COVID-19 Fallout and the Trouble Estimating Deposit Assumptions

What was the impact of COVID-19 Stimulus?

• COVID-19 stimulus was huge. ­COVID-19 stimulus of $4.6 trillion was nearly two times larger than the dollar adjusted amount infused into the financial system during the 2008 financial crisis.
• Stimulus drove surge deposits – The stimulus helped drive commercial bank deposits higher. At the end of the second quarter of 2022, banks held $4.8 trillion more in deposits than they did at the end of 2019, according to the Federal Deposit Insurance Corp., a 33% increase. This high deposit base kept interest rates on both loans and deposits low and put financial institutions in a strong liquidity position.
• Inflation – Eventually inflation started to spike because of low rates and easy money, reaching a 40 year high of nearly 9% in 2022. Fed Funds rate hikes started in 2022 to slow these price increases.
• Deposit Runoff – As consumers and businesses struggled with high prices and rising rates, they started to draw down on deposits throughout 2022, putting financial institutions in a tight liquidity position in 2023 and into 2024. At the end of the third quarter of 2023, commercial bank deposits had declined to $17.5 trillion from $19.4 trillion at their peak in 2022.

How did that impact the ability to calibrate deposit assumptions?
Through the end of 2021, even a 10-year historical lookback would have contained only low and stable interest rates – sufficient neither for estimating deposit behavior in the current high and rising rate environment nor for a potential falling rate environment in 2024.  However, incorporating 2022 and 2023 data into calibrations causes new issues:

Betas – Betas represent the correlation between deposit rates and the federal funds rate.  The issue with the massive stimulus, however, was that when rates started increasing, banks had so many deposits that they didn’t have any need to raise rates.  For the first 200 basis point hikes in 2022, there were nearly no changes in rate.  Incorporating this period of historical data would significantly underestimate betas. Conversely, when the Federal Reserve slowed hikes in 2023, s deposit rate increases due to competition for deposits in a tight liquidity market.

Decays – When financial institutions first started receiving COVID-19 stimulus deposits, the conventional wisdom was that it was ”hot money” and would leave the banks soon. The money was considered unstable and subject to high decay speeds.  This thinking softened as the deposits remained at the institutions well into 2022, and many risk managers started believing that this money was in fact stable. Any decay rate calibrations done during this time would have significantly underestimated potential decay.

What can Financial Institutions do?
The most important thing to do in this period of uncertainty is to stay calm.  It is not the time to abandon best practices or quantitative approaches.  Uncertainty is not a reason to keep stale estimates or to use static assumptions across rate shock scenarios.  Rates will continue to change, and decays, betas and stable balances will continue to change along with them.

1. What we can do is apply enhanced fundamental model risk management approaches to risk forecasting. Considering market and model uncertainty, financial institutions need to implement a methodical approach to managing model risk. Four key approaches that all institutions should follow can significantly enhance management’s ability to rationalize and act on asset and liability management results:1. Follow best practices for parameter calibration. Those include considering historical time periods, refreshing parameters frequently, aligning product segmentation with behaviors, matching drivers with products and adding lags to calbrations. Maintaining a methodical approach to calibration helps justify output with examiners, auditors and senior management.
2. Document assumptions, methodology, and rationale. Robust documentation assists with transfer of knowledge both internally and externally.
3. Monitor realized parameters. The best way to be sure that calibrated parameters are performing as expected is to monitor actuals. Ask, “what is the ratio of deposit rate increases to fed funds increases over the last year”?
4. Stress-Test Assumptions. Running scenario analyses with up and down betas and decays is a great way to build confidence in your output. Even if your assumptions are wrong, knowing that your risk profile is acceptable under both 50% up and down beta shocks can better equip management to make decisions using the calibrated parameters.