There is a new paper available at Cliometrica. It is co-authored by Mathieu Lefebvre, Pierre Pestieau and Gregory Ponthiere and it deals with how the poor were counted in the past. More precisely, if the poor had “a survival disadvantage” they would die. As the authors make clear “poor individuals, facing worse survival conditions than non-poor ones, are under-represented in the studied populations, which
pushes poverty measures downwards.” However, any good economist would agree that people who died in a year X (say 1688) ought to have their living standards considered before they died in that same year (Amartya Sen made the same point about missing women). If not, you will undercount the poor and misestimate their actual material misery.
So what do Lefebvre et al. do deal with this? They adapt what looks like a population transition matrix (which is generally used to study in-,out-migration alongside natural changes in population — see example 10.15 in this favorite mathematical economics textbook of mine) to correctly estimate what the poor population would have been in a given years. Obviously, some assumptions have to be used regarding fertility and mortality differentials with the rich — but ranges can allow for differing estimates to get a “rough idea” of the problem’s size. What is particularly neat — and something I had never thought of — is that the author recognize that “it is not necessarily the case that a higher evolutionary advantage for the non-poor over the poor pushes measured poverty down”. Indeed, they point out that “when downward social mobility is high”, poverty measures can be artificially increased upward by “a stronger evolutionary advantage for the non-poor”. Indeed, if the rich can become poor, then the bias could work in the opposite direction (overstating rather than understating poverty). This is further added to their “transition matrix” (I do not have a better term and I am using the term I use in classes).
What is their results? Under assumptions of low downward mobility, pre-industrial poverty in England is understated by 10 to 50 percentage points (that is huge — as it means that 75% of England at worse was poor circa 1688 — I am very skeptical about this proportion at the high-end but I can buy a 35-40% figure without a sweat). What is interesting though is that they find that higher downward mobility would bring down the proportion by 5 percentage points. The authors do not speculate much as to how likely was downward mobility but I am going to assume that it was low and their results would be more relevant if the methodology was applied to 19th century America (which was highly mobile up and down — a fact that many fail to appreciate).