Monday, September 7, 2015

When Averages Lie: VIX and CAPE

The historical average for VIX is 19. When the market is calm, "normal", VIX is usually much lower, like 12-15. But since the historical average is 19, people expect VIX to return to the average level - you can easily see it on VIX futures term structure.

Ok, VIX will someday spike and return to the average, but the question is how long you have to wait for this. From my research, there's no point to expect mean-reverting behaviour from VIX. You best assumption about close-future VIX is that it will stay the same as it is now. If it mean-reverts, it happens on much longer time-frame than the term structure implies.

Calm, "normal" markets show low VIX, but sometimes things happen and VIX spikes, that is why the average VIX is significantly higher than "market-normal" levels. But from the fact that something bad can happen, you cannot draw conclusions about the closest VIX behaviour. Yes, something bad will surely happen someday in future, but you can't know when. Most probably not tomorrow. Not anytime soon.

So when you truly believe in VIX mean-reverting behaviour, you will probably end up doing expensive things too early.

I strongly suspect that the same thing applies to CAPE (Shiller's P/E). When conditions are somewhat to be called "normal", maybe that's ok for markets to have CAPE above 20. And only when markets are stressed when something bad happens with the economy, only then CAPE goes significantly lower, which makes historical average float around 15.

Let's assume we have a "normal" market today, with CAPE higher than average, should we worry about high CAPE? Can we draw conclusions from the fact, that in US stock market history there were events that draw CAPE much lower than a "normal" level? Should we expect the stock market to go down soon just because something bad can happen someday?

If we consider CAPE case to be something like VIX case, the answer is no. Your best bet about the normal market is that it will continue to be normal. Yes, normal markets end someday, when some really nasty problems come out. But it can be very expensive to expect it to happen tomorrow. Or anytime soon. 


  1. Since the historical averages aren't helpful, perhaps looking at the median would be. Do you have any data on the median?

    1. Why do you think the median will make any difference here?