#robert murphy

Oh shit.

#OTD in 1978 – La Mon Restaurant Bombing | Twelve people, all Protestant civilians, were killed and 23 seriously injured when an incendiary bomb exploded at the restaurant of the La Mon House Hotel, Gransha, near Belfast.

The La Mon restaurant bombing was an incendiary bomb attack by the IRA on 17 February 1978 that is widely considered to have been one of the worst atrocities of the Troubles. It took place at the La Mon House hotel and restaurant near Belfast. The IRA left a large incendiary bomb, containing a napalm-like substance, outside one of the restaurant’s windows. There were 450 diners, hotel staff and…

In a free society where pieces of property are all assigned ownership to specific individuals, there would be no such thing as a person having his "right to walk around" revoked — because there's no such thing as a "right to walk around" in the first place. Rather, what could happen is that someone is considered so dangerous that all of the reputable health agencies place him at the top of their lists, and they hold news conferences, send out emails and faxes, etc. to alert the relevant owners to look out for this person. Major property owners would probably have prearranged agreements on how to deal with cases like this, so that the response could be coordinated.

Private businesses aren't stupid; they don't need the government to order them to keep lepers away. And if a particular church, say, wants to open its doors to such a person, that's perfectly within their rights. (As a matter of courtesy, we would hope this policy would be announced to others who might not want to visit the same building.) Indeed, the final repository for such people would be buildings where the owners thought they could safely contain the disease. And the common name people would use for these buildings is "hospital." In a free society, to be "quarantined" would simply mean that most owners (of roads, sidewalks, malls, hotels, factories, etc.) would refuse access, and so a contagious person would have few choices outside of treatment facilities.

In any event, the bogeyman of a self-centered fugitive carrying a highly contagious disease is a bit unrealistic. As Andrew Speaker himself has said, he was furious when he was told that he might be a danger to others — because had he known that, he wouldn't have been in the same room as his fiancée's young daughter! Those with communicable diseases are people too, and they don't want others getting sick. On top of that, if they want to be treated, they need to go to the hospital anyway.

CONCLUSION

The free market could deal more effectively with contagious diseases, just as it beats the government when it comes to computers, cars, and crops. The very idea that we should give the government the right to lock somebody up because it classifies him or her as a health risk sounds pretty sick to me.

#OTD in 1978 – La Mon Restaurant Bombing | Twelve people, all Protestant civilians, were killed and 23 seriously injured when an incendiary bomb exploded at the restaurant of the La Mon House Hotel, Gransha, near Belfast.

The La Mon restaurant bombing was an incendiary bomb attack by the IRA on 17 February 1978 that is widely considered to have been one of the worst atrocities of the Troubles. It took place at the La Mon House hotel and restaurant near Belfast. The IRA left a large incendiary bomb, containing a napalm-like substance, outside one of the restaurant’s windows. There were 450 diners, hotel staff and…

#OTD in 1978 – La Mon Restaurant Bombing: Twelve people, all Protestant civilians, were killed and 23 seriously injured when an incendiary bomb exploded at the restaurant of the La Mon House Hotel, Gransha, near Belfast.

The La Mon restaurant bombing was an incendiary bomb attack by the IRA on 17 February 1978 that is widely considered to have been one of the worst atrocities of the Troubles. It took place at the La Mon House hotel and restaurant near Belfast.

The IRA left a large incendiary bomb, containing a napalm-like substance, outside one of the restaurant’s windows. There were 450 diners, hotel staff and…

#OTD in 1978 – La Mon Restaurant Bombing: Twelve people, all Protestant civilians, were killed and 23 seriously injured when an incendiary bomb exploded at the restaurant of the La Mon House Hotel, Gransha, near Belfast.

The La Mon restaurant bombing was an incendiary bomb attack by the IRA on 17 February 1978 that is widely considered to have been one of the worst atrocities of the Troubles. It took place at the La Mon House hotel and restaurant near Belfast.

The IRA left a large incendiary bomb, containing a napalm-like substance, outside one of the restaurant’s windows. There were 450 diners, hotel staff and…

Robert Murphy and David Friedman are having a debate about “ordinal” vs “cardinal” utility. Murphy favours ordinal, Friedman favours cardinal. In this post I’ll explain what these terms mean and argue why I think Murphy is right and Friedman is wrong. I make this post with the greatest respect to David Friedman, whose writings have influenced me massively.

Utility as used by utilitarian philosophers

The term “utility” in economics is an unfortunate historical artifact. It was introduced by utilitarian philosophers to refer to a calculus of pain and pleasure. Even here, what “utility” actually means is a bit vague. Setting aside whether utilitarianism is really the “right” morality we should seriously wonder whether the quantity of “total utility” exists at all. With something like pain, we could try to lay down a rule which says that “quantity of pain = number of times C-fibre fires in the brain” and this way you could compare the amount of pain in one person with another.

But already right out the gate we are running into problems. Is it really accurate to identify pain with C-fibre firing or any other fixed brain process? For example, suppose we discovered octopuses don’t have C-fibres, their biology being very different from ours. Yet if you kicked an octopus (this blog does not condone animal cruelty) it might squeal and make other signs suggestive of pain. Would you say “it isn’t really experiencing pain because no C-fibre is firing.” Most people would say no, the squealing and other noises and movements would be plenty of evidence of pain. But if so, if instead when the octopus squealed chemical X was released in its brain, how to we compare C-fibre firing in humans with chemical X in octopuses? What exactly is the “exchange rate” between chemical X and C-fibre firings so that we get equivalent pains? Do not try to answer this question. The question is unanswerable because it is a nonsense question.

The same issue arises both within and between humans. What is the exchange rate between sharp stabbing pains and dull throbbing pains? Is a stabbing pain in your arm equal to a stabbing pain in your leg? And what about pleasure -- supposing that pleasure could be identified with a single chemical like dopamine or something. How many grammes of dopamine plus how many C-fibre firings equals zero net utility? Again, do not even attempt to answer this question.

I dislike the whole project of utilitarianism. But even if the aims of the project were acceptable it’s important to recognise that no such quantity as “utility” exists. No psychology or neuroscience textbook will mention it. There just plainly is no fact of the matter about what “total utility” is. It is not just “hard” or “impractical” to measure because brain scans are expensive. An omniscient being could not measure it.

Utility as used by economists

The reason people fall for the idea that there is some exchange rate between dopamine and C-fibre firing is because people do in fact make tradeoffs between different feelings. It is then falsely assumed our choices represent a genuine exchange rate that exists “out there.” It reminds me of Marx’s discussion of the differing grades of labour, where he says that because firms do make tradeoffs between different kinds of labour, labour must really be a homogeneous quantity after all.

But even if the sum over pleasures minus the sum over pains could be made meaningful, it simply would not be relevant to economics. The relevance of this quantity to economics is based on a pun. Pleasure and pain are only relevant insofar as a decision maker tends to choose pleasure over pain. The “utility theory” as it appears in economics textbooks is equivalent to a consistency assumption on behaviour, having nothing to do with psychology. Contra some Austrian economists, this consistency assumption isn’t tautological. As demonstrated by Samuelson’s work on revealed preference theory, consistency has empirical content and could be violated.

It was first recognised by Vilfredo Pareto around the turn of the century that all talk of a “utility function” could be translated without loss into talk of a “preference relation” \( \succsim \). The individual then prefers A to B if and only if the utility of A is greater than the utility of B, and is indifferent if the utilities are equal. Conversely, given a rank order of outcomes \( A_1 \precsim A_2 \precsim A_3 \precsim \dots \) we can just choose a sequence of numbers \( u_1 \leq u_2 \leq u_3 \leq \dots  \) and then say the utility of outcome \( A_i \) is \( u_i \). Any other monotone sequence would also work. Things get a bit more complicated when the possible outcomes form a continuum, then a continuous utility function may not exist without some continuity assumption on the preference relation itself, but this is a technicality.

The advantage of the “utility” point of view is that it allows you to use the tools of differential calculus. It gives you nice equations like (Marginal utility of X)/(Marginal utility of Y) = (Price of X)/(Price of Y), and nice equations expressing the demand function in terms of the indirect utility function. But as before, the utility function is merely a non-unique way of codifying the preferences. It is a bookkeeping device. It does not cause the preferences, only psychology and biology can do that. If you think utility does enter into causal relations with choices you have to explain why no scientist has ever discovered it.

As an aside, I would defend the idea of “indifference,” something Rothbard attacked. Indifference does not require any psychological assumptions to define, it is a limiting concept when one has a continuum of choices. If a person is indifferent between 1 litre of wine and 2 litres of water, this only means that any amount of water more than two litres is preferred to 1 litre of wine; any amount less the wine is preferred. The fact that people only make finitely many choices in their lives does not undermine this argument because preferences are a matter of how a person would choose in a possibly infinite number of different circumstances.

The nonsensical nature of interpersonal utility comparisons can be seen from the following example. Suppose individual A has preference relation \( \succsim_A \) and individual B has \( \succsim_B \). Now add their preferences. Well, you can’t. It’s clear that this is an undefined operation. You could, if you like, assign a utility function \( u_A \) to person A and \( u_B \) to person B, and add them \( u_A + u_B \). But this makes no more sense than adding your number of heartbeats per minute to the height of the height of the Eiffel tower measured in inches. You can do it. You might even obtain a number. But what is it a number of? In the same way \( u_A \)  encodes the preferences of A and \( u_B \) encodes the preferences of B, but what does \( u_A+ u_B \) encode?

But you might reply: but surely we do make such comparisons? You may prefer to give something to whoever you think "needs it more.” But by making such a choice you have in no way successfully added two utility functions. If person C gives money to person A rather than person B, this is a fact about C’s utility function and nothing more than that. Like utilitarians discussing pain and pleasure and like Marx discussing labour, the fact that C does make the tradeoff between giving to A or B produces the illusion there is a natural exchange rate out there to be discovered.

The closest we could ever get to an interpersonal utility comparison is something like this. We consider the ratio (Marginal utility of X)/(Marginal utility of income). This turns out to depend only on the underlying preference relation and not on its representation in terms of utility. It has units of money (£,$,...) and so could be added across individuals. Economists call it “value.” The “value” of X is how much money would you be indifferent to one unit of X. But in this setup £1 is necessarily worth exactly the same to everyone. It is common in "Marshallian” economic reasoning to treat the marginal utility of income as a constant and so treat valuations and marginal utilities as more or less the same. Indeed it is these valuations rather than utility that seem to be the main workhorse in Friedman’s writings.

Choice under uncertainty

In 1947 John von Neumann and Oskar Morgenstern proved a remarkable result relating to choice under uncertainty. The setup of this is a set of outcomes \( A_1, A_2, \dots \) and a probability distribution over the outcomes \( p_1, p_2, \dots \). The individual prefers some probability distributions to others. The von Neuman-Morgenstern result says that if we have an agent obeying some reasonable consistency assumptions then there is a function \( u(A_i ) = u_i \) such that the individual prefers probability distribution \( p \) to distribution \( q \) if and only if $$ \sum_{i \geq 1 }u_i p_i > \sum_{i \geq 1 }u_i q_i $$

We write \( \mathbb{E} u = \sum_{i \geq 1 }u_i p_i \) and call this the “expected utility” and \( u \) the individual’s “utility function.” Note that in this setup expected utility rather than utility is being maximised. It should be pointed out that in this case what’s ordinal is preferences over probability distributions (or “lotteries” in the technical jargon). The individual can just as easily be assumed to be maximising \( f(\mathbb{E} u ) \) for monotonically increasing \( f \), though nothing is gained by doing this. The von Neumann-Morgenstern result is rather striking because it says that the individual can always be assumed to be maximising a linear function of \( p \), which seems almost too strong.