After I ranted this to my friend Steve Roth, he suggested I write it up. In substance, it's a retread of my welfare economics series, especially the second installment, Perils of Potential Pareto.
That series is the best economics writing I've ever done! But this is shorter and more ranty.
In the beginning, there were they were just Benthamites. The greatest good for the greatest number, all of that. But they were always ambitious fucks. It was never enough for them to have a view, their view had to be writ, had to be science. You can claim that this or that action produces the greatest good for the greatest number, but how do you measure it? How do you prove it? Suppose you with the same resources you can feed ten people or entertain a thousand. Feeding a hungry person is a greater good, perhaps, than entertaining a fat one, but how much greater? Is it only 90 times the greater good, in which case the resources should go straight to Broadway, or 101 times the greater good, in which case the soup kitchen should get it? Benthamites could opine, but they could not measure. They had no means by which to persuasively contradict cretins who would make the opposite call.
So, they declared economics an empirical science. "Henceforth," they shouted, about a century ago, "we shall base our conclusions only upon what we can (at least in principle) measure!" Utility, the phlogisten of the Benthamite creed, the hypothetical substance of good that was to be maximized across a population, was recast as "ordinal". We cannot know whether one hungry person's meal is worth more or less than 100 feted people's entertainment. But we can know that the hungry person, given a choice, will choose the meal over the musical. We can order and rank the value of goods to an individual, even if we can't measure the distance or difference between them, or make comparisons across different people. And, thanks to our man Pareto, we can offer a scientific, empirical, can't-dispute-this-motherfucker account of the good. If some action or policy will make at least one person better off (i.e. they would choose their new circumstance over their old), and make no other people worse off that, that, is indisputable an improvement. Here is a criterion that involves only observable (at least "in principle") rankings, and that requires no comparisons of unmeasurable welfare across different people! Hurrah! Those cretins have to listen to us now.
Only. At the level of public policy, where economists aspire to rule (for the betterment of all, of course), there are almost no actions that meet the Pareto criterion. If you want to build a new public park for free, there is always some motherfucker who loves the wild shrubbery already there and hates the idea that loud, laughing children might interrupt his afternoon idyll. If (as economists always did) you want to open the borders to trade, there is always someone who will be put out of work by cheaper imports. The Pareto criterion was arguably scientific, but, as John Hicks famously put it, this "economic positivism might easily become an excuse for the shirking of live issues, very conducive to the euthanasia of our science." And without our science, all that will be left are the cretins.
So, Hicks, along with Nicholas Kaldor and Harold Hotelling came up with a workaround. Render unto the cretin the things which are the cretin's, but unto the economist the things which are the economist's. The economist cannot make claims about who should get what, the ten hungry people or the thousand fat circus-goers. That is a judgment of values. It is beyond science, so must be left to the cretins and their politics. However, we can observe whether some action would create a situation where if you taxed some of the people who would very much prefer the new situation and gave the proceeds to the people who would otherwise be grumpy about it, then everyone would prefer (or at worst be indifferent) to the new reality. If economists require the cretins — err, politicians — to actually perform this redistribution in order for it to be scored "efficient", then it is just the Pareto criterion. But, of course, economists knew very well that politicians would not in fact usually redistribute to make real some hypothetical Kumbaya. So they washed their hands of it. If everyone could have been made better, or at least no worse off, by an action if some redistribution had occured, then clearly the action itself "makes the pie bigger". We can score it as scientifically more efficient. That the cretins are unwilling to do the distribution? Hey, that's a values question, not our problem. Economists again could don the authority of science to tell us all what must be done. NAFTA is efficient! You don't understand arithmetic if you disagree.
But oh no! It turns out (curse the name of Tibor de Scitovsky!) this criterion of potential Pareto improvement is not an ordering. There are actions where, if we were to perform the action, the "winners" could compensate the "losers", and they would all be better off. But if we do not actually compensate the losers, reversing the same action and letting the initial losers compensate the undoing of the winners might also make everybody better off! (For a detailed treatment of this, follow the pictures here.) It is incoherent to declare something an efficiency improvement and then declare its reversal an efficiency improvement, but that is the cul de sac our poor scientists had backed themselves into. To this very day, economists do not have a practical criterion by which to declare an action or policy authoritatively "efficient", should cretins have the temerity to disagree. (There is the Pareto criterion, of course, in theory.)
To really twist the knife, one economist examined the cycles Scitovsky had discovered and concluded this problem was very general. In a Rand Corporation note that was the foundation of a celebrated, seminal result, this economist acknowledged, "the compensation principal does not provide a true ordering of social decisions. It is the purpose of this note to show that this phenomenon is very general."
That economist was Kenneth Arrow, and if you've taken any economics at all, you've probably learned about the result that emerged from the Rand note. It is known as Arrow's Impossibility Theorem. But when you learned about that result, I bet you didn't learn that its original object was the impossibility of authoritative technocracy. Social actions cannot in general be ranked under the ordinal utility constraint economists impose upon themselves. Instead, you learned that it was about the impossibility of a good voting system, if we impose the constraints economists invented for themselves upon how we conceive of voting systems!
In other words, after struggling for decades to come up with a basis to make authoritative claims beyond politics, the economics profession realized by about 1950 that, at least on the terms they had prescribed for themselves, it was impossible. They had no criterion, and could have no criterion, by which they could elevate their conclusions as "scientific" and above the fray. What was to be done? Well, we know what they did.
They took the weapon they invented, this theorem that damned their own authority, and they pointed it at the rest of us. They stopped talking about the impossibility of ranking social actions, but loudly proclaimed their theorem foreclosed the possibility of coherent democratic choice. If someone must be silenced by this discovery, it would not be our profession, it will be the cretins, the quarrelsome public with its crass politics.
Our voting systems as we actually understand them are not, in fact, touched by the impossibility theorem, because we generally conceive of votes in cardinal rather than ordinal terms. Economists forbid themselves interpersonal comparisons, but in politics we have no trouble asserting that my vote expresses a preference of the same weight as yours. What economists foreswore because they could not measure, in politics we impose as an axiom on normative grounds.
We are taught that Arrow's theorem condemns democracy when in fact it condemns economic technocracy. That doesn't mean voting systems can be made perfect: It remains true that no voting system meets all of the many, sometimes conflicting, criteria we might desire of one. In particular, no deterministic voting system is immune from strategic voting when there are more than two candidates. (We should consider stochastic voting systems!)
Still, I find it breathtaking. This very result that economists understood perfectly well meant they had no sound basis for judging social action with authority beyond other approaches, they (we?) marketed as damning the other approaches.
Economists are such scoundrels.