It's Mancur Olson's "Institutional Sclerosis" growth theory. He claimed to have shown that growth rates increase after a large shock to the existing order. His explanation for this is that institutions tend to be corrupted by special interests over time, feeding on the wealth of society. Shocks give people a chance to start afresh.
He does claim to show that the losing side in several conflicts experienced higher growth rates after wars than would be expected from historical data (the American south after the civil war, Japan after WWII, etc.). Critics point to counter-examples, but as far as I know it's considered a reasonable (though not air-tight) theory.
