I have a mathematical / theoretical question regarding the PV of an Overnight Index Swap (Floating Side) at the time of fixing.

Starting from this question:

How to compute Overnight Index Swap (OIS) fixed rate?

--> At each Fixing Date of the Floating Cash Flows will be the Floater PV at par? Similar to a Floating Rate Note? If I look at the math I would guess that you don't have the same effect (that counter and denominator are the same).

EDIT: To be more precise: A Floating Rate Note will be at time of fixing pricing at par (Cash Flow Term and Discount Term cancel each other out of the equation). The same effect doesn't hold for an Overnight Index Swap - Floating Side (because calculation method of the interest is in arrears and a compounded geometrical mean)?

Thank you very much in advance.

Best Regards.


Let's assume that the collateral rate on cash equals the overnight rate, that we have a schematic (lined/tiled up accrual periods and payments dates) strip of dates/times $T_0<T_1<\ldots <T_n$, accrual factor $\tau_t := \tau(t-1,t)$, and $c_t$ collateral rate at $t$ (overnight $t-1$ to $t$).

The floating coupon is then:

$$ \prod_{s=T_{i-1}}^{T_i}\left(1+\tau_sc_s \right) -1. $$

Let's further assume that we can live with approximating daily compounding by continuous compounding:

$$ \prod_{s=T_{i-1}}^{T_i}\left(1+\tau_sc_s \right) -1 \approx \mathrm{e}^{\int_{T{i-1}}^{T_i}c_sds} -1. $$

Then the time-$0$ present value of this strip of floating coupons is:

$$\sum_{i=1}^n \mathbf{E}^Q\left[\mathrm{e}^{-\int_{0}^{T_i}c_sds} \left(\mathrm{e}^{\int_{T{i-1}}^{T_i}c_sds} -1 \right)\right] = \sum_{i=1}^n \left( \mathbf{E}^Q\left[\mathrm{e}^{\int_{0}^{T_{i-1}}c_sds}\right] - \mathbf{E}^Q\left[\mathrm{e}^{\int_{0}^{T_{i}}c_sds}\right]\right) $$ $$ = \mathbf{E}^Q\left[\mathrm{e}^{\int_{0}^{T_{0}}c_sds}\right] - \mathbf{E}^Q\left[\mathrm{e}^{\int_{0}^{T_{n}}c_sds}\right], $$ that is, the difference of collateralized discount factors at stub time and last payment time (under assumptions made, we do have the 'telescopic' effect that makes FRN's 'at par').

Note: Let current time be $T_j$ (we are inside the strip timeline, not before it; $j\geq 1$). Under assumptions above, $T_j$ is also the fixing date (or rather the publishing date of the compounded index based on already fixed overnight rates) of the value of the $j$-th floating coupon. The current PV of the residual floating coupon strip will be:

$$\sum_{i=j+1}^n \mathbf{E}^Q\left[\mathrm{e}^{-\int_{T_j}^{T_i}c_sds} \left(\mathrm{e}^{\int_{T{i-1}}^{T_i}c_sds} -1 \right)\right] = \sum_{i=j+1}^n \left( \mathbf{E}^Q\left[\mathrm{e}^{\int_{T_j}^{T_{i-1}}c_sds}\right] - \mathbf{E}^Q\left[\mathrm{e}^{\int_{T_j}^{T_{i}}c_sds}\right]\right) $$ $$ = 1 - \mathbf{E}^Q\left[\mathrm{e}^{\int_{T_j}^{T_{n}}c_sds}\right]. $$

Note 2: If this strip of floating coupons were part of an FRN, we would add one extra cash flow to it at $T_n$ consisting of the reimbursement of the principal (set to $1$ here) of the note. So the PV of the extended strip would then show the strip being 'at par': $$ 1 - \mathbf{E}^Q\left[\mathrm{e}^{\int_{T_j}^{T_{n}}c_sds}\right] + \mathbf{E}^Q\left[\mathrm{e}^{\int_{T_j}^{T_{n}}c_sds} \cdot 1\right] =1. $$

Note 3: Under the same assumptions, time-$0$ par swap rate is then:

$$ K = \frac{P^{ois}(0,T_0) - P^{ois}(0,T_n)}{\sum_{i=1}^n \delta_i P^{ois}(0,T_i)},$$

where $P^{ois}(0,T):= \mathbf{E}^Q\left[\mathrm{e}^{-\int_{0}^{T}c_sds}\right]$, $\delta_i=\tau(T_{i-1},T_i)$.

  • $\begingroup$ Hi, thank you very much for your work. It was not my intention to generate such a workload. $\endgroup$ – DV01_KRD Aug 9 '20 at 11:15
  • $\begingroup$ No problem. I hope it helped. Does it answer your question? $\endgroup$ – ir7 Aug 9 '20 at 12:36
  • 1
    $\begingroup$ @rvignolo Basically, what I'm showing is that the par swap rate can be written as a difference of two discount factors (our expectations) divided by annuity, just like we have for Libor (note that we have a single curve context here; multiple discounting with collateral awareness is a separate discussion). So, bootstrapping can be used. There are further insights on the type of expectations we use above (measure play) in Mercurio et al recent paper risk.net/cutting-edge/banking/6732471/… $\endgroup$ – ir7 Sep 11 '20 at 18:17
  • 1
    $\begingroup$ @rvignolo Added an edit on the par swap rate mentioned above. $\endgroup$ – ir7 Sep 12 '20 at 21:41
  • 1
    $\begingroup$ @rvignolo RFR rates, aka IBOR replacement, is a very hot subject these days. I recommend you make it a question on the Stack Exchange (make sure you reference the paper). If I don’t get to it, others, perhaps more involved in this subject, will get to it. (Btw, good questions are as valuable as good answers on SE.) $\endgroup$ – ir7 Sep 12 '20 at 23:32

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.