The scatterplot shown here appears to show a strong positive correlation between population and GDP per capita. [EDITED to add: no, I'm an idiot and misread the plot, which shows a clear correlation between population and total GDP and suggests rather little between population and per capita GDP. Sorry about that. The Gapminder link posted by satt also suggests very little correlation between population and per capita GDP. So the context for my (unchanged) argument below is not "Increasing returns to scale are just one factor; here are a bunch more" but "Increased returns to scale are probably negligible; here are a bunch of things that aren't".] In any case, "increasing returns to scale" were just one example (and I think not the best) of how someone might be more productive on moving from (smaller, poorer, more corrupt, less developed) country A to (larger, richer, less corrupt, more developed) country B. Here, let me list some other specific things that might make someone more productive if they move from (say) Somalia to (say) France. * Better food and healthcare. Our migrant will likely be healthier in country B, and people do more and better work when healthier. * Easier learning. Our migrant may arrive in country B with few really valuable skills, but will find more opportunities than in country A to learn new things. * Better infrastructure. Perhaps our migrant is working on making things; country B has better roads, railways, airports, etc., for shipping the products around for sale. Perhaps s/he is (after taking advantage of those educational opportunities) working on computer software; country B has reliable electricity, internet that doesn't suck, places to buy computer hardware, etc. * Richer customers. Perhaps our migrant is making food or cleaning houses. People in country B will pay a lot more for this, because they are richer and their time is worth more to them. So, at least as measured in GDP, the same work is more productive-in-dollars in

That is a bit better, but even as a Bayesian argument it quietly rests on empirical priors which I find odd (as we're about to see). [...] In fact I do. gjm has listed a priori reasons to expect this. More empirically, I already know that epidemiologists talk about a "healthy immigrant effect", which suggests that immigrants are selected (or indeed self-select) to be healthier & wealthier than the average for their home country. I've also seen people bringing up the selectedness of immigrants in arguments about race & IQ, to rebut observations that Third World immigrants tend to do better in their new homes than the estimated mean IQs of their home countries would suggest. [...] It does, but it is a very weak suggestion. The correlation is not that great to start with, and doesn't account for other factors, like international differences in the working-age proportion of the population. Economists have tried to take a more systematic approach based on more involved regressions and/or fitting production functions, but the results here seem mixed, varying across industries and the level of aggregation. (My own prior that increasing returns to scale occur in manufacturing — the class of industries, AFAIK, most important to a country's development — seems broadly consistent with the available evidence. I may as well add that my point 3 invokes "increasing returns to scale & specialization" simply as an example, and the basic conclusion that there's "[s]omething like a fallacy of composition" going on stands even if returns to scale are everywhere decreasing.) [...] I have little to add to what NancyLebovitz & skeptical_lurker have already said.