And taking a different tack, studying lottery winners with modest sums:
http://www.econ.pitt.edu/papers/Mark_lottery.pdf - only 3 years after winning 50-150k people have regressed to the mean rate of bankruptcy
http://www.nber.org/papers/w19348 - land seized from Cherokee people in 1832 was given away in a lottery. The land was valued at between 20k and 150k in modern terms, depending on how you count - inflation vs relative value of unskilled labor. Sons of lottery recipients had no better adult outcomes than the sons of eligible men who received nothing.
So, all of these do find a significant regression to the mean and not a permanent upper class, but for large wealth and social status differences it can take hundreds of years. For more moderately sized random windfalls it happens rather quickly.
I'm out of time to write this comment further, but I'd like to look next for research on what happens a generation or two after a family makes a very large move, i.e. rags-to-riches and riches-to-rags.
http://www.econ.ucdavis.edu/faculty/gclark/papers/Sweden%202... - A study of Swedish surnames since 1700s suggested that wealth regressed to the mean quite slowly (the statistical measure b), finding significant differences between those holding noble and common names after hundreds of years http://www.eea-esem.com/files/papers/EEA-ESEM/2013/1862/SUwp... - Another study of Swedish intergenerational wealth transfer finds similarly. http://www.econ.ucdavis.edu/faculty/gclark/papers/Clark%20Cu... - A study of rare British surnames since 1800s finds a similar rate of regression to the mean.
And taking a different tack, studying lottery winners with modest sums: http://www.econ.pitt.edu/papers/Mark_lottery.pdf - only 3 years after winning 50-150k people have regressed to the mean rate of bankruptcy http://www.nber.org/papers/w19348 - land seized from Cherokee people in 1832 was given away in a lottery. The land was valued at between 20k and 150k in modern terms, depending on how you count - inflation vs relative value of unskilled labor. Sons of lottery recipients had no better adult outcomes than the sons of eligible men who received nothing.
So, all of these do find a significant regression to the mean and not a permanent upper class, but for large wealth and social status differences it can take hundreds of years. For more moderately sized random windfalls it happens rather quickly.
I'm out of time to write this comment further, but I'd like to look next for research on what happens a generation or two after a family makes a very large move, i.e. rags-to-riches and riches-to-rags.