In a paper just published in the Review of Economics and Statistics (gated, free version here), Hankins, Hoekstra and Skiba argue that the rags to riches to rags story has a systematic component.
The authors link records of lottery winners to bankruptcy records. The use of the lottery is a great randomization device, although obviously it restricts the sample to people who play the lottery.
The central finding is this: people who win large amounts are just as likely to end up bankrupt as people who win small amounts. People who win a large amount, $50,000 to $150,000, have a lower bankruptcy rate immediately after winning but a higher bankruptcy rate a few years later so the 5-year bankruptcy rate for the big winners is no lower than for the small winners. Amazingly, by the time the big winners do go bankrupt their assets and debts are not significantly different from those of the small winners.
--Alex Tabarrok, Marginal Revolution, on bankruptcy being driven by the person, not the person's wealth
Thus, our findings suggest that skepticism regarding the long-term impact of cash transfers may be warranted.
--Hanka, Hoekstra, and Skiba