In the 20th century, we have observed three periods of decoupling between bond prices and equity prices; once during the 1930s, once in the early 1960s and lastly a period that started at the end of the 1990s. Computing these correlations using daily, weekly or annual returns and over different time windows will impact the way the change in correlation is tracked and analysed. However, beyond these minor differences, this decoupling phenomenon can be explained by the way economic and financial conditions lead investors to review the key components of the pricing of these asset classes.
When we consider a conceptual framework based on the fundamental equation of stock and bond pricing (using a Gordon Shapiro framework for equities and Fisher work for bond prices), it’s clear why there should be a link between these asset classes, since both discount future cash flows with a specific rate. We can see this in the equations.
All of these financial items are expected. Some of course can be predicted with a much higher degree of certainty than others.
The main difference between equity and bond prices has to do with the incremental equity risk premium investors require to invest in equities, the variability in the dividend growth for equities in the numerator of equation 1, and the fact that equities have an indefinite life whereas bonds tend to have a limited maturity. Naturally, with bonds, the coupon and principal payments are usually determined at issue and backed by the credit worthiness of the issuer.
Focusing solely on the correlation changes themselves, the following three outcomes are possible when viewing the same periods for the computation, namely:
❑ Correlation remains positive when both asset classes rates of return are positive;
❑ Correlation is positive when both assets’ rates of return are negative;
❑ Correlation becomes negative when one asset class goes up, and the other down.
Moreover, it is important to bear in mind that correlations over a certain horizon between the two asset classes can be negative, but that both asset classes can still register positive or negative rates of return at the end of this horizon. We witnessed this negative correlation in 2003, 2004 and 2005 year to date.
If we wish to understand why a negative correlation between equities and bonds is possible, we need to focus on the economic and financial forces that allow the underlying pricing elements to move in such a way that the overall impact on the price makes them diverge during the individual sub-periods.
In a normal economic environment, we would expect dividends to grow in line with economic activity (even if there is empirical evidence showing that dividend growth is less than gross domestic product growth in real terms over the long run). However, in periods of economic weakness and low inflation, even deflation, the growth rates in dividends will become very small or even negative. Hence equity prices will be severely impacted by the decline in the elements in the numerator of equation 1; as the economic environment deteriorates the larger the decline in the expected equity risk premium (ERP). This premium will usually increase as investors require a higher expected return in order to be compensated for taking the extra risks associated with equity investing.
In a recessionary environment, as dividend growth declines and bankruptcy risk increases, the increase in ERP usually offsets the decline in long term yields commensurate with the decrease in forecast real activity and inflation.
In a steady business cycle, we would expect the bond risk premium to stay low and expected inflation and real rates to also remain steady. When the economy overheats the three parameters in the denominator in equation 2 usually increase, which explains why bond prices tend to decline in this environment, whereas equity may see some of the negative impact counter-balanced by higher dividend growth and lower equity risk premium.
Strong economic growth is positive for stocks and negative for government bonds, which explains why a negative correlation is possible at a time when equities register positive rates of return and bonds register negative returns.
Beyond the equations, the psychology of the market participant is also a driver of a negative correlation change. This was especially evident at the peak of the equity bubble in the nineties when investors had unrealistic expectations of dividend growth and equity risk premium – just as night follows day, this was corrected in 2001 and 2002. The equity de-rating was a consequence of the rise in expected risk premium resulting from the deterioration of equity fundamentals and the large rise in equity volatility over the period.
Since the turbulent period of the second half of the 1990s the concept of ‘flight to quality’ has illustrated the sudden re-rating of long-term government bonds in periods of market volatility (highly risk averse regime). This has reinforced the role of government bonds (especially long-term ones) as a safe heaven and therefore explained why investors have started to re-rate bonds over equities for their true defensive characteristics. A negative correlation is not surprising in this context.
Lastly, correlation between the two asset classes has also been positive in inflationary times. When inflation picks up, this has a negative impact on real earnings growth rates (and then dividends) and the inflation risk premium, which affects equity and bond markets in the same direction.
Investors’ preferences have become strongly influenced by the introduction of regulatory change, such as the financial assessment framework, or FTK, regulation in the Netherlands, the implementation of IFRS in Europe and FRS 17 in the UK.
These regulatory changes have modified the landscape for the relative pricing of both asset classes and influenced the demand for long-term fixed income instruments, bidding up their prices to such an extent that there is hardly a detectable risk premium currently priced into bonds.
We could expect the correlation between equities and bonds to increase and become less negative in the near future as there is evidence that we have moved back into a slightly higher inflation environment.
The correlation could eventually become positive again as monetary policy readjusts. Of course without a longer term perspective, we cannot judge if this is a temporary change, or if it will become more structural in nature.
On the one hand, if inflation picks up beyond levels that are unexpected nowadays, many investors may be cautioned to remind themselves of the events of 1994 when bonds and equities fell in sympathy, after a massive wave of unexpected increases in short term interest rates on both side of the Atlantic. In this case we could see a sudden and sharp increase in correlation as we witnessed in late 1993 and 1994 and this would be associated with a decline in both asset class prices.
Surprisingly, there may be a vested interest for many market participants in not observing a very positive correlation between equity and bonds in the short term as this has historically turned out to be bad for investors - as in 1994.
If the rebound in inflation is more subtle and gradual, then the readjustment will take place more slowly and we can expect the correlation to stay around zero, and eventually become slightly positive in the future.
It is safe to assume that the regulatory change currently reshaping the investment landscape will remain an important driver of correlations, if for no other reason than the fact that institutional investors must now favour assets whose correlations with liabilities are very high. This change in topology can only lend longer term structural support to the correlation of equities and bonds remaining in negative territory for many years to come.
Frederic Dodard is head of asset allocation and balanced portfolio management in Europe, State Street Global Advisors