This preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: IEOR E4731: Credit Risk and Credit Derivatives Lecture 05: Reduced form models. Notes originally written by Prof. Rama Cont Instructor: Xuedong He Spring, 2012 1 / 40 Reduced form models I Structural models have strong economic meaning but are not applicable I In practice, people use reduced form models I In reduced form models, people do not care about why defaults happen. They just model default times as random times. I Drawback of reduced form models: does not tell why defaults happen I Advantage of reduced form models: I Easy to implement I Easy to incorporate any market information I Can use the machinery of defaultfree termstructure modeling. I In view of riskneutral pricing, let us start from specifying information sets. 2 / 40 Information sets: modeled via filtrations I {H t } t ≥ : we observe default or no default only. Information at time t is: H t := σ { 1 τ>s , ≤ s ≤ t } I {F t } t ≥ : market information other than the default time (stock prices, interest rates, news,...) I {G t } t ≥ : full information, i.e., G t := F t W H t . I Example: Merton model. F t is the historical firm value up to time t . H t ⊂ F t , thus G t = F t . I Example: F t = {∅ , Ω } , i.e., no market information. In this case, G t = H t . 3 / 40 Risk Neutral Pricing I Starting point: the market value of a terminal cash flow H paid at T is given as Π t ( H ) = E Q h e R T t r ( u ) du H  G t i where Q is the risk neutral pricing measure (with e R t r ( u ) du ,t ≥ to be the denominator) I Example: for a zerocoupon zerorecovery bond that only defaults at maturity T , the payoff is H = 1 τ>T . I It is critical to compute the conditional expectation. 4 / 40 Conditional expectation I Formal definition of conditional expectation: let X be a random variable on a probability space (Ω , F , P ) and G be a sub σalgebra. Then, the conditional expectation Z := E [ X  G ] is a Gmeasurable random variable defined via E [ Z 1 A ] = E [ X 1 A ] , ∀ A ∈ G . I Example: let A 1 ,...,A n be a partition of Ω and G := { A 1 ,...,A n } , then E [ X  G ] = n X i =1 E [ X  A i ] 1 A i , where E [ X  A i ] := E [ X 1 A i ] P ( A i ) 5 / 40 Conditional expectation (cont’d) I For any B ∈ F , P [ B  G ] := E [ 1 B  G ] . I Example: if G := { A 1 ,...,A n } , then P [ B  G ] = n X i =1 P [ B  A i ] 1 A i where P ( B  A i ) := P ( B ∩ A i ) P ( A i ) I For any random variable X and Y , E [ X  Y ] := E [ X  σ ( Y )] where σ ( Y ) is the σalgebra generated by Y , i.e., the information one can obtain from Y . I Example: if Y = 1 A , then σ ( Y ) = { A,A c } . As a result, E [ X  1 A ] = E [ X  A ] 1 A + E [ X  A c ] 1 A c . 6 / 40 Compute Conditional Expection We start from the simplest information structure: no market information, i.e., F t = {∅ , Ω } . In this case, G t = H t ,t ≥ ....
View
Full
Document
This note was uploaded on 03/14/2012 for the course IEOR 4731 taught by Professor Hexuedong during the Spring '12 term at Columbia.
 Spring '12
 HeXuedong

Click to edit the document details