Opinion: Why are bond investors so bad at predicting interest rates?

This article and the graphs mentioned can be seen here: Citywire

Jack Roberts, manager of the Atlantic House Dynamic Duration fund, says human bias is largely at fault for flawed predictions and following certain rules can help investors.

Interest rates movements are notoriously difficult to predict.

Time and time again, there is a significant divergence between market-implied interest rates for the UK and the actual base rate.

Making decisions on overall portfolio duration is therefore very difficult and making the right rates call is as difficult as stock picking.

However, this is not because of a lack of supply and demand in the market, as short-term interest rates are highly liquid and tradable, and getting such a call right can be very profitable for investors.

Chart: Market-implied interest rates in grey vs actual base rate in blue. Source: Atlantic House Investments

Following the global financial crisis (GFC), investors who were used to higher interest rates expected the supportive monetary stimulus in the form of near 0% interest rates and QE to reverse.

Their predictions were clearly incorrect and were likely caused by the human bias known as anchoring. Anchoring bias causes decision makers to rely too heavily on past information, or an anchor, which in the period post-GFC was that the higher rates environment should return.

Fast forward to 2021, the low rates environment with negligible inflation which persisted for more than 10 years created a new anchor.

But when the elevated levels of inflation which were assumed to be transitory stopped looking so transitory post-pandemic lockdowns, central banks worldwide had to react very quickly by hiking interest rates, and once again, investors were caught out.

As of today, expectations appear anchored to a return to lower interest rates given the recent rapid rate rises. But is the upward trajectory in short-term interest rates finished? We’ll leave you to make that difficult call and instead focus on the long term.

Human bias seems to be largely at fault for incorrect predictions. To address this, instead of relying on human intuition, a combination of observing certain signals and implementing a systematic rules-based approach can introduce conviction and remove the influence of anchors. Ironically, rules can free us.

Choosing a relevant signal is the next challenge. Helpfully, economic theory can guide us on this point. It is well established that interest rates and inflation are linked, not only by correlation, but also by causality.

A change in one variable consistently determines the other’s fate. The economists who asserted this intuition, Taylor (rule) and Fisher (equation), found that it was strongly supported by data in developed economies.

The Taylor Rule links benchmark interest rates to levels of inflation and economic growth. The Fisher Equation, meanwhile, connects nominal and real returns with the expected inflation rate.

By applying these theories to a strategy which adjusts its interest rate exposure, assets can be allocated based on three signals, all with inflation at their core.

Each signal is different, but they are linked by the idea that if inflation is under control, then the allocation should hold bonds (increase duration). Alternatively, if inflation is threatening, the allocation should hold fewer bonds (reduce duration).

There are three signals which we believe to be particularly significant. First is the Central Bank Credibility signal, based on the Taylor Rule. This states that if core inflation is above a central bank’s target by more than 1%, the allocation moves away from bonds (lowers duration). If core inflation is within target, the allocation moves towards bonds (increases duration).

Second is the Market Inflation Credibility signal, based on the Fisher Equation. This states that if real yields in the market are positive, inflation expectations are under control, and the allocation moves towards bonds (increases duration). If real yields are negative, the allocation moves away from bonds (reduces duration).

Lastly is the Inflation Trend signal, based on momentum. This indicates that if year-on-year inflation is increasing, the trend is expected to persist. Higher inflation implies a worse environment for bonds, and therefore, the allocation moves away from bonds (lowers duration). Alternatively, if the trend in inflation is decreasing, the allocation moves towards bonds (increases duration).

Applying an equal weight to each signal offers diversification and simplicity. Moreover, inflation data is published monthly in the UK and US by statistical agencies, meaning a strategy can be rebalanced monthly. 

Fewer data points also suit a rules-based approach such as this, as opposed to an algorithmic approach.

We recommend this rules-based signal approach to managing portfolio duration, and those who have the necessary expertise to access derivatives can trade pure inflation, which is not available to cash investors wishing to implement such a strategy.

Pure inflation investing is not dissimilar to a bet on inflation over a specified period. The benefit of this versus investing in inflation-linked bonds is access to inflation without the duration, coined as a ‘golden ticket’ for bond investors.

Given this enhanced access to the inflation market, the asset allocation of a fund can move efficiently between bonds and pure inflation, presenting opportunities for investors. Such an allocation has powerful diversification benefits compared to a traditional bond holding, as inflation and bonds tend to be negatively correlated. Furthermore, an inflation uplift can increase returns at the same time as reducing risk.

Recent history has reminded us that being nimble to protect portfolios during changing inflationary environments is a necessity. However, reliance on human intuition alone to make the right calls may be misplaced. Instead, rules and processes, like regulation, can protect investors from fund manager misjudgements.

We do not suggest a bond allocation to be entirely composed of such a strategy. Traditional bond allocations with corporate credit risk, differing maturities, and yield curve strategies have their own unique risks and returns. However, there is value in a dynamic and rules-based approach to duration management.

Jack Roberts, CFA - Fund Manager