This paper examines the effects of introducing an auto-collateralization scheme in a simulated securities settlement system. Using artificially generated data, it shows that auto-collateralization can substantially lower liquidity needs without delaying settlement. By allowing for a more efficient usage of collateral, it also reduces intraday credit exposures. The paper further demonstrates how these efficiency improvements in usage of liquidity depend on the structure of the settlement cycle. The benefits of auto-collateralization are particularly large when system participants' incoming and outgoing payments are negatively correlated over time. Finally, the impact of participant defaults is assessed, and it is shown that auto-collateralization protects against liquidity risk in certain stress scenarios.