Join our community of smart investors

Getting rates wrong

Something unusual has happened in the last few months: interest rates have been higher than investors had expected.

There’s a simple way of assessing this from the basic maths of the gilt market. The expected returns on a two-year gilt should be equal to the expected return on a one-year gilt which we reinvest on maturity for another year. From this starting point we can infer the market’s expectations for the one-year in a year’s time from the level of one and two-year yields today.

A year ago the one-year yield was minus 0.15 per cent and the two-year yield was minus 0.14 per cent. That meant the market was forecasting that the one-year yield today would be minus 0.12 per cent. In fact, as I write, it is 0.6 per cent. That’s 0.72 percentage points more than expected.

My chart shows why this error is unusual. It shows the gap between actual one-year yields and those that the market was pricing in 12 months earlier. It is a measure therefore of the forecast error in interest rates. It tells us that the 0.72 percentage point higher than expected rate today is the biggest positive interest rate surprise since 2007. Which is why I say recent rate moves have been unusual.

What have been much more common are downside surprises – interest rates being lower than expected. Since the Bank of England was given operational independence in May 1997 one-year yields have on average been 0.46 percentage points lower than the market expected 12 months previously. Granted, some of this was because the 2008-09 crisis caused rates to plummet unexpectedly. But that’s not the whole story. Even since 2011 one-year yields have on average been 0.27 percentage points lower than forecast 12 months earlier.

For years, investors have been expecting interest rates to rise and have been consistently surprised by them not doing so – until the last few months. The forces of secular stagnation, which cause low growth and low “natural” interest rates, have been stronger than investors have realised.

In fact though, this tendency to over-predict interest rates is an old one. From 1973 to 1997 interest rates were on average lower than expected, because investors were surprised by the recessions of the early 1980s and early 1990s and by the fall in inflation in between.

It’s in this context that we should read the market’s current forecast, which is for one-year yields to rise from 0.6 per cent now to 0.9 per cent by December. It thinks rises in the Bank rate in response to inflation will push up yields. History warns us that such forecasts have in the past had an upside bias.

It’s easy to tell a story of how history will repeat itself. Real incomes will be squeezed by higher gas prices, a public sector pay squeeze and increased national insurance contributions in April. That will hit consumer demand and hence retailers’ ability to raise prices. With China’s weak monetary growth predicting falls in raw materials prices, and price increases last year dropping out of the inflation data from the spring onwards, inflation should fall. That could cause interest rates to surprise us on the downside, as has happened so often in the last 50 years.

But, but, but. For one thing, we must guard against the narrative fallacy: we attach too much credence to stories, and therefore trust forecasts too much. And for another, investors should (eventually) learn. Having realised that they’ve been very often wrong in the past to expect rising rates, they should now be more cautious. If so, the history of over-predicting rates doesn’t tell us much about the market’s next error.

There is, however, another important fact about interest rate forecasts. It’s not just that they have been biased, but that the mistakes have been large. Since May 1997 the average error regardless of direction has been 0.8 percentage points. Even if we ignore the 2008-09 period (which we should not!) and look only at the post-2011 period the average error has been 0.4 percentage points.

One interpretation of this is that, given the market’s forecast of a one-year yield of 0.9 per cent in December, there’s a two-thirds chance of yields being between 0.5 and 1.3 per cent; a one-in-six chance of yields being under 0.5 per cent; and a one-in-six chance of them being over 1.3 per cent. And this perhaps understates the uncertainty because it is based on an average error taken from a period that under-samples big surprises.

The point, then, is that interest rate forecasts are unreliable, even when made by people who have plenty of skin in the game. We must be prepared for surprises. As investors, our job should not be to predict moves in yields – which is impossible – but rather to ensure that our portfolios are resilient to surprise moves.