This DeLong vs Krugman post, and the comment from James Galbraith, raise an interesting issue regarding redistribution of income, free trade, and globalisation. Namely, as trade creates both winners and losers, but with a positive sum, it is both just and politically necessary that the winners compensate the losers. You could consider it a side-payment for the gains, or a necessary social duty. The cynic in me suggests you could consider it either, so long as it happens.
Economists Peter Timin and Frank Hailey at Voxeu.org - a site well worth reading, I suspect - argue that rather than changes in technology, or globalisation, rising inequality in the US since the 1960s is due to changes in institutions. That is to say, US society made a political choice to reduce minimum wages and union power, and cut taxes on the rich. Women with degrees were the only large group whose earnings kept pace with productivity growth, presumably reflecting the simultaneous rise of middle-class feminism and decline of working-class unionism.
As James Galbraith points out, Chinese labour relations is an issue that we have little power to influence, but the minimum wage is entirely ours to control. This raises my point.
Milton Friedman argued in favour of floating exchange rates on the grounds that it was better for one price, the exchange rate, to adjust in order to respond to variations in the relative price levels of economies (and in the animal spirits of speculators) than all the prices except that of foreign currency doing so. In the strong form of a fixed exchange rate, the gold standard, the adjustment process involves changing prices and wages in the entire economy including the nontradable sector.
Similarly, if you try to reduce inequality by protectionism, that involves altering thousands and thousands of investment and management decisions in favour of domestic production generally. And it's pretty obvious that the more decisions you fiddle with, the more you are going to bugger up. The upshot, naturally, is that you should prefer solutions that minimise the number of decision points for any one institution. That implies the best solution is just to, well, tax the rich and redistribute as directly as possible.
What does this say about my views on CO2 emissions control? Well, administrative controls or rationing are clearly right out. Cap-and-trade is a poor second best, especially given all the possibilities for pathology in there - imagine that there are two glassworks, one of which fuels with coal and one which drives an electric arc furnace with hydro power. Glass from A will have a really bad carbon rating, and vice versa. There will be a strong incentive for buyers of Glass A to misrepresent themselves as buying B. If the CO2 charge falls on the end user, rather than Glassworks A, A doesn't have much of an incentive to change rather than cheat.
The best solution? Tax carbon-rich fuels at the point of sale, and redistribute some of the revenue to the poor. There's an existing and highly effective administration to collect such a tax, too. Set a broad goal - perhaps in the form of an average over several years, to keep the dog from chasing its tail - and link it to the tax calculation so an overshoot has automatic consequences and an undershoot has rewards. (See also this post at Kleiman's.)