Implied volatilities in a number of asset markets have been recently trading close to their lowest levels since 2007 despite prevailing uncertainties about the durability of US economic growth (if weather was to blame for the weak Q1 data), the path of Fed policy (the timing and speed of rate hikes), China growth (whether the continued need to rebalance trumps short-term growth measures), broad asset market valuations (if extraordinarily loose money depressed the market price of risk) and geopolitical risks (e.g., Russia-Ukraine, Syria) to name a few.
Market prices should and do reflect these uncertainties on average; however, investors may wonder if the current level of implied is too low given its medium-term determinants. We find that although market vol may be depressed currently, it is not exceedingly so – and most of the deterioration can be explained by the decline in macro and policy uncertainty.
Over the medium term, asset market volatility (proxied by the VIX) is related to broader macroeconomic and policy uncertainties – higher uncertainty raises the price for insuring against such risks, leading to higher implied vols (Figures 1 and 2).
However, recently there has been a divergence between these two drivers– macroeconomic uncertainty (proxied by the dispersion in economists’ forecasts 2Q ahead in the Fed’s Survey of Professional Forecasters) has remained subdued, whereas policy uncertainty (a measure developed by Baker, Bloom and Davis (2012) that includes references to policy-related uncertainty in the press, the number and revenue effect of expiring tax code provisions and disagreement among economic forecasters about government purchases and inflation) has remained elevated.
A medium-term model for vol
Macroeconomic uncertainty tends to be the primary driver of market volatilities (Figure 3). An increased dispersion of 1% among economic forecasters tends to boost market volatilities by 10.5 vols and this effect is statistically significant at the 99% level. By itself, policy uncertainty has a statistically significant effect (at the 90% level) on vols; however, it is no longer significant in models incorporating both macroeconomic and policy uncertainty.
That being said, policy uncertainty does seem to affect the level of vol, but in an asymmetric fashion. The argument goes that policy uncertainty matters more during periods of high uncertainty when the need for a policy offset is greatest (Pastor and Veronesi (2011)).
Indeed, we find some evidence of this in the data – the interaction of policy uncertainty with a dummy for high uncertainty (VIX above its long-term average in the previous period) is found to have a positive and statistically significant effect on the level of vol.
Vol levels are low but not at extremes
The medium-term relationships do not seem to imply a particularly depressed current level of vol (Figure 4). We use the model that includes the asymmetric response to policy uncertainty as our benchmark (Figure 3). Whereas vol appears to have been systematically underpriced for most of the period during 2004-06 – about 3.5 vols on average – there does not appear to be a similar disconnect this time around (average vol levels have been below the model implied levels by 1.2 vols over the past four quarters).
As always, a caveat is in order. Given the asymmetry of the relationship of vol to policy uncertainty and its persistently high levels, any increase in market uncertainty could be very sharp and would not need to be accompanied by an increase in macro uncertainty. This dynamic may explain the elevated levels of steepness in volatility curves, e.g., in FX (Figure 5) and especially those in G10.