Proposed VIX ETNS Are Not a Volatility Bet

My original post on those VIX ETNs was republished over at Seeking Alpha. If you read that post and the two subsequent posts on the donfishback.com site, you know that I am not a fan of these products for a variety of reasons — the main one being CONFUSION.

A commenter over at Seeking Alpha asked a series of questions. I don’t know if they were rhetorical or not. But his first question really stuck out. He asked, “What do you suggest as a replacement for a volatility play?” I rest my case. His question is proof that people are confused about what these instruments really measure. So to clear things up, I’m going to drop this inflammatory bomb.

Because of the lack of arbitrage between the VIX index and its corresponding futures, VIX futures (and the proposed ETNs) are not a bet on volatility.

Here’s what they really are: they’re a bet on the market’s expectation of what volatility will be in the future. Actually, they’re not even that. They’re a bet on the market’s current expectation of what the S&P 500 Index option implied volatility will be on one specific day in the future — at the contract’s expiration!

For the vast majority of the time, these two concepts — volatility versus implied volatility expectations — move in unison. The providers of these products tout that the correlations are high, and they are. The actual correlation is about 88%. The key is, it’s not 100%. Now some may say, “88% is pretty high. That’s good enough” But it’s not good enough. The reason is WHEN the correlation breaks down. These products stop working the way you expect them at the precise moment you need them the most!

At the risk of repeating myself, I am republishing the table from my prior post:

Looking at that table, you can see that, had you bought November or December VIX futures on September 10, when VIX and the rest of the 2008 futures contracts were all around 24, you’d have been horrified at how little those futures moved during the next two weeks compared to the VIX. For on October 10, the VIX itself had skyrocketed to 69.95. But the December futures only moved up to 33.79. If you had been relying on those futures (or the ETN that is based on these futures) as a hedge, or if you had bought the ETN as a speculation, you’d wonder why the futures barely budged higher. Well, the future didn’t move because the expectation was that the volatility would die down by December. So the futures remained low, even though actual volatility, and index option implied volatility, skyrocketed.

But it got more bizarre. If, on October 10, you thought VIX had peaked and that it would be lower on December 2, you would have been right. Actual volatility declined quite a bit, and implied volatility declined a little bit from 69.95 to 62.98. But the December futures went up from 33.79 to 57.93 as traders realized that their expectations of volatility dying down were wrong. As a result, the futures had to eventually catch up to the actual implied volatility level of the index options. If you have gone short those futures (or those ETNs) in expectation of volatility dropping, you’d have been right, as implied volatility and actual volatility dropped. But you’d have gotten crushed because the vehicle you used for your “volatility play” wasn’t really a volatility play after all.

My point is, if you don’t know this and understand this, you shouldn’t be trading this product at all because VIX futures, and the proposed ETNs, are not a pure play on volatility. If they were, movements like those described above would be impossible. Instead, they’re a bet on the current expectations for index option implied volatility at some point in the future. For example, on October 10, the expectations for what VIX would be in December were extremely wrong; those expectations were way too low.

And that’s why I don’t like them. It’s not that they’re a flawed product. It’s how they’re being advertised. They’re being touted as being a mainstream way of investing in volatility. And judging by the ETF Expert who commented at Seeking Alpha, that’s what most folks believe.

Well, sometimes, when the correlations are perfect, these products might indeed be a volatility bet. But other times — the most critical times — they’re not.

Comments

[ETFexpert:]

Hey Don: ETF Expert here:
Good work!! I stand corrected.
I looked at your research and did a little of my own. The prices that you referred to in your graph were interesting. I went back and looked at the open interest on the VIX futures, same dates. (See graph)

And to my surprise we saw a huge unwinding in the contracts towards the end of the year. To me, this is interesting. Now to the credit of our friends with IndexUniverse’s article under “planned rotation” Barclay’s is offering two ETNs. One short term and the other a longer term, weighted at one month and five months respectively. Based on your graph do you think the end of year pricing would have been rectified in January? You know as contracts roll forward. Just asking??
Am I understanding your comment here that….. Because of the lack of arbitrage between the VIX index and its corresponding futures, VIX futures (and the proposed ETNs) are not a bet on volatility.
Per the Barclay’s prospectus: “The redemption feature is intended to induce arbitrageurs to counteract any trading of the iPath ETNs at a premium or discount to their indicative value, though there can be no assurance that arbitrageurs will employ the redemption feature in this manner.”

Can you help me here, so what you’re saying is that an ARB will not participant? I thought that ARBs prefer “products” that have a Futures contract behind them. That way they can lay off the risk simultaneously as they close the premium/discount spread down. Am I missing something here? Let’s agree to disagree on this one; my bet is they will “ARB” anything out there that makes them money.

Refer back to my comments about the Merrill Lynch Focus 20---that too was late to the party. At first glance I thought that these ETNs were a “VOL-Play” they may still be, the jury is out. But I agree with you one needs to be very careful when using products like these. People need to think of the wrapper--- a futures product that has no instruments behind it, traded like an equity. If that isn’t around Johnson’s Barn then nothing is. I still say…let’s see what the pros do with them.

Jan 25 06:26 PM

[ETFexpert:]

Darn, the graph that I made didn't print. SHOOT!! Anyway, you were correct. My graph showed a huge sell off in Ocotber...a little more in November and basically flat in December.

Jan 25 06:28 PM

[vix switch ...:]

although December was priced lower than spot vix when it was a back month; as the dec future became a front month it converged to the spot vix price quickly. this is a common occurrence if one was to study past front month convergences.

Jan 25 09:45 PM

[Don Fishback:]

Thanks for the comments.

To vix switch trader, you're sort of right about that convergence. Convergence happens. In fact it's a must. [Well sort of. You should see how they calculate the settlement values.] But to say "December was priced lower than spot when it was a back month" is not correct. Notice in the top row of the table that when VIX was 17.83, the December futures were nearly 5.00 points higher at 22.72.

As I said in prior posts, the futures tend to have a greater central tendency than the VIX index itself. As you get closer to expiration, correlations go up and you eventually get the convergence you spoke of. But not always.

What is unusual is the *size* of the departure from spot that occurred in October. As the table above shows, at the end of June, July, August, and even as recently as September 10, the difference between VIX and its futures were mere pennies. But after Lehman's bankruptcy, the difference in the futures and the "spot" went haywire, which would have caused the ETNs to go haywire compared to what most people would have expected.

And that's the key. When VIX makes an extreme move -- up or down -- the normal relationship between VIX and the futures breaks down. At the precise time you want profit from an extreme -- either via a hedge or a speculative position -- the product behaves differently than it normally does.

Why that happens is the lack of arbitrage. The VIX futures and VIX itself are extremely difficult to arbitrage. As I hear it, only Susquehanna even tries. There is a formula for the VIX futures' fair value. It's actually not too difficult, but it is extraordinarily tedious. At any rate, you can still figure out what the fair value should be and then compare that to the VIX futures. This process is extremely common with S&P 500 index futures. The thing about VIX futures is that, unlike other index futures, the difference between fair value and the price can get extremely large for extended periods of time due to the inability to arbitrage.

And if you want evidence as to the ease/difficulty of the arbitrage play, think of this. If it was easy to replicate VIX itself, why not launch an ETF that mimicked VIX instead of the futures? How popular do you think THAT would be? I know I'd find it tempting. But there are zero plans to launch a product like that, because nobody can create a portfolio of tradeable securities that consistently replicates VIX itself. If they could, they would ... and then they'd easily arbitrage the futures divergence. The implications of this are: The divergences that exist for weeks at a time are proof that arbitrage doesn't exist in any meaningful size.

Now it is true that the ETN's prospectus mentions arbitrage. But it's important to realize they're talking about the arbitrage between the ETN and the futures contracts, not the ETN and the VIX itself.

One final thing, and this is important regarding convergence. Because these products are perpetual, they may never converge. Unlike futures which have an expiration date and therefore *must* converge to spot, these ETNs represent a "rolling" 30-day futures contract, so there is no guarantee of convergence at expiration, because there is no expiration.

As for the other questions you ask, I simply don't have the answer. And I have to say, I don't want to take the time to investigate until these things prove more valuable than I think they'll be.

-- Don

Jan 26 08:57 AM

[vix switch ...:]

i wish i had more time; but suffice to say well written as i agree with your assessments. what i do is to trade the front month and back month spreads when i feel they are mispriced or when i see a market move imminent that could effect the spread. nice post.


On Jan 26 08:57 AM odds wrote:

> Thanks for the comments.
>
> To vix switch trader, you're sort of right about that convergence.
> Convergence happens. In fact it's a must. [Well sort of. You should
> see how they calculate the settlement values.] But to say "December
> was priced lower than spot when it was a back month" is not correct.
> Notice in the top row of the table that when VIX was 17.83, the December
> futures were nearly 5.00 points higher at 22.72.
>
> As I said in prior posts, the futures tend to have a greater central
> tendency than the VIX index itself. As you get closer to expiration,
> correlations go up and you eventually get the convergence you spoke
> of. But not always.
>
> What is unusual is the *size* of the departure from spot that occurred
> in October. As the table above shows, at the end of June, July,
> August, and even as recently as September 10, the difference between
> VIX and its futures were mere pennies. But after Lehman's bankruptcy,
> the difference in the futures and the "spot" went haywire, which
> would have caused the ETNs to go haywire compared to what most people
> would have expected.
>
> And that's the key. When VIX makes an extreme move -- up or down
> -- the normal relationship between VIX and the futures breaks down.
> At the precise time you want profit from an extreme -- either via
> a hedge or a speculative position -- the product behaves differently
> than it normally does.
>
> Why that happens is the lack of arbitrage. The VIX futures and VIX
> itself are extremely difficult to arbitrage. As I hear it, only
> Susquehanna even tries. There is a formula for the VIX futures'
> fair value. It's actually not too difficult, but it is extraordinarily
> tedious. At any rate, you can still figure out what the fair value
> should be and then compare that to the VIX futures. This process
> is extremely common with S&P 500 index futures. The thing about
> VIX futures is that, unlike other index futures, the difference between
> fair value and the price can get extremely large for extended periods
> of time due to the inability to arbitrage.
>
> And if you want evidence as to the ease/difficulty of the arbitrage
> play, think of this. If it was easy to replicate VIX itself, why
> not launch an ETF that mimicked VIX instead of the futures? How
> popular do you think THAT would be? I know I'd find it tempting.
> But there are zero plans to launch a product like that, because nobody
> can create a portfolio of tradeable securities that consistently
> replicates VIX itself. If they could, they would ... and then they'd
> easily arbitrage the futures divergence. The implications of this
> are: The divergences that exist for weeks at a time are proof that
> arbitrage doesn't exist in any meaningful size.
>
> Now it is true that the ETN's prospectus mentions arbitrage. But
> it's important to realize they're talking about the arbitrage between
> the ETN and the futures contracts, not the ETN and the VIX itself.
>
>
> One final thing, and this is important regarding convergence. Because
> these products are perpetual, they may never converge. Unlike futures
> which have an expiration date and therefore *must* converge to spot,
> these ETNs represent a "rolling" 30-day futures contract, so there
> is no guarantee of convergence at expiration, because there is no
> expiration.
>
> As for the other questions you ask, I simply don't have the answer.
> And I have to say, I don't want to take the time to investigate until
> these things prove more valuable than I think they'll be.
>
> -- Don

Jan 26 09:23 PM