Economic Derivatives Auction Trader Motivations
Jason Ruspini November 27th, 2006
The National Association for Business Economics’ experimental auction data for the October payrolls release seems to be unavailable, but let’s compare their forecasts to CME’s economic derivatives auction for the September release.
NABE:
CME:
The CME numbers had more variance and fatter tails. Specifically:
NABE: Less than 5% in the less-than-0k and greater-than-250k tails.
CME: 10.1% in the less-than-0k and greater-than-250k tails.
NABE: Approximately 75% between 75k and 175k
CME: 47.7% between 75k and 175k
One possible interpretation is that the tails of the CME auction are overbought (by hedgers?). I have not seen the CME data recently, but from what I hear it does not actually show a longshot bias over time. This suggests that the tails were underbought in the NABE auction. We don’t have enough NABE data to analyze, but since their participants were fully-funded we might expect them to herd around the consensus number. Justin Wolfers has also suggested that what we see in the NABE auction is overconfidence, which would dissipate in an auction with larger stakes.
In any case, hedging against the payroll release has recently been an absurd exercise. The number reported for September was 51k, but this was later revised to 148k. This revision was reported on the day of the October payrolls release, along with a second revision to the August number, which now stands at 230k. That is, the numbers on which the CME auctions for August and September were settled were low by a total of approximately 200k. Revised numbers were a well-known inevitability to the market’s designers and traders, but this had to be exasperating for anyone actually hedging a fixed income position with the October release auction. If this theoretical person had been long fixed-income, he probably would have hedged by buying the upper tail of the payroll outcome: a result likely to embolden the fed and send bonds lower. The October number actually came in on the low side, and our theoretical trader would have happily accepted his hedging loss, but wait a minute… the numbers for the earlier two months were simultaneously revised much higher, causing the largest one-day sell-off in bonds in months.
This begs the question of whether our theoretical hedger exists. A recent article by Howard Simons at TheStreet.Com echoes the earlier opinions of JC Kommer and myself that the economic derivatives markets mainly represent a new avenue of speculation (which is perfectly admirable). Using the risk-based definition of gambling (which I am not endorsing), Simons writes, “The risk, however, is attached to particular markets, not to the number. [...] But in the absence of a tradable instrument created for the sole purpose of assuming that risk — our definition of gambling above — the number carried no actual risk itself.”
The massive effects of revisions only reinforce the argument against the existence of substantial hedging in economic derivative auctions, but even without hedging utility the auctions still fulfill an information discovery function.
- All Best Posts Ever , All Guest Authors's Posts , Analysis (Data) , Analysis (Meta) , Exchanges & Markets , Finance
- Comments(2)
I think your problem is a rather limited defintion of risk…
To a large degree for most fixed income “traders” as opposed to investors who don’t count.. the risk is volatility, not direction…
Although the CME exchanges are far too small, and so poorly designed that no one takes them seriously, in theory if you had enough time to waste attempting to do an odd lot on the contract..
One would compare it and trade it against the OTM probabilities on the FED FUND futures options.. which is where the real risk transfer takes place..
No body has risk to the Non Farm Payroll number per se, they have risk as far as that number is an input into the Central Banks forecast and more importantly the perception that the market has of that forecast.. (a la Keynesian beauty contest)
You should ask yourself who rationally will sell the wings of those NON FARM payroll trades?
Other than my uncle sitting in Florida trading an odd lot and watching CNBC.. only someone who can convince himself that he has an offsetting hedge…
Yesterday’s ISM number was a good example.. within seconds of that numbers realease.. (forgetting the 15 minute leak).. the probability of the FED easing by January essentially doubled…
Out of the money calls quadrupled in value and more…
The key is weighting.. the mapping should be.. (assuming infinite wisdom) what will be the change in probabilities for a given number..
This week has been trivial.. and more so because by now the game is so well established at least in probability terms that it has its own “rulebook”… with live commentary on Bloomberg .. and referees from the Federal Reserve Bank.. who actually comment directly on the “bets” being made in both the Fed Funds and Fed Funds options marketplaces.. (see Donald Kohn..William Poole..and Janet Yelling [who actually integrates the Case/Shiller Prediction market projects into FED Forecasts])
For the last weeks the Bet is/was… how likely is the FED to change its Bias at the December FOMC meeting.. not ease or tighten but to change their revealed “forecast”..
This meta event, should it happen, would of course imply a change in the expectations that the next FED move would be an EASE and if the bias Statement comes in December, by definition the probability of a move increases ex ante at the January FOMC…
raising form zero percent where it was to X%..
As a side note.. this has nothing of course to do with the question as to whether the FED will actually EASE or not in March… a) you’ll know the answer to this.. on December 12th,
b) actually doesn’t have much to do with whether the FED will ease in January either..
Just like Keynesian Beauty contest.. its a question of how many levels of backwards induction you want to run.. there is now a market convention..
If the ISM was below 50%.. etc. etc.. the probability that the FED would change the bias..would be say 50%.. too.. An Easing bias is modeling rationally at implying 50/50 odds that the FED will EASE in the future.. i.e half way between on hold and ease..
So not surprisingly on the number the probability of the FED easing by January moved to around 25%… or 1/2 way to 50%.. which is where it will be if the FED shifts their bias..
In a Bayesian World the key is can one map ex ante a given economic data point to a change in expectations..
Sometimes the number is relevant and sometimes its not..
Post the massive revisions in Non Farm Payrolls for the time being nobody is paying much attention to them..
Even algorithmic traders now now you have to read the caveats to figure out whats what.. and once you do that the data is meaningless..
Everyone agrees that the numbers carry no risk in themselves, and that the relevance of the releases shifts over time.
My notion of risk was not limited to directional risk though. After seeing the two distributions above, my initial idea was precisely that CME traders were buying volatility (the “straddle” on the numbers) and thus overpricing the tails. But this does not seem to be borne-out by the data (which I haven’t seen myself.) In these auctions, buying one outcome is a sell of all other outcomes, which helps explain who is supplying the wings.