The Trade is:

  1. You have USD 100m funding
  2. Swap USD for YEN equivalent at today's spot, agree to swap back in 12 months at the USD/JPY forward rate
  3. With the YEN buy a 12 months Japanese Government bond
  4. In 12 months, the JGB matures, get the YEN proceeds from the principal repayment, give the YEN back to the FX swap counterparty, get us your USD back.

Here is a numerical example of a practical implementation. The deal is price each month for 12 month until it unwinds:

Arbitrage trade 1 with pricing details and calculations

As you can see in col V the swap makes USD 12 million, the FX revaluation of the bond (which is held at historical cost, so no MtM for the bond in this deal apart from the FX revaluation of the purchase cost) makes USD -7.4m, and the total is a profit of USD 4.9m. This is a yield of 4.9% on the initial $100m of funding we had. In the meantime when the trade was placed, USD 1y OIS was at 5.4% and YEN 1y OIS was at 0.52%. So the yield differential was at 4.88%. Very close to the 4.91% obtained. For full disclosure I am using JYSO Index and USSO Index for the OIS rates.

This led me to believe that by doing this arbitrage and holding to maturity, one can simply use the bond as a hedge for FX risk, and capture what seems to be the yield in the FX swap cash flows.

However here is another deal at a later time in history, and as you see the logic is broken:

Arbitrage trade 1 with pricing details and calculations

Now this trade yield 1.61%, whilst I could have lent $ at 0.47% for a yield and borrowed yen at -0.14%. The numbers don't add up anymore.

What is happening here? Sometimes this lets lend USD at higher rate than the US rate, and sometimes at a lower rate. What is the PnL of this trade actually capturing?

  • $\begingroup$ Without any actual data, is it because you have coincidentally picked one trade (out of 1000s) that has had a trend for a consistent period of time and you have (in sample) chosen to be on the right side? Here's another... From 2007, receive a 5y IRS every month for the following 10 years. Probably well in the money. $\endgroup$
    – Attack68
    Nov 10, 2023 at 20:56
  • $\begingroup$ Thank you @Attack68, I realised my question was not clear so I rephrased and added calculations examples above. $\endgroup$
    – tweedi
    Nov 13, 2023 at 23:17


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.