Economists often say that the law should be written to promote efficient outcomes. That’s more ambiguous than it sounds.
Suppose I want to take an action that causes you harm; for example, I want to cut down a tree that you like looking at. How do we tell if that action is efficient?
Definition 1. The action is efficient if my willingness to pay exceeds your willingness to accept. For example, if I’m willing to pay $100 for the privilege of harvesting the tree, and if you’d accept less than $100 to part with it, then the tree-cutting is efficient.
Definition 2. The tree-cutting is efficient if it would occur in a world with no transactions costs (i.e. a world in which there are no impediments to bargaining).
In many circumstances, these definitions are equivalent, and economists often pretend they’re equivalent always — but sometimes they’re not.
Example 1. I want to punch you in the nose non-consensually. (The non-consensuality is a big part of my enjoyment.) I’d pay $100 to punch you in the nose, and you’d accept $50 to take the punch. By Definition 1, the punch is efficient. But the punch would be unlikely to occur in a world with no transactions costs, because it would require bargaining, hence consensuality on your part, which kills my interest. So by Definition 2, the punch is inefficient.
Example 2. I am willing to pay $100 to cut down a tree; you are willing to accept no less than $150 to part with it. By Definition 1, the cutting is inefficient. But part of the reason I’m willing to pay only $100 is that I’m credit constrained. In a world with no transactions costs, I’d borrow more, and would be willing to pay $200 to cut down the tree. So by Definition 2, the cutting is efficient.
Example 3. I am willing to pay $1000 to cut down a tree; you are willing to accept $500 to part with it. By Definition 1, the cutting is efficient. But the only reason I’m willing to pay so much is that I make an excellent living in my job as a mediator who helps people overcome transactions costs. In a world with no transactions costs, I’d be much poorer and would be willing to pay only $200 to cut the tree. So by Definition 2, the cutting is inefficient.
I believe that a version of Example 1 was first suggested by Richard Posner, and Example 2 was first suggested by Steve Margolis; Example 3 (which sort of riffs on the Margolis example) might be new.
Question: When we say that the law should encourage all and only those actions that are efficient, what, exactly, should we mean?
Change the first “inefficient” to “efficient” in example 3.
Daniel R. Grayson: I beat you to it.
SL, I get what you’re asking but I’m not yet convinced that this comes up a lot in the real world. Care to give some more examples?
“When we say that the law should encourage all and only those actions that are efficient, what, exactly, should we mean?”
Hmmm. What are you willing to pay for the answer?
Is example 3 a pecuniary externality and thus not a concern for economists?
I think definition 1 and definition 2 are sufficient for this purpose. Of course these examples illustrate that these definitions don’t always give clear guidance, but I could certainly live with laws guided always by those two definitions compared to what we have now.
CC (3): To say example 2 doesn’t come up in the real world is to say nobody’s credit is constrained by transaction costs, which is to say transaction costs don’t exist.
Going back to your post of 3/20, in Definition 1, don’t I also have to account for Granola McMustardseed, who would pay $30 to keep the tree just on principle? And on Farley Treehater, who believes trees reduce property values, and is willing to pay $10 to anyone willing to cut down a tree in his neighborhoood? I would think that efficiency should account for anyone who had a financial interest, no matter how tenuous.
Sez who? I have it on good authority that people punch each other in the nose all the time without acquiring consent. Indeed, my sources suggest that consensual nose-punchings are much rarer than the non-consensual variety.
Yes, you’d likely incur some liability for the offense. Would the liability be more than $100? Don’t know. So the problem with this hypothetical is not an inability to engage in unilateral nose-punchings; the problem is the failure for people to publish tariffs declaring the price at which they’d accept such a punch.
In contrast, we encounter precisely such tariffs in sports. Consider the following scenarios:
A. The opposing center is driving for the basket.
B. The opposing receiver is preparing to catch the ball and run it in for a touchdown.
C. The opposing coalition’s Ultimate Ascent robot is loading and shooting Frisbees through the slots without ever leaving the zone in which interference by rival robots is prohibited.
In each case, the rules limit the things defensive players may do in response. But the rules also specify the consequences for violating those rules. And often, it’s advantageous to go ahead and violate the rules and incur the penalty. Foul the guy driving for the basket, ‘cuz you know that the penalty is to give him the opportunity to make free-throws, and you know he sucks at that. Foul the receiver going up for the ball; his team may take possession of the ball at the point of contact, but that’s still better than letting him score a touchdown. Slam into the rival team’s robot; you’ll take a penalty, but it may still be better than letting the other team fire off shot after shot.
(Especially if the rival robot’s aiming mechanism is “fidgety;” one good slam will keep them from shooting for a solid minute. And then you can slam ‘em again!
Meet me in St. Louie, Louie – and bring your shoulder pads!)
We don’t need to make up examples as long as
real life is stranger.
Yes, the tree got cut down, and Olerud did not have to pay anything like the value (by his own estimate!) he got from doing it.
Landsburg collects scenarios that purport to pose a conundrum for public policy, suggesting that there is ambiguity in identifying the “efficient” outcome. But if you put each of these scenarios to a vote (from a microecon perspective), I suspect there’d be near unanimous agreement about what the public policy should be – or, at least, what they shouldn’t be.
Yes, the act of securing consent may reduce the utility of some actions to some people. But given a choice between a world in which non-consensual nose-punching is subject to reasonable penalty, or a world in which it is subject to no penalty, I doubt there’s much disagreement about which world would likely achieve better results in aggregate.
Yes, imperfect credit markets represent a market failure that distorts price signals and leads to sub-optimal social outcomes. But given a choice between a world with our imperfect credit markets, or a world in which people can borrow without limit, I doubt there’s much disagreement about which world would likely achieve better results in aggregate.
I don’t mean to discount the possibility that people could articulate some better policy between “status quo” and “no holds barred.”
In contrast to Examples 1 and 2, Example 3 is merely a riff on the ceteris paribus assumption – comparing two scenarios “all else being equal,” but incorporating two changes between the scenarios – one endogenous to the other — thereby frustrating any attempt to link the change in the independent variable to the change in the dependent variable. (Compare: On a quiz show Paddington Bear was asked, “If it takes two men, each with one bucket, one hour to fill a tub, how long would it take one man with one bucket to fill the same tub?” Paddington answered, “No time at all; the tub’s already full.” The second scenario incorporates two changes relative to the first scenario – change in the rate a tub could be filled, and change in the initial status of the tub – which then frustrates any attempt to connect the change in the dependent variable to the change in the independent variable.)
Again, given a choice between a world in which we could achieve efficient outcomes with lower labor costs, or a world in which we must incur higher labor costs to reach those outcomes, I doubt there’s much disagreement about which world would likely achieve better results in aggregate.
Note that I specify “from a microecon” perspective. I can’t discount the possibility that, for example, the economic drag incurred to hire mediators proves to be less than the economic drag incurred to pay to support a social safety net for legions of unemployed mediators. Hard to know how stylized Landsburg intends his examples to be.
It seems the standard we’re using here is really what people are *able*, as opposed to “would be willing”, to pay. (E.g. in ex. 3 do I really value something more just because my income is greater?) Of course the latter raises practical measurement issues, but I’ve always been a little unclear about how this distinction plays into the economist’s appeals to the efficacy of “revealed preferences”. And at least the law does attempt to value things like psychic harm (where granted) irrespective of the victim’s bank account.
Example 1 is not so hard, provided definition 2 is understood properly. The definiens should be read as a subjunctive conditional (usually counterfactual, since the premise, that the bargain has occurred is usually false.)
Consider a similar subjunctive definition and a similar apparent counterexample.
Definition: an object O is fragile if and only if O would break if a force Fi (per m2) were applied to it (for some force we figure is small enough to mean O is fragile if it breaks under Fi per m2).
Now imagine a case where O would break if Fi/m2 were applied except for the fact that O is attached to a machine that will instantly toughen O if Fi/m2 is applied to it.
We do not want to say that O is not really fragile. Of course it is: Fi/m2 has not been applied to it and the de-fragilizing machine has not yet done its work. O is a flimsy little object, just as fragile as a similarly fragile object which has not been hooked up to the machine in question.
The way to save the definition is to add the condition that O retains the property of fragility while Fi is applied. In other words, the definition should be that:
O is fragile iff O would break if Fi were applied while O were still fragile.
Or, more pedantically,
O is fragle iff O has a property P, such that O would break if Fi were applied while O had P.
This rules out cases where Fi/m2’s being applied stops O having the property P (such as a certain molecular structure) which makes it fragile.
Now return to the bargaining definition of efficiency. Bargaining could, in special cases, change the valuations of the bargainers so as to change which outcome is efficient. To save the initial insight of the bargaining definition, we need only add the condition that this not actually occur. Or, in other words, definition 2 becomes:
“The tree-cutting is efficient if it would occur in a world with no transactions costs (i.e. a world in which there are no impediments to bargaining”, *and bargaining does not change the bargainers’ valuations of the outcomes*).
If the bargaining really does happen, and the valuations really do change in the way specified, then cutting down the tree (or punching the nose) is not, really, the efficient outcome. But the revised definition does not say it is, because the extra condition is not satisfied. And when the bargaining does not occur, or does occur but does not have this (unusual) effect, it also gets the answer right.
In short, example 1 creates no special problem for definition 2. Such supposed problems can be created for all “response dependent” definitions (as philosophers call them) and they can all be solved in the way I have described (not that many philosophers recognise it!)
Question: When we say that the law should encourage all and only those actions that are efficient, what, exactly, should we mean?
My guess is that we should mean definition 1.
To answer the question needs some theory of ethics to define “shouldness”. If we say “X should happen if it makes the world, on average, better off” or some approximation to this, then “X should happen iff X is efficient” only holds under definition 1, not definition 2.
On another note, are there transactions that are efficient under definition 2, but not under 1? What about “you want the tree kept, but would accept $200 to chop it down. I want it chopped down, but am only willing to pay $100. However, in a world with no transaction costs, I’d be much wealthier, due to the economic rents I’d be able to earn from my inborn talent at basket-weaving, and would happily pay $500….”
On another
A side issue here, but slightly disturbing to me. If I would accept $50 for you to punch me on the nose, and you would offer $100, then the punch is efficient. The punch remains efficient even if there is no actual transaction. Therefore you can efficiently punch me without paying me – is this right?
@Harold 14:
Only if you work for The Wall Street Journal.
Ha!
I hadn’t really thought about this much, but I suspect for most people bargaining is a HUGE transaction cost. I suspect most people are so accustomed to being price-takers and have such anxiety around money (Do I look cheap? Do I look naive?) that they resist entering into explicit negotiation situations.
Auto dealers regularly advertise their “no haggle pricing” as a benefit for consumers. eBay is finding that people are much more drawn to “Buy It Now” offers than their traditional bidding structure.” And I’m reluctant to ask the guy standing next to me at the bus stop to refrain from smoking. If he wore a button saying, “Will stop smoking for $0.25/hr,” we’d have a deal. Heck, for all I know, he’d be perfectly willing to acquiesce for free — but I’m unwilling to initiate the negotiation. And so I live in smoke.
(But I may feel the compunction to haggle over a recent financial aid offer….)
I surmise that economic theories based on “low transaction costs” in practice must refer to situations in which prices arise through a very structured process — especially, from a menu.
Harold – I think the terms of the transaction define the efficiency of the punch so the two are tied together. You perceive $50 of harm so you would not accept $49, much less $0.
On the broader issue, I think if the idea is that *the law may serve as a way to effectively reduce transactions costs*, then it seems we do want to think seriously about a world that conforms to Def 2 (beyond just tweaking the definition).
Counter-example #1 seems too circular / contrived so not that compelling – we want a situation where somehow we can ensure consent exists (and value it!) without making the puncher aware.
#2 involves a credit constraint (but not necessarily an issue of income or wealth), so something we’d potentially want to try to look through to determine value / benefit under Def 2.
#3 is a “pure” look at how varying income level should affect the determination of value.
This gets me back to my previous question – is the right basis one’s practical ability, or (theoretically unconstrained) “willingness”, to pay? I suppose ways in which utility varies with wealth would make this a trickier problem.
Sorry about the wonky stuff in my above comment about force and the areas it is applied to. DON’T SUBMIT COMMENTS WHEN YOU GET HOME FROM A BIG NIGHT OUT! Anyway, the substantive point holds. But it can be better expressed.
Assume that Force = Mass x Acceleration. Then an object has mass 1kg iff it would accelerate at 1m/s2 if 1 newton were applied to it.
But here is an apparent problem. Imagine an object with mass 1kg that is so fragile that, if a force of 1 newton were applied to it, it would not acclerate at 1m/s2 because it would dissintergrate.
But it is not really a problem because, in such cases, the object stops having the propoerty of weighing 1 kg when the force of 1 newton is applied to it (a 1kg item that disintegrates is no longer a 1kg item). So it is not really a counterexample to our Newtonian definition of mass. To make this clear, we restate the definition thus:
an obejct has mass 1kg iff it would accelerate at 1m/s2 if a force of 1 newton were applied to it AND it retained the property of having a mass 1kg.
(For those concertned about circualrity in definitions, state it thus:
x has mass 1kg iff x has a property F such that if a force 1 newton were applied to it while x had F then x would acclerate at 1m/s2.) Some may worry about the counterfactual nature of the definition, about the fact that x would NOT have mass 1kg if 1 newton were applied. But that does not matter. Provided the relationship between a force beling applied to an objectand its disintegrating is contingent, as it is, then the following statements are consisitent:
If Fx and Mx, then Ax.
If Fx then not-Mx.
The connection between bargaining with Sam to punch him in the nose and how much I value punching the unwilling Sam in the nose is contingent. So the above solution to the apparent problem presented by example 1 should work.
@nobody.really #16 I’d also like to know what an economist means by “transaction costs”.
Do they, for example, include shipping? the extra money I forked out for a car with a large boot? Gasoline? Do they include Apple’s 30% cut on the App Store? Do they include (surely yes) the internal discomfort you must overcome in order to initiate bargaining?
What does a world without transaction costs look like?
Would all externalities disappear in such a world because of the instantaneous negotiations possible? Eg, between 1000 farmers stretched across a nation and a railroad company over sparks from trains? Between an entire generation and myself over my choices on further education? Between the 9 billion residents of 2075 and the 7 billion of 2013 over our Carbon Dioxide emissions?
I am wondering if “no transaction costs” means we must all be omniscient beings in constant instantaneous communication with each other across both space and time – some kind of eternal “hive mind”. Not recognisably human.
I suspect, rather, that when economists talk about “a world with no transaction costs”, they mean this in a quite limited sense. This limited sense may well mean Steve’s Example 3 is a straw man example.
Iceman #17 – I suffer $50 damage, but Steve gains $100 benefits – surely “efficient” to the tune of +50? If Steve were to pay me $50, then I would be even, and Steve would lose $50 cash but gain $100 benefit – “efficient” to the tune of +$50. It is the same whether or not the transaction takes place.
Mike H is on to something. If there were no transaction costs at all then externalitites disappear. What about costs of information? The actors are not omniscient, but is this lack of information a transaction cost? If so, then the efficient outcome is what would have happened IF everybody were fully aware and informed.
My view is that the world without transaction costs defined in this very broad way is the *real* efficient outcome. It does get a bit messy when we consider trading with the future the future – after all, this is not so much a “cost” as an impossibility.
So back to the examples with this definition in mind. In example 1, it would be efficient for the transaction to occur (i.e. the punch) because that leaves everybody better off afterward than before. How we would arrive at this oucome is a practical difficulty.
Example 3 depends whether “transaction costs” mean the costs of that particular transaction, or more generally. It would I think be valid to take either approach as long as you are clear about it. Thus we could say given that I have this much money, what is the eficient thing – then it would be cutting the tree. Or we could say we must include what would have happened in all transactions ever if there had been no transaction costs – in which case we are unlikely to have arrived at the tree cutting scenario at all. It makes little sense to single out the fact that my money has been made through transaction costs without considering the other effects.
In example 2 I am not sure how much of the credit constraint is due to transaction costs.
I’ve been pondering this, too. How do “perfect” credit markets operate, such that we’d be able to distinguish them from existing credit markets? Presumably these markets would not simply let people borrow 100% of all available assets upon request, right?
Would a “perfect” credit market require perfect knowledge — at least, perfect knowledge of the past and present?
(And if you believe in determinism, does perfect knowledge of past and present also imply perfect knowledge of the future? And if people had such perfect knowledge, how much lending would occur? Wouldn’t people with money to lend simply make equity investments instead?)
The answer to “What is an efficient policy” is always dependent on the distribution of income, at least when preferences are different. It is not the case that tree cutting is always efficient or not efficient. Efficiency is contingent. If a tree hugger wins the powerball, it becomes efficient to cut less trees. I think the law can only promote the policies that are efficient given the current facts, so one should not look for timeless rules.
Harold – but we called this a “transaction”, not a license to steal anything you think you might value more than the holder. (Perhaps this is yet another trap for the utilitarian, but that seems like a different topic.) Any outcome that involves you not fully compensating the punchee would not fall within the set of *mutually* agreeable outcomes.
All – I’m not sure we have to require perfect information etc. to ask (what I think the question boils down to) if we can do any better than relying solely on ability to pay and revealed preferences, as opposed to ‘pure’ willingness to pay and maybe valuing things in terms of other things rather than units of income. Maybe we simply can’t. But as long as the goal were to facilitate transactions that would be mutually agreeable, I don’t think it would take me in directions I don’t wish to go.
It seems eminent domain is an example of the law attempting to reduce transactions costs by looking past ‘revealed preferences’? Unfortunately addressing the holdout problem has opened the door to govt abuse a la Kelo. But a practical example of transactions costs is where negotiators attempt to conceal their true preferences. The ‘solution’ is not omniscience, but the law applying a “reasonable person” standard.
Iceman 24. There are two issues here – agreement and efficiency. If I would accept $50 to be punched, and you punched me, then I would be able to agree to you paying me $50 not to prosecute. However, does this only affects the distribution, not the overall value, or the efficiency. How are we to decide in instances where the compensation cannot be paid because of transaction costs? Using the tree example, it may not be possible to identify hundreds people who would accept $0.5 each to allow you to chop it down. The economic argument I think would still say that it was efficient if you were prepared to pay more than the objectors were prepared to accept, even if they were never actually paid.
I thought the initial talk about bargaining and transactions costs presumed consent. In that context the obvious problem with “paying me $50 not to prosecute” is at that point I’m motivated to conceal my true preferences (transactions costs) and sue for more. Especially if I can invoke psychic harm.
But yes I agree that if we set aside all notions of individual / property rights and thereby dispense with the niceties of agreement, any redistribution that increases total utility could be viewed as efficient. I flirted with this a little myself in exploring a distinction between ability and “pure” willingness to pay (still wish I better understood the economist’s take on this). But I hope that idea makes anyone squirm a little. In fact, it set off a firestorm in another recent post when taken to conclusions. Even a die-hard utilitarian can recognize longer-term costs and so construct a notion of property rights purely for pragmatic purposes. It also seems part of the appeal of Friedman’s argument is that one needn’t declare sides on the rights issue to say we can minimize real-world transactions costs by applying a property rule in some situations and a liability rule in others — e.g. the former to you taking an axe to a tree, the latter to you meandering through a crowd with it.
Iceman: “I flirted with this a little myself in exploring a distinction between ability and “pure” willingness to pay” I am not sure if I am on the right lines here about what you mean here, but SL’s definition of efficiency above was that my willingness to pay was greater than your willingness to accept. Thus we are not competing over who has the most money directly. I could be penniless but still not accept a very large sum to give permission for a tree to be cut down. However, this has problems, because it is quite possible that one person would not accept all the money in the world to cut down a tree, in which case its removal could never be efficient. It is of course not possible to determine exactly how much anyone would actually accept without a genuine offer.
If we go back to McCrankypants and his pornography dislike, which would be the efficient outcome? Either we could ask how much McCrankeypants was willing to pay to prohibit pornography, or we could ask how much he would be willing to accept to allow it.
But the post talks about credit constraints or level of income affecting one’s “willingness” to pay. I agree asking how much one would *accept* sidesteps the issue, but by focusing on one side of a transaction which also seems to take us back to assigning a property right (otherwise why is “permission” required?). And since the post was about what our legal system can do, one thing it’s probably not going to do in the real world is disregard ownership issues. Your tree lover as “utility monster” is (only) problematic if you’re a utilitiarian. Of course if the guy owns the tree he is entitled to refuse all offers…at least until we invoke eminent domain and force him to take a “reasonable” offer for the greater good. I previously offered that as an example of the law looking through revealed preferences to reduce transactions costs.
My question is basically whether the economist’s focus on “revealed preferences” — which it seems by definition reflect ability to pay not necessarily unconstrained willlingness i.e. potential utility — is based on theoretical or practical considerations?