NBER paper explores drivers of collateral prepositioning

New research explores why US banks choose to pledge some, but not all, of their available collateral at the Federal Reserve.

The “largest, most sophisticated” banks pledge 28% of their unencumbered collateral at the Fed, authors Gary Gorton, Chase Ross and Sharon Ross find. Prepositioning assets allows the US central bank to value them for quick use as collateral in the discount window during a crisis.

“At face value, it is surprising that banks have quietly pledged so much of their assets to the Fed,” the authors say in the working paper, published this week by the US National Bureau of Economic Research. “But simultaneously, it is surprising that they don’t pledge more, or all, of their unencumbered assets to the Fed.” They say this discrepancy implies there must be a cost to prepositioning.

The authors build a model around three drivers of prepositioning: a bank’s expectations about the future probability of a liquidity shock; the opportunity cost of prepositioning, as the assets cannot be used as collateral elsewhere; and stigma.

The authors test each element of the model. They find prepositioning varies throughout the business cycle, rising when credit spreads are higher. They show haircuts and rates in other markets, such as the repurchase market, also matter.

Additionally, they find evidence that both drawing on the discount window and simply prepositioning collateral carry stigma. Many banks choose not to voluntarily disclose prepositioning, and those that do tend to be riskier. The authors suggest this is because the benefit of signalling a willingness to pay for liquidity insurance outweighs the cost of stigma for riskier banks.

Lastly, the authors demonstrate a causal link between outflows of uninsured deposits and greater prepositioning of collateral. The link is “completely absent” for insured deposits, they say.