Examining the tangled relationships between VIX, VIX futures, and VXX
The more I think about this CBOE Volatility Index (VIX) versus iPath S&P 500 VIX Short-Term Futures ETN (VXX) discussion, the sillier it seems. If there's a steep slope between the nearest two VIX futures, it puts pressure on VXX, thanks to the fact it has to roll out in time every day to maintain the constant 30 days' duration. Everyone concurs on that point, as far as I know. The debate is whether the level of VIX itself versus the futures impacts the path of VXX. My opinion is that it does, as I laid out yesterday.
In the real world, though, these are not mutually exclusive. If VIX is at a big discount to VIX futures, it almost certainly implies that the front two months are in steeper-than-normal contango. The steepness of the curve will, of course, vary. If you want to credit/blame only the front two futures for the action in VXX, you are mathematically correct. But it seems remiss to neglect to mention that VIX itself is likely the main driver to the futures themselves. How much of a driver?
Going back to the beginning of actual VXX trading on Jan. 30, 2009, the daily percentage change in VIX has a 0.8843 correlation to the percentage move in VXX. Thus, we can explain 78% of the move in VXX on a given day via the move in VIX.
That's pretty strong, considering there are factors that go into options pricing that won't impact VXX. An imminent holiday comes to mind … it should exert greater downward pressure on VIX than VXX. A day of the week influences VIX more than VXX, as well. Not to mention, of course, our friend contango. If VIX does literally nothing in a given day, but VIX futures sit in contango, we'd expect modest downwards pressure on VXX.
This speaks nothing of whether VIX trading at a discount to VIX futures weighs on VXX, but it's not a real leap to say that the level of VIX to VIX futures affects the relationship of the futures to both VIX and to each other -- and that all impacts VXX.
As to whether you can then say "well, 'X' affects VXX, too," that's very true -- provided 'X' is SPDR S&P 500 ETF Trust (SPY) (or the S&P 500 Index (SPX), obviously). The daily change in VIX has had a -0.76 correlation to the daily change in SPY since Jan. 30, 2009. The daily change in VXX actually has a higher correlation to the SPY move (-0.8). If you think about it, though, it makes some sense. VIX is both more anticipatory to moves than VXX, and it has the calendar quirks.
It's also important to note that despite that, SPY independently "explains" 64% of the VXX move, versus the 78% relationship of VIX and VXX. Does any of this prove anything? No. There will be times when VIX futures and VIX move somewhat independently of one another. This just suggests that's not all that common.
Disclaimer: Mr. Warner's opinions expressed above do not necessarily represent the views of Schaeffer's Investment Research.