You have been misled! The VIX is often referred to as a volatility index, or sometimes even a ‘fear’ index. This implies it is used as a gauge of Mr. Market’s perception of the likelihood of a stock market crash. This is not the case, however.

First, what is the VIX? Not to get too technical, but first we need a little option pricing tutorial. The price of an option, be it a put or call, is a function of five factors: the price of the underlying asset, the exercise price of the option, the risk-free rate of return, the time to expiration, and the volatility of the underlying asset. When pricing an option we can look up values for the first four factors – they are “known” in other words. But the volatility must be estimated. What most option traders do is look at market values of options (the premiums) and calculate the volatility using those premiums and some model of option values, like the Black-Scholes. The volatility, so calculated, is labeled ‘implied volatility.’

The VIX is an average of implied volatility of options on the S&P 500 index futures contract. The options chosen are puts and calls that are close to expiration (within 30 days) and close to the money (with an exercise value close to zero). Using market prices on these options, the implied volatilities are calculated and then averaged to get the VIX. Note, by the way, that there also derivative contracts directly on the VIX itself. But that’s a different story.

Now, I’ve heard and seen several different stories telling us that the VIX is relatively low, and so Mr. Market doesn’t think a crash is in the offing. I ask a simple question: does the VIX actually tell us anything about the probability of a crisis? As prima facie evidence that it does NOT, I offer the following graph:

fredgraph

In the above graph I use the Wilshire 5000, because FRED only had the S&P 500 back to 2004. I don’t know why that’s changed, they used to have it way back, but not so anymore. There doesn’t seem to be much warning coming from the VIX regarding crashes. At most, they appear to be coincident indicators, no lead-lag relationship.

To be more scientific, I used the S&P 500 index to calculate crashes. Basically, a crash was defined as a drop of 20% or more in the value of the S&P 500 in one month. The two major ones in the data I used (1990 – 2014) were, of course, the dot.com crash and the real estate crash. Then, I modeled the probability of a crash using a logit model. I won’t bore you with the econometrics, but basically I checked a few different factors and how they affect the probability of a crash. The factors I used are the P/E ratio, the trailing twelve months’ total return, the bull-bear spread in investor opinion from the American Association of Independent Investors (AAII), and the VIX.

My results? P/E and and returns matter, the bull-bear spread and the VIX do not. The higher the P/E and the higher recent returns have been, the more likely we are to see a crash (makes sense). VIX does not help in predicting a crash. It goes up when a crash occurs, but it happens after the crash begins.

Long story short: the VIX doesn’t help predict crashes!