Low Volatility as a Magus of Financial Crises
The magus is an ancient sorcerer or magician, and the title implies that low volatility could posses the magic to predict a financial crisis. With aftershocks of the 2007-2008 global financial earthquake still being felt in some parts of the world, it would be useful to have a set of “early warning indicators” to tell us what countries are most vulnerable. Indeed, the Global Financial Crisis in 2007-2008 bushwacked every observer and market participant alike. Markets had come to believe that crises were part of history and that the regulations and institutional structure in place were shielding the markets. Alas, the crisis proved all of this wrong. Since then, passionisti like pundits and me alike have been trying to search for techniques to foretell that a crisis might be underway so the prudential authorities and private institutions can prepare themselves ahead of time.
Note: Obviously, it would be better if the financial institutions and policymakers would create a system whereby crises would occur less frequently and with less severity .. but that is a different story altogether.
One of the critical tools for which I always had a weak spot is Early Warning Indicators (EWI) of crises, which earned some prestige in the wake of the Asian crisis of 1998. Those tools that were in place in 2007-2008 did not work well (cf. even though a variety of indicators had been suggested). At the risk of oversimplification, I could label these EWIs in two main groups. First is a group of EWIs that are exclusively data-driven, and based on market data indicators, such as bond yields, stock prices, CDS spreads, commodity prices, macroeconomic indicators, and volatility. The second group utilizes economic theory to shape the indicator design, looking more at heart at how risk surfaces in the first place.
The Basel Committee indicates that EWIs are an essential component for the implementation of time-varying macroprudential policies, such as countercyclical capital buffers, that can help reduce the high losses associated with banking crises. EWIs in this context must not only have sound statistical forecasting power, but also need to satisfy several additional requirements. For instance, signals need to arrive early enough, so that policy measures have enough time to be effective, and they need to be stable as policy makers tend to react on trends.
Economic theory is utterly clear on where we should look for symptoms of crisis. Both Keynes (1936) and Hayek (1960) indicate that perceptions of risk – mainly when they differ from what economic agents have come to anticipate – affect risk-taking behavior. Maybe the best expression of this view is Minsky's (1977) instability assumption, where economic agents observing low financial risk are encouraged to increase their risk appetite, which in turn may result in a crisis – the premise of his famous axiom that ‘stability is in essence destabilizing.’
The recent writings achieve similar conclusion, for instance, Brunnermeier and Sannikov's (2014) ‘volatility paradox’ and Bhattacharya’s et al. (2015) study of Minsky's assumption. Similarly, Danielsson et al. (2012) consider a general equilibrium architecture with risk constraints, whereupon observing low volatility, agents are endogenously incentivized to take more risk. The fundamental mechanism for the relationship of risk to a crisis can be noted in the following graph.
Graph 1: The correlation between financial crises and volatility
At inception, economic agents observe that volatility is lower than what they have come to anticipate. This gives way to optimism, making the agents more bullish to take more risk. This is manifested within the financial services sector, which will both lend more than it did before and also underwrite riskier loans. Eventually, loan defaults increase, a banking crisis ensues, at which time volatility goes up discernibly.
This chain of events suggests that a considerable time takes place between observation of a period of low volatility and the timing of a crisis. By contrast, a representation of high volatility directly erupts from a crisis, implying that by the time volatility peaks, it is already too late to react.
Most empirical literature on EWI takes the contrarian view that the probability of crises increases along with quantified risk, typically measured by ex-post findings in financial markets, price movements or volatility, CDS spreads, VIX, volumes, and other activity measures (cf. good illustrations are both the ECB’s systemic risk composite indicator and the SRISK (Acharya et al. 2012)).
Graph 2: Systemic Risk on a global scale
Note: SRISK measures the systemic risk contribution of a financial firm. SRISK measures the capital shortfall of a firm conditional on a severe market decline, and is a function of its size, leverage and risk. (Source: IDEAS)
Note: VIX is the annualized 30-day S&P 500 volatility that justifies option prices, so it can be thought of as the market’s prediction of volatility. As you can see, the prediction is quite accurate, except that the VIX is consistently high by about four percent. That’s not a flaw in the VIX, or not mainly a flaw. The VIX is based on the amount investors are willing to pay for protection against all market moves, which is rationally greater than what you would extrapolate from average market moves. (Source: Bloomberg)
Graph 3: the VIX
The academic writing suggests that such indicators are most likely to be of limited use. They indicate a high risk once a crisis is underway, but do not provide much regarding ex-ante prediction, limiting their effectiveness.
A counter-argument might be that a specific indicator provided a strong signal for the events of 2007-2008. However, that is one observation with multiple degrees of freedom for optimization, calculated ex-post. It is plain sailing to make use of the benefit of hindsight to come up with a measure that forecasts a single known crisis event in history. However, such signals are unlikely to capture the proliferation of systemic risk that results in future crises.
The theoretical literature has inspired a couple of recent studies in the setting of EWIs. The most encouraging ones are those that connect credit expansions to crises, as Schularick and Taylor (2009), or Baron and Xiong (2016), where over-optimism leads to increases in the probability of a financial crisis.
In a recent paper, one step further was taken and formally investigated the mechanism in Figure 1 above (Danielsson et al. 2018). They take a historical perspective, utilizing data going back to 1800 were deemed available, and sixty countries, examining the probability of crises as a function of volatility. As a starter, they study the relationship between current volatility and the probability of future crisis – what could be labeled as the linear volatility crisis view, as seen in Graph 4. This is rebuffed by the data sets. When linked with macroeconomic variables to capture the state of the economy, volatility itself is not a crucial magus of crises.
Graph 4: Volatility as a predictor of crises
Nonetheless, by decomposing volatility into a trend, and deviations from trend, a disparate picture surfaces. The underlying reason is that economic agents look at the pattern as ‘usual/expected’ volatility, and when it deviates from the trend, risk-taking behavior is impacted. That leaves two kinds of volatility: low volatility when it is below trend; and high volatility when it is above the pattern. The link between low and high volatilities and the probability of crises can be noted in Graph 5.
Graph 5: Low and high volatilities as forecasters of crises
The slope of the curve, which converts into regression coefficients in a statistical model, shows the strength of the relationship. The higher the slope in absolute numbers, the more likely a crisis will take place as volatility moves away from its trend. In Danielsson et al. (2018), they find substantial proof for low volatility but only weak evidence for high volatility resulting in crises. The macroeconomic impact is the highest when the economy stays in the low volatility environment for five years: a 1% decrease in volatility below its trend corresponds to a 1.01% increase in the likelihood of a crisis. We also indicate that the chain of causality as explained in relation to Graph 1 above, with low volatility leading to increased lending and excessive risk, holds empirically.
These results suggest that volatility would not be a good crisis indicator. It might tell policymakers that a crisis is underway, but we reckon that they will realize that anyway. A traditional method to evaluate the quality of EWIs is the area under the receiving operating curve (AUROC). An AUROC value of 0.5 indicates that an EWI is no better than random noise, while a number of 1.0 implies stellar predictability. Low volatility has an AUROC value of 76%, with a 95% confidence interval of [72%, 80%]. This means that it yields a reliable signal to policymakers for the incidence of an upcoming crisis.
As a conclusion, within the post-GFC macroprudential agenda, policymakers are actively looking for indicators of future financial and economic instability and developing policy instruments to mitigate the most unfortunate results. Most early warning indicators of crises are based on ex-post market data, typically volatility in some form or another, where increasing volatility highlights that the financial system has become riskier, and a crisis is, therefore, more probable. However, a finding of high risk is the result of a stress event where financial instructions are in grave danger, not the cause. It will be more fruitful to come up with EWIs amongst the fundamental drivers of financial instability. A significant cause of financial crises is excessive risk-taking by economic agents. When they perceive a low-risk environment, they are endogenously incentivized to take more risk, which ultimately culminates in a financial/economic emergency.
This view has been voiced by policymakers. "Volatility in markets is at low levels...to the extent that low levels of volatility may induce risk-taking behavior...is a concern to the Committee and me."
Source: Former Federal Reserve Former Chair Janet Yellen. 2014
The data confirms this view. Observation of low risk is a significant crises predictor and suitable for an early warning indicator. The policy authorities and private institutions would, therefore, benefit from utilizing low volatility as a crisis signal since an observation of current low volatility means that a future crisis is more probable.