Quantopian's community platform is shutting down. Please read this post for more information and download your code.
Back to Community
Future premium on the ES

S&P 500 futures (ES) tends to trade at a discount of more than 3%. When annualized this amounts to a discount of more than 12%. As far as my knowledge goes, the real interest rates in the US are not -12%. No amount of inflation expectation would justify this? Any explanation for this behavior?

5 responses

Futures don't pay dividends.

Nominal interest rate: Lets take it as 0%
Dividend yield : 2%
Inflation expectation: 2%

Still doesn't explain the future premium of -12%. am I missing something?

I think the magnitude of the discount is much less than you've calculated. Right now the June 17, 2016 ES futures are at 2062 and the SPX is at 2101, a difference of 39 points which is 1.9% of 2101. Normalizing for a year, that's pretty close to explained by Simon's comment. Note that interest rates work in the other direction, for most of history ES has been in backwardation but is in contango only because of current low interest rates. For those who like equations, carry-ES=dividends-SPX with carry being the interest money saved from posting bond for the futures vice buying the SPX. Also, minor technical point to quibble over, but it is interest rates not inflation expectation. Expectation makes it sound subjective, but this is a very fixed arbitrage controlled futures price based on interest of a risk free bond of equivalent tenor of the future. I only bring this up because the mistake is often made, even by supposed experts, that ES somehow represents sentiment about the future price of the S&P 500, which it most definitely does not. (See http://www.zerohedge.com/article/es-futures-curve-hits-whopping-10-point-6-month-backwardation for an example of this type of mistaken logic)

Sorry. I had got it wrong.

Kevin beat me to it. http://www.quora.com/How-do-you-calculate-the-implied-open-from-futures has an example of how to calculate it, I haven't verified all the conventions he mentions (ACT/360), nor does he address any convexity adjustments, but it already comes out pretty close, as well it should.