The swap market Bergomi model
The combination of two popular volatility models sharpens the hedging of exotic rate derivatives
Kenjiro Oya builds a forward variance model for the co-terminal swap market model. With the model, the position management of exotic
interest rate products, eg, Bermudan swaptions, can be performed in a more sophisticated and systematic manner
It is common practice to hedge the volatility exposure of exotic derivatives products with vanilla options. However, the methodology behind volatility exposure hedging has not yet been fully established. The difficulty lies in the fact that the direct modelling of vanilla option prices or implied volatilities is technically challenging (Bergomi 2015). Therefore, practitioners often deal with the issue by calibrating diffusion parameters. These parameters, which are assumed to be constant with regard to model dynamics, are adjusted on a regular basis so the model can reproduce the market prices of vanilla options. However, diffusion parameters calibrated at different times will be inconsistent. Consequently, in a profit-and-loss (P&L) analysis of a derivatives contract, the P&L will contain an additional contribution from the change in diffusion parameters, which is considered difficult to manage. In particular, when the P&L results have an unexpected trend, knowing how we should modify the model assumptions is not straightforward.
In equity modelling, a promising approach to dealing with the issue is the forward variance model, introduced in Bergomi (2005), for which a forward variance curve is considered to be a model state variable. For this model, the market rate changes are understood as model state variable variations, and the calibration of model parameters is not required. This feature makes the P&L formula – in terms of market observables – quite simple and the risk management of the derivatives contract more comfortable. However, to the best of the author’s knowledge, an equivalent approach has not been presented for interest rate modelling. This might be because there is no liquidity in the variance swaps of interest rates. In reality, the forward variance model can still be useful if forward variance curves are computed using the market prices of vanilla options (Bergomi 2015). In this article, we build a forward variance model for the co-terminal swap market model such that (1) all the market price changes of our hedging instruments are interpreted as state variable changes, (2) the model has flexible parameters that are solely used for controlling the model dynamics and (3) the P&L formula becomes quite simple so we can easily understand the reason behind a material P&L trend when it appears before we consider how to modify the model parameters. With the model, the position management of exotic interest rate products, eg, Bermudan swaptions, can be performed in a more sophisticated and systematic manner.
The swap market Bergomi model
In this section, we first review the forward variance model (the Bergomi stochastic volatility model) introduced in Bergomi (2005). Next, we consider a variance swap contract on a swap rate, via which we discuss how to apply the forward variance modelling approach to the swap market model.
A forward variance curve and the Bergomi stochastic volatility model
The Bergomi stochastic volatility model uses a forward variance curve as its modelling object. A variance swap is a contract that pays the realised variance of the log return of a tradable asset less a strike rate at maturity . We assume that the payoff at is given as:
where is set at to make the value of the variance swap zero. Namely, must satisfy:
(1) |
is the -conditional expectation under the risk-neutral measure. In this subsection, we assume the risk-free rate process to be deterministic.
A discrete forward variance swap rate is defined (using ) as:
We can confirm that is a martingale under the risk-neutral measure; for , we obtain:
An infinitesimal forward variance swap rate is obtained by taking limit for ; .
A continuous Bergomi model is specified by assuming lognormal dynamics for infinitesimal forward variance swap rates in a forward variance curve :
(2) |
where is a static model parameter, is a Brownian motion under the risk-neutral measure and is the model terminal date. Note has zero risk-neutral drift because the forward variance swap rate is a martingale under the risk-neutral measure. The underlying process follows:
(3) |
where is a Brownian motion under the risk-neutral measure. With (1) and (3), we obtain . Thus, we get .
The Bergomi model assumes and to be model state variables. Consider that we are managing a derivatives contract using the underlying asset and the infinitesimal forward variance swaps on as our hedging instruments. We denote the value of the derivatives contract by . The pricing equation is written as:
(4) |
with the correlation functions:
Consider a hedged contract that is a portfolio of a unit of the derivatives contract , a bank account , the underlying asset and the forward variance swaps on . We require to satisfy , , and . Then, the P&L formula of the hedged contract is computed using (4) as:
(5) |
Equation (5) provides the term-wise breakeven condition for the P&L of . Therefore, when a non-negligible P&L trend appears, we can easily understand which term causes it and how we should modify the model assumptions. Note all of the hedging instrument price changes can be interpreted as variations of the state variables; thus, we can use the model parameters , and solely to control the breakeven condition.
Remark 1
In this article, we consider the risk management of a fully hedged contract of an exotic derivatives product. In practice, however, an exotic derivatives product is not always fully hedged. For such a case, we regard the partially hedged contract as the portfolio of the fully hedged contract and the hedging instruments, and we leave the risk management of the hedging instruments to vanilla models, which are beyond the scope of this article.
Setup
Before proceeding to our discussion on the application of the Bergomi model to interest rate modelling, let us define the basic variables first. Consider the discrete time grids , , with accrual factors , where is the terminal date of the model. Denote by the discount factor at time with maturity date . We denote a continuous bank account process by , where is a risk-free rate process. Swap rates and associated annuity factors are given as:
respectively. We denote by when no confusion can arise, and we assume empty sums denote zero while empty products denote 1.
A variance swap contract on an interest rate
Here, we introduce an interest rate variance swap contract such that the variance swap rate will be a martingale under the associated annuity measure. Owing to the martingale property, we can build a two-factor model of a swap rate and a forward variance curve under the associated annuity measure. In addition, we are able to compute a European swaption price using the two-factor model. As we will see later, this simplifies the relationship between the forward variance curve and the market European swaption price, and the P&L interpretation in terms of the market observables becomes clear.
Consider a contract that, at , pays the sum of quadratic variation of a swap rate for the period , where . The quadratic variation is multiplied by at the end of each observation time grid and rescaled with the factor ; the payoff at is given as:
Also consider a variance swap on a swap rate that pays and receives at , where is fixed at so the value of the contract is zero. This means:
(6) |
where and denote the -conditional expectation under a terminal measure and an annuity measure, respectively. From (6), we obtain:
(7) |
This indicates is a martingale under the annuity measure. For , we get:
The swap market Bergomi model
Next, we consider the dynamics of a swap rate process . Because is a martingale and, hence, driftless under the annuity measure, we assume:
(8) |
where is a stochastic process and is a Brownian motion under the annuity measure. We define an infinitesimal variance swap rate as . This leads to:
(9) |
Thus, we get .
We assume lognormal dynamics for , which is a martingale under the measure:
(10) |
where and are static model parameters.
In this article, we assume a variance curve is driven by a single Brownian motion for simplicity. The extension to the multi-factor setting is straightforward. In order to obtain a low-dimensional Markov representation, we introduce the Ornstein-Uhlenbeck (OU) state variable:
With , the dynamics of is given as:
(11) | ||||
(12) |
The co-terminal swap market Bergomi model
In the last section, we built a two-factor model of a swap rate and a forward variance curve under the associated annuity measure. In order to evaluate the exotic derivatives products that depend on multiple swap rates, such as Bermudan swaptions, we need to know the joint dynamics of the swap rates and the forward variance curves. In this article, we use the co-terminal swap market model (Jamshidian 1997). Using this model, we discuss the second-order P&L formula and breakeven levels for a derivatives contract with a hedge portfolio. Then, we analyse the P&L formula using the factor reduction method.
Model dynamics
The swap market model is classified by the underlying swap rates to be modelled (Jamshidian 1997; Oya 2018). In this article, we consider the co-terminal swap market model for which the yield curve dynamics are determined by modelling swap rates that share a common terminal date (Jamshidian 1997). We will work with a terminal measure. To specify the co-terminal swap market model, we omit the end index for a swap for ease of notation, ie, . We introduce the state variable vector as:
with , .
The dynamics of under the terminal measure are given by adding no-arbitrage drifts to (8) and (12):
(13) | ||||
(14) |
where are the Brownian motions under the terminal measure, defined as:
We define the correlation functions as .
Proof.
Compute the no-arbitrage drift using the standard measure-change technique as:
∎
Forward variance curve computation with swaption market prices
In the previous subsection, we formulated the model dynamics using the state variable vector . Remember the elements of the state vector are directly related to market observables. A forward swap rate is evaluated from spot starting swap rates, while is computed from a variance swap rate using (11). In reality, the interest rate variance swap contract is illiquid in the market; thus, we use European swaptions as our hedging instruments instead. For this purpose, we discuss how to compute using the market price of a European swaption.
We use the different computation procedures for and . At , the initial forward variance curve is computed in order for the model to reproduce the market prices of European swaptions assuming . For , European swaption market price changes are reflected in the state variable . In other words, the model parameter calibration to market prices of European swaptions is performed only at . After that, the market price changes are interpreted as model state variables variations.
The initial forward variance curve is parameterised as . is the model parameter for the control of the curve shape of . is calibrated to a European swaption market price at and then fixed for .
Consider the computation of at . We will obtain so the model reproduces the market price of a European swaption on with strike using the efficient numerical computation scheme presented in Bergomi (2015). Here, we use a two-factor model for a swap rate and a state variable under the associated annuity measure for the computation of the model swaption price. Denote by the expiry time of a swaption on . Note becomes driftless under the annuity measure:
(16) | ||||
(17) |
where is a Brownian motion that is independent of .
Next, let us look at a payer swaption price on a swap rate :
where we use the tower rule. Because and are independent, conditioned on a path for follows a normal distribution with the following moments:
(18) | |||
(19) |
Thus, can be calculated analytically using the normal Black-Scholes formula as with:
where is the standard cumulative normal distribution function and . Then, the swaption price is obtained as .
We assume the market swaption price for a European swaption on as of is given as a normal implied volatility . We require the model price of the swaption to be equal to the market price:
(20) |
Equation (20) is solved using a one-dimensional root-finding algorithm in terms of for and for .
Remark 2
Owing to the martingale property of under the annuity measure, the expectation depends on the dynamics of and but not on or . As a consequence, the state variable is related to market observables in a simple manner; namely, (and, equivalently, the forward variance curve ) relies on and but not on or .
Once (20) is solved, the first-order sensitivities of with regard to and are computed as follows:
(21) | ||||
(22) |
The P&L formula
Here, we analyse the P&L formula with the co-terminal swap market Bergomi model (SMBM) and discuss how we could adjust our model parameters using the results of the P&L analysis. The derivation of the P&L formula proceeds in the standard manner, and we obtain a well-known form of the P&L formula. The novelty of the co-terminal SMBM approach is that this theoretical P&L formula can be used for risk management when both interest rate swaps and European swaptions are used as hedging instruments. On the other hand, the existing interest rate models need to rely on calibration procedures to reproduce the market prices of hedging European swaptions in such a context. Consequently, the theoretical P&L formula does not hold for the P&L number computed as the price difference of two different model parameter assumptions.
Assume we hold a derivatives contract and the value of that contract is denoted by . The pricing equation of is given as:
(23) |
where is the no-arbitrage drift of under the risk-neutral measure. We do not provide the explicit form of here because it is irrelevant to the discussion of a hedged contract. is given as:
Next, we consider a hedged contract for the derivatives contract:
(24) |
where are the hedge weights. The prices of hedging instruments are given as:
(25) |
Namely, we hedge the derivatives contract with interest rate swaps, payer European swaptions of strike and a bank account . We build the hedged contract to satisfy the equations below:
(26) | ||||
(27) |
The hedging conditions (26) and (27) are given as a linear system and solved using standard linear algebra.
Now, we consider the second-order P&L formula for the hedged contract . Using a Taylor expansion of order 2 in , we obtain:
(28) |
Equation (28) indicates the second-order P&L is given as the sum of the differences of the realised quadratic cross variations and the breakeven level multiplied by the gamma term , as in (5). Therefore, we can carry out a term-wise analysis in the same manner as for the Bergomi model. The calibration procedure is not required for the co-terminal SMBM, and we can use the model parameters solely to control the breakeven levels. Let us summarise the model parameters as follows:
- •
determines the shape of the initial forward variance curve , which affects , which appears in the breakeven level expression as for .
- •
controls the lognormal volatility of . In terms of the state variable, controls the scale of . Because always holds for , the control of the scale of is equivalent to the control of the breakeven levels for .
- •
controls the correlations between state variables.
Factor analysis of the P&L
Because an interest rate model often contains a large number of state variables, practitioners frequently use this model driven by a limited number of Brownian motions to reduce its complexity. This approach is also useful with the co-terminal SMBM to aid in intuitively understanding the economy of the product. In this article, we consider as an example the co-terminal SMBM for which swap rates and forward variance curves are driven by three Brownian motions:
(29) |
where we parameterise as , .11 1 In practical applications, the number of Brownian motions and coefficients is determined based on historical analysis of market movements (eg, principal component analysis) in order to make the P&L contribution from the second term of (33) small enough. We assume the correlations are given as for , with .
We consider the P&L formula associated with the above Brownian motion changes. Define the value function of a derivatives contract parameterised with reduced factors as , with and . is given as:
(30) |
With , we can write the theta term for the hedged contract in (28) as:
(31) |
where . Also, the gamma term in (28) becomes:
(32) |
where is defined so the -norm of is minimised:
With (31) and (32), we obtain the second-order P&L formula with reduced factors:
(33) |
If the model with these reduced factors is able to reproduce the realised dynamics of the state variables well, the contribution from the second term of (33) will be negligible compared with the first term. If this is not the case, we should review the assumption of the factor reduction. Here, we assume the reduced factors can reproduce the realised dynamics well and the P&L can be explained accurately enough using the first term of (33). Then, to understand the P&L of the derivatives contract, we should check which term of is large; the breakeven level is reasonable compared with the realised quadratic cross-variation in the same manner as standard P&L analysis. The benefit of factor reduction is that lesser P&L component terms are involved in the P&L formula; we only need to look at six terms in our example.
A numerical experiment with the co-terminal SMBM
Bermudan swaptions have been traded for a long time and are one of the most popular exotic interest rate products. However, position management of the product is still a challenging task. In this section, we perform a numerical experiment for a Canary swaption, which is a Bermudan swaption with only two exercise dates, and illustrate how the P&L analysis can be performed in a systematic manner using the co-terminal SMBM.
Canary swaption pricing
We assume the two interest rate swaps underlying a Canary swaption have a common terminal date . The price of a receiver Canary swaption is given below:
(34) |
where , , are the time indexes for the start time of the underlying swap rates, and is the strike. Because is the conditional expectation at , the exact simulation requires a nested Monte Carlo method that takes a huge amount of valuation time. Therefore, in practice, an approximation method is often applied for the valuation of Bermudan swaptions such as the least square Monte Carlo method. Here, in order to see the model behaviour with fewer numerical errors, we use the semi-nested Monte Carlo method described below:
- •
We sample for a reasonably wide range of values. For each sample of , we perform a Monte Carlo simulation to compute and calculate the implied normal volatility . Then, we build a spline function using the samples.
- •
In a Monte Carlo simulation for pricing, we diffuse the model up to , and we obtain as . Then, we evaluate using a normal Black-Scholes formula with , and .
We use the reduced-factor model (29) with common parameters among the swap rates , and . The correlation functions are parameterised as , , , and:
Monte Carlo simulations are performed using the quasi-Monte Carlo method with the number of paths being .
Numerical exercise of P&L analysis
The aim of the P&L analysis is to understand the root cause of non-flat P&L when it appears and to have an idea of how to improve the P&L behaviour. Here, we provide a numerical example of the P&L analysis for a hedged Canary swaption , which is the price of the hedged contract (24)–(25) of given in (34). We parameterise as:
using and defined in the section titled ‘Factor analysis of the P&L’. Remember that we build the hedge portfolio as follows. We compute the sensitivities of to all the elements of the state vector , and then evaluate the hedge weights so the price of the hedged contract should satisfy the hedging condition (26)–(27). The hedge weights are computed at and remain unchanged for . We assume the time difference is and the state variable differences are fully expressed with the reduced factors, namely with .
We analyse the realised P&L using the below P&L approximation formula, which is obtained from (33) by omitting the second term:
(35) |
The analysis proceeds as follows. First, we evaluate the price of a hedged Canary swaption using two different timings and market rates, which are and . We then compute the realised P&L as the difference of the prices, . Second, we compute the residual term as:
and confirm the term is small enough. If this is the case, we now have a high-precision decomposition of the P&L into the components:
Next, we check which P&L component contributes to the non-flat P&L the most. The non-flatness means the realised quadratic variation is materially different from the model-implied breakeven level for that component. Finally, with this information, we can update the model assumptions based on a historical analysis of for better P&L behaviour.
Table B shows the analysis results. First, we note the materially non-flat realised P&L of basis points (bp per ). Second, we can confirm the approximation (35) worked fine; out of the total realised P&L of bp, bp was explained by the right-hand side of (35). Third, we can clearly understand why the non-zero realised P&L appeared: the covariance dynamics of and for are in disagreement between the model assumptions and the realised numbers , so these P&L components contributed most to the non-zero realised P&L. If we perform the P&L analysis for a larger time window and a similar disagreement between model assumptions and realised numbers in the covariance dynamics persists, we may do better to change the model parameters for these covariance pairs. For example, to make the P&L of the component close to flat, we need to assume larger values for , which means we are required to decrease and then re-compute .
1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | |
---|---|---|---|---|---|---|---|---|---|
2.53% | 2.57% | 2.57% | 2.57% | 2.58% | 2.59% | 2.60% | 2.60% | 2.62% | |
0.658% | 0.698% | 0.718% | 0.729% | 0.739% | 0.740% | 0.739% | 0.736% | 0.724% |
A | B | C | D | |
(1,1) | 0.765% | 0.55e4 | 0.38e4 | 0.17e4 |
(2,2) | 0.124% | 0.04e4 | 0.06e4 | 0.02e4 |
(3,3) | 0.011% | 0.00e4 | 0.01e4 | 0.00e4 |
(1,2),(2,1) | 0.199% | 0.10e4 | 0.00e4 | 0.10e4 |
(1,3),(3,1) | 0.084% | 0.04e4 | 0.01e4 | 0.03e4 |
(2,3),(3,2) | 0.017% | 0.01e4 | 0.00e4 | 0.01e4 |
0.80e4 | 0.34e4 | 0.45e4 | ||
Realised P&L | 0.48e4 | |||
Conclusion
In this article, we applied the forward variance modelling approach to the co-terminal swap market model. The model is advantageous in that diffusion parameter calibration is not required to take the market price changes of the hedging instruments into account. As a result, the P&L formula of a hedged contract becomes quite simple. We numerically illustrated how the P&L analysis was performed with the co-terminal SMBM for a Canary swaption and confirmed we could clearly understand the P&L and easily determine which model parameter should be changed to cope with non-flat P&L.
Kenjiro Oya is an executive director at Nomura in Tokyo. He sincerely thanks Kei Minakuchi, Paul McCloud and the anonymous referees for their thoughtful suggestions, which greatly contributed to the improvement of this article. The opinions expressed in this article are the author’s own and do not reflect the view of Nomura Securities Co., Ltd. All errors are the author’s responsibility. Email: kenjiro.oya@nomura.com
References
- Bergomi L, 2005
Smile dynamics II
Risk October, pages 67–73 - Bergomi L, 2015
Stochastic Volatility Modeling
CRC Press - Jamshidian F, 1997
Libor and swap market models and measures
Finance and Stochastics 1(4), pages 293–330 - Oya K, 2018
The swap market model with local stochastic volatility
Risk July, pages 82–87
Only users who have a paid subscription or are part of a corporate subscription are able to print or copy content.
To access these options, along with all other subscription benefits, please contact info@risk.net or view our subscription options here: http://subscriptions.risk.net/subscribe
You are currently unable to print this content. Please contact info@risk.net to find out more.
You are currently unable to copy this content. Please contact info@risk.net to find out more.
Copyright Infopro Digital Limited. All rights reserved.
You may share this content using our article tools. Printing this content is for the sole use of the Authorised User (named subscriber), as outlined in our terms and conditions - https://www.infopro-insight.com/terms-conditions/insight-subscriptions/
If you would like to purchase additional rights please email info@risk.net
Copyright Infopro Digital Limited. All rights reserved.
You may share this content using our article tools. Copying this content is for the sole use of the Authorised User (named subscriber), as outlined in our terms and conditions - https://www.infopro-insight.com/terms-conditions/insight-subscriptions/
If you would like to purchase additional rights please email info@risk.net
More on Banking
Volatility shape-shifters: arbitrage-free shaping of implied volatility surfaces
Manipulating implied volatility surfaces using optimal transport theory has several applications
Infrequent MtM reduces neither value-at-risk nor backtesting exceptions
Frequency of repricing impacts volatility and correlation measures
SABR convexity adjustment for an arithmetic average RFR swap
A model-independent convexity adjustment for interest rate swaps is introduced
Joint S&P 500/VIX smile calibration in discrete and continuous time
An arbitrage-free model for exotic options that captures smiles and futures is presented
The carbon equivalence principle: minimising the cost to carbon net zero
A method to align incentives with sustainability in financial markets is introduced
Leveraged wrong-way risk
A model to assess the exposure to leveraged and collateralised counterparties is presented
Neural joint S&P 500/VIX smile calibration
A one-factor stochastic local volatility model can solve the joint calibration problem
Most read
- Breaking out of the cells: banks’ long goodbye to spreadsheets
- Too soon to say good riddance to banks’ public enemy number one
- Industry calls for major rethink of Basel III rules