Economists, financial engineers and traders have long relied on the assumptions that asset returns are “normally” distributed with correlations that are
slow to change. However, the New Neutral, the current environment of slower global growth and lower neutral policy rates, and new geopolitical events are
leading to price action that is less apt to follow this rule. Returns (and price action) are likely to be less “continuous,” more violent and more random
in the future given the structural changes in initial conditions and end investor reaction functions.
There are many reasons that the familiar underlying assumption of normally distributed returns and stable correlations may not describe realized asset
returns going forward, including:
- Changing and erratic central bank policy (including negative interest rates)
New global regulations
- Less balance sheet availability from market-makers to provide liquidity (lower leverage ratios)
- Global uncertainty over growth prospects
- Divergence in global central bank policy – as bankers respond to regional economic conditions with the backdrop of a multi-speed global recovery
- Increasingly narrow client mandates leading to “herding” action in asset classes
A “normal” distribution meant that the odds of a market move up or down were equal and that over the long term, the average expected return was symmetric
around the “trend” return (similar to flipping a coin). During and after the financial crisis, there had been increased talk of “fat tails,” “black swan
events” (Taleb 2007) and “bi-modal” distributions – all pointing to the fallacy of normal distributions (or the higher probability of more violent market
But these concepts were not widely accepted in 2008, and certainly most market participants had not planned or reserved capital for them. As a result, when
the market repriced in 2008, the knock-on effects were disastrous: Very few participants had contingency plans in place for the market’s new, more volatile
The same could be said today: Volatility that affects returns across markets still catches many participants by surprise. But based on our analysis of the
current market versus historical performance, such events are now part of the New Neutral landscape. Take, for example, the number of
multiple-standard-deviation events that have occurred to date in 2015 assuming that this pace continues for the remainder of the year versus the 10-year
history of these events, as shown in Figure 1. In the major markets PIMCO tracks daily, all except credit have seen dramatic increases in these large
moves. (Note that if we assume returns are normally distributed, then a two-standard-deviation event will occur approximately 23 out of 1,000 observations,
a three-standard-deviation event approximately 1.4 times out of 1,000 observations, and a four-standard-deviation approximately three times out of 100,000.)
Additionally, there have been at least seven four-standard-deviation events in 2015, several of which have been driven by central
bank policy changes, including the Swiss National Bank’s recent removal of its currency peg, easing of monetary policy in Canada and Russia and
larger-than-expected asset purchases by the European Central Bank. Changes in the yield curves in Japan and Europe in response to policy changes have also
played a part in these events.
Today, market-makers and risk takers are well versed in fat-tail analysis and stress-test methodologies to manage the risk of multiple-standard-deviation
events. Is the world better off? Yes and No.
Initial conditions have changed so much and central banks have become so much less predictable since the financial crisis that market participants should
have much less confidence in estimating the size or direction of market moves.
As discussed earlier, structural changes in initial conditions have led to higher probabilities of multiple-standard-deviation events that are larger in
size and have different relationships to each other than in the past. This is why PIMCO believes that a fat-tailed event to either the left or the right is as likely to occur as a non-event is – we now have flat distributions and correlations that are much less stable. As illustrated in Figure 2, the probability of small market moves that would otherwise be expected when considering a normal distributed variable,
shown by the green line, is reduced and instead larger moves may occur with higher probability, shown by the purple line.
In our view, portfolio construction should be particularly mindful of hidden (or assumed) correlations and regime transitions, and traditional normal
distribution assumptions for pricing risk are no longer fully valid. We believe active management that recognizes and adapts to such fat tails and changing
correlations will be essential to seeking strong returns on capital – as well as the return of capital itself – going forward.