13 Sep 2023 4 min read

Should higher cash rates decrease interest in equities?

By John Southall , Yikai Shen

Interest rates have shot up over the past 18 months, increasing the return available on 'risk-free' assets such as cash. Could this impact equity returns?


All else equal, traditional theory[1] says that the relative attractiveness of asset classes doesn’t depend on the risk-free rate, so they shouldn’t be impacted by this development. The idea is that different asset classes compete for investors’ capital. Equities, for example, ought to earn a premium over cash to compensate investors for the far greater uncertainty in their returns.

But could such models be overstating the efficiency of the multi-asset universe?

It’s natural to feel sceptical, especially after recent market experiences. It also has become common to hear opinions such as ‘bonds are now more attractive’, ‘annuities are now a viable option for DC retirees’ and ‘TINA (there is no alternative to stocks) is over’. Are they right? Should investors dynamically shift their risk-on/off allocation through time?

Empirical evidence

Some papers, such as this one, point out that historic stock returns appear to have virtually no relationship[2] with cash returns. As an illustration, the scatter-graph below plots rolling 12-month US equity returns against rolling 12-month cash returns for US investors since 1926:


However, we must be careful not to be ‘fooled by randomness,’ as Nassim Taleb might put it. Although the graph feels compelling – it contains more than 1,000 data points over almost a century – we shouldn’t jump to conclusions. You need to rule out the possibility the apparent lack of relationship is luck.

To find out, we ran our strategic economic scenario generator for 100 years[3]. The generator assumes the ‘cash-plus’ model is correct[4]. This means we would expect the overall beta of equity returns to cash returns to equal one, but it might not be in any given simulation[5] due to chance. The chart below shows our results as a histogram:


The average beta is 1.0, consistent with the model’s assumptions. But the dispersion of the underlying results is wide. Indeed, 327 out of 3,000 simulations, that's more than 10% of them, recorded a negative value. This means that the observed ‘non relationship’ in Fama-French’s data could plausibly be random, and that it isn’t therefore statistically significant evidence[6] against a cash-plus model.

We should also remember that markets have tended to become more efficient through time, so even if we found strong evidence of inefficiency in the past it may well not persist into the future.

How big could the benefits be?

Let’s suspend our disbelief and suppose expected nominal equity returns are constant, having nothing to do with the cash rate. How good an opportunity would that offer?

To find out, we back-tested two potential strategies, both combining cash and equity, since 1926. In the first, a constant 50% in equity is held. In the second, expected equity returns are assumed constant. Correspondingly, when cash rates are high the strategy tilts out of equity (even allowing shorting) and when low it tilts into equity[7]. On average this dynamic allocation is also 50% in equity. These are shown below:


The results? The dynamic strategy boosted the return achieved by 0.7% p.a. but incurred a tracking error of 5.0% p.a. This meant it offered an information ratio of just 0.7%/5.0% = 0.14 before costs (obviously less after costs).

This is surprisingly modest, but achieving even this result could be challenging in reality. The strategy assumed the investor knew the average cash return over the period and so could identify when cash rates were ‘high’ or ‘low’. The reality is tougher. For example, you might think cash rates are currently high, so underweight equity under the assumption you can switch to overweight equity when cash rates fall. But there is no guarantee that day will ever come, especially if you don’t have the luxury of an ultra-long investment horizon!

This suggests a crude strategy of using only the prevailing cash rate for market timing may not be particularly profitable. Taking account of more information – such as equity valuation metrics[8] – might improve matters. That said, investors could be better off tilting between asset classes with similar long-term expected returns. For example, if they believe US equity is currently overvalued relative to UK equity, they might sell some US equity and buy some UK equity, leaving the overall risk on/off split broadly unchanged. Many such tilts can be performed at the same time, offering diversification benefits.

A Bayesian mindset

High equity volatility and the decreasing relevance of historic data to today’s markets means it is impossible to prove either the relevance or irrelevance of cash rates statistically[9]. As such you need a prior belief or starting point. We think the best and most natural prior belief is market efficiency. However, you can then apply the idea of Bayesian updating.

Evidence such as the above is far from conclusive but suggests equity markets are relatively less attractive when cash rates are higher. A modest tilt to reflect this, or preferably an analysis of fundamentals, is not foolish.

But market timing is hard – so it may be worth prioritising other ways to boost risk-adjusted returns. Examples include multi-asset diversification, accessing alternative risk premia and tactical tilts that keep long-run expected returns the same.


[1] Such as the Capital Asset Pricing Model (CAPM) or Arbitrage Pricing Theory (APT), provided there is no risk impact

[2] There is also no apparent relationship if you allow for the level of equity volatility

[3] For a US investor.

[4] In each of 3,000 simulations, monthly cash returns were modelled stochastically, and US equity returns were modelled as the sum of the cash return and a random excess return with constant mean.

[5] When calculated across time as opposed to across simulations.

[6] Our modelling allows for the high autocorrelation (or persistence) of cash returns through time. This may help explain why we don’t find statistical significance whereas Blitz does. Another potential approach is to test whether equity and cash returns are cointegrated in the long run. We found no evidence of cointegration using a Dicky-Fuller test.

[7] The tilt is sized proportionately to the estimated risk premium.

[8] For example, the CAPE ratio

[9] There is some tentative evidence that risk-free yields over longer horizons (using gilt or treasury yields) have been more predictive of equity returns over the same horizon, but this also struggles with statistical significance.

John Southall

Head of Solutions Research

John works on financial modelling, investment strategy development and thought leadership. He also gets involved in bespoke strategy work. John used to work as a pensions consultant before joining LGIM in 2011. He has a PhD in dynamical systems and is a qualified actuary.

John Southall

Yikai Shen

Quantitative Associate

Yikai works as a quantitative associate in LGIM's Solutions group. Yikai assists in financial modelling and investment strategy development work. Yikai joined LGIM in August 2020 from Duff & Phelps, where he was an associate within the complex asset solution team and assisted in the valuation of credit derivatives. He obtained a master's degree in Maths and Finance from Imperial College London.

Yikai Shen