Should dividend yields be higher than government bond yields?

For most investors alive today, this will be an unusual state of affairs. Since the mid-1950s, US dividend yields have been almost exclusively lower than bond yields.

But this wasn’t always the case. As the chart above shows, a higher dividend yield was the normal state of affairs for most of the first half of the 20th Century (and for many years before that).

Why is this important? The key role of the regime

For an investor who had only seen the last fifty years of markets, a knee-jerk reaction to the higher dividend yield today would be to think that equities are ‘cheap,’ bonds are ‘expensive,’ or both.  The only other two periods in the last fifty years where dividend yields were higher (in early 2009 and mid-2012) proved to be good entry points for equity.

But such rules of thumb are dangerous. This is particularly true today, since while equity weakness was the cause of higher dividend yields in 2009 and 2012, this year it has been falling bond yields that have done most of the work.

The more important questions are about the regime: what led to dividend yields being higher for the first half of the 20th Century (and for many years before that)? Why did this reverse? And have we now come full circle? We cannot properly answer the questions of whether equities or bonds are ‘cheap’ or ‘expensive’ without considering these issues.

Perceptions of risk

In 1997, as the tech bubble was beginning to take hold, Peter Bernstein wrote an excellent piece on the nature of the long run which touches on the importance of regime. In it, he explained why bond yields had been lower than dividend yields for much of history:

“Until 1950, most conservative investors were convinced that bonds were the superior asset, and surely the less risky asset, and they had 150 years of history to support their view…Now we stand at a juncture where about fifty years of history support the view that stocks are the superior asset and, over the long run, surely the less risky asset.”

So what changed? As Bernstein suggests, the huge change in perceptions of risk was driven by experience. According to the Barclays Equity Gilt Study the real returns on US Treasuries (around 20 year maturity) between 1950 and 1981 was an annualised -2%, for UK Gilts this was an even worse -3%. This experience, compounded by the fact that the most extreme losses had been associated with the inflation of the 1970s, explains why the early 1980s represented the period when equities were seen as at their least risky relative to bonds.

Since then, we have seen a steady reversal in perceptions of risk as bonds have delivered exceptional returns. The starting point of high interest rates and declining inflationary pressures have driven this, and have also altered the correlation dynamic between bonds and equities.

We are now used to bonds in a portfolio providing safety when equities are weak. Bonds are now seen as less risky in their own right and an insurance policy in conjunction with other assets.

Looking ahead

Valuation signals are always conditional on the environment. A dividend yield that is higher than the bond yield should not be thought of as the screaming buy signal that it might have been in recent decades. We cannot say that one yield ‘should’ always be higher than the other.

But that is not to say that the signal be ignored. The effects of compounding suggest that equities and corporate bonds are priced today such that even though they may continue to be more risky in a volatility sense, they offer the chance of higher prospective returns. This is true in most global markets today.

Moreover, the transition from the regime of the 1970s seems to be complete: market fears of inflation are very subdued and interest rates are at all-time lows.

This suggests that government bonds are far less likely to continue behaving as they have for the last thirty years. We have already seen a couple of phases in recent history where correlation patterns between equities and bonds have shifted.

It might therefore be tempting to say that the world looks more like the first half of the twentieth century than the second half and that we should expect dividend yields to be higher from here.

But taking the historical analogy too far would be risky. Current nominal interest rates are already lower than they were then, and to achieve anything but disappointing returns from many negative or low yielding government bonds would require further declines. While dividend yields may remain above bond yields, it seems less likely that you would get the outperformance of government bonds that characterised the long periods prior to 1950.

The reality is that while history may ‘rhyme’ it never offers a playbook for dealing with the present.

_{*For simplicity’s sake the charts in the post show nominal bond yields versus dividend yields, which are expected to keep up with inflation to some extent. This will exaggerate the gap in the 1960s and 70s but does not invalidate the broader observations, particularly in today’s environment where inflation is low.}