1
$\begingroup$

For a specific coupon, if TBA TOAS is greater than a Specs TOAS, does that imply that the Specs are trading too rich to TBAs (given liquidity risk, payup risk, etc)? I realize relative value comparisons in Specs are usually done against Specs, but am trying to understand what this could mean. Could there be another explanation for why this could be the case?

$\endgroup$

1 Answer 1

3
$\begingroup$

Added later The first answer I came up with was embarrassingly incorrect; I have pointed out why. Please feel free to uncheck it.

Technically, this should be impossible: the Spec pool is always deliverable into the TBA, so the TBA OAS is a floor for the Spec pool valuation. This is plain wrong -- the TBA price is a lower bound on the price of the Spec pool, but there is no such constraint on the respective OASs of the two instruments. We can definitely ask the question as to why one would hold a Spec Pool at a richer valuation than TBAs and the discussion below offers some sanity checks to first ensure that this relationship really holds.

Sanity Checks:

  • Is there an issue with the prepayment model? The model may not be recognizing the prepayment protection offered by the Spec pool and/or the true callability of the TBA

  • Are the correct collateral characteristics of the TBA being input into the OAS model? This can be tricky to get right: Dollar rolls, float, issuance patterns, and Fed purchase activity all play a role (among other factors).

  • Is there a pricing error? It is hard to get TBA prices wrong but Spec pool payups can sometimes be stale because of an absence of deep markets (or because the pricing source does not have an accurate idea of the market).

In the unlikely event that the OAS anomaly still persists after these factors are accounted for, the only somewhat contrived scenario in which Spec Pool OASs are lower than TBA OASs that I can think of is if there is a strong bid for Spec Pools from the CMO desk for a particular settlement month. Even this would be pretty irregular because CMOs are generally created from cheap not rich collateral, but occasionally there could be a short squeeze.

$\endgroup$
7
  • $\begingroup$ Thanks, that answers the question. I noticed this using the Yield Book Production Model. I use the generic YB cohort for TBA and the collateral characteristics of that appear to match pretty closely with the TTB ones and I ran this close to Class A Settle. And, I'm using the spec payups available from a Trading desk, so it seems like the issue could be with the prepayment model itself. But given its the YB Model, wouldn't this issue already be well-known? Could you potentially suggest what next steps I could take based on this, if I just want to value Specified pools? $\endgroup$
    – Jojo
    Jun 22 at 16:45
  • 1
    $\begingroup$ Suggestion (1) Anecdotally, the YB Production Prepayment Model always appears to be about at least 6 months to a year behind the current reality of prepayments so it's possible that market participants are aware of this bias. YB usually makes an "experimental" model available that fits current trends better; perhaps worth checking the valuations on this experimental model to see if the anomaly goes away. $\endgroup$
    – Sharad
    Jun 22 at 20:29
  • 1
    $\begingroup$ (2): Look at the prepayment model tracking error over a cross-section of different collateral characteristics and "dial" the production model to get something more realistic. Doing this properly requires a solid understanding of the prepay model and the structure of prepayments. Nevertheless, it's a good exercise in understanding what's driving collateral valuations. $\endgroup$
    – Sharad
    Jun 22 at 20:34
  • 1
    $\begingroup$ (3): Revalidate payups: it's possible with the big move in Rate and Volatility markets, the Trading desk may not have updated payups. Ask them to make a market in the specific Spec sector. $\endgroup$
    – Sharad
    Jun 22 at 20:37
  • 1
    $\begingroup$ See p3 of the PDF at this link: arx.cfa/~/media/5F70FBDC7DE44E5598BA77B0BCD994E0.ashx to see the full set of Prepayment Model dials. Yes: you "dial" the model to get prepayments with lower tracking error. $\endgroup$
    – Sharad
    Jun 23 at 20:21

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.