Friday, March 21, 2014

Can the Illogic of Orthodox Economics Be Overcome By Being Comparative or "Practical"?

-- Posted by Neil H. Buchanan

In a post here on Dorf on Law a few weeks ago, I returned to the question of whether there is any coherent meaning to the concept of economic efficiency.  There isn't.  A few years ago, after having spent many years trying to get people to stop calling things efficient or inefficient, because the terms have no content, I gave a talk at a conference in which I announced a change of strategy.  Specifically, I told people to call everything that they like efficient and everything that they dislike inefficient.

This suggestion, although admittedly cheeky, was based on two ideas.  First, this is what orthodox economists do all the time, apparently in most cases without being aware that they are doing so.  And second, a person who takes my advice would always be right (even though she would also always be wrong).  That is, there are defensible assumptions and baselines that can make any situation or policy appear to be efficient or inefficient.  If I call Policy A efficient, I am thus both right and wrong.  So why be a pessimist?!  If it can be right to call something I like efficient, then why not do it?

This would, I hope, have the additional benefit of devaluing the word efficiency, so that perhaps over time people would stop using it, because it would no longer communicate anything useful.  Kind of like "freedom," "hero," and "ideas guy."  Orwell would be proud.

As I discussed in that recent post, my anti-efficiency argument proceeds from the same premise that motivated the now-famous Murphy/Nagel argument about taxes.  Because there is no economy without a government to define and enforce the laws of property, contract, and so on, and because different (completely legitimate) choices about what those laws might entail lead to wholly different social and economic outcomes: (1) It is logically incoherent to talk about a "before-tax" amount of income, meaning that no one (not the individual, not the society, not the government) owns before-tax income, because there is no such thing as before-tax income, and (2) There is no single, meaningful baseline from which to measure efficency.  Point (1) is the Murphy/Nagel point (versions of which others have made).  Point (2) is my point (versions of which others have made).

One possible objection to my point (and possibly to Murphy and Nagel as well) is that I am being too demanding.  Even if there are different hypothetical worlds in which we could live, each of which could lead to different conclusions about what is efficient or not, we only live in one world.  Can we not define efficiency in a meaningful and unique way in the here and now?  The answer is no, for two reasons.  First, taking current endowments, laws, and so on as the proper baseline simply locks in the very rules that we are supposed to be debating and considering changing.  If we define efficiency against a baseline that takes current distributions as presumptively correct, then of course any deviations from that baseline (most obviously, redistributive fiscal policy) will necessarily be inefficient.

More fundamentally, the second reason that we cannot simply take "today's real world" as the proper baseline is that doing so requires deciding which parts of today's reality are up for debate, and which are not.  Unless we simply rule out any changes at all, in a Panglossian everything-is-for-the-best sense, then we have to have some decision rule to say that some property laws can be changed, while others cannot, or that antitrust enforcement will be aggressive or not, and so on.  Economic theory provides no such decision rule, and the economists who make grand pronouncements about the inefficiency of government policies (like the guy who was upset about the ACA's effect on labor supply decisions, whose arguments I ridiculed in my earlier post) simply make unspoken choices about what is changeable and what is not.  That is not science.  It is merely opportunistic advocacy hiding behind a pretense of objectivity.

Perhaps, however, there is an even more stripped-down version of efficiency that could be available.  Rather than obsessing over baseline endowments of laws, talents, opportunities, inherited wealth, and so on, maybe there is a way to compare two policies, or two choices that an individual might make, such that one could say that one choice is more efficient than the other.  That is, even if it is not meaningful to say that A is efficient and B is inefficient, maybe it could be meaningful to say that A is more efficient than B, without getting into the whole baseline mess.

For example, people often think of efficiency in the colloquial sense rather than the economic sense.  A person can say, “My car is using its gasoline efficiently” without making assumptions about underlying tort laws or international treaties, for example.  Of course, even in this example, the speaker is making background assumptions, such as how many miles her car—or perhaps cars in general—can travel on a mile of gasoline.  But as a friend recently put it to me: "Sure, we can’t say whether a Prius is efficient in an absolute sense, but we can say that a Prius is more efficient than a Hummer."

Actually, we cannot even say that.  Or, more precisely, the sense in which we can say that is true only under a different use of the word efficiency.  In orthodox economic theory, there is an important distinction between "technical efficiency" and "economic efficiency."  Technical efficiency means using available technology in the way that gets the most output from the least input.  If, for example, it is possible to securely seal food cans using 3 ml of molten metal, then using 4 ml is technically inefficient.  A Prius is more efficient than a Hummer in that sense, which means that you can get from Point A to Point B using fewer gallons of gasoline.

The problem is that these stylized comparisons only work, even on their own terms, if we have already agreed that the only thing we care about is the amount of gasoline burned per mile traveled, for that particular task.  A Hummer can be more efficient, even on technical grounds, if the task is different, e.g., not just getting a person from point A to point B, but for getting a huge load of material from A to B.  It depends on what you are trying to accomplish.  Still, one could object that it is possible in each of those discrete situations to describe one choice as more efficient than the other.  For moving one person, a Prius is more efficient than a Hummer, whereas for moving a load of iron ingots the same distance, a Hummer is more efficient than a Prius.  Even that, however, does not save the notion of economic efficiency.

As I described in a Dorf on Law post last August, one can describe something that is much more expensive as being more efficient, if one cares about a different set of criteria.  In an example that I noted there, it is possible that the (very expensive) traditional military academies are more efficient than Officer-and-a-Gentleman-style officer training schools, if one believes that the academies provide a different product.  Which is more efficient?  It is not enough to say that one is more efficient for one purpose, and the other is more efficient for another purpose, because you have to agree on the purpose to decide whether you want to continue to support the more expensive option.  Certainly, it is not meaningful (even in the narrow economic sense) to describe the academies as inefficient -- or even less efficient -- merely because of their price tag.  The Prius/Hummer example at least allowed us to say, "If I know the technical facts, I can tell you which one will use less gas for the task at hand."

Even in the military academies example, one could make a similar move: "Well, one training system is more efficient for one set of goals, and the other is more efficient for the other set of goals."  But that is precisely the difference between technical efficiency and economic efficiency.  Economic efficiency means that one chooses from among the technically efficient options to maximize net social welfare.  Comparative efficiency means choosing between the options under consideration based on which option would have higher net social welfare.  But measuring net social welfare in a consistent way requires an agreement about what to include in our measure of social welfare.  And then we are back to the baseline problem.

Another way to put this is that it is possible that choosing between two policies can be misleading, because one can choose a policy on the basis of an incomplete understanding of the full implications of that policy.  For example, I have argued in the past that the U.S. should increase infrastructure spending.  What if that spending is used to build more roads?  Is that more efficient than not spending at all, or not?  Is it more efficient than using it to build better public transportation?  Than using it to allow more people to work from home?

I think that I could make defensible arguments about whether and how to spend infrastructure dollars, but that is only because I know my prior commitments.  But the point is that I could plausibly say, as a comparative matter, that we should spend more on roads rather than spending nothing, only to find that I have moved the economy in the wrong direction, increasing pollution and sprawl, reducing the usage of existing public transportation, and so on.

Yes, there are situations in which it will seem easy or obvious to compare and choose between a limited number of options.  Those situations, however, will be easy or obvious only because we will have (probably unconsciously) decided that the full consequences of those decisions -- the spillover effects -- are unimportant or are likely to be insubstantial.  And we might be right.  But this kind of atheoretical move is not really a matter of saying that one policy is more economically efficient than another.  It says only that we can sometimes agree to close our eyes and hope that we are not making matters worse.

Efficiency in an economic sense is, therefore, not even meaningful as a comparative term.


Sam Rickless said...

Hi Neil, I wonder whether I might make a distinction that preserves at least some of what you want to say without ending up with as radical-sounding a conclusion. As philosophers might put it, the distinction is between metaphysics and epistemology, between reality and our knowledge of it.

I'm not sure exactly what it means to say that a *concept* is meaningful, but it seems to me that the phrase "economically efficient" has a meaning. As you put it, "economic efficiency means that one chooses from among the technically efficient options to maximize net social welfare". The thought is that the world, such as it is, fixes a set of means (or tasks) that will maximize NSW. An economically efficient state is defined by the most technically efficient of those means (relative to the end of maximizing NSW). I don't find anything incoherent in this. Of course, it matters greatly how NSW is actually defined. As you say, we need agreement on what to include in our measure of social welfare. But I don't see this as a "baseline problem", unless social welfare is defined relative to a baseline. And, at least at first blush, I don't see why social welfare should be so defined.

You say that "another way to put this" is that it is possible to choose a policy on the basis of an incomplete understanding of the full implications of that policy. But this is an epistemological point. That is, assuming that the concept of economic efficiency is coherent, it is going to be difficult for us to *know* (or determine) which of several alternative policies is the most technically efficient way of maximizing net social welfare. But this doesn't mean that "economically efficient" is not a meaningful term.

Here's an analogy for you. There is surely a fact of the matter concerning the heaviest salmon swimming in Pacific Ocean right now. The phrase "heaviest salmon swimming in the Pacific Ocean right now" is certainly coherent and meaningful. At the same time, it is really really difficult to *know* which salmon is in fact the heaviest one in the Pacific Ocean right now.

Neil H. Buchanan said...

Thanks for this input, Sam. You've identified one side of the impossibility, which is that we cannot hope to correctly measure all of the factors that determine an economically efficient outcome, even if one exists. BUT, the problem is that even if one exists (and a lot of high-powered mathematical modeling has been devoted to showing that unique equilibria do NOT necessarily exist), the problem is that one exists for every possible baseline, and there are an infinite number of possible baselines.

That's why the notion of economic efficiency is incoherent -- not because it's impossible to describe what an economically efficient situation would entail, but because there are an infinite number of economically efficient situations, one (at most) for each baseline. So, when I say that "economically efficient" is not a meaningful term, I'm not saying that one cannot describe an economically efficient situation, based on a given baseline. I'm saying that it's meaningless to call one situation efficient and another inefficient, because the efficient one is inefficient under other baselines, and each inefficient one could be efficient under possible different baselines.

Jimmyd said...

I am amused and bemused. I am amused because it was this precise problem that was one of the motivators for me to leave my professional career as an economist 20 years ago. I'm bemused because I didn't realize that rational people needed to keep talking about this issue. People who insist economic efficiency is a meaningful concept are talking out of their ass. That phraseology is more vulgar but has the virtue of conciseness.

By the way, I'd add that there are those who would argue that even "technical efficiency" is a meaningless term insofar as we live in a world where events are not precisely reproducible. So the baseline problem exists there too. If one wants to get physicists all in a lather ask them how atomic reality hooks up with macroscopic reality (quantum mechanics vs classical physics). There is no agreement.

For me, I'm willing to accept that a Pirus is more efficient than a Hummer in terms of gas mileage because that is what I care about. That is not an economic argument but one that ultimately sounds in a personal value hierarchy. And the fact that we care about gas mileage as a society (see the window decals at your dealership) sounds in democratic politics. But of course such notions of efficiency are far to amorphous and hand-wavey for a profession that has scorned its roots in political economics and fallen in love with harlot mathematics.