Cyclical investment behaviour of investment funds: its heterogeneity and drivers

ABSTRACT

Do open-ended funds amplify negative feedback loops of declining asset prices through procyclical selling and reduced purchases of securities experiencing falling prices? Leveraging security-level data of Czech funds, this study reveals procyclical sales of government bonds and held fund shares. Therefore, funds may amplify market downturns and financial stress in government bonds, which are crucial for price discovery and financial stability. Cyclical sales of cross-fund exposures then show adverse network effects. Furthermore, the study examines factors influencing procyclical sales. The results show that cyclical sales are intensified by the extent of funding constraints and liquidity mismatch.

KEYWORDS:

• Asset management
• emerging markets
• externalities
• interconnectedness
• investment behaviour

JEL CLASSIFICATION:

• G11
• G23

Acknowledgments

The author acknowledges support received from the Prague University of Economics and Business through grant: F1/14/2023.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/17520843.2024.2380565.

Notes

1. Illustratively, the 2020 International Monetary Fund Global Financial Stability Report indicated that mutual funds accounted for more than 40% of issued high-yield bonds.

2. Additionally, the NAV of funds investing in illiquid assets is extremely stale, resulting in substantial dilution risk for buy-and-hold investors, see Choi, Kronlund, and Oh (2022). Therefore, opportunistic fund managers may withdraw capital from funds with unaligned NAV to exploit staleness in NAV.

3. Furthermore, Hanson et al. (2015) show the traditional banks as contrarian traders since they load on illiquid bonds during stress periods owing to their “sleepy “liabilities as government deposit insurance makes depositors insensitive to the value of the banks’ assets. Apicella, Gallo, and Guazzarotti (2022) also study the asset allocation of insurers’ unit-linked portfolios, which, like open-ended funds, are closely affected by the decisions of policyholders who can redeem their unit-linked policies at short notice. The authors show that insurers do not act as shock absorbers for assets relating to unit-linked policies, as investments from these portfolios behave procyclically.

4. The primary sources available for studies of US funds, Thomson and CRSP, have portfolios reported as of the end of calendar quarters, and the SEC filings contain portfolios aligned with the fund’s fiscal year. Moreover, Thomson portfolios contain only long common equity positions. CRSP portfolios have some coverage of nonequity and short equity positions; however, this coverage appears relatively sparse (Schwarz and Potter 2016).

5. For example, the Czech funds studied here grew by 400% over the last ten years, representing one of the EU countries’ most dynamic growth rates.

6. Information about split dates and split factors is obtained from the financial data providers. Without the adjustment, splits in stocks would be falsely recorded as a trading activity due to changes in split quantities.

7. Here, the „Other“category consists of bond funds that provide exposures in both corporate bonds and government bonds or follow an exotic and less represented investment strategy, such as ESG-oriented. There are also bond funds for which we could not obtain a detailed investment strategy from the data providers (“Unknown”). All the results shown later are robust to omitting these unmatched funds completely or merging them into one category with the “Other“.

8. Given the log-level models, the reported parameters have to be converted using $100\left(e^{\mathit{\beta }\cdot \mathrm{\Delta }x}-1\right)$ before being interpreted as a percentage change.

9. We note that any time-variant fund-level exchange rate risk hedging would not be captured by the fund fixed effects, which may attenuate exchange rate movement elasticity and hence bias the respective $\mathit{\gamma }$ estimate towards zero.

10. Consisting of hedge funds, mixed funds, etc. Less compressed results are available but no additional statistically-important relationships were found.

11. Results stay unchanged even if we omit ETF funds, which can be presumably traded more aggressively with a limited risk to financial stability; see Antoniewicz and Stahel (2020).

Additional information

Funding

The work was supported by the Vysoká Škola Ekonomická v Praze [F1/14/2023].

Notes on contributors

Milan Szabo

Milan Szabo pursues his doctoral degree at the Prague University of Economics and Business. He also works at the Czech National Bank in the the Financial Stability Division. His professional and research activities focus on financial markets, stress testing of financial institutions, interconnectedness in the financial system and sovereign risk.