Saturday, November 18, 2017

The New York Fed Is Puzzled By Low Volatility

From the Federal Reserve Bank of New York's Liberty Street Economics blog, Nov. 15:

The Low Volatility Puzzle: Is This Time Different?
As stock market volatility hovers near all-time lows, some analysts are questioning whether investors are complacent, drawing an analogy to the lead-up to the financial crisis. But, is this time different? We follow up on our previous post by investigating the persistence of low volatility periods. Historically, realized stock market volatility is persistent and mean-reverting: low volatility today predicts slightly higher, but still low, volatility one month and one year from now. Moreover, as of mid-September, the market is pricing implied volatility of 19 percent in one to two years’ time. This level contrasts with the pre-crisis period when the term structure of implied volatility was relatively flat, which suggests this time may indeed be different, at least as measured by market participants’ pricing of risk.

Realized Volatility Forecasts When Volatility Is Low
When realized volatility is low, does it tend to stay low the next month? What about twelve-months ahead? The chart below answers this question by plotting current realized volatility on the horizontal axis against realized volatility one-month and twelve-months ahead on the vertical axis. Realized volatility is computed as the sum of squared daily Center for Research in Security Prices (CRSP) value-weighted returns obtained from Kenneth R. French’s website, reported in annualized volatility units.

The chart shows that realized volatility is persistent and mean-reverting. To see this, note how low volatility today forecasts low volatility one-month and twelve-months ahead. The volatility forecasts are above the 45 degree line when volatility is low but below the 45 degree line when volatility is high. Moreover, there is no evidence of a nonlinearity when volatility is at the lower end of its historical distribution. On average, extremely low volatility today predicts low volatility in the future, not higher. This evidence weighs against the narrative discussed in the previous post that low volatility, in and of itself, may be a concern.
Nonetheless, it may be that this analysis masks an increased probability of a jump in volatility when volatility is extremely low. To consider that possibility, the next chart shows the probability of moving into a high volatility state, defined as realized volatility above 16 percent (the seventy-fifth percentile of the historical distribution), based on the current level of realized volatility. Similar to the volatility forecasts, we find no evidence that being in a low volatility environment raises the probability of jumping to a high volatility state, as compared to a “normal” volatility environment of, say, 15 percent.

The Low Volatility Puzzle: Is This Time Different?

The Term Structure of Implied Volatility...