I'm a rule utilitarian, so I think that the morality of actions should be assessed by their consequences. If someone offered me R50 000 for the privilege of breaking my nose I'd take it; I'd be happier with the R50 000 than an unbroken nose. So the actions of this weird stranger are good as long as he has my consent. The money raises my "good" meter by more than the pain etc lowers "bad" meter.
There are others who would also prefer the money, including Joe. Suppose our mystery stranger punches me in the nose but gives money to Joe. Total utility still goes up but I can't agree that the action is good.
Intuitively, most of us don't accept that benefiting somebody is a good reason for hurting someone else, even if the benefit is quite great. This is a general principle that we'll ignore under certain circumstances but it's still a principle that I think is common (and good).
Public policies normally create winners and losers so they involve the above kind of trade off. How can we choose between policies? One way is checking the levels of coercion. If two policies result in the same amount of utility then the less coercive one is better and highly utility doesn't necessarily justify coercion.
If this point has any value, it is relevant to the minimum wage question. Economists agree that a minimum has a negative effect on employment, but many argue that those who benefit benefit more than those whose employment prospects are (often very slightly) worse. Others argue the other way. They disagree about which policy will maximize utility.
The non-coercive option is the no minimum wage option. A minimum wage involves punching some in the face and giving cash to others.
This post was prompted by this post on The Economist blog. They suggest that on average the minimum wage helps the wrong (i.e. well off) people and hurts those most in need.
Happiness research also suggests that unemployment makes people particularly miserable; this should be included in "utility calculations".