Treasury bonds tell a tale of great expectations
Investors in US Treasury bonds enjoyed spectacular real returns of more than 8 per cent annualised between the autumn of 1981 and the summer of 2012, as the yield on the 10-year dropped from an all-time high of almost 16 per cent to a record low below 1.5 per cent.
Astonishingly, a dollar invested in risk-free Treasury bonds outgrew a dollar invested in the stock market over this period – $11.54 versus $11.02 in real terms.
The bull market has been nothing short of extraordinary, but it was interrupted last May by Ben Bernanke hinting that the Fed could soon take the first step toward the normalisation of monetary policy. Bernanke’s comments sent bond markets into a tailspin; the 10-year yield surged from below 2 per cent the day before the former Fed chairman’s remarks to almost 3 per cent by early-September.
The sell-off could well spell the end of the 30-year plus bull market in Treasury bonds. Investors need to be reasonably sure of what constitutes a “normal” level of long-term interest rates, and how soon yields can be expected to reach that level.
According to the expectations hypothesis, the expected return from holding a 10-year Treasury bond until maturity is identical to the expected return from the continuous rolling over of a series of short-term bonds over the same 10-year period. Thus, a “normal” level of long-term yields must be a function of an equilibrium short-term policy rate.
The equilibrium real rate or natural rate of interest is one consistent with stable inflation and output equal to potential. Not surprisingly, it is a function of the economy’s potential growth rate.
A growth accounting framework is typically used to estimate the economy’s long-run potential growth rate. This framework decomposes real GDP into labour input (hours worked) and labour productivity (output per hour).
The growth in labour input or hours worked averaged about 2 per cent during the 1980s and 1990s, and stemmed primarily from increases in the working age population (WAP) and a higher labour force participation rate (LFPR) – the positive growth contribution attributable to the latter was due to an increasing share of working-age females seeking work.
A notable slowdown in the annual rate of increase in the WAP, combined with a marked decline in the LFPR, contributed to an average growth rate in hours worked below 1 per cent in recent years. The unfavourable trend reflects the adverse impact of an ageing population on growth in the WAP and the downward pressure placed on the LFPR, due to the growing importance of the labour supply above 55 years – a group that exhibits participation rates substantially below average.
The annual growth in labour input should slow to just 0.5 per cent in future. This downward shift is unlikely to be offset by an upward shift in the growth rate in labour productivity. Indeed, labour productivity growth dropped from an average rate of about 2 per cent during the 1980s and 1990s, to 1.5 per cent during the first post-millennium expansion, and slipped to just 1 per cent over the last two years. In this context, a future growth rate of 1.5 per cent seems reasonable.
The equilibrium real rate is about 2 per cent, but a “normal” nominal rate of interest will also include an inflation rate that is consistent with the economy expanding at its potential growth rate. The Fed’s inflation target is 2 per cent, a rate that has been accompanied historically by the best growth outcomes. This puts the “normal” policy rate at 4 per cent.
A “normal” 10-year yield will tend to be higher again, since investors typically require a premium to hold a 10-year bond until maturity, instead of a series of short-term bonds. The premium’s size depends on a number of factors including the level of inflation uncertainty and central bank credibility, not to mention the role of Treasury bonds as a deflation hedge and safe haven in times of stress.
Estimates of the term premium show it was close to 4 per cent in 1981, reflecting high inflation risk and low central bank credibility. The premium trended downwards through the 1980s and 1990s as the Fed re-established its credibility, and visions of permanent prosperity pushed it down to zero before the crisis struck.
The premium dropped below zero in the wake of the financial crisis, due to quantitative easing and robust private-sector demand for safe assets. The recent yield increase has seen the premium move back into positive territory, but well-anchored inflation expectations – and the portfolio insurance provided by Treasury bonds – mean that it is likely to remain not far above zero going forward.
Thus a “normal” 10-year yield is about 4 per cent – well above current levels. However, a violent sell-off from current levels appears unlikely given that the economy should not reach potential until 2016 at the earliest, while the term spread already anticipates a return to “normal” policy rates within the next five years.
The secular bull market in Treasury bonds may well be at an end, but analysis suggests yields are likely to be range-bound and trend sideways for now.