�ݺ�ߣ

International Journal of Business Marketing and Management (IJBMM)
Volume 8 Issue 2 Mar-Apr 2023, P.P. 01-08
ISSN: 2456-4559
www.ijbmm.com
International Journal of Business Marketing and Management (IJBMM) Page 1
Research on Credit Default Swaps Pricing Under
Uncertainty in the Distribution of Default Recovery Rates
Yuetong Fan1
, Peibiao Zhao*1
1
School of Mathematics & Statistics, Nanjing University of Science and Technology, China
Email: 2863175374@;pbzhao@njust.edu.cn
Abstract: In this paper, we construct a Credit Default Swap pricing model for default recovery rates under
distributional uncertainty based on a structured pricing model and distributional uncertainty theory. The model
is algorithmically transformed into a solvable semi-definite programming problem using the Lagrangian dual
method, and the solution of the model is given using the projection interior point method. Finally, an empirical
analysis is conducted, and the results show that the model constructed in this paper is reasonable and efficient.
Keywords: Credit Default Swap, Default recovery rates, Distribution uncertainty, Structured pricing model
I. Introduction
In recent years, Credit Default Swaps(CDS) have become one of the most active financial derivatives
in the financial markets because of their good characteristics, and the pricing of Credit Default Swaps is one of
the hot and difficult issues in the field of financial economics. Over the past 20 years or so, researchers have
focused on solving for default probabilities and have obtained many important results, however, during the 2008
financial crisis, researchers found that default recovery rates have a profound impact on the pricing models of
Credit Default Swaps. There are three general approaches to default recovery rates in the current industry: (1)
Recovery of Face Value where creditors are compensated for a percentage of the value of the bond’s face value
following a debtor’s default.(2)Recovery of Treasury i.e. after an event of default, the creditor is able to be
compensated for a percentage of the value of a default-free Treasury bond that is equivalent to the bond, this
percentage is usually set at 0.4. (3) Recovery of Market Value i.e. the default recovery is assumed to be a
fraction of the market value of the bond prior to the event of default.
Such an approach, while easy, is unrealistic and can have a significant impact on the pricing model for
credit derivatives.So in recent years some scholars have begun to model the distribution of default recovery
rates, initially with Frye (2000) [1]
who fitted the default recovery rate with a normal distribution.Pykhtin (2003)
[2]
improved Frye’s model by applying a log-normal distribution to describe the default recovery rate.Rosch
(2005) [3]
proposed fitting the default recovery rate with a Logit-normal distribution. Subsequent research by
Moody showed that real-life default recovery rates do not exhibit a simple single-peaked distribution, but rather
a bimodal state, with recovery rates either in the vicinity of b eighty percent or in the vicinity of twenty percent,
suggesting that the credit default swap prices obtained by choosing the mean of the default recovery rates in the
model have a large deviation from the actual prices.In this way, the Beta distribution is more suitable for fitting
the default recovery rate because it takes values in [0,1] and can make the probability density function curve
show a double-peaked pattern by adjusting the two parameters in the density function, and later scholars mostly
use this as the basis for modelling the default recovery rate,Chen(1999)[4]
used the Beta distribution in
constructing his credit risk model to fit the default recovery rate.Some scholars have also used the kernel
function in non-parametric estimation to fit the default recovery.Brown (2004) [5]
used the Beta-Bernstein
polynomial smoothing technique to construct a smoothing kernelto fit the recovery density curve.
The Distributional Uncertainty Method is a method for dealing with parameters with uncertainty,
which is solved using optimisation theory by constructing an uncertainty set containing all possible distributions
of the parameters, constraining the uncertainty set, and then transforming the original problem into a robust
problem corresponding to the uncertainty set.One of the earliest uses of uncertain stochastic models was by
Scarf (1958) [6]
who used a distribution uncertainty model to solve inventory control problems where only the
mean and variance of demand were known. Delage and Ye (2010)[7]
constructed distribution uncertainty sets for
mean vectors and covariance matrices from historical data to study the loss function in best worst type
* Corresponding author

Research on Credit Default Swaps pricing under Uncertainty in the Distribution of Default …
distribution robust optimization problems.Calafiore (2006)[8]
studied the transformation of the problem into a
quadratic cone constraint under conditions where only the mean and variance of the parameters are known,
solving the linear chance control problem under conditions of uncertainty in the distribution.
This paper constructs a Credit Default Swap pricing model for default recovery rates under
distributional uncertainty, mainly based on structured models and distributional uncertainty theory.
Algorithmically, the Lagrangian dual method is used to transform the distributional robust chance constrained
model into a solvable semi-definite programming problem, and the interior point projection method is used to
give a solution, which has certain theoretical and practical significance.
The paper is structured as follows. In the second part, a model for pricing Credit Default Swaps under
uncertainty in the distribution of default recovery rates is constructed based on a structured model. The third part
is the empirical analysis, where we use Matlab to numerically calculate the price of Credit Default Swaps under
uncertainty in the distribution of default recovery rates based on real-life cases of Credit Default Swaps and
compare it with the actual price to verify the validity of the model. In the last section, we summarize the theories
and models covered in this paper and give directions for further in-depth research on the Credit Default Swap
problem based on distribution uncertainty.
II. Model formulation
In a credit default swap transaction, the purchaser of the default swap must pay a periodic fee (known
as the credit default swap spread) to the seller of the default swap, usually at the end of every quarter, every six
months or every year. In the event of a credit-type event (e.g. the bond host is unable to pay), the purchaser of
the default swap has the right to demand full or partial compensation for losses from the seller of the default
swap, and if no credit event occurs during the life of the contract, the seller of the default swap does not have to
pay any money to the purchaser of the default swap and the contract terminates.
From the above description of the rationale for trading single-asset credit default swaps, we have the
following pricing model: Consider a single-asset credit default swap contract with an underlying bond of face
value V.The following are the necessary parameters we have defined.
(1)The expected recovery rate of the bond in the event of a default by the company is: R.
(2)The risk-free rate in the market is: r.
(3)The density of the firm’s probability of default at any moment t during the term of the bond contract is:
q(t).
(4)The annual premium to be paid by the purchaser of a default swap to the seller is: s
(5)The maturity date of the credit default swap contract is: T.
(6)The time of default of the enterprise occurs at the moment when: .
Assuming that premiums are paid quarterly, the premium payment moment is
, so the amount of each premium payment is .At the moment of default , the
present value of all premiums payable by the purchaser of the default swap to the seller of the default swap is
The expectation of the present value of premiums paid by purchasers of default swaps is
At the same time, the present value of the payout to the purchaser of the default swap upon the occurrence
of an event of default at time τ is ,Then the expectation of the present value of the payout is
The pricing problem for credit default swaps refers to finding a "fair" price for the current contract before
the underlying bond defaults, so our objective is to match the payout at default as closely as possible to the
present value of the premium paid by the purchaser of the contract, i.e. to minimise the hedge. Therefore, the
objective function in this paper is based on the minimisation of hedging as follows

Because of the excellent characteristics of the bimodal Beta distribution, most academics currently use it to
fit the default recovery rate, but comparative studies have found that the Beta distribution is very sensitive to
parameters, and the robustness of the model based on the Beta distribution is poor. The true default recovery
rate is not far from the value obtained from the Beta distribution. Therefore, we use the default recovery rate
based on the Beta distribution as the criterion to make a CDS pricing model under the uncertainty of the default
recovery rate distribution.The following assumptions are made first.
Definition 1. Random variable R in the asymmetric uncertainty set
Here
 T
N
R
R
N









 

]
i
[
1
i
]
i
[
)
(
1
1



N
R
N 1
i
]
i
[
1


The parameters , control the size of the uncertainty set and the degree of uncertainty We make the
following chance constraints.
where is the default recovery rate estimated from the bimodal Beta distribution, b is the degree of
investor distrust in the bimodal Beta distribution, and is the probability of safety, given in advance by the
investor.
For the chance constraint (2.8), the following proposition can be equivalently transformed into the form of
a matrix inequality, and in combination with the objective function (2.4) , the initial distributionally robust
optimization model with an uncertain set of uncertain parameters can eventually be transformed into a
computationally solvable semi-definite programming problem. Compared to traditional pricing models, this is
more realistic and more in line with the actual characteristics of the market.
Theorem 2.1. The chance constraint with the set as an uncertain set is equivalent to the set of
inequalities in the following equation.
Proof. The default recovery rate R may be a discrete random variable or a continuous random variable, which is
treated as a continuous random variable in this paper in order not to lose generality.
For ease of analysis, we split the uncertainty set .We defines
and defines a family of distributions as
Thus for the left-hand side of the chance constraint inequality (2.8) there is an equivalent form as follows.

For and its constraints, the following Lagrangian dual form can be
obtained.
Further organizing (2.13) yields
where , , are Lagrange multipliers. Given , , to obtain the minimum of the inner function, the
product function in equation (2.14) must be non-negative, otherwise the minimum will be taken to negative
infinity, so for any random variable R we have
After sorting, equation (2.15) is transformed into the following programming problem with constraints.
(2.17)
where the feasible domain of the constraint is
Equation (2.18) can be equivalently translated into the following matrix inequality form.
Equation (2.19) can be equated as
The equivalent pairwise form to that shown in equation (2.20) is
Equation (2.22) is equivalent to such that satisfies the following condition.
(2.24)

As evidenced above, the left-hand side of the chance constraint (2.8) is equivalent to equation (2.12),For
and its constraints can be transformed by Lagrangian methods into an
optimization problem with equation (2.20) and equation (2.24)as constraints and equation (2.4) as the objective
function of the optimization problem, so that the chance constraint equation (2.8) is equivalent to equation
(2.25) below as well as equations (2.20) and (2.24).
From equation (4.16), the mean and variance under satisfy
(2.27)
that is
(2.29)
Bringing equation (2.28), (2.29) into equation (2.25), equation (4.32) is equivalent to
The Lagrangian dyadic approach, combined with the objective function (2.4), leads to the following pricing
model.
III. Empirical analysis
In this chapter, we use one credit risk mitigation warrant in the market for our empirical analysis: the
18 Origin Water cp002 Joint Credit risk Mitigation warrant.
To investigate the effect of the parameters of the uncertainty set on the price of the credit default swap
pricing model, the following parameters are defined Number.
where and denote the prices of credit default swaps under and .
Consider the parameters of the ambiguous set varying in a range of 20 %, i.e.
According to the Bootstrapping method, the estimates of and can be obtained as

Then the parameters and vary in the range and the
result is shown below
Figure 1. The effect of ambiguous set parameters on CDS prices
As can be seen from the graph, the prices of CDS are more sensitive to changes in and change faster in the
direction, correspondingly the prices of CDS are less sensitive to changes in and the slope of credit
default swap prices in the direction is significantly greater than the slope in the direction.
We use the linear matrix inequality toolkit in Matlab to solve the semi-definite programming problem shown
in equation (2.31), where we set the investor’s distrust of the bimodal Beta to 0.1, i.e. b = 0.1, the risk-free
return is set to 0.035, and the probability of investor distrust of the bimodal Beta distribution is set to 0.05, and
the results obtained are shown in Fig2.
Figure 2. The effect of parameter on the optimal strategy
The blue curve in the graph indicates the price of a robust CDS when the distribution of default recovery rates
is uncertain, while the red curve indicates the price of a CDS when default recovery rates follow a bimodal Beta
distribution. It is clear that the price of a robust credit default swap lies below the price of a credit default swap
under the bimodal Beta distribution, indicating that the bimodal Beta distribution overestimates the price of a
credit default swap, which can result in larger losses to investors in the event of a bias in the default recovery
rate.
This is next compared to the actual price, and the stock price for Origin Water from November 19, 2018 to
November 19, 2019 is chosen for this paper, and then estimates of other parameters are obtained based on
previous research, with the following results.

Measurement results
CDS prices under a bimodal Beta distribution CDS prices under distributional uncertainty Market offer
1.32 1.11 1.08
IV. Summary and prospect
In the pricing of Credit Default Swaps, the estimated value of the default recovery rate is obtained by
estimating the probability distribution assumed in advance, but in real markets, the distribution of default
recovery rates is affected by a variety of factors and investors are often unable to use the limited information
available to obtain the exact probability distribution of default recovery rates. It is based on these practical
factors that we introduce uncertain distributions into portfolios and develop a pricing model for credit default
swaps under uncertain distributions of default recovery rates. Finally, based on Matlab numerical calculations,
we solve for the prices of Credit Default Swaps under uncertain distributions in conjunction with actual cases,
giving optimal results and comparing them with actual results to verify the validity of the model.
Due to the limitations of the author’s own professional level, the problems studied in this paper still
have shortcomings and need to be improved: for example, when constructing the distribution uncertainty set,
higher-order and more refined uncertainty sets can be introduced; this paper assumes that the default probability
and default recovery rate are independent of each other for the sake of computational simplicity, and the
correlation between these two random variables can be considered; in addition, a more complex process of
enterprise asset value can be considered. For the above shortcomings, we hope that future scholars will do
further research.
Reference
[1] Frye J. Depressing recoveries[J].Risk London Risk Magazine, 2000, 13(11): 108-111.
[2] Pykhtin M.Recovery rates: unexpected recovery risk[J]. Risk London Risk Magazine, 2003, 16(8): 74-
79.
[3] Rösch D, Scheule H. A multi-factor approach for systematic default and recovery risk[J].The Journal of
Fixed Income, 2011: 117-135.
[4] Chen, Song Xi. Beta kernel estimators for density functions[J].Computational Statistics and Data
Analysis,1999,31: 131-145.
[5] Brown B M,Chen S X.On the way to recovery:A nonparametric bias free estimation of recovery rate
densities[J].Journal of Banking and Financial,2004,28(12):2915-2931.
[6] Scarf H E. A min-max solution of an inventory problem[M]. Santa Monica: Rand Corporation, 1957.
[7] Delage E, Ye Y. Distributionally robust optimization under moment uncertainty with application to
data-driven problems[J]. Operations Research, 2010, 58(3): 595-612.
[8] Calafiore G C, Ghaoui L E. On distributionally robust chance-constrained linear programs[J]. Journal
of Optimization Theory and Applications, 2006, 130(1): 1-22.
[9] Zhilin Kang, Xun Li, Zhongfei Li & Shushang Zhu (2019) Data-driven robust mean-CVaR portfolio
selection under distribution ambiguity, Quantitative Finance, 19:1, 105-121.
[10] Kang Z, Li X, Li Z, et al. Data-driven robust mean-CVaR portfolio selection under distribution
ambiguity[J]. Quantitative Finance, 2019, 19(1): 105-121.
[11] Friedman C, Sandow S. Model performance measures for expected utility maximizing investors[J].
International Journal of Theoretical and Applied Finance, 2003, 6(04): 355-401.
[12] Cox J C, Huang C F. Optimal consumption and portfolio policies when asset prices follow a diffusion
process[J]. Journal of Economic Theory, 1989, 49(1):33-83.
[13] Liang J, Zou H. Valuation of credit contingent interest rate swap with credit rating migration[J].
International Journal of Computer Mathematics, 2020, 97(12): 2546-2560.
[14] Zhang Y, Gao J, Fu Z. Valuing currency swap contracts in uncertain financial market[J]. Fuzzy
Optimization and Decision Making, 2019, 18(1): 15-35.
[15] Black F, Cox J C. Valuing corporate securities: some effects of bond indenture provisions[J]. The
Journal of Finance, 1976, 31(2): 351-367.
[16] Merton R C. On the pricing of corporate debt: The risk structure of interest rates[J]. The Journal of
Finance, 1974, 29(2): 449-470.
[17] Shimko D C, Tejima N, Van Deventer D R. The pricing of risky debt when interest rates are
stochastic[J]. The Journal of Fixed Income,1993,3(2):58-65.
[18] Wang A. The pricing of total return swap under default contagion models with jump-diffusion interest
rate risk[J]. Indian Journal of Pure and Applied Mathematics, 2020, 51(1): 361-373.
[19] Chen, L., He, S. and Zhang, S., Tight bounds for some risk measures, with applications to robust
portfolio selection. Oper. Res., 2011, 59, 847–865.

[20] Jarrow R A, Yu F. Counterparty risk and the pricing of defaultable securities[J]. The Journal of
Finance, 2001, 56(5): 1765-1799.
[21] Liang J, Zou H. Valuation of credit contingent interest rate swap with credit rating migration[J].
International Journal of Computer Mathematics, 2020, 97(12): 2546-2560.
[22] Zhang Y, Gao J, Fu Z. Valuing currency swap contracts in uncertain financial market[J]. Fuzzy
Optimization and Decision Making, 2019, 18(1): 15-35.
[23] Marple T. The social management of complex uncertainty: Central Bank similarity and crisis liquidity
swaps at the Federal Reserve[J]. The Review of International Organizations, 2021, 16(2): 377-401.
[24] Jansen J, Das S R, Fabozzi F J. Local volatility and the recovery rate of credit default swaps[J]. Journal
of Economic Dynamics and Control, 2018, 92: 1-29.
[25] Siao J S, Hwang R C, Chu C K. Predicting recovery rates using logistic quantile regression with
bounded outcomes[J]. Quantitative Finance, 2016, 16(5): 777-792.
[26] Lin J H, Chen S, Huang F W. A contingent claim model of life insurer-bank swap default pricing:
strategic substitutes and complements[J]. Applied Economics, 2021, 53(36): 4166-4177.
[27] Peel D, McLachlan G J. Robust mixture modelling using the t distribution[J]. Statistics and Computing,
2000, 10(4): 339-348.
[28] Jiang R, Guan Y. Data-driven chance constrained stochastic program[J]. Mathematical Programming,
2016, 158(1): 291-327.
[29] Mohajerin Esfahani P, Kuhn D. Data-driven distributionally robust optimization using the Wasserstein
metric: Performance guarantees and tractable reformulations[J]. Mathematical Programming, 2018,
171(1): 115-166.
[30] Kim S I, Kim Y S. Tempered stable structural model in pricing credit spread and credit default swap[J].
Review of Derivatives Research, 2018, 21(1): 119-148.

�ݺ�ߣ

Research on Credit Default Swaps Pricing Under Uncertainty in the Distribution of Default Recovery Rates

More Related Content

Research on Credit Default Swaps Pricing Under Uncertainty in the Distribution of Default Recovery Rates