From the reviews: SHORT BOOK REVIEWS "The text is no doubt my favorite on the subject of interest rate modeling. It perfectly combines mathematical depth, historical perspective and practical relevance. The fact that the authors combine a strong mathematical (finance) background with expert practice knowledge (they both work in a bank) contributes hugely to its format. I also admire the style of writing: at the same time concise and pedagogically fresh. The authors’ applied background allows for numerous comments on why certain models have (or have not) made it in practice. The theory is interwoven with detailed numerical examples…For those who have a sufficiently strong mathematical background, this book is a must." From the reviews of the second edition: "The book ‘Interest Rate Models – Theory and Practice’ provides a wide overview of interest rate modeling in mathematical depth. … The authors found a good approach to present a mathematically demanding area in a very clear, understandable way. The book will most likely become … one of the standard references in the area. … if one were to buy only one book about interest rate models, this would be it." (David Skovmand and Michael Verhofen, Financial Markets and Portfolio Management, Vol. 21 (1), 2007) "This is the book on interest rate models and should proudly stand on the bookshelf of every quantitative finance practitioner and student involved with interest rate models. If you are looking for one reference on interest rate models then look no further as this text will provide you with excellent knowledge in theory and practice. … is simply a must for all. Especially, I would recommend this to students … . Overall, this is by far the best interest rate models book in the market." (Ita Cirovic Donev, MathDL, May, 2007) "This is a very detailed course on interest rate models. Its main goal is to construct some kind of bridge between theory and practice in this field. From one side, the authors would like to help quantitative analysts and advanced traders handle interest-rate derivatives with a sound theoretical apparatus. … Advanced undergraduate students, graduate students and researchers should benefit from reading this book and seeing how some sophisticated mathematics can be used in concrete financial problems." (Yuliya S. Mishura, Zentralblatt MATH, Vol. 1109 (11), 2007)
The modeling of interest rates is now a multi-million dollar business, and this is likely to grow in the years ahead as worries about quantitative easing, government budgets, housing markets, and corporate borrowing have shown no sign of abatement. The approach that the authors take in this book has been branded as too "theoretical" by some, particularly those on the trading floors, or those antithetic to modeling in the first place. The authors though are aware of such reactions to financial modeling, and actually devote the end of the book to a hypothetical conversation between traders and modelers (but omitting some of the vituperation that can occur between these groups). The book is written very well, with calculation steps for the most part included in detail. Since it is a monograph, there are no exercises, but readers will find ample opportunities to fill in some of the calculations or speculate on some of the many questions that the authors list in the beginning to motivate the book. These questions are invaluable for newcomers to the field, or those readers, such as this reviewer, who are not currently involved in financial modeling but are very curious as to the mathematical issues involved. There is also an excellent list of "theoretical" and "practical" questions in the preface that the authors use to motivate the book, along with a detailed summary of upcoming chapters.
The first part of the book sets the tone for the rest of the book, and can be considered as an elementary introduction to the theory of contingent claim valuation. In this discussion the authors focus on a portfolio consisting of riskless security (bond) and a risky security (stock) that pays no dividend. The object is to follow the time evolution of the price of these two securities. The time evolution of the riskless bond is merely exponential, as expected, but that of the risky security is random according to a geometric Brownian motion. The `trading strategy' consists of holding a number of units of each of these securities at each time. All changes in the value of the portfolio can be shown to be entirely due to capital gains, with none resulting from the withdrawal or infusion of cash. The authors refer to this as a `self-financing' strategy, and the initial investment results in a pattern of cash flows that replicates that of a call option. This option is attainable by dealing only in a stock and a bond. This leads to the question as to what class of contingent claims a group of investors can actually attain, where a contingent claim is viewed as a nonnegative random variable which is measurable with respect to a filtration of a probability space. This filtration can be viewed as essentially a collection of events that occur or not depending on the history of the stock price. The bearer will obtain a payment at expiry, the size of which depends on the prior price history.
A contingent claim is said to be `attainable' at a particular price if there exists a self-financing trading strategy, along with an associated market value process that equals the initial prices and equals the contingent claim at expiry. It is shown that every contingent claim is attainable in a complete market. The goal is then to find conditions under which arbitrage is impossible, i.e. conditions that prevent the occurrence of a zero investment and through some trading strategy is able to obtain a positive expected wealth at some time in the future. The authors show that a market is free of arbitrage if and only if there is a martingale measure, and that a market is complete if and only if the martingale measure is unique.
It was primarily the interest of this reviewer in analytical models rather than Monte Carlo simulations, even though there is a thorough discussion of the latter in this book, including the most important topic of the standard error estimation in simulation models. For analytical modeling, the Vasicek model is usually the first one discussed in the literature, and this book is no exception. But the Vasicek model allows negative interest rates and is mean reverting. The authors want to go beyond this model by searching for one that will reproduce any observed term structure of interest rates but that will preserve analytical tractability. One of these, the Cox-Ingersoll-Ross (CIR) model, is analytically tractable and preserves the positivity of the instantaneous short rate. Ample space in the book is devoted to a discussion of this model, which is essentially one where one adds a "square root" to the diffusion coefficient.
Physicists who aspire to become financial engineers may find the discussion on the change of numeraire to be similar to the "change in gauge" in quantum field theory. In the latter, a clever choice of gauge can make calculations a lot easier. The same goes for a choice of numeraire for pricing a contingent claim, and the authors give a detailed overview of what is involved in doing so. Of particular importance in this discussion is the role of the Radon-Nikodym derivative, a concept that arises in measure theory, and also the use of Bayes rule for conditional expectations. To fully appreciate this discussion, if not the entire book, readers will have to have a solid understanding of these concepts along with stochastic calculus and numerical solution of stochastic differential equations.
Interestingly, the authors devote a part of the book to the connection between interest rate models and credit derivatives, wherein they argue that credit derivatives are not only interesting in and of themselves, but that the tools used to model interest rate swaps can be applied to credit default swaps to a large degree. Of particular importance is the appearance of copulas in chapter 21, which have been criticized lately for their alleged role in the "financial crisis". The authors give an overview of these entities for the curious reader but do not use them in the book.
Some readers may find when first exposed to `reduced form models' that they might seem too extreme or judged to be inapplicable because default is viewed as being essentially independent of market observables. Instead default is modeled by an exogenous jump stochastic process. The authors spend a fair amount of time explaining why these models are suitable for credit spreads. In particular, they show that the probability to default after a given time, i.e. the `survival' probability, can be interpreted as a zero coupon bond and the intensities as instantaneous credit spreads. Positive interest short-rate models can therefore be used to do default modeling. The lack of an economic interpretation for the default event is to be contrasted with term structure models, and the authors discuss this in detail.
Structural models on the other hand are tied to economic factors, namely the value of the firm, i.e. its ability to pay back its debt. If this value drops below a certain level, the firm is taken to be insolvent. The authors give a brief overview of structural models, emphasizing their similarities to barrier-free option models, but do not treat them in detail in the book, since they do not have any analogues to interest rate models. For credit risk, the defaultable zero coupon bond is the analog of the zero coupon bond for interest rate curves. The forward rate for credit default swaps also has an analog with LIBOR and SWAP rates. Readers interested in counterparty risk will be exposed to an interesting assertion, namely that the value of a (generic) claim that has counterparty risk is always less than the value of a similar claim whose counterparty has a probability of default equal to zero. The authors give a rigorous formulation of this assertion by proving a general counterparty risk pricing formula.
Poisson processes, used heavily in network modeling and queuing theory, are discussed here in the authors' elaboration of intensity models, along with Cox processes where the intensity is stochastic. Detailed examples are given which illustrate how to use reduced form models and market quotes to estimate default probabilities.
Monte Carlo simulations, which are the bread and butter of financial modeling (along with many other fields of modeling) are used to simulate the default time. The authors address the problem of large variance and the consequent large number of simulations needed if the standard error is just one basis point. Techniques of variance reduction in Monte Carlo simulation are well-known, and the authors discuss one of these, the control variate technique.
Also discussed is a hybrid model where both interest rates and stochastic intensities are involved, and the authors show how to calibrate survival probabilities and discount factors separately when there is no correlation between the interest rates and intensities. The calibration must then be done simultaneously when this is not the case. One is led to ask in this case, and in general, whether interest rate data can serve as a proxy of default calibration, and vice versa. Not really, but the authors do explain how the correlation can be ignored, since it has little impact on credit default swaps.
Ensuring that interest rates remain positive is thought of as an important side constraint by many modelers, who point to the large negative rates that may occur in Gaussian models of interest rates. One model that particularly stands out in this regard is due to B. Flesaker and L. Hughston, and which is discussed in one of the appendices in the book. Their strategy is to enforce positivity via the discount factor, and doing this in such a way so as to eliminate the possibility of "explosions", i.e. situations where the payoff can become infinite in an arbitrarily short time. Their model can essentially be characterized by an integral representation for discount bonds in terms of a family of kernel functions. The members of this family are positive martingales, and this ensures the required positivity. Their behavior under a change of measure involves a ratio called the `state-price density' or `pricing kernel', and this shows that the Flesaker-Hughston model can be interpreted as a general model of interest rates. Arguments are given as to whether all choices of kernel can result in viable interest rate models. Examples are given illustrating that not all can be, but the Flesaker-Hughston model is interesting also in that it does not depend on possibly highly complex systems of stochastic differential equations for interest rate processes. The authors unfortunately do not include a discussion on how to calibrate this model to market data, but instead delegate it to the references.
In the late nineties I went through Brigo's innovative work on stochastic nonlinear filtering with differential geometry techniques. I was favorably impressed by results and style, particularly in his dissertation and in his 'geometry in present day science' very readable overview. Interesting results are found and nicely told with accurate - but not pointlessly complicated - advanced mathematics for the problems at hand, I reasoned.
I've followed a similar path from control to finance, and having worked with interest rate models, I couldn't help but order this Brigo-Mercurio book. I had high expectations 'cause these two guys are working in a bank on the real thing.
Sure enough I'm not disappointed.
1-factor models are handled with great care, a ton of formulas and recipes are given. I've never seen this kind of analysis of pricing with Gaussian 1-f models. The new upgrade of the CIR model is interesting and accurate. "CIR++" is now my favorite 1-f model. I like the treatment of lognormal 1-f models and the explanation of Monte Carlo and trees -- the flow-chart for Bermudan swaptions is crystal clear! Plots of market implied structures and volatility calibration are useful additions.
The chapter on 2-f extensions has one of the best discussions on volatility, and two tons of useful formulas/recipes. Two dimensional trees!
The HJM chapter size is OK. I agree - the useful models embedded in HJM are short rate models and market models.
Market models - these three chapters alone are worth the book. You'll find yourself nodding as you read the guided tour. They make it look easy all the time. The exposition is focused, clear, intuitive, detailed. There's also new stuff, just check the calibration discussion! Smile modeling begins with a brilliant tour and ends with Brigo-Mercurio's new approach - the mixing dynamics - deserving a whole chapter if expanded.
The detailed explanation on products is a much welcome original addition. Cross currency derivatives!
Quotes - as in Brigo's old work - are a pleasant diversion while reading. The 500 and more pages are a treat given the competitive price.
Still there's room for improvements - more "CIR2++"! Something on 3-f models. Historical estimation of the correlation matrix and low-rank optimized approximations. Expand smile modeling! More hedging. Something on structured products. Cross currency libor model. chapter 9 - other interest rate models - sounds out of place and can be suppressed for other things.
This book rings true and has useful teachings for students, academics and practitioners. Although it requires some background in stochastic calculus, it's hard to beat on the pricing front. Kudos to Brigo and Mercurio! It only harms there aren't enough books like this.
Product Details :
Hardcover: 1038 pages
Publisher: Springer; 2nd edition (August 2, 2006)
Language: English
ISBN-10: 3540221492
ISBN-13: 978-3540221494
Product Dimensions: 6.1 x 2.1 x 9.2 inches
More Details about Interest Rate Models - Theory and Practice: With Smile, Inflation and Credit 2nd edition
or
Download Interest Rate Models - Theory and Practice: With Smile, Inflation and Credit 2nd edition PDF Ebook
No comments:
Post a Comment