On the 2nd of September, 2019, ESMA published the final guidelines on liquidity stress testing (LST) in UCITS and AIFs. The final guidelines follow on consultations initiated in April 2019 and will take effect from 30 September 2020.

UCITS and AIFS (including ETFs that operate as UCITSs or AIFs), as well as MMFs and leveraged closed-ended AIFs, are within the scope of the guidelines. Despite the fact that asset managers expressed a strong preference for the exclusion of MMFs from the guidelines, ESMA only decided to reduce the scope of applicable provisions to MMFs, focusing on those parts of the guidelines not already covered in the MMFR rules.

Contextually, ESMA released an economic report – "Stress simulation for investment funds 2019" (STRESI) – documenting a comprehensive study on liquidity resilience of bond funds to liquidity shocks. The report also describes some of the methodologies and tools applied during the stress exercise which could potentially help asset managers to implement their internal LST framework.

What does it mean for asset managers?

Asset managers are given a challenging 12-month implementation timeline to comply with the new requirements.

LST will have to be carried out on an annual basis (at least) but ESMA, however, recommends quarterly testing, given that there are specific situations (for instance, high dealing frequency) which will increase or decrease the frequency required.

In order to decide on the appropriate frequency as well as the LST features, proportionality will be the founding principle underpinning all relevant decisions.

Asset managers are also required to document the framework in an LST policy within the RMP. This represents a substantial (and improved) change compared to the consultation paper which required asset managers to document the LST twice – namely in both an LST policy (stand-alone) and the RMP itself.

Following consultations, ESMA has also clarified that reverse stress testing (RST) – albeit useful especially for funds exposed to high-impact and low-probability events – will not be mandatory for all funds in light of the complexity of the implementation, and the little added value for the majority of fund.

As a result, the LST should take into account both historical and hypothetical scenarios and, if necessary, RST.

To ensure the soundness of the overall framework, an initial validation of the LST models and assumptions will need to be carried out by an independent – but not necessarily external – entity.

Lastly, and most importantly, the guidelines will require the asset management industry to demonstrate or build up an acceptable level of substance in terms of liquidity risk management and measurement knowledge.

The asset perspective

Stress testing on the asset side should take into account both historical (Lehman's crisis, for example) and hypothetical (say, rising interest rates) scenarios and, if relevant, reverse stress testing.

One innovative step forward outlined in the guidelines stipulates that while assessing the time and cost of asset liquidation, the manager will need to ensure compliance with investment objectives and restrictions.

As a result, the simulation of asset liquidation would likely result in a sort of "slicing" of the fund composition as opposed to a "waterfall" approach (which can potentially translate in portfolios as not compliant with existing investment restrictions and policies).

The slicing approach would offer a more realistic picture of how a manager would liquidate assets under normal conditions and determine the "true" liquidity of the funds' assets. On the other hand, the presence of even small pockets of illiquid assets might impair the ability of the fund to obtain liquidity in a short amount of time.

The liability aspect

Considerable attention has been paid to stress testing of liabilities – something which used to be put on the back burner. Scenarios for stress testing should change according to the market conditions considered. For normal conditions, averages and trends of historical outflows observed in the past could be used. For stressed conditions, however, the final guidelines suggest using historical severe outflows or hypothetical/event-driven scenarios as well as reverse stress testing.

If the historical approach is used, severe outflows might be calibrated by simply relying on the empirical distribution of net flows, or by initially fitting a parametric distribution and then using a VaR approach or an expected shortfall approach.

Moreover, as the calibration focuses on the tail of net outflow distribution, two separate distributions might be used to model the tail and to model the body in line with the extreme value theory (EVT). Lastly, expert judgment is the third possible approach to be considered in the context of a pure redemption shock.

Looking at redemption from another interesting perspective, fund performances can be a valuable predictor of future outflows. The flow-return relationship can be estimated from simple regression analyses, and can also be a suitable field of application for more advanced data analytics techniques.

In this context, historical time series of subscription and redemption become an invaluable asset in order to set up and calibrate the liability side stress.

A holistic view

Once stress testing has been carried out on both assets and liabilities, the asset manager should combine the results to assess the overall impact on the fund. Eventually, the LST would be translated into a common metric such as the redemption coverage ratio (RCR) or, in case liquidity is estimated by the time-to-liquidity approach, into the fund liquidation coverage ratio (FLCR) so that comparability between funds of an asset manager is ensured. It is likely that measures including RCR and FLCR will become the most prominent indicators internally monitored and used to trigger escalation processes.

The text of the final guidelines can be obtained here.

The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.