By this (we think) they mean, among other things, remarkably static tri-party haircuts across a variety of asset classes while the financial crisis was swirling around them and lack of granularity in tri-party schedules that allowed for some volatile paper to slip in under the asset-backed category. Mitigating the dealer-side risk could be longer dated financing, thus reducing the vulnerability that comes with maturity transformation (especially on less liquid collateral)..
Much of this has been fixed or is in process of being corrected. And finally the authors noted that capital and liquidity regulation could play a part to reduce dealer vulnerability.
The paper takes in interesting look at liquidation timing.
Calculating the number of days it would take to sell off a hypothetical portfolio of $200 billion (the sample size of a large dealer’s tri-party financing book), they find it could range, based on typically traded volumes, between 9 days for US Treasuries and Strips to 30 days for ABS. Overlaying Va R to that process results in some idea for the magnitude of losses that could be incurred.
Another concept mentioned was a repo resolution authority with the power to claw back losses (from liquidation).
Finally, “an ex post emergency central bank liquidity backstop for the dealer’s creditors” was examined.
Large, concentrated financed risk paper positions increase vulnerability to a fire sale.