Overseas ETFs, time zones, fx (WTF)

Would like to discuss how NYSE ETFs track their foreign indices. Been trying to reconcile this so I can have some level of confidence/understanding (tracking error, foreign currency movements, premium/discount to NAV, etc). Specifically I’d like to discuss China, and their complex HK/SH/US situation… There only seems to be one ETF that semi-accurately tracks the China A-shares, and that’s Deutsche’s ASHR… CSI 300 Index (Shanghai, 300 largest A-shares, company shares trade in RMB) vs ASHR (NYSE ETF, trades in USD, tracks CSI300) * Dec 4th Asia - per Shanghai exchange the CSI300 closes up 4.6%. * Dec 4th USA - ASHR opens in premarket up 5.0%, closes postmarket 8.0%. * Dec 5th Asia - CSI300 down 2% as of my post. So how do the mechanics work here? The underlying was up 4.6%, but the US market opened up 5% (which cannot be explained by FX since USD/RMB didn’t move that much). Perhaps it is tracking error, or people bidding it up in premarket? But by the end of the day in the US it was up 8%! Is this the US market pricing in their 3.6% expected increase for Dec 5th?. But what if in Shanghai on Dec 5th it closes down -2%? When the ETF opens on the NYSE on Dec 5th do the fund managers feed a -5.6% open price to the system? Or does this require the market participants to arbitrage the difference back in line with the underlying? Sorry if this is obvious, I think it gets into prem/disc to NAV? Here’s another example: MSCI Japan Index closed up 0.67% on Dec 4th, then the US market opens and HEWJ is down -0.50%. EWJ tracks the MSCI Japan index, so I find this 1.1% delta rather peculiar. It’s not explained by FX, and besides the sucker is supposed to be hedged. Confussing. What’s up yo?

Do the ETFs own the underlying stocks or are they using equity swaps/forwards?

Deutsche owns the underlying thru a relationship with Harvest Global Investors (asset management firm in China).

I suspect the primary driver of the difference is something to do with NAVs and pre/disc but I just don’t know how this crap works. My desk used to be next to a mutual fund department and I did everything in my power to ignore these idiots and their 5am NAV talk, hmm maybe I should have learned a few things.

I was under the impression the spread between HK and mainland shares was due to different rights for each type of share…similar to spreads between other types of share classes or claims on a part of the capital structure. The spread could certainly be overestimating the risk between the two.

As far as EWJ and the index goes, I’m not sure there’s a reason they should be the same and I don’t think it’s tradeable without the proper infrastructure. It’s equivalent to saying the SP500 index return is different from the SPY which doesn’t match up to the ES…it exists for a reason and isn’t tradeable.

You could even take a look at trading IPAC against EWJ+EPP…it’ll probably have an even higher delta and it still won’t be tradeable.

There are multiple factors that lead to tracking error and that in total probably sum to approximatley the tracking error of the ETF.

  1. Subtract the ending premium/discount from the starting premium/discount.

  2. Subtract the prorated expense ratio.

  3. Subtract the fair value pricing impact. Fair value pricing is used to accound for the different closing times of non-US exchanges from the NYSE.

Where I used to work we used VWO to equitize cash occasionally and this process would generally explain any tracking error withing +/- 10 bps.

Thanks, I’ve got a really simple and probably incredibly stupid question to ask. When the foreign index (say Shanghai CSI-300) closes up 4% like it did today, what determins the opening price of the NYSE ETF which tracks this index (ASHR)? Obviously the investors already know what happened in Asia, they can’t be allowed to start buying at yesterdy’s closing price. If someone can dumb the math way down for me, I just don’t get it.

This pg shows the premium on the ASHR ETF to be a massive 5.5% right now. http://etfs.morningstar.com/quote?t=ashr

This looks ike it may be a bit more of a special situation, due the oversubscription. If you look in the link you sent the volume has spiked enormously and they are probably having troble creating shares fast enough, so the ETF has been trading at a permanent premium. Maybe due to some of the costs of creating the units, the ETF desk is taking its time arbing this away. I would not buy this ETF if you have access to others trading a less of a premium.

I’m not really an ETF guy, so really your best bet would be to just call them. ETF places will sometimes put you right on the phone with a trader.

I would emphasize the importance of time zones on daily international data with a focus on your point about EWJ and MSCI Japan.

If you were to measure the daily return, the Japan and U.S. closes are not happening at anything like the same time. Japan closes at something like 1AM EST. So Japan closes, then you have the entire European and U.S. sessions. If something significant happens in the U.S., then this might lead people to think that something might happen tomorrow in the Japanese market. So EWJ moves, even though MSCI Japan, which is based on the changes in individual Japanese stock prices, has already closed.

Another convenient way to think about it is that Japan’s close today, depends on the U.S. close yesterday. In the example above, the news from the U.S. will then impact the MSCI Japan index the next day. From a time series perspective, this means that you need some sort of AR or MA model to get the right daily covariance matrix. If you regress daily MSCI Japan on its lag, the MSCI USA and its lag, then the biggest coefficent will be on the lag of MSCI USA.

If you don’t like this behavior, it mostly works itself out on a monthly basis.

Thanks guys, yeah after watching it for a few more days and doing some math, I think I get it now. There is a massive premium (probably over-buying and/or problems keeping up creating units) that is expanding/contracting and causing weird daily returns that don’t match the performance of the underlying index. Also people in the US time zone keep pricing in their expectation of the following day, and their expectation keeps being super wrong (since the SH market only has a .1 correlation with the US market, and does not trade on fundamentals).