A Policy Shock Can Reprice the Tail Before Volatility Moves

This chapter uses a historical Federal Reserve policy episode to test the measurement distinction developed so far. The focus is what the BIS study actually observed when it reported a fall in tail-risk perceptions: changes in the market price of downside insurance, changes in broad volatility, or changes in the physical probability of a crash. Separating those outcomes lets the reader understand why policy phase, option measure, and timing matter before drawing any causal conclusion.

The case covers unconventional Federal Reserve policies from the financial-crisis and early post-crisis period. It is useful not because it settles how all central-bank actions work, but because it places several risk measures beside one another around the same events. The result is a concrete example of the title’s logic: the price of a particular downside state can move before a symmetric volatility gauge shows an equally clear response.

Announcements changed the price of downside protection first

The BIS researchers study 17 Federal Reserve announcements associated with QE1, QE2, the Maturity Extension Program, and the combined QE3/MEP phase. These announcements included asset-purchase decisions, maturity changes, and forward guidance about short-term interest rates. The study compares risk measures before and after the events using several windows and supplements the simple comparisons with event-study regressions and a stationary time-series bootstrap.

The event-study design is deliberately local in time. The researchers compare average readings in windows such as five or ten trading days before and after an announcement, and also display the path over 22 days on each side. The simple before-and-after comparison is useful for seeing whether the market measure moved in the expected direction, but 17 events are a small sample for conventional large-sample inference. The stationary bootstrap preserves dependence in the time series, uses a 30-day average block length, and generates 5,000 replications. The resulting p-values ask how often similarly large declines would appear if the announcements had no relation to the risk measures.

The regression exercise adds controls for contemporaneous S&P 500 returns, lagged U.S. economic conditions, and the lagged dependent variable. It therefore asks whether the announcement indicators explain an incremental movement beyond some other observed drivers. This is stronger than simply pointing to a chart, but it does not remove all confounding information. A single announcement can reveal the central bank’s assessment of the economy, alter expected rates, change the supply of bonds, and signal a commitment at the same time. The study estimates what moved around the event; it cannot assign every part of that movement to one policy instrument.

Their main downside measure is the risk reversal constructed from out-of-the-money S&P 500 options. A 25-delta risk reversal compares an out-of-the-money call and put at one degree of moneyness; a 10-delta risk reversal reaches farther into the tail. Since puts are usually more expensive in a negatively skewed equity market, the absolute value of the reversal approximates the market cost of insuring against a sharp fall. It is a price of protection, not a count of how many crashes investors expect.

That distinction determines how the results should be read. When the BIS paper says that perceived tail risk declined, it means that the option-implied price of downside skew and crash insurance declined. The evidence does not, by itself, show that the physical probability of a crash fell by the same percentage.