1
$\begingroup$

Bloomberg Barclays index returns (e.g. LF98TRUU Index "index_total_return_mtd" & "index_excess_return_mtd") and sub-index returns (e.g. BCBATRUU Index "index_total_return_mtd" & "index_excess_return_mtd") are published by Bloomberg. Index constituent returns, market-values, and excess returns are also available through Bloomberg.

I have access to the constituent-level data from Bloomberg and would like to replicate the monthly returns from the bottom up. Bloomberg publishes their methodology here: https://data.bloomberglp.com/professional/sites/10/Index-Methodology-2019-07-10.pdf

However, I am unable to exactly tie-out the returns. I can get within 4bps of tracking error to LF98TRUU-index_total_return_mtd & index_excess_return_mtd, and about 27 bps of tracking error to BCBATRUU total/excess returns.

Question: Is it possible to exactly replicate these indices using the Bloomberg Barclays constituent data (edited to specify this constraint)? Or does replication somehow require other data?

If the answer is yes, then I must be doing something wrong and may create another question to go over my understanding of the methodology and where an error may be introduced.

$\endgroup$
2
  • $\begingroup$ Just an idea: Are you using the "correct" constituents per historical month, i.e. are you fetching historical constituents? $\endgroup$ Sep 16 at 13:51
  • $\begingroup$ Yes, I have the "Returns Universe" constituents each month. $\endgroup$ Sep 16 at 13:52
3
$\begingroup$

The devil's in the details. Here are a few things to check off the top of my head:

  1. Index constituents: The index is rebalanced only once a month, at the end of each month. We'd switch to the the constituents of the forward-looking "statistical universe," which becomes the "returns universe" at the start of the next month and is then used for the entire next month.
  2. Constituent weights: The index is market-value weighted, but has a ton of additional adjustments (SOMA holdings). Make sure to use the same weights as the index.
  3. Pricing: Make sure to use exactly the same price source as the index, particularly for less liquid issues. The timing of the snap also matters. For US securities, 3pm bids were historically used, but the index has switched to 4pm snaps this year.
  4. Settlement Convention: When computing accrued interest, use T+1 regardless of the actual convention used by a constituent.
$\endgroup$
8
  • $\begingroup$ +1. Point 3 can be a real pain for the replication task. $\endgroup$ Sep 16 at 13:57
  • $\begingroup$ Thanks for your response, Helin. Regarding 1) I am using the "Returns Universe" to calculate monthly returns, which I believe is correct. 2-3) I don't understand this second point. I have the Bloomberg Barclays constituent level data with both beginning_mktval and pricing coming from the Index Provider itself. 4) Not sure I understand this point. I have Total and Excess return numbers from Bloomberg at the security level. So I don't think i need to get into these details. Please correct me if I'm wrong. $\endgroup$ Sep 16 at 13:58
  • $\begingroup$ Is your understanding, Helin, that my basis to the index is a sign that I'm doing something wrong? My first question is: is perfect replication possible using the Bloomberg Barclays constituent data? $\endgroup$ Sep 16 at 13:59
  • $\begingroup$ I edited my question to clarify that I have the Bloomberg Barclays official constituent data and associated fields, and want to know if replication of the monthly index returns is possible with that data. $\endgroup$ Sep 16 at 14:00
  • 1
    $\begingroup$ @Lepidopterist Yes I used to do this type of thing a lot, so it's definitely doable. The only problem is that these index data used to be on Lehman Live/Barclays Live, so I don't know what the corresponding fields are on BBG. At this point, I recommend just contacting BBG's index team and ask them to send you a sample calculation sheet. (Back when the indices were still Lehman/Barclays assets, the team was always happy to send over sample calc sheets; it's basically the same team of people and I'm sure they'll help out.) $\endgroup$
    – Helin
    Sep 17 at 0:44

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.