Can both good buying and good selling cause a bond to go special on repo?

Question

A bond is known to go special when its repo rate gets particularly low relative to the GC (General Collateral) repo rate.

In my mind, this can be caused by two scenarios:

1. Institutional interest to short the bond

Note short here is different to selling a bond that is already owned. In this scenario, institutions will sell the bond to bank dealing desks, who will then subsequently be long. Since it is not allowed to be 'naked' short, the institutions will need to cover their short in the repo market. The added demand for the bond on repo will push the repo rate lower (technically the reverse repo rate, but same thing). This is because those who actually own the bond (who will repo it out to the institutions shorting it) can demand a lower borrowing rate on the cash in the repo transaction. This can clearly push a bond to trade special if short interest is high enough.

2. Institutional interest to buy the bond

In this scenario, institutions will buy the bond from bank dealing desks. If the desk did not already own the bond, they will be naked short and need to cover it in the repo market. For the same reasoning as scenario (1), demand from dealing desks for the bond on repo will push repo rates lower.

Is this correct? It seems odd (but clearly intuitive) to me that both demand to buy a bond and demand to short a bond can have the same impact on repo rates and pushing a bond to trade special. It suggests that only plain boring bonds will be the ones that ever have a GC rate.

Am I missing something?

Possible explanations

I'm thinking that it's not as simple as (1) and (2), and rather depends on the current market positioning of the dealing desks.

Example 1: Institutional interest to short a bond when dealers are long: In order for dealers to buy the bond, they will need to repo out the bond they already own so that they have the cash finance buying the bond from the institution. This demand for the other side of repo (i.e. demand to borrow cash) will have the counter-acting impact on repo rates outlined in (1). This could thus stop a bond from going special.

Example 2: Institutional interest to short a bond when dealers are short: In this case, dealers can just close their short position by buying the bond back from the institution. This will reduce the demand for repo, so the demand for reverse repo to cover shorts from the institutions is one sided, thus pushing a bond towards trading special.

Answer 1

From my experience you are thinking about this the right way, although I think the positions of dealers are somewhat of a red herring here. Dealers aim to run roughly flat positions in recently issued Treasuries , which are the ones that sometimes go special. Secondly , it’s clear that if all buyers of Treasuries lend them to the repo market, then there will always be enough bonds and there won’t be any specials. So the problem comes when there is (a) institutional demand to short the security and also (b) presence of cash buyers who do not lend the bond to the repo market. (An example could be a central bank who is reinvesting dollars acquired through trade surplus). Having said that the market is quite opaque so consider this a strong theory rather than established fact.