It’s well understood that if you see the world through sufficiently Keynesian eyes, you might welcome a destructive hurricane or the threat of an alien invasion (together with the frantic spending it would stimulate) as just the ticket to lift the economy out of a recession.
What seems to have been largely overlooked is that even in a thoroughly non-Keynesian world where markets work perfectly (or as perfectly as they can in the presence of a distortionary income tax), and recessions cure themselves, we might still want that hurricane.
Or, because we can’t always call forth hurricanes when we need them, we might want our government to simulate their effects by diverting funds from useful to destructive spending projects — or just occasionally showing up at people’s houses and trashing their furniture.
Here’s why: Hurricanes make us collectively poorer. When we’re poorer, we work more. When we work more, the government collects additional income tax revenue. But — taking total government spending as given — the government can’t continue to collect additional revenue forever; sooner or later it must lower tax rates. (This assumes we’re on the good side of the Laffer curve, where the way to collect less revenue is to lower rates, not raise them.) When tax rates fall, labor markets work more efficiently. So much so, in fact, that the efficiency gains can more than compensate for the initial destruction.
I only realized this recently, and it surprised me (along with several others I showed it to) enough that I wrote it up as a short paper. (Update: A more recent version of the paper is here.) I also looked back through my blog archives to see how badly I’d gotten this wrong in the past.
Here’s what I wrote last year in response to the notion that Hurricane Sandy could be “good for the economy” by creating a lot of construction jobs:
If you find yourself in an argument about this, ask your opponent whether it’s “good for the ants” when you put a stick down their anthill, wiggle it around and destroy their infrastructure. Go ahead and acknowledge that this can sure put a lot of ants to work.
Or, for that matter….
Ask if spilling ink on the living room rug is “good for your household’s economy” because of all the cleanup work you’ll do.
Of if a flu epidemic is “good for the economy” because it keeps doctors working overtime.
Well, let’s see now. Ants, as far as I am aware, don’t pay income taxes. So I stand by the assertion that you do them no favor by destroying their anthill. Spilling ink on your living room rug won’t affect your income tax rate, so I stand by the assertion that it can’t be good for your household’s economy. On the other hand, if we all spill ink on our rugs, and then work a little overtime so we can afford to pay the cleaning bills, income tax rates must fall and we can be better off. A flu epidemic, on the other hand, is likely to lead to less work, not more, so I’m still firmly anti-flu.
But while I hate the phrase “good for the economy”, I do have to admit that a destructive hurricane can be good for people, which is of course what we really care about.
On the other hand, the reason destruction can be good for us has nothing to do with the commonly cited arguments about “putting people to work”, because those arguments do not invoke the existence of a distortionary tax structure, and so would apply mutatis mutandus to an ant colony (or, in those rare instances when the arguments are more carefully constructed, to an ant colony with high menu costs).
Perhaps the real moral, though, is that our tax structure is so destructive on its own that a hurricane can actually improve matters. Rooting for natural disasters — or replacing national parks with holes in the ground — is not the only conceivable way to address that problem.
It’s interesting that your argument breaks down precisely when there is a shortfall in demand for labor. Therefore you’ve argued that hurricanes can be of net benefit at precisely those times when Keynesians are least likely to agree.
Fascinating.
I wish I had time to read your paper in more detail, but I’m busy working to cover my tax bill.
@MikeH
Why does it fall down when there is a shortfall in demand for labor?
Income (and thus income tax receipts) rise regardless of what the demand situation is like. In fact, given that it’s easier to hire workers in a high-unemployment situation, wouldn’t the impact (adjusted for inflation) on total income be higher in a situation of high unemployment compared to a period of full employment?
Part of the problem with Keynesian politicians is that they forget about reducing spending when times are good.
#1 and 2. Don’t forget, in the world described above markets work perfectly, so there is no such thing as a “shortfall” in the demand for labor.
Steve,
Good post. Thanks for giving us something to think about. I have one point. To the extent the hurricane destroys the capital stock, wages will be lower. Then, you have offsetting income and substitution effects and the sign on the change in hours is ambiguous.
In Fig 1, taking just the line where 100% of Govt spending is useful. When the tax rate reaches about 0.66, we reach the shaded portion. So this means that up to 66% income tax, we are better off spending usefully, but then we are better off diverting a tiny fraction to waste. As the tax rate increases further, presumably we are made better off by transferring more to waste.
This is interesting, but since tax rates are less than 66% for most people you could say of academic interest only.
But then we get into the multiple period curves, which show that welfare is improved if tax rates are above 25%, and at least some Govt spending is useful – if income and capital are taxed the same.
For any level of useful spending, we get to the anomalous zone sooner if income and capital are taxed equally than if only one is taxed. This makes it difficult to predict what other mixed tax-rates might look like. Out of interest, can you do a curve for capital tax and income tax approximately what they are in the USA, with capital taxed at a lower rate than income?
@5
I think SL assumes this away in the paper by assuming the government flushes the “wasted” revenue down the toilet.
Harold (#6):
Out of interest, can you do a curve for capital tax and income tax approximately what they are in the USA, with capital taxed at a lower rate than income?
I’ll try to get around to this, but of course the resulting graph will still be unrealistic in that it’s quite impossible to capture all the idiossyncracies of the US tax code — and even if it were possible, we’d still be pretending that all taxpayers are identical.
#8. No worries. It just struck me as interesting that the anomalous zone moved left then right again. (I typed that before I realised what I had just said). Is there any reason why 50% would represent the extreme position, or could mixed tax regimes cover even more of the graph?
@ Landsburg,
I don’t have time to do the math now, but I’m just curious. What would happen to your result if it was possible for government spending to be more useful than what it would have be spent on in the private sector? Does your result entirely depend on us living in a world of perfect competition, and only distorted through taxation?
Daniel (#10): The inequality (10) continues to apply when X/G > 1, which means that when govt spending is more useful than it would have been in the private sector, we might still want less of it.
@Landsburg,
Very interesting. Thanks for that. I’ll take a more detailed look at the paper tonight :)
The large fallacy in this argument is this statement:
“But — taking total government spending as given — the government can’t continue to collect additional revenue forever; sooner or later it must lower tax rates.”
Total government spending would not stay at the same level; more revenue would lead to more spending, with more projects funded and at higher levels. And while more government spending may be seen as a good thing (providing jobs, etc.), an increased budget almost always leads to increased laxity on spending; in other words waste increases.
Thus more revenue leads to more spending; sustained increases in revenue lead to wasteful spending. More revenue does not lead to less taxes.
I have a bleg for Steve. In those periods where you go a while without blogging why not invite a guest blogger or two? In particular Jamie Whyte. You had a guest post from him once that was terrific, as have been his infrequent comments.
“Here’s why: Hurricanes make us collectively poorer. When we’re poorer, we work more. When we work more, the government collects additional income tax revenue. But — taking total government spending as given — the government can’t continue to collect additional revenue forever; sooner or later it must lower tax rates. (This assumes we’re on the good side of the Laffer curve, where the way to collect less revenue is to lower rates, not raise them.) When tax rates fall, labor markets work more efficiently. So much so, in fact, that the efficiency gains can more than compensate for the initial destruction.”
Both the second and third sentences are true ceteris paribus, but the third doesn’t hold when the second occurs. That is, when we’re collectively poorer and decide to work more, the government doesn’t necessarily get more in tax reciepts.
A significant shortfall to capital stock that a society collectively must spend about x labor hours replacing puts them x labor hours behind the growth rate of a society that never experienced that distruction. Because growth is exponential, that absolute gap should, ceteris paribus, steadily increase over time.
Furthermore, ceteris paribus, tax receipts by the government should grow faster under the non-destruction situation as well, meaning that it will fall earlier if the mechanism you claim is operating actually comes into effect.
I’m not seeing how you could presume that the destruction could be good for people, on the net. The reduced marginal productivity that results from the destruction of capital requires either a sacrfice of leisure time or a sacrifice of income (or both). The only way to end up with more income than normal is to greatly sacrifice your leisure time and work more hours… but had you wanted that same level of income pre-destruction, you wouldn’t have had to work as many extra hours than you do now anyways, so prefering now to pre-destruction is irrational.
This problem doesn’t go away if you ignore destruction of capital and focus only on destruction of consumption-based property (cars, houses, toys, food, etc.). The reduced standard of living due to a destruction of personal property requires time be spent replacing the property which results in, under Bastiat and the Broken Window Fallacy, a loss of what you would have spent your income on if you had not needed to replace the lost property. This loss is present iff you value any of the property you own, and this value can only be recovered if, again, one sacrifices leisure time. And the only way to get more money than you had before would be to sacrifice extra leisure time, an amount you wouldn’t have needed to sacrifice
In summary I’m not sure how you conclude that having to sacrifice more leisure time than normal just to get the same result somehow makes everyone better off. The idea that “people” are better off only applies to those productive workers that are irrational (e.g. they would choose to have less leisure rather than more for the same level of income) and thoe who don’t have to work for a living (e.g. those who are dead, or who mooch) such that they enjoy the benefits of those who sacrifice leisure for a higher standard of living without having to make any sacrifices themselves.
The Djinni:
I’m not seeing how you could presume that the destruction could be good for people,
You can’t. That’s why you need to prove it.
Ken B: I have no doubt that any guest post from Jamie Whyte would raise the tone around here considerably, and I trust he knows this.
I confess to not yet having read the paper before posting but, here goes: Does Dr. Landsburg’s result mean that by extension other progressive fallacies might also be less false? I’m thinking of the liberal entreaties to do X in order to decrease unemployment where X is:
a) Banning some labor saving technology like industrial robots
or
b) Taxing low cost imports
Knowing minimal economics (scarcity, supply and demand, incentives is about it), and virtually no math, and therefore no understaning of the the Landsburg equations, it is with great hesitation that I suggest that Euler’s equation, (a+b^n)/n = x, may be relevant.
@Advo #2 Why does it fall down when there is a shortfall in demand for labor?
I was going to say that Steve argues that when we are poorer, we work harder. In the article, he explicitly states that this is a choice made by the worker, with a negative example: “An increase in X makes the agent wealthier, which means he chooses to work less” So, when we are poorer, we work harder explicitly to earn more money.
However, if there is a shortfall in demand, all that would happen is that more people start looking for work – not that the work is available. I can’t work overtime (and get paid for it) if the boss says no. Hence, if there is a shortfall in demand for labor, the hurricane produces no increase in earnings, no increase in government revenues, no decrease in the tax rate.
Harold (#4) may have provided a better answer: #1 and 2. Don’t forget, in the world described above markets work perfectly, so there is no such thing as a “shortfall” in the demand for labor hence, the argument breaks down in worlds where markets don’t work perfectly, and shortfalls in demand can exist.
Steve, so if more people getting HIV leads to more economic activity, does that mean…less sex is safer sex?
It is not the increase in destruction that makes the taxpayer better off, it is the resulting decrease in the tax rate. And this depends on your assumption that increased destruction induces the government to lower the tax rate (because G is fixed). This assumption is arbitary, and somewhat implausible given that government is wasteful and clearly not motivated to benefit the taxpayer. You need to provide a behavioral model for the government to explain your assumption that G is fixed rather than (say) government waste G-X.
“Taking total government spending as given” means freezing government spending. I propose that usual growth in GDP would increase tax income, so “sooner or later the government must lower tax rates”, or else it would collect too much taxes, which it supposedly would not want to spend.
An intelligent, limited government (utopia) would reduce tax rates without the gimmick of destroying a few hundred buildings. A rapacious, self-serving government (reality) would react to greater tax income from disasters by creating more destruction itself.
Is it realistic, do you have examples, where a government lowered taxes because it collected more money? Absent that, does a complex mathematical analysis matter?
My fear for reality is that big-government types will not care about any unrealistic assumptions of your paper. They may still cite it to “prove” that disasters and physically destructive plans (like building highways and railroads) are intrinsically beneficial before we even get to ride on or in them.
#23 “An intelligent, limited government (utopia) would reduce tax rates without the gimmick of destroying a few hundred buildings.” This would be bad. Reducing tax rates makes everybody better off, so they work less, further reducing tax revenues. Govt would then have to increase tax rates. The overall effect on welfare would be negative (for some conditions).
@16 Steve Landsburg
Right, and so I’m asking for your proof. Not empirical (that’s simply absurd to even contemplate for such a topic), but a deductive proof.
Why does the destruction scenario experience decreased taxation when the non-destruction scenario doesn’t? How do you reconcile the irrationality in assuming that people value less leisure/consumption/property to more?
You can’t even make the statement that people would be better off if you completely ignore (or assume away) the value of what is destroyed to the people who own it. They’d simply be just as well off as they were before.
Let me try to make my problem clear:
“On the other hand, if we all spill ink on our rugs, and then work a little overtime so we can afford to pay the cleaning bills, income tax rates must fall and we can be better off. A flu epidemic, on the other hand, is likely to lead to less work, not more, so I’m still firmly anti-flu.”
Why don’t people just not spill ink on their rugs, but then work a little overtime anyways and increase their purchases elsewhere? That way, they keep the rug, get extra money to buy something else, and the taxation levels decrease too.
And why would people prefer your plan over mine? In yours, they’re either out the rug (and the ink), or out the income used to replace it. In mine they get both. My plan is just as good or better than yours, regardless of how much or little the rug is valued.
And why would they even consider your plan, if “markets work perfectly”? Why go through the extra step of destroying their own property? If they preferred more income to more leisure, wouldn’t they just go and work more anyways?
TheDjinni:
Right, and so I’m asking for your proof.
The proof is in the paper linked form the post.
#26. I understand that the paper concludes that welfare is increased by the useless spending, at least under certain tax/spend regimes. However, I think Djinn may have a point. In his scheme, the tax rate is a part of the utility function, i think. People in Djinn’s world realise that by working more, they will increase tax revenues, and thus end up with lower tax rates. I think in the world described in your paper the effect of extra work on the tax rate is not included in the maximisation of the utility function. I may have this completely wrong.
Harold:
People in Djinn’s world realise that by working more, they will increase tax revenues, /i>
But if they “realize” that, they are mistaken. No one person can have a significant effect on tax rates by working more. (Just try the experiment: Work a few extra hours this week and see what happens to your tax rate.)
#28. Ok, so that is why it makes no sense to have tax revenue feedback as part of the utility function. Your own choices affect you directly, but you can have almost no effect on “aggregate” choices.
This reminds me of prisoner’s dilemma. We have three states. Status quo, destructive ink spilling and extra work without the ink spilling. Without any external impetus, the status quo will continue, but you have shown this is not the best we can have. If there is a destructive event that affects lots of people, their responses by working more may improve welfare overall. The third outcome would be even better – working harder without destroying the carpet, but only if everyone does it. However, we have no obvious way to bring about this outcome about.
I was thinking about PD in relation to voting. A rational person would not vote, just as a rational person in PD would not cooperate. However, social welfare is maximised if everyone does vote, or everyone cooperates in PD. This is perhaps an argument for compulsory voting.
Harold: Yes, your invocation of the prisoners’ dilemma is spot-on.
Harold – or how about limiting the things we are supposed to vote on to those that are (absolutely) ‘necessary and proper’? Rational ignorance is a strong argument for limited govt
Forcing people to vote does not ensure they will be informed voters
Maybe stupid but had to ask — if we all decide to work more does this ensure we will in fact be wealthier? Because supply creates its own demand? If the answer is that a natural disaster does create demand to reconstruct what was lost, does that address Mike H’s question? Does it translate to the scenario of the govt diverting useful to wasteful spending?
iceman:
if we all decide to work more does this ensure we will in fact be wealthier?
We could, I suppose, all work more performing completely useless tasks, but why would we choose to do that?
The wider implications of the PD are that rationality does not get us to the optimum social welfare position.
In the PD, social welfare is maximised if both parties cooperate, but it is irrational for any individual to cooperate. Therefore social welfare can only be maximised by an external compulsion or incentive to cooperate. There are variations depending on whether it is a one-time game etc. but I think this is a fundamental conclusion.
Isn’t this essentially the whole Keynes’ argument? People rationally do not invest because everyone else is rationally not spending. Anyone who unilaterally invests is irrational, and doomed to failure. A stimulas cause people to spend, so the rational person then finds it worthwhile to invest, which leads to more spending etc.
With voting, if everyone votes (cooperates), the system works best since politicians must try to appeal to every constituency. Steve has demonstrated that it is irrational for nearly every individual to vote since you will always be better off not voting (defecting). If only irrational people vote, the system does not work so well since politicians only need to appeal to irrational people. Currently we have irrational appeals to “duty” and “they fought for our right to vote” to provide the external impetus for otherwise rational people to vote, but this is imperfect. A simple vote or pay a small fine may be enough to prompt all rational people to vote, and then we may start hearing rational arguments from our politicians.
Although looking at Australia, perhaps not.
35 – Except again forcing people to check a box doesn’t make it more rational for them to spend time studying the issues. Arguably we’re better off having fewer uninformed voters. Yours seems simply a recipe for spreading more pork around.
34 – Oh I was just thinking out loud b/c saying everyone suddenly wants to do something to increase their income (vs. create wealth) set off a little alarm in my head, like are we co-opting Keynes by invoking Say’s Law? I guess if everyone engages in net new productive activities and intends to spend all of the remuneration, e.g. to replace trashed furniture, there can be enough hiring to go around and no paradox of thrift (e.g. vs. I plant a garden at the grocer’s expense). Not sure if that would apply to all forms of government waste? Like I said not necessarily a brilliant insight.
#36. “Arguably we’re better off having fewer uninformed voters.” I don’t see how we can be better off if only irrational people vote.
@26
“The proof is in the paper linked form the post.”
I read that before posting, and I don’t see you account for the value of the destroyed property or the harmful externalities of relying on random acts of nature to do your destruction for you.
A proof that’s missing two key variables is no proof at all.
@28
“No one person can have a significant effect on tax rates by working more.”
Then I guess spilling ink on your carpets doesn’t help either.
Unless, you know, people are rational and markets are perfect (as per the assumptions), in which case all labor suppliers collude to drive tax rates to zero. Whether they do that by spilling ink and then working more, or just working more, the effect is the same.
@29
“This reminds me of prisoner’s dilemma. We have three states. Status quo, destructive ink spilling and extra work without the ink spilling. Without any external impetus, the status quo will continue, but you have shown this is not the best we can have. If there is a destructive event that affects lots of people, their responses by working more may improve welfare overall. The third outcome would be even better – working harder without destroying the carpet, but only if everyone does it. However, we have no obvious way to bring about this outcome about.”
You’re mischaracterizing the states. If you want to talk about superior outcomes, then I’d argue there are five. They are (arranged from worst to best):
-Undirected Destruction of Random Property by Natural Disaster
-Status Quo
-Directed Destruction of non-Capital Goods Through Government Dictat or Voluntary Consumer Choice
-Directed Increase to Labor Through Through Government Dictat or Voluntary Consumer Choice
-Directed Decrease to Tax Rates Through Government Dictat or Voluntary Consumer Choice
The first can randomly destroy capital stock, or destroy too much property, and therefore will almost always lead to a reduction in living standards. You won’t see decreased tax rates when your marginal productivity is cut.
The second is alleged to be bad because taxes. I’ll won’t argue otherwise.
The third is stupid because you’re destroying valued property to indirectly reduce tax rates. Generally, either everyone has to be willing to destroy their own goods AND work more to replace them, or else some dictator has to come around and force them to.
The fourth is an undeniably good outcome, except for the fact that people have to sacrifice leisure that they wouldn’t normally choose to do. This again can only be accomplished if everyone chooses to work more or a dictator forces them.
The last is most likely the best outcome, because you get what you wanted in the first place with few unwanted side effects. Everyone just chooses to not pay the higher tax rate and government is forced to go along with it, or the dictator simply declares lower taxes for all. The reduction in tax rates then increases taxes in the future. In this situation, government spending can’t be fixed and must decrease in the short term, but that’s arguably better than destroying property and/or sacrificing leisure involuntarily, particularly in terms of being “better off”.
If you try to naively argue the merits of the first scenario, you’re simply failing to account for the value of the destroyed property and capital. If you try to argue the third, you have to allow for the fourth and fifth.
In no scenarios is “destroying property” an unequivocably superior outcome that ALSO leads to decreased taxation due to an increase in work. If a model assumes that the destruction will be coercively mandated or voluntarily implemented, then you get a better outcome by simply NOT destroying the property to begin with and then coercively mandating/voluntarily implementing whatever effect you were hoping to come out of it. If your model assumes that the destruction will happen randomly, then you’re naive for expecting anyone to be better off.
TheDjinni:
I read that before posting, and I don’t see you account for the value of the destroyed property or the harmful externalities of relying on random acts of nature to do your destruction for you.
You might want to go back and read again. Consumption is equal to (1-theta)f(L)+X = f(L)-(theta f(L)-X). That theta f(L)-X is of course the value of the destroyed property.
Wow. I hadn’t considered how a backwards bending labor supply curve might intersect with the Laffer Curve.
Laffer proposed his curve to argue that reducing taxes might induce people to earn (or report) more income, and the government revenues produced by the growth in taxable income would offset the loss of revenues produced by the decrease in the tax rate. Landsburg suggests that the causation might work in reverse – increasing labor and taxable income may induce a reduction in marginal tax rates – and that an anti-wealth effect might trigger the initial increase in labor.
Ok, I’ll bite: All kidding aside, what policy implications to you see from this? Is this merely a clever anomaly — a fun math slight-of-hand that seems to show that 0=1 or whatever — and I’m just not clever enough to spot the error? Or is this a (new category of?) market imperfection, whereby counter-intuitive extrinsic events can predictably move an economy from a suboptimal point of equilibrium to a better point?
Or is this a (new category of?) market imperfection, whereby counter-intuitive extrinsic events can predictably move an economy from a suboptimal point of equilibrium to a better point?
There is, really, nothing new about the observation that if you’ve already got one distortion, a second distortion can improve matters. I was surprised to notice this instance of that general principle, but the general principle is well understood.
“There is, really, nothing new about the observation that if you’ve already got one distortion, a second distortion can improve matters. I was surprised to notice this instance of that general principle, but the general principle is well understood.”
Fair enough. But the distortion in question — marginal taxation — is pretty much universal. So statements such as “Obama’s policy will distort incentives sub-optimally — unless there is already some existing distortion, in which case, all bets are off,” can now be shortened to “All bets are off.”
I guess the strength of my reaction is in proportion to the strength of my belief that the classical model provided some useful guide for public policy. While I’m quick to cite market failures, I still harbor the notion that they’re the exception, and that the classical model still signals the likely direction in which the answer lies.
I’m no geographer, but I’ve long thought that if I keep driving east I’ll hit the Atlantic. You’ve just said that if I’m not careful I might end up on that land bridge to Europe which academics have long known about. It’s going to take me some time to fit this insight into my current map.
Maybe it’s just caffeine, but I’m really finding this post unsettling.
SL: “We could, I suppose, all work more performing completely useless tasks, but why would we choose to do that?”
To earn a sociology degree?
#37 – “I don’t see how we can be better off if only irrational people vote”
Hard to see how we’re better off with more people voting either randomly or out of pure self-interest either. Voting “irrationally” out of a sense of moral duty doesn’t seem like the worst thing (as long as you don’t actually believe you may be the tiebreaker), and certainly more conducive to making the larger effort of developing a coherent philosophy, analyzing the issues etc. Of course some people may also follow politics for the entertainment / water cooler value (i.e. not irrational). Perhaps your hope is that many of them are then making the rational decision not to stand in line at the polls; to me it seems more likely they’re voting already and more of the ‘nudged’ fall into the first category.
I assume you’re not really serious but compulsory voting also seems either unenforceable or very intrusive, maybe even an affront to free speech; e.g. what if I genuinely don’t like any of the candidates, or want a protest non-vote (which may be equally irrational)?
Compulsory voting already happens in Australia and Argentina and Singapore among others – it is not a joke. You are not forced to endorse any particular candidate (as I understand) – you are entitled to spoil your vote, you just have to cast one. I think this invalidates the freedom of speech argument, since you don’t have to “speak”. It may also be possible to register as a non voter if you have idealogical reasons for not voting. It is of course an infringement of some liberty, but I seriously think it may be a good idea.
Then I’m afraid I will have to resign as your unofficial campaign manager.
I’m late to this party, but I believe Carlton (#13) and Neil (#22) both identify the flawed premise in the argument. Why should we assume that the government will lower tax in order to keep total revenue remains fixed, rather than finding more uses for the additional revenue?
Correction: … lower tax *rates* in order to…