From policy to portfolio: an inquiry into the impact of monetary policy decisions and low interest rates on pension fund asset allocation
This study confronts micro-level data on asset allocation with low interest rates and monetary policy for a broad set of defined benefit pension funds in the United States. Findings indicate that funds increase their investments in equity assets and decrease their investments in fixed-income assets in response to interest rate reductions. Strikingly, the passive portfolio movements through mechanical return effects explain the most significant part of the observed change.
The ageing population and low interest rate environment have put pension funds’ stability and asset allocation strategies under pressure. Yet, the literature on portfolio management decisions within pension funds tends to neglect the role of monetary policy and low interest rates as possible drivers of asset allocation. This study confronts micro-level data on asset allocation with low interest rates and monetary policy for a broad set of defined benefit pension funds in the United States. Findings indicate that funds increase their investments in equity assets and decrease their investments in fixed-income assets in response to interest rate reductions. The effect can be decomposed into passive portfolio movements driven by returns and active rebalancing decisions. Strikingly, the former explains the most significant part of the observed change: portfolio variation originates from the mechanical return effects following monetary policy and the absence of active rebalancing mechanisms.
Background
Pension firms constitute one of the three industries within the conventional institutional asset management sector, alongside insurance companies and mutual funds. Pension funds in the United States (US) alone managed 35,011 billion USD of assets in 2021, corresponding to 152.6% of the United States Gross Domestic Product (Thinking Ahead Institute, 2022). In a persistently low interest rate environment, funds in the US and other advanced economies face the challenge of ensuring their financial stability for future generations. As fund liabilities towards pension beneficiaries typically have a longer duration than fund assets, a decrease in interest rates will disproportionately increase the present value of liabilities, weakening the fund’s solvency or funding ratio and affecting its long-term viability (Breuer et al., 2019).
Sound investing within pension funds is vital. Its importance is once again emphasised in the United Nations Sustainable Development Goals (SDGs) and Agenda 2030 (United Nations & Sustainable Development, 2015). In the realisation of SDG 10, realising reduced inequalities, future generations’ wealth and capital accumulation are explicitly considered. Sound investment strategies are pivotal in realising reduced inequalities (SDG 10) and can funnel decent work and economic growth (SDG 8). To support financial sustainability and intergenerational solidarity, pension funds should build up capital responsibly without taking too much risk (Statistics Netherlands, 2023).
The persistent low interest rate environment constrains this responsible build-up of capital. First, in response to the ageing population, low interest rates and low returns, the United States is transitioning its pension system. Defined benefit funds transform into defined contribution funds, transferring interest rate and market risks to individual contributors (Ghio et al., 2021; Myers & Topoleski, 2021). Second, the low interest rate environment has driven investors in general to change their portfolio composition towards riskier assets. This risk-taking behaviour of investors occurs for three reasons. (1) Lower interest rates boost asset and collateral values, implying wealth increases. This increase in wealth may induce a reduction of risk aversion. (2) Lower interest rates also interact with more “sticky” return targets, creating a search-for-yield effect when there is a differential between low realised returns and high return targets. (3) Finally, policy decisions that enhance the perceived transparency and commitment to future policy decisions (e.g. forward guidance) can increase risk-seeking through a reduction of uncertainty (Borio & Zhu, 2012).
Previous empirical literature on risk-taking within institutional investors has paid specific attention to search-for-yield effects within mutual funds. For instance, Hau and Lai (2016) examined the risk-taking behaviour of equity and money market mutual funds in eight Eurozone countries from 2003 to 2010. Leveraging variations in the real short-term interest rate (the policy interest rate minus the inflation rate), the study found that fund investors adjusted their portfolios by moving away from the money market and towards the riskier equity market when real interest rates decreased (Hau & Lai, 2016). Supporting evidence is also present for Eurozone bond funds. Following expansionary monetary policy from 2007 to 2023, funds moved towards riskier asset classes, such as corporate or high-yield bonds (Giuzio et al., 2023). Finally, results confirmed Eurozone evidence for United States bond and equity funds: unexpected tightening of monetary policy led to a shift away from safer bond assets (Banegas et al., 2022).
The evidence for pension funds is less clear. While flow-of-funds data have been employed to scrutinise the response of the pension fund sector as a whole, microlevel data detailing the reaction of individual funds is not often used. Limited available evidence for the United States has shown that expansionary monetary policy shocks (from 1998 to 2013) increase pension funds’ allocation to equity. However, this conclusion is constrained to 151 defined benefit funds for public sector employees (Boubaker et al., 2017).
Methodology
This study aims to contribute to the research gap described above. Asset allocation data from individual pension funds is leveraged to examine whether funds respond to lower interest rates and expansionary monetary policy in a way that can be explained by increased risk-taking or a search-for-yield effect.
Main hypothesis
(Low interest rate environment and the bond-equity split): In response to a low interest rate environment and expansionary monetary policy measures, pension funds increase their investments in riskier asset classes, such as equity investments, while reducing their investments in lower-risk bonds.
Data
This paper focuses exclusively on defined benefit funds. In the US, these fall apart in funds for public sector employees (hereafter “public sector funds”) and funds for private sector employees (“private sector funds”). Annual data on asset allocation strategies for public sector funds (2009 - 2022) are obtained from the Public Pension Plans Database of the Center for Retirement Research at Boston College (2023). After pre-processing, 210 out of 229 funds are maintained in the dataset. Data on private sector funds (2009 - 2021) are taken from the Private Pension Plans Database of the United States Department of Labor (2023). After selecting only primary, defined benefit, large funds (> 100 members) and after removing funds with missing asset allocation data, 453 private sector funds are maintained in the dataset. The dataset contains information on asset allocation (percentage of assets invested in equity and fixed income) and fund-specific information (number of contributors, assets under management, fundings’ ratio) for each of the funds mentioned above.
The Shadow Rate serves as a proxy for the stance of monetary policy. This short-term interest rate, introduced by Wu and Xia (2016), bundles the effects of conventional and unconventional monetary policy and is not constrained by the zero lower bound. Lower (higher) levels of the Shadow Rate represent more accommodative (restrictive) monetary policy. Other macroeconomic control variables, such as the VIX index as a proxy for stock market volatility, market returns, the credit spread on investment-grade rated corporate bonds, local inflation and GDP growth, are taken from the Federal Reserve Bank of St. Louis (FRED, 2022), Thomson Reuters Datastream (2023) and the United States Bureau of Economic Analysis (2023).
Methods
The effects of monetary policy and interest rates on asset allocation were estimated using a panel regression model, following the work of Hau and Lai (2016). The regression equation regresses current period equity (EQi,t) of fund i in year t on previous year’s investment strategy (EQi,t-1, …,EQi,t-p). The inclusion of previous year’s investment strategy on the right-hand side of the equation is used to model inertia in investment strategies. Most importantly, the regression includes on the right-hand side the effect of changes in the Shadow Rate (SRi,t). The coefficient associated with this variable is the coefficient of causal interest, . Finally, the regression includes past market returns (Ri,t-1) as a control variable. The error term consists of an individual fixed effect ui and an idiosyncratic error term vi,t .
The main challenge in estimating is that monetary policy is set in function of underlying business cycle conditions. That is, the Federal Reserve Bank (Fed) determines the stance of monetary policy considering their dual mandate, targeting price stability and maximum employment (Board of Governors of the Federal Reserve System, 2021). Possible correlations between low interest rates and asset allocation decisions might, in fact, be explained by business cycle variables and not by the effect of interest rates. To overcome this challenge, this research uses an identification strategy introduced by Hau and Lai (2016). It was originally developed to analyse mutual fund behaviour in Europe, but it naturally lends itself to application in the US. The following paragraph explains the strategy.
Investors make investment decisions based on real interest rates, not on nominal interest rates. The real interest rate is defined as the nominal interest rate minus inflation compensation. In the US, the Fed determines important (nominal) policy interest rates, which apply to the entire US and do not exhibit cross-sectional variation by state. However, inflation differs per state. As a result, the real interest rate investors face also varies per state. When adjusting the nominal Shadow Rate with state-level inflation, we obtain a real interest rate that exhibits variation across states.
Furthermore, and very importantly, it is improbable that the Fed will change its monetary policy stance only because one state in the US is facing more extreme inflation levels. We can, therefore, see variations in the real interest rates between states as a source of variation that does not depend on the business cycle of the United States as a whole. In fact, these variations represent the effect of interest rates while being independent of underlying US-level business cycle conditions. This independent variation mitigates the concern that we measure the effect of business cycle fluctuations instead of the true effect of interest rates. Using the local real Shadow Rate in equation 1, a panel fixed effects estimator can now estimate the parameter .
In robustness checks, the stability of this result is verified by adding extra macroeconomic and financial control variables, by using different proxies for monetary policy and interest rates (the Federal Funds Rate and the 10-year Treasury Yield) and by estimating the equation with a dynamic panel estimator (Arellano-Bond). The latter also resolves possible remaining biases in the fixed effects estimator (such as correlation with local business cycles).
Results
After analysing the effects of local interest rates, it was discovered that pension funds indeed shift their portfolio to riskier equity assets as interest rates fall. For public sector funds, changes in the Shadow Rate are significantly and negatively related to the allocation to equity assets (). A reduction of the Shadow Rate, thus, leads to an increase in equity investments. Introducing macroeconomic control variables decreases the size of to , yet its significance remains unchanged. Across the United States, on average, the real Shadow Rate has dropped from 5% in 2006 to -2% in 2014, a reduction of 7 percentage points. This fall translates into a cumulative increase of equity investments ranging between 1.2 and 2.5 percentage points. The impact of interest rates on equity investments for private sector funds generally lacks significance.
For all funds, the coefficients quantifying the effects of Shadow Rate changes on fixed-income investments are positive, suggesting a reduction in such investments following a Shadow Rate decrease. The economic significance of this coefficient is negligible: portfolio decreases in fixed-income investments attributable to the interest rate changes from 2006 to 2014 are around one percentage point. The results are robust to alternative specifications of monetary policy, addition of control variables and alternative econometric specifications.
Decomposition
The previous analysis has uncovered an overall effect of monetary policy. However, it did not allow us to disentangle if risk-taking indeed drove this effect. In fact, two distinct sources of variation could have generated changes in the outcome variable. The first component is the mechanical effect of returns on the portfolio composition. For example, suppose no active portfolio management decisions are made, and the return on equity investments is higher than that on fixed-income investments. In that case, the proportion of equity investments in the portfolio mechanically increases. This change is referred to as the passive change in the portfolio, as it originates mechanically from return effects. On the other hand, the second component of change in the outcome variable measures all active portfolio management decisions. This component is named the active change in the portfolio. With increased risk-taking, funds are expected to increase their equity investments actively. The decomposition is also schematically represented in Figure 1.
In the next step, the new definitions of passive and active change serve as outcome variables in Equation 1. Results are presented in a condensed format in Table 1. This table only shows the results for the coefficient of interest . Each cell thus represents a separate regression. Results are shown for equity and fixed-income investments, as well as the fixed effects (FE) and dynamic panel data estimator (DGMM).In the next step, the new definitions of passive and active change serve as outcome variables in Equation 1. Results are presented in a condensed format in Table 1. This table only shows the results for the coefficient of interest . Each cell thus represents a separate regression. Results are shown for equity and fixed-income investments, as well as the fixed effects (FE) and dynamic panel data estimator (DGMM).
For equity investments, passive allocation is negatively related to changes in the Shadow Rate. In periods where the Shadow Rate decreases, returns on equity investments are higher than the return on the portfolio as a whole, implying an automatic increase in the allocation to equity.
Contrary to passive equity allocation, active equity allocation and the real Shadow Rate are positively related. However, the effect of negative passive portfolio relationship outweighs the active portfolio reallocation, creating the previously documented overall negative relationship between the real Shadow Rate and allocation to equity. The dynamics are reversed for fixed-income investments.
These findings align with earlier empirical research, as is visually represented in Figure 2: in periods of expansionary monetary policy (i.e. with decreases in the real Shadow Rate) the stock market’s performance is documented to increase. Lower discount rates, higher dividends and higher excess returns over the riskfree rate lead to increases in stock prices and thus to higher realised returns, outperforming riskless assets or fixed income instruments (Bernanke & Kuttner, 2005; Chava & Hsu, 2019; Sekandary & Bask, 2023; Thorbecke, 1997). Higher returns on stock markets affect a fund’s portfolio composition through a mechanical effect, automatically increasing the share of equity investments. The absence of active rebalancing mechanisms explains why portfolio composition does not revert to the original strategic allocation. This inertia in response to stock market performance was documented before (Bikker et al., 2010), but is now also found to hold for changes in the interest rate environment.
Conclusion
In comparison with existing empirical literature, the findings of this study corroborate and extend significant insights. While a growing body of literature confirms a shift in risk appetite among mutual funds in response to low interest rates (Banegas et al., 2022; Giuzio et al., 2021; Hau & Lai, 2016), evidence regarding pension funds remains relatively scarce and limited to the United States public sector (Boubaker et al., 2017). Leveraging dynamic-panel data analysis, this study confirms the assertion that changes in monetary policy rates induce a substantial increase in pension funds’ allocation to equity.
Limited previous literature explains this phenomenon and puts forward a structural risk-shifting incentive favouring riskier investments (Boubaker et al., 2017; Konradt, 2023). This study’s results suggest an alternative explanation. The decomposition analysis highlighted the importance of passive and mechanical return effects and revealed an absence of significant active rebalancing towards the original portfolio composition.
These results put forward an explanation not rooted in an active “search-for-yield” effect or active reduced risk aversion. Rather, these coherently combine previously documented empirical observations on financial market responses to monetary policy.
If the increased riskiness of pension funds’, mutual funds’ and insurance companies’ portfolios is robust to further analyses with higher frequency data observations, this would advocate for broader monetary policy incorporating financial stability in the monetary policy framework. Furthermore, it would more broadly acknowledge an important trichotomy in economics: economic policy is a constant balancing act between redistribution, stabilisation and allocation policies (Musgrave & Musgrave, 1989). Although monetary policy has typically been associated with the objective of stabilising the business cycle, its impact extends beyond that, influencing our long-run allocation and redistribution.
Auteurs
