Inequality, Leverage and Crises
Although the details of their actual model will require some knowledge of economics to follow, the paper contains some lengthy non-technical sections, including one showing how the 1920's run-up to the Great Depression was similar to the 2000's runup to the Great Recession and one explaining the mechanism by which their model works.
Basically, they posit a shift in bargaining power (think decline in unionization rates, offshoring of jobs, etc.) from a working class (95% of the population that earns its money from wages) to an investor class (5% of the population that owns most of the capital) and then assume that the extra revenue coming to the investor class as a result of their improved bargaining power is lent back to the workers. This allows the workers to maintain their relative share of consumption, and provides an additional source of income for the investor class.
Over time, the debt level of the working class increases and the vulnerability of the system to a debt crisis increases along with it.
The authors find that widespread defaults during a crisis will help by reducing debt levels of the workers, but because the underlying cause is left unaddressed (the lack of bargaining power for the workers), this is a weak and short-lived solution, with crises repeating regularly. The quicker, more sustainable solution is measures to restore the bargaining power of the workers so that the incentive for workers to borrow and investors to lend is removed or at least reduced.
Anyway, it's just a model, but it's one of the few that actually seems to present a plausible theory of how (certain types of) debt crises happen that seems to mostly fit the facts of what we've observed in the Great Depression and our current Great Recession.