I like your analysis of the changing Easterlin paradox. The "countries as cases" model always bothered me in almost any research I've seen it in - and in this happiness work more than in others. Why would we have any reason to believe that happiness scores are independent between countries?

On the other hand, while I buy poor people in poor countries comparing themselves now to rich people in rich countries, I'm not sure that the poor people in rich countries are comparing themselves to people in poor countries. It makes sense that reference groups are changing, but are we really starting to believe the "you should be happy because a child is starving in Africa" line of reasoning that always worked so well when we were kids? It seems unlikely. I wonder if there's any detailed, qualitative work (perhaps by social psychologists or sociologists?) on why and how people decide if they are happy, and who they end up comparing themselves to?

The only work I know in this vein is Daniel Gilbert's "Stumbling on Happiness" and the various works he cites. But I think Gilbert's take is that most people (although I don't know remember if he goes beyond the US in his discussion) compare themselves to their own (poorly remembered) past and (poorly projected) future. I wonder if that holds less true for poor countries right now?

Also interesting is philosopher/historian Jennifer Michael Hecht's "The Happiness Myth" for a nice historical take on how we got into this whole happiness debate and where else that debate has been.

Conceptually this makes a lot of sense, if we accept the premise that Americans are actually aware of what is going on elsewhere in the world. A survey might be in order - "How Happy Are You" vs. "How well are you doing compared to people in 3rd world nations"

This happiness literature is not only sketchy in methodology, but dangerous in practice by providing the justification of "scientific research" to those wishing to push some ideology. Every example of policies they advocate (limiting the work week, for example) to correct for things like "positional externalities" have deep conceptual flaws. For instance, suggesting that we would work extraordinary hours to keep up with our neighbors is a zero-sum games ignores the gains to employers and their consumers.

The hypothesis that the reference group(s) that people are comparing themselves to may have been changing over time due to changing communications technology etc is interesting. It is presumably testable if one had happiness survey data across countries going back far enough. I doubt that available data are adequate to test the hypothesis. If so, then we can only speculate. My speculation is that this hypothesis would probably not hold up well, because I think that we tend to underestimate the degree of exposure to "lifestyles of the rich and famous" in past decades. Africa, for example, was largely colonized until 1960, and many Africans had European colonists living among them - that is a kind of contact that would, I think, produce a much sharper feeling of relative impoverishment than watching shows on TV. We should also keep in mind that exposure to TV in the early TV era (say 1960s) was more widespread than one might think from simple statistics like TVs-per-capita: many villages throughout the developing world would invest in one TV that they would watch communally. I think that knowledge of developed-world lifestyles may have been more widespread in the past than we might assume today.

One interesting experience related to this hypothesis that might be observable and testable with happiness data is the opening of the communist bloc in the early 1990s. Average happiness fell sharply in these countries in early transition. This may have been at least in part due to sudden exposure to knowledge of what the relative western standard of living was. Then again, real incomes were plunging for most people in this period and unemployment increased dramatically, so it would be challenging to diesentagle the effects.