The AMF publishes a study about potential explanatory variables for the record outflows that French MMFs faced in March 2020
Building on a detailed database on French money market funds (MMF) flow and performance during the March 2020 crisis, the AMF shows that the heterogeneity of redemption behaviors cannot be explained by differences in terms of fund performances, or in terms of the structure of the portfolio (asset type, maturity, liquidity ratios). It would thus appear that redemptions would have been the result of the individual decisions of some investors in need of cash rather than a mistrust towards money market funds.
A significant heterogeneity in redemption patterns
Between 12 March and 30 March 2020, French MMFs recorded an aggregated net outflow of € 48.6 billion. Redemption requests were concentrated on standard MMFs while short-term MMFs rather benefitted from subscriptions. However, one could observe a significant heterogeneity within each category. The AMF tried to shed light on the factors that could explain that, within a given category, some funds faced less investor redemptions than other, and even sometimes received inflows.
An heterogeneity that some attribute to fears on specific funds
Several analyses considered that the larger magnitude of the redemptions observed on some funds could be explained by stronger initial vulnerabilities, assuming investors can detect them, e.g. using the information on the various asset liquidity ratios. According to this interpretation, in March 2020, investors would have withdrawn cash primarily from some MMFs by fear that these funds be less able than others to withstand future redemption requests (run). Another hypothesis was that MMF investors were highly yield-sensitive, and thus redeemed in priority those funds with a lower relative performance. None of these theories could be confirmed by the data on the French market used by the AMF in this study.
A statistical analysis on flow data to explain redemptions
The study applies various econometric technics to model the relationship between redemption flows and fund performances, while introducing explanatory variables accounting for funds’ asset caracteristics (liquidity ratios, portfolios’ liquidity, cash holdings, …). The data is first analyzed at a daily frequency and then at a weekly frequency.
Redemptions were most likely motivated by investors constraints rather than by funds’ characteristics.
None of the fund-specific variables introduced in the model appear to play a significant role in explaining the fund’s redemption (whether the variable relates to the fund’s performance or to the structure of its asset side). The analysis thus demonstrates that, as far as French MMFs are concerned, the explanation of a run mechanism with first mover advantage is not valid, and that the investors’ high-yield sensitivity hypothsis is not observable. These results tend to prove that other factors, such as the liability structure, must have had a predominant influence during the record-high redemption episode. The decision to redeem, taken by some investors (insurers, non financial corporates, other funds…) might have been driven by their own cash needs (whether actual or anticipated), including increased margin calls during the crisis period.
On the same topic
