Very often, I hear that in a free market system, all transactions are mutually beneficial and non-coercive more or less by definition, as all participants increase their utility in the transaction. Whether highly regarded theorists of markets say such things, I do not know. If so, it would surprise me, as it is immediately obvious that such a definition is unsatisfactory.
Let's take a paradigm instance of coercion, say Tom is cornered in a quiet alley with a gun to his head and a demand for his wallet in his ears. Understandably, Tom surrenders his wallet, as he values his life more highly than the contents of his wallet. Both participants, the mugger and Tom, have gotten what they wanted from the transaction: the mugger got the wallet, and Tom got to live. If you like, you can spell out the inequalities of their utility functions which capture this entirely rational transaction.
Clearly, defining coercion with respect to any particular transaction is inadequate. Rather, Tom's net utility is of interest, as the circumstance of his being put in that dilemma presumably constitutes a lose/lose choice, say, in the sense that both outcomes (presumably) have negative utility. But it also seems that `coercion', normally employed, also applies to `forced choices' that have positive utility, albeit utility less than that which might have been possible in `unforced' circumstances. But this idea captures virtually any system of incentives, and the entire operation of the market becomes a system of coercion. A laborer attracted by a higher salary to a less enjoyable occupation is also being coerced. A company which makes donations to charity for tax reductions is being coerced. In almost any circumstance, one might imagine a more preferable possibility. So, something more restrictive than `circumstances forcing a net loss in utility as compared to other possibilities' is needed, if the concept is to be at all useful. But it gets even worse: if I force a person to do what she already wanted to do, is that not also coercion? As Hitchens described a similar experience in Cuba in Hitch-22: "A cat may stay contentedly in one spot for hours at a time, but detain it in that spot by grasping its tail and it will tear out its own tail by the roots." (Note: I tested this on my cat. It didn't work, but then, my cat is not a representative cat.) So the prospects for a conceptually adequate restriction of net utility reduction are most not good as well.
The next step fails to be obvious, which is probably why the concept was usually treated as a primitive until recently. Though I often rail against coercion, I have treated it more-or-less as such as well. Perhaps this is OK, since many concepts are not susceptible to precise analysis, especially the value-laden ones. Yet, I really need a workable concept, since it is often difficult to explain to right-libertarians why I feel that their ideal is more coercive than a left-libertarian one, other advantages aside and ignoring my strongly held - and I think strongly evidenced - suspicion that a right-libertarian society is deeply unstable and will collapse into forms of obvious tyranny. But for this to be formalized, coercion needs to be susceptible to some form of measure, so that the `coerciveness' of differing outcomes may be contrasted. What we need, then, is something analytically similar to what we need for spelling out guidelines which define particular ethical systems, i.e. an extension of this account which also measures properties of events in addition to their probabilities and utilities, and a way of relating each of the quantities now in question. In other words, to formalize a goal like `minimize coercion', we need a rule ascribing utilities to decisions with `degree of coercion' as an input. Further, as we need to evaluate differing circumstances simultaneously to speak of things like `net coercion', the account should be extended to sequences of decisions, instead of focusing on a particular action.
Clearly, such an extended formalism could accommodate a lot of other ideas, like `maximize the general happiness'. I'm fleshing out the details, and with this motivation in place, I'll put up a post soon. But first, I want to briefly survey other accounts of coercion, so that we might see how they could be analyzed in practice.