October European Issue Focus - Where Will Global Interest Rates Peak




Yields low, but for how long?

Global interest rates have been unusually low for several years. Short-term rates in most advanced economies are close to zero (and negative in real terms) and longer-term rates are also very low, by recent historical standards.

As time has gone on and interest rates have failed to rise back towards ‘normal’ levels, it has been increasingly argued that these ‘normal’ levels have themselves shifted down significantly. So how much of an adjustment has taken place, and where is the likely long-term or ‘terminal’ level for global interest rates?

Short-term rates: We look first at short-term interest rates. Here, a popular concept is the ‘natural’ rate of interest (or ‘R*’) which is defined as that real (inflation-adjusted) short-term rate, consistent with output being in line with potential output, and inflation being constant.

This variable cannot be viewed directly, but it can be estimated. Generally, it is believed that R* is strongly influenced by trend economic growth (the idea being that, in equilibrium, interest rates should be close to the rate of return in the economy). So, if trend economic growth has fallen, then so should R*.

Estimates of trend economic growth in the U.S., Eurozone and the UK have indeed declined over recent years (see Chart 1) and this probably explains a large chunk of the drop in estimates of R*. However, some estimates suggest that R* has fallen by a lot more than just declines in trend growth could explain.

For example, estimates by economists Holston, Laubach and Williams (2016) suggest the real natural rate for the U.S. may be just 0.4 per cent, and -0.5 per cent for the Eurozone. These figures are 1.0-1.5 percentage points below estimated trend growth. Thus, ‘other’ factors (including debt levels, changes in lending spreads and global factors) may be significantly compressing short-term natural rates.

Long-term interest rates: Another element of the story is the relationship of long-term interest rates (e.g. government bond yields) to short-term rates. Long-term yields are, of course, partly a function of the current and expected path of short-term rates. However, they are also affected by a ‘term premium’, which is usually thought to reflect uncertainty about the future path of short-term rates as well as factors such as fiscal risk.

This term premium has historically added around 1.0 percentage point to 10-year U.S. yields, on average. Although, it has been close to zero recently. Why has this happened? In our view, there are two main reasons.

The first reason being very large scale purchases of government bonds by central banks. We forecast that in 2016, combined bond purchases by the European Central Bank, Bank of England and Bank of Japan will easily outstrip combined net government bond issuance in the Eurozone, UK, Japan and the U.S. (see Chart 2).

These purchases have contributed to the second factor, which is a global shortage of ‘safe assets’. Our estimates suggest that the supply of ‘safe’ (i.e. highly rated and liquid) government bonds and close substitutes has contracted from 45.0 per cent of world GDP in 2000 to 32.0 per cent in 2015.

This shortage has also resulted from some ‘safe’ assets dropping out of this category (e.g. AAA-rated mortgage-backed securities) and from a mismatch between supply and demand, resulting from the pattern of global growth. Rapid growth in emerging countries, such as China and India, has boosted demand for safe assets, however, these countries are often unable to supply them so demand is concentrated on ‘core’ advanced debt markets.

Many forecast uncertainties!

So what are realistic levels for peak short- and long-term interest rates in the major economies? In our view, there is great uncertainty about forecasts for these, given the many moving parts. These include what has happened to R*, and the term premium as well as the outlook for inflation.

R*: As noted above, some estimates for the natural short-term rate are now well below the level of potential growth. Should we expect this to remain the case, or should the unobservable ‘other’ factors influencing R* ease over time?

If we assume the Holston et al. estimates of R* are the bottom of the possible range, while the pre-crisis levels of ‘other’ factors (generally close to zero) are the upper end, we can get a range for R* for the U.S. of 0.5-1.2 per cent, -0.4-1.0 per cent for the Eurozone and 1.6-1.9 per cent for the UK.

Inflation: Our baseline forecasts assume that inflation returns to close to the target level of 2.0 per cent in the U.S., UK and Eurozone by 2026. However, there are arguably considerable downside risks to this. Inflation has been undershooting these targets in recent years, and market-based expectations of long-run inflation are generally below 2.0 per cent, being closer to 1.0 per cent in the Eurozone, for example. 

If we assume that long-run inflation is between 1.5-2.0 per cent, then adding this to our range for R*, we can get a range of 2.0-3.3 per cent for the terminal nominal short-term for the U.S., 1.1-2.9 per cent for the Eurozone and 3.1-3.9 per cent for the UK.

The term spread: We expect central bank bond purchases to wind down from 2018, so the downward pressure on long-term yields from this source should ease. However, we think the ‘safe’ asset shortage related to the mismatch between the pattern of global growth and the supply of safe assets will endure.

Therefore, we think a realistic range for the long-run-term premium is 0.5-1.0 per cent. The bottom end of the range suggests some increase from current levels, while the top end is the longer-term historic average.

Adding this to our ranges for nominal short-term rates, we get a possible range for terminal 10-year bond yields of 2.5-4.3 per cent for the U.S., 1.6-3.9 per cent for Germany (the base for the Eurozone) and 3.6-4.9 per cent for the UK (see Table 1).

Our forecasts: The above ranges are very wide, and imply rather large potential price ranges for government bonds over the next decade, but where do our forecasts sit within these ranges?

In general, we are around the midpoints of the ranges. We expect Eurozone and U.S. short-term rates to peak at 2.75 per cent (in line with the Fed’s current estimate for the U.S.) and UK rates at 3.5 per cent. For 10-year bond yields, we expect yields to peak at 3.5 per cent in the U.S., 4.3 per cent in the UK and 3.1 per cent in Germany. The term spread in Germany is narrower than in the U.S. and UK, reflecting Germany’s relatively strong fiscal position.

We anticipate central bank rates will rise as output gaps are closed, and inflation moves back up towards target levels, but we forsee a slow process of rate normalisation. We expect yields to peak in the U.S. and UK only in 2026, and in the Eurozone in 2028.

A key reason for the slowness of rate normalisation is the level of uncertainty about where the ‘natural’ short rate, R*, actually is. The range of estimates is wide, and the confidence intervals around those estimates are also wide (as much as 1.0-2.0 percentage points on either side).

This is likely to make central banks quite cautious about tightening policy in the years ahead, waiting longer than they might otherwise do before adjusting policy in response to incoming economic news. 

This update was researched and written by Oxford Economics, 121 St. Aldates, Oxford, OX1 1HB, England, as of 10 October 2016.