Pension funds cautioned on equity-bond correlation

by Luke Smolinski
Risk.net
30 September, 2016

Asset managers are disregarding the chances of a lasting shift in correlation between bond and equity prices, according to experts, with pension funds particularly unprepared.  Negative correlations have been the theoretical underpinning for investors diversifying their assets for 20 years. While some are optimistic about pension funds' resilience to correlations spiking, saying their liabilities drop more than the value of the stocks and bonds they own, others suggest pension funds are less prepared for falling bond prices than at any time in the past eight years.

Jack Goss, consultant for Imagine Software, a risk management technology firm in London, thinks many on the buy side are assuming a stable correlation in their value-at-risk models and will be ignoring stress tests that forecast big losses in the eventuality that correlations shift.  "We see people at the moment fearing that correlation will suddenly start to spike… If you get high inflation, low growth, both equities and bonds will fall together and that is an issue playing on people's minds," he says.  "The market in general is assuming correlation will not return to the levels of the 1970s and 80s."

Goss suggests that rather than plugging a different correlation number into VAR, risk managers should use some form of multivariate test, in which they can assume a range of probabilities of different correlations. Expected losses here would be determined mostly by the probability of a shift, he thinks.  For example, with a model assuming a one-in-five chance of correlation moving to +0.4, the risk manager's degree of conviction would have a greater effect than the precise change in correlation expected, he points out.

Concerns are heightened by the correlation turning positive in recent months. Three-month correlation between the S&P 500 and 10-year US Treasury bond prices switched from –0.66 immediately after the UK's EU referendum to +0.27 in the last month, according to analysts from Bank of America Merrill Lynch.  In early September, both bonds and equities sold off heavily, with the S&P 500 falling 2% on September 9 and US 10-year treasury yields hitting their highest levels since June.

Meanwhile, rolling yearly world equity-bond correlations have been negative with one exception since 1998, at an average of –0.35, before which they were mostly positive with an average of +0.24 since August 1962, according to quant fund Winton.  Goss also suggests risk managers are quietly disregarding stress tests they perform, as correlation spikes are thought to be unlikely. "[Stress testing] is being done at the moment," he says, "But a lot of people sort of say, ‘Well if it happens, we're in a fairly bad way', then they shrug their shoulders and move on. A lot of the time the result is so bad, what else can you do?"  Stress tests alone in this situation fail to impel asset managers to reallocate, whereas multivariate analysis would, he thinks.

A chief risk officer (CRO) at a fixed-income hedge fund says he tends to simply compare the losses in two correlation forecasts. "One can assign probabilities to different correlations, but it's very difficult to combine [them] together in a meaningful fashion that would be easy to explain to others," he says.  A lasting shift in correlation will affect pension funds most, suggests another hedge fund CRO.

Con Keating, head of research at BrightonRock Group, a London-based insurer for pension funds, says many pension funds might overlook the history of positive equity-bond correlations before 1998 as "the market's memory really does not extend beyond about seven years".  Asked how prepared pensions funds would be for a shift in correlation, Keating says: "Not at all. Quite the opposite: bond exposures have never been as high as they are now."  "All those who did liability-driven investment or moved their bond allocations up dramatically – which would be about 80% of them – are going to be hurt, because they're going to be sitting holding a lot of bonds that have performed badly. Their discount rates will go up, which will lower their liabilities, but the aggregate effect is not going to be good," he says.  He thinks pension funds cannot defy pain by diversifying into corporate bonds, commodities or hedge funds, and suggests they hedge bond exposures with options.

Hans den Boer, chief risk officer for the UK's Pension Protection Fund (PPF), which insures around 6,000 defined-benefit pension schemes, is more upbeat about pension funds' resilience.  "If interest rates go up and equity markets stay stable, that would mean asset values would go down, but the liability side would come down quicker, so their deficit situation would improve significantly," he says.  Many pension funds insured by the PPF have fewer than 100 members and would not have the scope to model correlation probabilities, he thinks.

The UK private pension deficit rose to £408 billion ($529 billion) in July, from £222 billion in January. Eighty-four percent of UK company pensions were in deficit and total assets stood at £1.4 trillion at the end of July.  Duncan Lamont, London-based head of research and analytics at Schroders, says high inflation led to positive equity-bond correlations before 2000.  Typically since 2000, interest rates rose as growth was revised up, so bond prices fell at the same time as investors valued company stocks higher. The prospect of central banks putting up rates after years of investors bulk-buying bonds is causing concerns about the correlation switching back to positive.  "Within pension funds, they are working hard to find asset classes which are not correlated," says Lamont, pointing to commodities and insurance-linked securities as non-correlated assets.