The silver standard for policy analysis is the second best weak Pareto improvement --- find the winners of a policy change and have them compensate the losers in one way or another. National healthcare is one such policy pay-off to people who have borne increased volatility, competition and minimal real wage changes for the past thirty years despite overall net growth. We don't seek this standard for everything; for instance pimps are not compensated for loss income if social services and criminal justice services reduce the number of prostitutes in an area. Also, breaking a monopoly will harm the holders of the rights to monopoly revenue and profit streams while producing significant society wide gains. These are standard policy analysis questions and techniques.
Prof. Landesburg in the New York Times says 'screw Pareto' in regards to free trade. Workers in impacted professions should be thankful that despite the significant losses in income through the combination of job loss AND the cost of not having perfectly substitutable skills to find a new job without retraining, they can get cheaper goods at Wal-Mart.
What we lose through lower wages is more than offset by what we gain through lower prices. In other words, the winners can more than afford to compensate the losers. Does that mean they ought to? Does it create a moral mandate for the taxpayer-subsidized retraining programs proposed by Mr. McCain and Mr. Romney?
Um, no. Even if you’ve just lost your job, there’s something fundamentally churlish about blaming the very phenomenon that’s elevated you above the subsistence level since the day you were born. If the world owes you compensation for enduring the downside of trade, what do you owe the world for enjoying the upside?
He is making a fairly weak analogy ridden 'moral' case that has at its core assumption that the overarching goal in life is slightly more than mere subsistence and that the games and rules of society are single iterations. Furthermore, the cost of unemployment due to structural outsourcing from the combination of direct wage losses, direct re-training/job seeking expenses and indirect future losses due to the combination of a lower base wage and lower future trend growth is borne either universally, or if it is not borne universally, the losers are implicitly and quickly compensated by lower prices. This is empirically absurd for structural shifts within the labor market.
Some people are significant winners in trade, some people are general winners via lower prices, and others are significant losers as their positions with attendant human capital accumulation is destroyed. He admits that the winners are in aggregate better off than the losers, and in almost all cases I will agree with him. However if you assume that policy and politics are multi-iteration games and people in pain will scream louder than people in mild, background contentment, then compensating people who are losers through no fault of their own of a specific policy regime is the smart, and economically efficient thing to do. People in pain that is not mitigated will organize to prevent more pain from being inflicted upon themselves or others like them, which means aggregate gains are being left on the table. If you take a utilitarian approach and seek the greatest aggregate gain for the relevant analytical unit, then not compensating people who have lost for previous, similiar policy changes is immoral, and dumb at the same time.