Uncertainty shocks are changes in beliefs about probabilities. They are perhaps the most powerful driver of financial markets. Uncertainty comes in various forms, such as macro uncertainty, firm-specific uncertainty and uncertainty about others’ beliefs. However, empirical and theoretical research suggests that different types of relevant uncertainty shocks have one common dominant origin: updated beliefs about disaster risk. Hence, when markets give greater probability of downside tail risks, all sorts of uncertainty would rise, with a profound impact on macro trading strategies, whether they are directional or based on relative value.

Kozeniauskas, Nicholas, Anna Orlik and Laura Veldkamp (2016), “The Common Origin of Uncertainty Shocks”.

The post ties in with this sites lecture on setback risk. In particular, disaster risk re-assessment seems to be a good qualitative indicator of high exit risk from dominant current positioning.

The below are excerpts from the paper. Headings and some other cursive text has been added for context and convenience of reading.

The nature and types of uncertainty shocks

Uncertainty in economic theory usually means that agents believe that the value of a relevant parameter is a random drawing from a specific probability distribution. An uncertainty shock is a change to that belief in form of a re-assessment of the probability distribution. This can affect any of its moments, such as mean, standard deviation, skewness (bias towards upside or downside) and kurtosis (probability of extreme events).

“A recent literature…demonstrates that uncertainty shocks can explain business cycles, financial crises and asset price fluctuations with great success.”

“The nature of uncertainty shocks varies…

  • Macro uncertainty…means that an aggregate variable, such as GDP, becomes less predictable… Conceptually we think of macro uncertainty…as the standard deviation of an agents’ beliefs about GDP growth…a proxy for macro uncertainty…is based on…implied volatility of the S&P100 from options prices.
  • Micro uncertainty…describes an increase in the variance of idiosyncratic shocks to firms, which manifests itself in an increase in the cross-sectional difference in firm outcomes…We measure micro uncertainty with the cross-sectional interquartile range of firm sales growth.
  • Higher-order uncertainty describes the uncertainty about others’ beliefs that arises when forecasts differ… We measure higher-order uncertainty with the cross-sectional standard deviation of real GDP growth forecasts from the Survey of Professional Forecasters.”

“Various measures of uncertainty are countercyclical [i.e. uncertainty tends to negatively correlated with economic growth].”

The common origin of uncertainty shocks

“While macro uncertainty comes from aggregate shocks, micro uncertainty depends on firm-specific shocks, and higher-order uncertainty arises from private signal noise…These shocks should be independent. But they are not: these different types of uncertainty strongly covary, so much so that they are sometimes conflated. The fact that all three are not obviously related and yet covary strongly suggests that they may have a common cause.”

“Risk and uncertainty are moments based on beliefs…One reason why the precision of prior beliefs varies over time, which is an integral part of the learning mechanism, is that agents do not know the parameters [of the economy]…and must re-estimate them each period.…When agents use…real-time data to re-estimate parameters that govern the probability of disasters, the result is that micro, macro and higher-order uncertainty fluctuate and covary…The many types of uncertainty shocks are not distinct phenomena. They are outcomes of a quantitatively plausible belief updating process.”

“When weak macro outcome make agents re-assess their beliefs about the skewness of the shocks they face, uncertainty of all types move in a correlated, volatile and counter-cyclical way. The skewness of the distribution is a crucial part of the story.”

“The reason that disaster risk can affect all three uncertainty measures originates from the fact that disasters arise infrequently, making their probability difficult to assess and the scope for disagreement large. When the risk of disasters rises, these errors and disagreements are amplified. Less accurate forecasts raise macro uncertainty. Divergent forecasts create higher-order uncertainty. Firms with divergent forecasts choose different inputs and obtain different outputs. This rise in the dispersion of firm output shows up as higher micro uncertainty. All three forms of uncertainty and their covariance can be explained in a unified framework that ties all three to disaster risk. This unification brings us one step closer to understanding what causes business cycle fluctuations.”

The model analysis

“To explain why re-assessing tail risk can cause many forms of uncertainty to co-move [our paper] presents a production economy [model] with firms that are uncertain about aggregate productivity.”

“Each period our agents observe past data, re-estimate the parameters of this non-normal model and then receive private signals to update their beliefs about total factor productivity growth. On the basis of their beliefs firms choose their labor input for production and forecast GDP growth. Since total factor productivity growth has stochastic volatility and agents must re-estimate the model each period, the precision of agents’ prior beliefs, and macro uncertainty, varies over time…the combination of learning and disaster risk can explain why micro, macro and higher-order uncertainty fluctuate together.”

“Because disaster probabilities are difficult to assess, a rise in disaster risk creates both uncertainty about aggregate outcomes (macro uncertainty) and disagreement. When firms disagree, they make different input choices and have heterogeneous outcomes (micro uncertainty). They also make different forecasts (higher-order uncertainty). Thus, it is that fact that disasters are rare and difficult to predict that drives all three forms of economic uncertainty.”

“Because tail probabilities are so sensitive to changes in skewness, small differences in skewness estimates produce large disagreement about disaster risk…Skewness…causes higher-order, macro and micro uncertainty to fluctuate more, covary more negatively with the business cycle, and covary more positively with each other. Thus, the message is that, while types of uncertainty are not obviously related, real-time estimation of disaster risk can produce fluctuations in the many types of uncertainty, in a way that explains the uncertainty data.”