Skip to main content Skip to navigation

How do demographics affect interest rates?

The middle-aged invest and the young borrow and the reationship between the two will affect interest rates, writes Dr Arie Gozluklu, Associate Professor of Finance at Warwick Business School. How important is tracking the slow moving changes in demographics when modelling forecasts for the long-term?

Interest rates“Every man gotta right to decide his own destiny,” sang Bob Marley in 1979.

But when it comes to long-term interest rates trying to decide their destiny has become increasingly difficult for the US Federal Reserve and central banks around the world. In 1979, when Marley sang about Zimbabwe’s revolution, the US base interest rate was at 16 per cent and the following year reached a record high of 20 per cent, while in the UK they reached 17 per cent.

But since the global financial crisis interest rates have been at an all-time low across the Western world. The base rate has been hovering around 0.5 per cent in the UK for the past decade, and it has been the same in the US before climbing to above two per cent in 2018.

The power of demographics

It had been thought these incredibly low levels were the result of the global financial crisis as the developing world tried to limit the damage with the extreme measure of quantitative easing.

But policymakers are increasingly coming round to the idea that something more powerful is at play, something they are powerless to control; something that controls real interest rates despite their best efforts to control their own monetary destiny – and that’s demographics.

Former US Treasury Secretary Larry Summers has popularised the secular stagnation theory to explain this era of low growth and low interest rates, while Andrew Sentance, Professor of Practice and former member of Bank of England Monetary Policy Committee, similarly talked of a “new normal” after the Great Recession of 2007 to 2009, with low interest rates failing to budge spluttering GDP growth. Both suggest demographics is a potential factor, pointing to low population growth and increasing life expectancy.

But my research, along with other colleagues (Carlo Favero, Andrea Tamoni, Haoxi Yang and Annaig Morin), indicates a population factor that has been largely ignored in the thinking of why real interest rates – that is the nominal or base interest rate minus the inflation rate – are so low in the US and much of the Western world, and that is the ratio of the number of middle-aged (40 to 49 year-olds) to young (20 to 29 year-olds), a variable first introduced in a model by Geneakoplos, Magill and Quinzii.

The MY ratio

Demographics has been used in other models before to explain long-term interest rates, but this has predominantly looked at the size of the population. We have found the composition of the working population, specifically the middle-aged to young (MY) ratio, is a more important factor.

Franco Modigliani and Richard Brumberg’s life cycle investment hypothesis suggests that people borrow when young, invest for retirement when middle-aged, and live off their investment once they are retired. So the middle-aged are the savers and if there is more of them than the young spenders, that means there is more demand for financial securities, pushing prices up and yields or interest rates down.

When the MY ratio is small there will be excess demand for consumption by a large cohort of young and therefore the price for bonds and stocks decreases, so the yield rises and saving is encouraged for the middle-aged.

When you look at interest rates over the very long term, over the last 100 years, you can see that the low rates the US is experiencing today is not just a cyclical story, they had been falling for nearly 20 years, long before the 2007-08 crisis, and this is just a continuation of that trend. Just why has been vexing economists. And although demographics has been cited, our research suggests the MY ratio seems to be the telling factor in helping us determine the future path of interest rates.


The MY ratio goes up and down in waves over time, as different size bulges work their way through the population structure. Right now we are feeling the tailend of one particularly large bulge in the US demographic.

After the Second World War the US, along with many Western countries, enjoyed a baby boom, which created a giant wave, MY ratio wave, running through the country’s demography. The 1960s thus saw a big rise in the MY ratio, pushing up bond prices and sending yields low, but by the 1980s, as Bob Marley’s battle cry echoed through the decade, this had swung round the other way.

It saw yields were high and prices low, so the MY ratio was low. As the baby boomers have gradually fallen out of the equation over the last 20 years that has swung back again, with slowly more middle-aged people compared to the young seeing a rise in savings, boosting financial asset prices and bringing down yields.

According to our data, and using the Bureau of Census projections, the MY ratio is coming down, so a standard forecasting model add the MY ratio as a variable to assess whole term structure of interest rates, from the short-term one-month bonds to the 20-year gilts, and produce a more accurate prediction of real interest rates.

We have found that the MY ratio can be used in models not only to help predict interest rates more accurately, but it can also help improve forecasting models of stock prices. We used the MY ratio in several long-horizon forecasting models comparing it to the S&P 500, including the traditional dynamic dividend growth model, and found strong predictive results.

How weighty are demographics?

When we started researching this 10 years ago there wasn’t much interest in demographics, but this is becoming an increasingly important discussion point among economists, with the Barack Obama administration citing our research in a report on long-term interest rates in 2015.

Now the question is not whether demographics determine long-term interest rates or not, but how much weight should be put on them in any model being used.

And having created models that use the MY ratio to better forecast interest rates and stock market prices in the long-term, we are now looking to see if demographics plays a part in affecting inflation. Real interest rates are unobservable, but if we can see that demographics affects inflation then we can have a better idea at what weight to give it in modelling forecasts for long-term interest rates.

The nominal interest rate should reflect the expected inflation in the future, but past inflation is not a good predictor of future inflation, it is very hard to predict, so it would be valuable to policymakers if we know the link between demographics and inflation.

However, our results in a recent paper with Annaig Morin suggest that demographics does not affect inflation as much as it affects the real rate, and there is no robust empirical evidence, even though there are papers suggesting otherwise.

Central banks react to transitory events in the economy by moving the short-term interest rate, in an effort to stabilise the long-term yield or interest rate of bonds, which are critical for business investment and households’ mortgages as they care about the next 10 years or more.

Our research shows that using the MY ratio in a standard forecasting model, like those used by the US Federal Reserve, not only provides improved long-term yield forecasts, but can also aid long-term horizon investors in stocks and bond allocation.

The destiny of interest rates might still not be in the hands of policymakers, but we can at least improve our inference of the future by taking into account slow-moving changes in demographics.

dr_arie_gozluklu.jpgDr Arie Gozluklu is Associate Professor of Finance at Warwick Business School. His research interests include Empirical Asset Pricing, Financial Econometrics,
Market Microstructure and Experimental Finance.

Terms for republishing
The text in this article is licensed under a Creative Commons Attribution 4.0 International License (CC BY 4.0).

Creative Commons License