Richard Layard points out that one's perceived position in the income distribution is a better predictor of self-reported well-being than one's absolute income level, given that a certain minimum income threshold has been reached. So, every time you move up in relative income, someone else moves down. This makes you happier, but makes everyone with a diminished relative position less happy, even though their absolute income has not changed, or may even have increased, but less than yours.
Layard interprets your gain in relative position as a straightforward negative externality -- in the book he actually calls it "pollution" -- and prescribes a straightforward Pigovian tax to minimize its harm.
...Consider the Jim Crow American South, or apartheid South Africa. Suppose it was the case that any increase in income among blacks leads to a reduction in self-reported subjective well-being among whites, a reduction that totally swamps the utility gain to blacks. Suppose further that a reduction in income among blacks causes a increase in "happiness" among whites that totally swamps the utility loss to blacks. If "we should set our other policy instruments at whatever level is optimal for the state of mind which currently prevails," then it is pretty obvious that an immiserating tax on blacks is optimal for the state of mind that prevails. Racist oppression is obligatory.
I like the point that taxing Jones merely to make Smith feel better puts you on quite a slippery slope, since there are any number of reasons for which Smith may resent Jones.
But Wilkinson's post says quite a bit more. Read the whole thing.