Regardless of whether the empirical claim that it would be possible to cause a net increase in well being by lower the average standard of the middle class and raising that of the lower class, I'm just not sure I'm comfortable with this kind of utilitarian reasoning.
Consider the following scenario. A small subsection of the population is compelled (economically, politically, w/e) to do shoulder the vast majority of negative utility... a statistically small slave class, say. However, the negative utility of this slave class is outweighed by the positive utility conferred on the rest of the population, which constitutes the vast majority. Furthermore, let's say -- hypothetically -- that this set up leads to a net increase in quality of life over spreading out the negative utility across a wider section of the population. Is the slave caste system therefore morally correct, or good?
I'd hesitate to say it would be. In fact, I'd say it was radically immoral. But, you can't justify that conclusion with utilitarian reasoning alone. You'd need reference to something else... say basic human rights that hold regardless of any net increase/decrease in quality of life. Now, I guess I want to say that something like a minimum wage is likely such a right. Respect for the worth of labor should not, on my view, ever be allowed to fall below a certain value. If you put in a hard days work, I believe you have a right to some minimum amount of compensation for that labor. What an appropriate value might be, I won't speculate about, but it seems to me this right holds independent of maximizing overall quality of life.
Of course, overall quality of life should be a very important concern nevertheless. I just don't want to say that it is the ultimate principle on which to judge policy