I ran into this issue from a simple problem, so I think I am missing something. I need to get implied dividend curves from futures (yes, I could use div swaps, but no arbitrage should match div swaps). The simple F = Se^(r-q)t is what I'm using to solve for implied div yield. The example I'll use if for KOSPI2 futures:243 = 244.05 * e^((.00185 - q) * (14/365)) => q = 11.4%All data is as of 5/31/2012, and am using the USD 2w deposit rate as risk free. I notice this most in the first 3 futures contracts. My thoughts are below, but any comments would be very helpful.Potential sources of error:- Tax- Daily MTM impacting the futures price, so I need to adjust futures price- Roll dates impacting futures prices

I don't know about your futures, but the same exercise done on the SPX futures can produce nonsense.The reason is that the futures close a session 15 min later than the last index print, and they can fall several points during that time frame. edit: earlier, I said this would produce nonsense for yesterday, but that was a mistake. The implied q for spx wouldhave been 1.9%, which is not nonsense.

i think your problem here is that you are annualising a small rate over a small period of time... another way of looking at the cash and carry is to write it asF = (S-D)*exp(rT) where S is spot, D is dividend (in index points) and r and T are rate, T and F is futures pricehence D = S - F*exp(-rT)plugging in your numbers gives a dividend of 1.06724 index points.Are the dividends seasonaly in Korea ? If memory serves me right they all get paid once per annum

Why are you using the USD 2w rate instead of the KRW 2w rate ?

Sorry for the delayed response. In short, all of these questions lead back to the original sources of error I mentioned, but add sec lending carry to that list (is much more important now that it was even 5yrs ago). Sec lending may make a lot of sense due to the more illiquid indicies typically have larger error. This is a simple concept that has been around as long as futures, which is why I'm struggling with it not holding. Any other ideas?MCarreira: Correct, rates should be tied back to funding currency. In this case it doesn't really matter, just provided USD deposit rate because it's readily available. KRW 2w deposit rate is not. If you were to do an interp on the KRW curve you could say the KRW 2w rate is something like 3.35% which yields an implied dividend of 14.55%.daveangel: At first glance this made a lot of sense to me, however, at second look you're de-annualizing the rate by e^(-rt) so your dividend (in pts) is also de-annualized. So you need to multiply that by (365/Act) to get back to an annualized div in pts. When you do that you get approximately the same thing (since ln(p2/p1) ~= (p2 - p1)/p1). Alan: This is not only for KOSPI2 but quite a few other indices. Actually, SPX has been one of the better indices under this methodology. UKX, SX5E, and EAFE have also been very poor. But I will look into syncing times. This could lead to a large error, esp. in low rate environments with small time to maturities.My new list of potential sources of error:- Tax- Daily MTM impacting the futures price, so I need to adjust futures price- Roll dates impacting futures prices- Need to incorporate credit spreads in borrowing cost (but should be collateralized so should not be a large credit charge...)- Sec lending carry

Quote daveangel: At first glance this made a lot of sense to me, however, at second look you're de-annualizing the rate by e^(-rt) so your dividend (in pts) is also de-annualized. So you need to multiply that by (365/Act) to get back to an annualized div in pts. When you do that you get approximately the same thing (since ln(p2/p1) ~= (p2 - p1)/p1). there is no such thing as annualised rate in index points.... you seem to have a problem understanding the basiscs

daveangel: What I mean is that your dividend is going to be for the period from now til future expry. Viz., D is in terms of T, whether it be index points or yield (q). Therefore, D - expressed as a dividend in index points over 2 weeks - can be converted to a yield, then annualized. That part is very simple. Ultimately, q and D are interchangeable... Take the original q = 11.4%, and convert to index pts over that 2 week period:(qT * Index Value) => 11.4% * (14/365) * 244.05 ~= 1.067129This is what you have in your explanation. I don't believe I am missing anything, but if you do please explain so we're on the same page and can get some resolution out of this.

QuoteOriginally posted by: AlanI don't know about your futures, but the same exercise done on the SPX futures can produce nonsense.The reason is that the futures close a session 15 min later than the last index print, and they can fall several points during that time frame. edit: earlier, I said this would produce nonsense for yesterday, but that was a mistake. The implied q for spx wouldhave been 1.9%, which is not nonsense.Alan is correct, synchronizing spot and future is the first thing you need to do (you want a sawtooth pattern on the time series of the spread).

QuoteOriginally posted by: MCarreiraQuoteOriginally posted by: AlanI don't know about your futures, but the same exercise done on the SPX futures can produce nonsense.The reason is that the futures close a session 15 min later than the last index print, and they can fall several points during that time frame. edit: earlier, I said this would produce nonsense for yesterday, but that was a mistake. The implied q for spx wouldhave been 1.9%, which is not nonsense.Alan is correct, synchronizing spot and future is the first thing you need to do (you want a sawtooth pattern on the time series of the spread).you dont need to rely on the index provider to calculate the spot - any decent index arbitrageur will calculate the index themselves at any time using the bid side and offer side of the market.

QuoteOriginally posted by: quantnyudaveangel: What I mean is that your dividend is going to be for the period from now til future expry. Viz., D is in terms of T, whether it be index points or yield (q). Therefore, D - expressed as a dividend in index points over 2 weeks - can be converted to a yield, then annualized. That part is very simple. Ultimately, q and D are interchangeable... Take the original q = 11.4%, and convert to index pts over that 2 week period:(qT * Index Value) => 11.4% * (14/365) * 244.05 ~= 1.067129This is what you have in your explanation. I don't believe I am missing anything, but if you do please explain so we're on the same page and can get some resolution out of this.no you arent missing anything but nobody annualises index points...

Either you take the approach that you accrue the dividend over the period under which you hold the equity (the assumption of futures pricing), convert the accrued dividend to an annual dividend, and convert that annual dividend to a dividend yield. Or, alternatively you look at the yield and convert the yield to an annual dividend yield. They are the same thing - it makes no difference how you think about it so long as you keep the assumption on constant accrual of the dividend (which in this case of futures pricing we agree on). However, rather than argue about the existence of differing opinions/perspectives, some resolution on the issue at hand would be positive.I am back to:- Tax- Daily MTM impacting the futures price, so I need to adjust futures price- Roll dates impacting futures prices- Need to incorporate credit spreads in borrowing cost (but should be collateralized so should not be a large credit charge...)- Sec lending carry

Quote I am back to:- Tax- Daily MTM impacting the futures price, so I need to adjust futures price- Roll dates impacting futures prices- Need to incorporate credit spreads in borrowing cost (but should be collateralized so should not be a large credit charge...)- Sec lending carry1. tax - in most cases tax will make futures price higher not lower2. mtm - unlikely 3. roll dates - unlikely. 4. credit spreads - if this is a factor then it should be minute5. sec lending - most likely

That makes sense. The question is then how do we cleanse the implied div. yield? Use TRS quotes to proxy lending revenue?