Understanding the ETF/ETN Decay (Part 1)
Location: Checklist34's CAPS Blog
Its widely realized that many ETFs and ETNs decay. Fees may contribute to this, as may managers just sitting at a desk somewhere and hitting the bid or the ask on various things all day with no concern for the nickel in between. After all, its not their money. But there are two large fundamental causes for decay that outweigh these little things. The first is the effect of contango/backwardation which affects many futures based ETFs and ETNs, primarily commodity and volatility types (USO, UNG, VXX, TVIX, etc.). The second is the compounding/decay effect of daily, weekly, or monthly resetting of leverage on various leveraged ETFs.
For this post, lets take a look at the first. First, lets look at what contango and backwardation really are.
What is a future? A commodity future is a contract between 2 parties wherein one agrees to buy something at some date at some price, and the other agrees to sell that same something at that same date at that same price. So if I buy December 2011 wheat futures today (if the marmket was open), I attain the right to have someone deliver me some amount of wheat for $9.09 a bushel. See here:
Other futures contracts, like futures for the S&P or the VIX trade at whatever price they trade for, and on the day they close they are worth what the underlying is worth on that day. So a VIX future for June may be trading at 30 bucks today (I think its about 17.10 today) but if the VIX is 49 on the day the contract settles, the buyer (today for 30 bucks) would make 19 bucks times the # of contracts times the # of units per contract. Etc.
What are contango and backwardation?
See how in the link above, as you go out in time, the cost of the futures contracts goes up? That is called contango. Contango refers to a futures curve (with the nearest dated futures value on the left and the farthest dated on the right) that slopes up. September futures are more expensive than July futures and December futures are more expensive still. Wheat futures, right now, are in contango.
So what is backwardation? The opposite. Check out corn futures (link below), corn futures are in backwardation. As we go out in time the price of a futures contract falls, this is backwardation.
Why do contango and backwardation exist?
Imagine that tomorrow someone found a 3 month supply for the world of corn in their closet and decided to sell it. What would happen to the cash price on the market? It would drop, maybe by alot. But would a smart trader drop the price of a December 2013 contract by as much? Probably not, because by December 2013 that 3 month supply is likely to be absorbed. If prices drop, farmers will plant less corn, which will cause future prices to rise. Contango, in commodities, is generally a function of ample current supply and the expectation of higher prices in the future. Many mechanisms could lead to this, but in general ample current supply exists in contango'd commodity futures curves. Backwardation in commodities tends to be the opposite. Imagine that Libya has a big old war or something, and people freak out and start worrying that the supply of Brent crude is going to be restricted. They bid up teh price for near term futures to reflect this fear. But what are the odds taht supply is still restricted from this in, say, January 2015? Not so high, so the long dated futures don't rise as much and the curve goes into backwardation.
In volatility or other futures, contango and backwardation are probably largely a function of expectations as their is no "supply" of volatility. If the VIX today is 10, the June VIX futures probably won't be trading at 10, they'll be trading at some higher value on the expectation that volatility is more likely to rise than not from that very low level. If we are very close to the June futures expiration (as we are now) maybe those futures trade for a small premium to the current VIX price, say 12 or 13. But waht about the July futures? Well, frankly, that's a long time from now and a lot can happen and the market will probably price those higher assuming that volatility will revert to its mean (which is higher than 10), so maybe they are priced at 16. The curve is now in contango, as the July futures are priced higher than the June futures.
On the flip side, imagine that on Monday morning Checklist breaks into the NYSE with a gigantic rolling pin in hand and starts smacking traders left and right, this causes a huge selloff as the freshly smacked traders liquidate their portfolios, and the VIX spikes to 35. Will June futures go to 35? Probably not, as once Checklist is arrested or escapes back home, the market is likely to calm down and the VIX will probably drop. It is likely that the June futures rise a lot, but maybe they only rise from 17 to say 25. Now, what about the July futures? They aren't likely to rise as much as the June futures, because the odds that the market calms down by July are fairly good, so maybe they only rise to 22 (from about 19 or whatever they are now). Now the curve is in backwardation.
How do contango and backwardation affect ETFs and ETNs? Imagine you are running an ETF...
Many ETFs or ETNs attempt to track the price of something by buying front-month futures. Of all the futures contracts, the front month ones are most closely correlated with the spot price of the underlying. If wheat spikes in price by 3 bucks a bushel, the July futures (nearest month) will probably rise more than the September 2012 futures. So many ETFs or ETNs including VXX, TVIX, USO, UNG and more work with the front month futures.
Now imagine that we merry CAPSters start an oil ETF so that investors everywhere can flood it with money panicking about something or other while throwing fits and slapping the family dog. Then we can all get rich by charging a 2% management fee, 2% of $1B is a lot of money each year.
we keep a constant 1-month duration of our futures. So on the last day of one contracts trading, we are 100% long the next contract. Say that the June contracts are expiring today, we are 100% long the July contract at the end of the day. Over each of the next, say, 22 trading days until the July contract expires we sell off 1/22ndof our July futures and buy August futures. This keeps us at a constant 30 day exposure level.
Now imagine that the curve looks like this:
July = $100
August = $110
We aren't very famous, so we only get $100,000 to manage, so all we can do is buy 1000 July oil contracts (I realize an actual contract isn't $100, but just pretend). And now say that there are 20 trading days until the July contract expires.
Each day we have to sell 1/20th of our contracts and roll them into August, per the laws of our ETF. So, one day, we sell 50 July contracts and we get $5,000. But our $5,000 only buys 45.5 August contracts. Now our fund is only long 995.5 contracts, not 1000. Like this:
Day 1, start with 1000 July futures @$100 each.
Sell 1/20th or 50 contracts @$100, take in $5000
Buy $5000 worth of August contracts, but that only buys 45.5 contracts
End the day with 995.5 contracts.
Rinse, repeat (and imagine that the July contracts remains 100 bucks all month and the August remains $110 all month) all month long and you wind up losing 91 contracts. You are now long only 909 contracts.
At some future day, oil future are going to be whatever they are going to be, but contango results in our fund owning less and less contracts as time goes by, meaning we have less money at that future date when oil is whatever it is. This is exactly the source of decay in UNG, USO, VXX, etc. This is exactly why the USO wasn't as high when oil hit #100 earlier this year as it was the first time oil hit $100. It owned way less contracts per unit of money it had under management.
Backwardation has the opposite effect, it would INCREASE the amount of contracts that our little ETF had, causing anti-decay.
And here is the fantastic irony... (it really is good)
Contango is most likely to exist when prices are low, below their mean or what the market expects to be their "mean".
Backwardation is most likely to exist when prices are very high, above their mean or what the market expects to be their "mean" going forward.
When prices are extremely low, we want to buy, betting on a reversion to that mean. So if oil prices were extremely low we would want to go long an oil ETF, not short it. But those extremely low prices probably led to contango, so going long that ETF we have the positive of low prices and the likelihood of higher prices in the future, but we have the negative of contango which will cause decay and make our returns lower, even if we are right we may in fact not make that much money.
When prices are extremely high, we may want to short something, expecting prices to go back to some kind of mean. But that is exactly when we are likely to have backwardation in one of these ETFs, which will harm our short position by causing anti-decay.
Screwed both ways, sort of.
A real example.
After Lehman, volatility sky-rocketed, and VIX futures sky-rocketed with it. But the market didn't anticipate that the VIX would stay at insane levels like 80 for 2 months, so in, say, October 08 the November 08 futures didn't rise nearly as much as the October futures rose. VIX futures went into backwardation. VXX, had it existed, would have anti-decayed at an amazing rate. At one point the amount of backwardation was close to 50%/month, meaning on that day you would have bought 1.5 times as many new contracts as you sold, causing extreme anti-decay. This extreme backwardation contributed to the meteoric rise than VXX and TVIX would have seen had they existed in that time.
Another real example.
Natural gas is about 4.40, Natural gas was about 4.40 in July of 2010. UNG was about 14.50 in July of 2010, its about 11.50 now. Why did it drop so much? Contango.
An important note.
These things do not track the underlying, they track the FUTURES of the underlying.
USO does not track the price of oil, it tracks the price of 30 day duration oil futures.
UNG does not track the spot price of natty, it tracks the price of 30 day duration oil futures.
VXX does not track the VIX. It tracks the price of 30 day duration VIX futures.
This is why the VXX doesn't go up 5% on a day that the VIX goes up 5%. Heck, it may one day go DOWN on a day when the VIX is up 5%. It may well go UP on a day when the VIX is down. Its NOT TRACKING THE VIX, ITS TRACKING VIX FUTURES.