It looks like investors are willing to pay more for options exposure to Treasury bonds than they ever have before.
It is a well-known fact that, across many different markets and consistently over time, options tend to be priced today at volatility levels greater than the actual statistical volatility that occurs in the underlying asset. This phenomenon is known as the volatility risk premium (VRP). Think of an investor who wants to hedge her stock portfolio and buys some puts on an equity index to achieve that purpose. If those puts were perfectly priced then, on average, the implied volatility of the puts as given by market prices would match the volatility that occurs in the underlying index over the life of those put contracts. But that’s not what happens; instead, we see options on equities priced consistently higher than we would expect. There is a large academic literature detailing the magnitude of the volatility risk premium, and it’s also pretty easy to observe empirically.*
There are intuitive explanations for why investors in stocks and commodities would be willing to pay a premium to transfer unwanted risk via options markets. Whether the same explanations apply in fixed income markets is an interesting question, but not the focus here. Instead, I want to note that the
First, some history. The VRP in Treasury bond options has not, historically, been that large. It is positive, meaning that net option sellers who hedge deltas dynamically have a positive expected return, but where Simon (2009) notes this fact, he also mentions that the edge is typically smaller than the bid/ask spread. That’s very small: it means the expectancy flips negative if you don’t successfully negotiate the spread.
Fig. 1. TLT and SPY 1M volatility risk premia, 2003 – 2012. Source: Condor Options, CBOE
Fig. 1 shows the VRP for TLT and SPY since 2003. I’m estimating the premium as the ratio of implied volatility one month ago to the one-month historical volatility today.** TLT is in red and SPY is in blue in the background. A reading of 1.00 indicates that options one month ago were priced at exactly the level of volatility that occurred in the underlying. A negative reading indicates that options underpriced subsequent volatility (see the big dip in the fall of 2008), while a positive level indicates that options were overpriced.
You can see that, over the last decade, the TLT VRP has only rarely broached the 1.50 level, while readings at that level or higher are very common for equities. The median TLT value for this period was 107%; for SPY, the median was 133%. And since we’re measuring the ratio of implied to historical volatility, it doesn’t matter that bond volatility tends to be lower in nominal terms.
What’s interesting is that, in 2012, the risk premium in TLT options has broached that 150% level twice:
Fig. 2. TLT and SPY 1M volatility risk premia, 2012. Source: Condor Options, CBOE
Not only have options on Treasury bonds traded at historically rich levels, but they are trading more richly than options on the S&P 500. Since 2003, that has happened less than 25% of all trading days; since the March 2009 market bottom, TLT options have proven to carry higher risk premiums than SPY options one month later only 17% of the time.
The collapse in VRP this spring and summer has occurred not just in stocks, but in options on EUR/USD, WTI crude oil, and gold as well. So what explains the resilience of IV in TLT options? I’m not sure there is a good explanation.
One plausible story is that, given high expectations for some sort of QE3 announcement from the FED at the most recent meeting, traders were bidding up TLT options more than they otherwise would in anticipation of a large move in the underlying. That story isn’t supported by the data, though: ahead of QE2 in October and November of 2011, the TLT 1M VRP was actually negative-to-normal, not elevated like we’ve seen this summer.
Hard money enthusiasts and sandwich board men will no doubt wonder whether this high risk premium in Treasury bond options counts as evidence for a looming invasion by bond vigilantes (when all you have is a hammer, every problem looks like a nail, etc.).
Fig. 3. TLT and SPY 1M implied volatility, 2012. Source: Condor Options, CBOE
It doesn’t. TLT implied volatility has been more or less tracking equity IV this year. What divergences there have been have occurred more on the statistical volatility side, with bond price volatility falling steadily this summer. If the collapse of Treasuries long heralded by the Peter Schiffs and Ron Pauls of the world was looming and there was going to be evidence of that fact in this options market, we would expect to see bond implied volatility rise inexplicably and/or stop correlating with equity IV.
These pricing relationships could just be historically unusual blips, or they could be part of a larger story. Whatever the explanation, we will be watching this data with interest: we do not often trade volatility arbitrage strategies on Treasury options as the premium is rarely large enough, but if these anomalies keep occurring, it may be a sign that short volatility traders should give this asset another look.
* The volatility risk premium is the concept underlying most of what we do in the managed client accounts and strategy offerings. For more on the concept and related academic research, see Options and the Volatility Risk Premium (FT Press).
** 1M IV-30 / 1M HV0 is a more “perfect” estimate of VRP, since it compares two measurements of the same time period. This method is also only available after the fact, since any forward-looking volatility estimate implied by the market today cannot be evaluated until all of the return data for the period has come in. In “live” situations, we look at current IV / HV or GARCH-type models.