The Reserve Bank of Australia has published a research discussion paper assessing the informational efficiency of the Secured Overnight Funding Index Australia, which is currently in its beta phase. The paper finds that noise in the benchmark increases when related-party transactions are included, market concentration is high and transaction volumes are low. It also concludes that alternative methods for trimming the transactions used to calculate the rate, relative to the initially proposed approach, could improve the benchmark’s efficiency and robustness. The analysis uses a state-space model to separate time-varying noise in the benchmark from the underlying efficient rate and examines drivers of noise in both the daily time series and transaction-level data. The paper presents the findings as evidence on how to optimize the benchmark’s design, and notes that the Australian Securities Exchange has already revised the SOFIA methodology to reflect most of the results as it prepares for a possible transition from beta to a live benchmark.