The Great Disconnect: 5 Reasons Why Volatility Is Detached From ‘Chaotic Uncertainty’

Ok, listen: we live an uncertain world. And part of that uncertainty is directly attributable to the increasingly cartoonish cast of the characters upon whose dispositions our fate depends. We gave you a brief overview of what we’ve called “The Austin Powers World” earlier on Sunday. In short, the geopolitical scene is a fusion of a James Bond film and an Austin Powers sequel. There’s real danger but at the same time, the central figures are so laughable that it’s difficult to accept it as reality.

In addition to this, or maybe because of this, it’s nearly impossible for anyone to figure out what everyone else is going to do. In the piece linked above, we quoted Immanuel Wallerstein who, in a recent commentary, described the current state of the world as “chaotic uncertainty.” To wit:

Are you confused about what is going on in the world? So am I. So is everyone. This is the underlying and continuing reality of a chaotic world-system.

What we mean by chaos is a situation in which there are constant wild swings in the priorities of all the actors. One day, from the point of view of a given actor, things seem to be going in a way favorable to that actor. The next day the outlook looks very unfavorable.

Furthermore, there seems to be no way in which we can predict what position given actors will take on the next day. We are repeatedly surprised when actors behave in ways that we thought impossible, or at the very least unlikely. But the actors are simply trying to maximize their advantage by changing their stance on an important issue and thereby changing the alliances they will make in order to achieve that advantage.

The world-system has not always been in chaos. Quite the contrary! The modern world-system, like any system, has its rules of operation. These rules enable both outsiders and participants to assess the likely behavior of different actors. We think of this adherence to the rules of behavior as the “normal” operation of the system.

It is only when the system reaches a point in which it cannot return to a (moving) equilibrium that renews its normal operations that it enters into a structural crisis. A central feature of such a structural crisis is chaotic uncertainty.

Needless to say, this manifests itself not only in uncertainty about who will or won’t launch a preemptive strike on the Korean peninsula, or who will or won’t move aggressively to curb Iraqi Kurds’ push for independence, or who will or won’t blink in the still simmering South China Sea standoff (backburnered for the time being by the necessity of dealing with Pyongyang), etc., but also in uncertainty as it regards fiscal policy and other domestic concerns that affect economic outcomes. To the extent there’s uncertainty around domestic policy making, that ends up magnifying uncertainty around things that have no connection to any of the above other than the fact that responding to them is the job of policymakers – things like natural disasters.

Here’s a short list compiled by Deutsche Bank of sources of uncertainty for US investors:

Currently, all of this uncertainty isn’t showing up in market-based measures of volatility and that disconnect has been the subject of vociferous debate this year. In fact, as Deutsche Bank writes, that disconnect (between market vol and various survey- and news-based measures of uncertainty) was “the subject of a roundtable discussion of researchers at the Federal Reserve Board last week.”

As the bank notes, “there is an intuitive feeling that EPU and measures of market implied volatility should be reasonably closely linked.” After all, “shouldn’t greater uncertainty about key economic policy issues lead to a larger dispersion of likely potential outcomes for companies?”

Well, not necessarily. For one thing, market-based measures like the VIX measure are conceptually different from news-based indicators. Here’s Deutsche:

There are a number of reasons why the two measures should not be expected to move in lock-step. First, they measure two different concepts — as Steve Davis noted, the VIX is about the arrival rate of news in a pre-determined timeframe that could affect expectations for stock returns, while EPU is a timeless measure of uncertainty about economic policy, which may or may not directly impact large corporations. Second, the time horizon on the resolution of economic policy uncertainty is likely longer than the typical 30-day VIX measure. This can be seen by the fact that longer-horizon VIX measures are more elevated. Third, the VIX is tied to the outlook for a subset of companies in the US that are not necessarily representative, for example they are larger and more internationally-exposed than the average US company. Thus, we should not be surprised if the level of uncertainty as measured by newspaper articles that are read by the average household differs greatly from the outlook for large corporates.

But that barely scratches the surface. Here’s Deutsche’s effort to list five reasons that explain the disconnect between the EPU and implied vol.:

1. Implied volatility is closely linked with realized volatility and the latter has been near historically low levels (Figure 16).
2. High implied volatility tends to be associated with economic downturns. Solid and steady macroeconomic momentum and corporate earnings in recent quarters is thus a key factor keeping vol low.
3. Monetary policy transparency is at historical highs, limiting the degree of surprise from central bank decisions. This is especially true for the Fed. Moreover, expanded central bank balance sheets have flooded markets with liquidity, helping to damp volatility and compress risk premiums.
4. More speculatively from our side, elevated levels of EPU have been driven significantly by issues that point to upside risks to growth, namely tax cuts / reform, de-regulation, and increases in defense and infrastructure spending. Given the inverse link between macroeconomic momentum and vol, we may expect that uncertainty which skews risks towards stronger growth actually may depress vol.
5. Relatively unaddressed in the round table was the influence of any changes in market behavior and market structure on measures of implied volatility. Our equity derivatives team has noted that in recent years any short-term volatility spike has been viewed as an opportunity to sell vol, and assets under management in inverse (short vol) VIX exchange traded products has soared. This has reinforced the low volatility environment (see August below). 
   

Ok, so a couple of things there. You’ll note that a lot of what’s in that list seems to suggest that the current environment is characterized by a number of factors that are keeping vol. artificially suppressed.

There’s transparency from the Fed, which as Deutsche’s Kocic recently noted, makes it impossible for vol. to sustain a bid by trapping markets in a communication loop that keeps everyone from taking a long-term view. And then there’s the proliferation of VIX ETPs and other modern market innovations that are exacerbating the feedback loops created when central banks become vol. sellers themselves.

As to the point about how policy uncertainty in the U.S. seems to be centered around an agenda that, if it ever materializes, would be growth-friendly, what can we say but “you better hope you’re right if you’re selling vol. on that assumption.”

Of course all of the above is rather superficial. The bank’s own Aleksandar Kocic took a deep dive into the disconnect earlier this year. Below, we excerpt that piece for anyone who might have missed it…

Via Deutsche Bank

Can market complacency be quantified?

Complacency is a conditional category. In some sense, it reflects the absence of risk aversion, although not only that — it is almost synonymous to taking an unhealthy amount of risk. But unhealthy relative to what? In the same way risk aversion requires comparison of two independent measures of risk, complacency is defined relative to a perceived level of (unrealized) risk.

Consider a coin toss as an example. The physical probability between heads and tails is 50/50. However, any bets that involve coin toss as a decision instrument will always overweight probabilities of unfavorable outcomes. For example, if the payout of the gamble is $10 for the heads and $0 for the tails, no one would pay $5 (the expected value = 50% * $10 + 50%* $0) to play, but less than that, say $4. So, the actual price is evaluated as an expected value with 60/40 probability assignment (40%*$10 +60%*$0 = $4). This is the essence of risk aversion. Because the actual risk involved in a coin toss can be accurately determined, reweighting of probabilities which incorporate risk preferences can be measured precisely and quantified. The price of risk aversion is $1, the risk premium that is the difference between the expected price of $5 and the $4 at which it is traded.

We approach the problem of quantifying complacency in the same way as we did for the risk premium in the coin toss example by comparing the two different measures of economic uncertainties: Economic policy uncertainty index and VIX (implied S&P volatility).

Economic Policy Uncertainty index (EPU) was proposed last year by Baker, Bloom, and Davis (Quarterly Journal of Economics, 1593, (2016) Vol. 131) and has been since implemented both as global and the US in various forms (Bloomberg ticker: EPUCGLCP Index). In our analysis, we choose the global EPU; use of other indices lead to the same qualitative conclusions.

The index is constructed by counting the frequency of articles in ten leading US newspapers that contain three of the target terms: economy, uncertainty; and one or more of Congress, deficit, Federal Reserve, legislation, regulation or White House. These numbers are properly normalized by their means and standard deviations of occurrence and combined into an aggregate index (see p. 1599 of the article for details). As such, EPU is completely market independent (in the same way the mechanics of a coin toss is relative to any particular gamble).

As an alternative measure of risk, one that reflects the market prices, we choose equity volatility. Similar, although noisier, results follow for almost any other market implied volatility (e.g. rates or FX). The history of the two measures of uncertainty is shown in the Figure.

Until 2012, for the most part (but not always) both levels and spikes tend to be coordinated across two measures. When VIX is in tune with EPU, the market is acknowledging the levels of risk through the prices. When VIX is low and EPU high, markets are complacent — they are underpricing risk. Spikes across different events are summarized in the Table (chronologically, from left to right). After 2011, the two measures of risk decouple with VIX consistently low despite growing uncertainty. The breakdown is structural, and it is visible across all market sectors, not only equities.

In order to quantify market complacency, we compare the two measures in the following way. We regress EPU onto VIX until 2011 und treat the residuals as the measure of complacency. The two Figures show the EPU index overlaid with the regression scaled VIX and the residuals.

In this measure, it appears that the markets have made a structural shift towards higher levels of complacency in the last six years. Current levels of complacency are alarming. This is what everyone is talking about. Despite growing uncertainties and tensions, the market volatility refuses to rise. Persistence of low volatility is increasing the penalty for potential dissent and reinforces one sided positioning. As a consequence, the risk of disorderly unwind is growing. And the longer this regime continues, the lower the threshold of painful unwind. Currently, VIX at 15% is perceived as a problem although before the crises it had traded above 20% most of the time. Similar observations hold for rates gamma, currently around 60bp, compared to pre-crisis averages around 100bp. It appears as if the markets lost their capacity to deal with uncertainty