Y and rewrite this in lowercase letters, to get
1 = c + i + g ( + x - m)
I guess you guys want to ignore x-m so I'll set it to zero
1 = c + i + g
=> g = 1 - c - i
The deficit? We can express government revenues as a fraction of Y as well (t), so the deficit is just (g - t)
Suppose we want to stabilize the debt/GDP ratio. Let's call that ratio b (for the stock of bonds.) IIRC, in the long run we'll need
g-t < b*(r-d)
where r is the real effective interest rate on the stock of debt
and d is the growth rate (drift) of Y

What about all those recessions and gap misperceptions that Nick talked about? Those are just temporary wobbles in our variables.

Good post. I the United States, we are 3 years into the recovery. Output has not been growing at trend, and as far as I know most experts do not expect above trend growth over the horizons they forecast (several years out.) Given those facts, it's not surprising that state and local governments are setting their budgets under the assumption that we are in a "Great Stagnation."

Scott: Whatever rate of growth you're forecasting several years out is trend growth.

Simon,

d>r so the ratio will always stabilize provided that the deficit per GDP ratio is capped.

Ooops.....better make that
t-g > b*(r-d)

RSJ: depends what country you're thinking about, doesn't it?

Simon,

Not really. Not if r is the risk free rate, which is the only rate that makes sense in this context. The only question is whether the country's debt is risk free or not.

On the other hand, wired things can happen in a low IR environment such as Japan. I would argue that ex ante rates have been less than g even though ex post rates have been greater.

JV: "Nick, I think that you just restated the old "structural (supply side) vs. demand side" argument of the nature of recessions. If "the cycle" means real shocks, then we are in RBC world and pro-cyclical policy is good. If we have demand induced recession, then procyclical policy is bad."

I don't *think* I'm doing that (I was trying to be agnostic on the cause of the shocks to GDP.)

Some economists do say that monetary shocks have only temporary real effects, so that all permanent shocks are supply side. And they use this as an identifying restriction in econometric work. Supply shocks can be temporary or permanent. Demand shocks might or might not have *some* permanent effects, even if prices eventually adjust. My example: take a standard AK growth model, add money and temporarily sticky prices. A monetary shock affects investment, and even when prices adjust the stock of Kapital will never recover fully to where it would have been otherwise, because in AK models, unlike the Solow growth model, there is no steady state that is invariant with respect to history.

Simon: "Why do you think this is unrealistic? Think about it; the predictability in GDP movements disappears after a few quarters; at the one and two year horizons, most models, forecasts, surveys, etc. have no detectable predictive ability. But with an annual budgeting process, the fiscal policy maker works at planning horizons of one year and up."

Hmmm.

"That means that when we judge whether fiscal policy is procyclical or counter-cyclical, we could get different answers depending on whether we use what people thought the gap was at the time, or what we now think the gap was. Alesina et al. used ex post measures of the gap. Cimadomo's survey finds that fiscal policy looks much less procyclical when we use contemporary measures of the gap. That suggests Alesina's political story is much weaker than we thought."

Hmmmmm again.

Important points.

Thanks Scott!

By "trend" I take it that Scott means "previous trend". The longer the US economy stays away from that previous trend (due in my and Scott's opinion to shortage of AD/NGDP) the further away from that previous trend it is likely to stay.

“Any change in government spending on newly-produced goods and services (G) and/or taxes (net of transfer payments, which are negative taxes) (T) is fiscal policy.” Well, keeping gov’t spending and taxes *the same* should also count as *fiscal policy*. And what about gov’t spending on things other than newly-produced goods and services—say, on land, or second-hand goods, or stocks and bonds, or its own previously-issued debt? None of that (not even the last item) seems like monetary policy! (What is the term for gov’t policy that is non-fiscal and non-monetary?)

James: " And what about gov’t spending on things other than newly-produced goods and services—say, on land, or second-hand goods, or stocks and bonds, or its own previously-issued debt? None of that (not even the last item) seems like monetary policy! (What is the term for gov’t policy that is non-fiscal and non-monetary?)"

Good point. That sort of government policy goes below the radar, given our conceptual scheme. You might like my old post on that subject.

RSJ: "Not really. Not if r is the risk free rate, which is the only rate that makes sense in this context."

Okay, so what is the risk-free rate currently in Greece? I've tried to think about how I should measure this, but without success. I know that I should not use the rate on govt debt, because that's risky. I know that I should not use the rates for Greek inter-bank borrowing and lending, for the same reason. As a Greek citizen, I'm not sure whether my consumption basket going forward will be priced in EUR or some other currency.

So how would you estimate their risk-free rate? (Then we get to worry about their expected growth rates ;-) )

RSJ: Please disregard the above....it was the result of a brain cramp.

I should have asked whether you're not confusing the conditions for dynamic efficiency (from the discussion last week) with the conditions for the stability of government debt. In my comments above, I was talking about conditions for the stability of government debt (in the absence of default, I guess.) The r in this context is the rate of interest on government debt, risky or not.

"if there is a unit root in GDP (so that some part of any fluctuation in GDP lasts forever), or if there is perceived to be a unit root, might that explain why fiscal policy is, or appears to be, pro-cyclical?"

Good question, Nick. It does seem at least judging by current American fiscal policy, it does appear pro-cyclical, which certainly doesn't seem optimum.

This is partiuclarly true at the state and municipal level where you see perpetual cuts. Even judging the by the language used in analysis of state and city economies seems to presume pro-cyclicality as a matter of course.

Agree with JV Dubois on the normative angle. A supply shock should be met with Keynesian response and a demand shock maybe not. Agree Nick, that there might be permanent divergence as a result of Keynesian-ing a coin flip. It doesn't seem that bad if you assume bubbles pop. There is a chance that lost GDP could've been a loan for someone's optimal career picker AI software prototype, or that some derivatives nutcase doesn't get hired on the happy flipside. I assume the divergence is logarithmic rather than exponential; ie. affects the timing of recessions more than anything. When credit becomes more available in a recession banks introduce even more credit, and that's where I hide most of the divergence...

Nick: I may have misunderstood you, but I really don't think so. It may be true what say that "in a wide class of macroeconomic models, ranging from Old Keynesian through to Classical models, pro-cyclical fiscal policy is a bad policy". I do not think this is true for RBC macroeconomic model. So if you speak with RBC economists, they may even tell you that countercyclical fiscal policy is a bad thing as it may delay "inevitable" restructiring of production to technology shock.

And I know that you understand this. The sole purpose of MM as I observe it is to restore the Say's law so that we get rid of the demand issues and end up solely with supply side issues. Our goal is to get into the RBC world. If there are no demand side recession (if for instance all CB follow NGDPL targeting and we will enjoy another era of Great Moderation), the only serious part left for macroeconomics is the long-run growth.

Now even with this setup RBC economists may not be right. There may be something into a "anti-cyclical fiscal" policy. If we assume that government has a role to prevent dynamic inefficiency, and that this inefficiency increases during economic downturns (or as RBC economists would say when we experience decrease in trend growth) then there may be some case for government interference. All I want to say is that this is a way category difference to what we usually think of as business cycle and/or recession. Government should always look for free lunches out there and help with the supply side - be it with better institutional or even physical infrastructure. Otherwise you end up with very confusing debates, as the ones where supply siders model aggregate demand falls as news about future aggregate supply falls, or where good monetary policy can be viewed as institutional infrastructure akin to better property rights, or better labor laws - promoting the supply side in the long run.

PS: Anyways, if you are nit-picky every demand induced recession (or any government policy for that matter) that has any - even temporary - impact on real economy reduces the "long run" economic outlook. There are calculation like Evan's here: http://esoltas.blogspot.sk/2012/07/how-much-did-recession-cost.html that set opportunity cost of the recession in USA somewhere between 7 - 13 trillion dollars. Mismanagement of this recession permanently reduced potential "lifetime" production/income of american nation by that amount. This sums corresponds to millions of people being idle, not producing goods and services. Maybe some of this wasted resources could have been used to research of new technologies that could have permanently increased humanity's "potential".


The comments to this entry are closed.

Search this site

  • Google

    WWW
    worthwhile.typepad.com
Blog powered by Typepad