Swap Spread Arbitrage & Rates/STIRT Vol

Question

Concerning the classic swap spread arbitrage trade where you (as far as I understand it):

1. Buy a treasury and borrow in GC repo, paying repo rate and funding the haircut in short term unsecured funding market.
2. Enter into a pay fixed swap with maturity matched to the treasury, borrowing initial margin @ OIS

On net, you pay the swap spread, and earn LIBOR - GC repo rate - haircut/margin financing.

Specifically, are there any papers that have done work on the effect of volatility in the short term rates markets on this trade? Given that LIBOR/GC repo rate are components in the trade’s return, are attempts ever made, in practice, to hedge the volatility in these components?

A)Think an event like the repo spike in sept 2019.

B)Or, hypothetically speaking, if you are doing this trade in the long dated treasury space (30y), could increasing UST vol, for example have an indirect effect on the trade by increasing haircuts?

C)Moreover, is it plausible to suggest that a severe steepening in the cash curve would not be accompanied by a widening in swap spreads. What is the dynamic observed in practice? Post 2019, When the cash curve steepens, do swap spreads tend to widen or tighten?

Answer 1

"are attempts ever made, in practice, to hedge the volatility in these components?"

You could try to fund the position via term repo rather than O/N which means no vol in your UST funding rate.

if you are doing this trade in the long dated treasury space (30y), could increasing UST vol, for example have an indirect effect on the trade by increasing haircuts?

My guess would be that the financing provider would build in some margin of error (excuse the pun) to the haircut so you'd need a significant pick-up in vol for the haircut to increase. But certainly possible and if you remember back to the GFC, there were meaningful changes in margin requirements (SI and CL spring to mind) at short notice which, at the time, were considered possible catalysts for position liquidations.