A useful proprietary database available in Eikon is the Refinitiv monthly survey on global portfolio allocations of capital across different asset classes and geographic regions. Below is a snapshot of the global portfolio allocation of UK institutional investors and the historical tracking for global bond (government and private) and equity holdings. Equities and bonds dominate other asset classes but with significant variation over time. For instance in the run up to the Brexit vote, equity holdings dropped presumably reflecting diminished risk appetite. Given that bonds and stocks are the principal asset classes, the gains for one should come at the expense of the other. While it is true that bond allocations did not rise in the face of the downward trend in equity holdings in 2015, this may have reflected a general move to cash as concerns of Brexit filtered into the market. In any event the correlation for flows remained negative throughout this entire period.
Current and Historical Asset Allocation of UK Portfolio Managers
Source: Eikon – Click to Request a Free Trial
The chart on the next page shows the same information on global portfolio holdings of U.S. investors. The pattern of equity holdings roughly parallels the UK experience with a dip in 2016 that reversed in 2017. Bond holdings however remain far more stable which could reflect either less impact from Brexit or a different rate environment. It appears that, as with the UK, there is an ongoing negative relation between bond and stock holding. But in this instance the eye is misleading.
Don’t Trust What Your Eye Tells You
The green lines in the following charts below show the same 12-month correlation for changes in stock and bond allocations for U.S. investors. While the correlation is negative on average for the whole period, there are cases of significant and, at times prolonged, positive correlation. It is certainly possible to imagine situations when investors might simultaneously be attracted to or shun both bonds and stocks. If interest rates were unusually low due to weak economic demand it would be natural for investors to flee both bonds and stocks. Conversely, if rates were coming off highs and the economy is gaining momentum stocks and bonds would both be attractive. The common factor in both these examples is that you would expect positive correlation to be associated with broad cross-market turning points.
Current and Historical Asset Allocation of U.S. Portfolio Managers
Source: Eikon – Click to Request a Free Trial
The first chart below suggests that high correlation is associated with major turning points in the U.S. bond market. Both the rate peak in 2014 and the bottoms in 2015 and 2016 were accompanied by a surge in correlation toward positive territory. The 2016 rate peak came at the end of an extended period of positive correlation but the dip back into negative territory came ahead of the rates downturn so correlation still served as a signal. Notably, the correlation has remained negative since 2017 suggesting that the rising trend in U.S. bond rates has not yet ended.
The second chart below shows the same 1Y correlation but in this case against the annual change in the stock market; but note that stock price changes are inverted so that a lower number indicates a market gain. High negative correlations, not surprisingly, are associated with sharp market moves – positively in 2014 and 2017 while negatively in 2015 – since it implies major desires to shift holdings into or out of bonds. Unlike bonds, however, it is not clear that moving into positive correlation provides any useful market signals. But regardless of signs, it appears that low absolute correlation suggests a non-trending market; the low negative correlation this year is consistent with a market that has been moving sideways.
US Stock/Bond Portfolio Correlation and: 10Y Rate Change Stock Market Change
Source: Eikon – Click to Request a Free Trial
At Least be Skeptical
The second chart below shows the same relationship between correlation of portfolio flows and performance in bonds and stock for the European market (we use German 10Y bunds as a proxy for European government bonds because some countries have significant credit risk). Europe has had only one extended period of positive correlation but the instance of positive correlation in 2016 was associated with a shift in the trend in long-term rates. Moreover, the correlation touch of zero in 2014 also signaled a shift in the rate trend.
Notably, the European period of positive correlation does not coincide with a U.S. positive period. This suggests the decisions driving global allocations between stocks and bonds are more sensitive to local than international market factors. While periods of positive correlation are less frequent, as in the U.S., there does appear to be some value in market signaling. Both the 2015 move into positive territory and the brief touch of zero in the prior year were associated with the onset of a declining rate trend. In line with the U.S. market, the correlation pattern in Europe suggests the rising trend in bond rates is not nearing an end.
While the European stock market did plunge with the 2016 correlation surge this appears to be a one-off as there was no comparable reaction to the correlation touch of zero in 2014. The absence of any ongoing relationship between correlation and equity market performance may be because of the broad array of individual national markets embedded in the European index that may be subject to varying local economic factors.
Euro Stock/Bond Correlation and: 10Y Rate Change Stock Market Change
Source: Eikon – Click to Request a Free Trial
An Island Nation and Mirror Image Connection?
The chart on the next page shows the portfolio snapshot and history of bond and stock holdings for Japanese portfolio managers. Allocation shifts in Japan are much more volatile than is generally true for the other countries assessed above. But the gyrations in stock holdings are a virtual mirror image of equity holdings which is not surprising because allocation to other assets classes – cash, property and alternatives – is minuscule. As with the UK, correlation between stocks and bond flows remain deeply negative through the period surveyed. And there is no evidence that shifts in correlation from less to more negative – or vice versa – has any meaningful signal for market performance.
Current and Historical Asset Allocation of Japanese Portfolio Managers
Source: Eikon – Click to Request a Free Trial
The Bottom Line
Flows into and out of an asset class are linked with movements in asset prices so have little predictive value. Flows into stocks and bonds are usually negatively correlated, but in some markets – here the U.S. and Europe – there are periods when the flows coincide. It appears that periods of positive correlation are associated with major turning points in long-term interest rate trends and, for the U.S. market, extreme negative correlation is a signal for a substantial directional move in equities – though which direction is indeterminant. It is also notable that the period of positive correlation between allocations to a global bond and stock portfolio for European and U.S. managers occur at different times suggests these decisions are more sensitive to local rather than global conditions. In a future Market Voice we will examine the correlation patterns for portfolio sub-components (e.g., corporate vs government debt) to see if we can glean additional market signals.
As for market implications: The modest but negative correlation in the U.S. market suggests that the uptrend in bond yields is not near a turning point and that equity markets are likely to continue moving sideways. Europe’s negative correlation is declining but must get closer to zero before Europe’s positive bond trend is likely to reverse.
______________________________________________________________________________
