ADRs and Yield

As was noted on other threads in the forum there was a recent change to the Yield calculation for ADRs. This effected one of my strategies fairly significantly, so I decided to investigate the ADR Yield calculation within P123. I focused on the numerator Indicated Annual Divided (IAD), which is a forward looking factor which is calculated as the Most Recent Quarter’s Divided * Number of Dividends Per Year (excludes special dividends). In order to analyze how P123 is handling Yield betweent ADRs and Regular US Stocks I downloaded the historical divided data from Fidelity for a handful of large ADR and non-ADR stocks.

Non-ADRs - CAG, BBY, ABBV, XOM

ADRs - RIO, BP, RDS, SSL

Using a simple algorithm I calculate what the IAD should be over time for each of these stocks based on the Fidelity data and charted it against the number reported by P123. In the first attached image below you will see that for Non-ADRs the Fidelity and P123 numbers match almost exactly. However, in the next image you will see that for ADRs that the match between Fidelity and P123 numbers is quite inconsistent. Hopefully some one at P123 can have a look at this, because I find it to be troubling. Some (all?) of the differences are I expect due to the recent changes. Note for example, how P123’s IAD fails to fall when BP and RDS cut their dividend in 2020. Based on the description of the recent ADR Yield calculation change, I am suspicious that change is the source of the problem. As you might imagine, keeping the IAD high when it should be cut… could lead to some extremely high erroneous Yield numbers. I am sure there are all kinds of complexities in attempting to calculate the ADR Yields, but the current method seems like it might have some holes.

Would appreciate someone taking a look at this issue.

Best Regards,

Daniel



We’re just reporting the data from the data provider in this case.

Well damn… looks like Compustat is doing a terrible job with this one. I’m curious if FactSet differs… will check when I get some time this afternoon.

Thanks,

Daniel

FactSet appears to be exactly the same as Compustat, so I am at a loss. I assume there is something I am not understanding about the ADR IADs…

It especially doesn’t make sense that IAD would stay elevated after a company cuts its dividend, there is no way that is the appropriate way to handle the new information.

-Daniel

I’ve noticed that Compustat ignores some forward dividend projections and will sometimes calculate IAD based on dividends paid. I’ve taken this issue up with them, but with no success. Perhaps, as a Compustat subscriber, you’ll have better luck? But since they’re P.I.T., it won’t change past data.

How do we know if Fidelity is right?

I’ll add a new factor called IADNTM which simply adds all regular dividends paid in the next 12 months from the corporate action stream. In other words a factor with 100% look-ahead bias. This way we will know what the right answer was.

We could even introduce factors with 80% look-ahead …

Thanks Yuval, that’s what it looked like to me also (regarding using dividends paid instead of forward looking announced dividends).

Marco, fair point. Will be interesting to see what the IADNTM looks like.

I am playing around with a few things to see if I can come up with something that works based on the available dividend data on P123. Will post if I come up with anything different.

Thanks,

Daniel