Customer Reviews

Notes On The Theory Of Choice (Underground Classics in Economics)

5 star:		(5)
4 star:		(1)
3 star:		(0)
2 star:		(0)
1 star:		(1)

 

 

I think this book certainly deserves its good reputation as a primer on the variations and developments of utility theory. It must be noted that the book requires a fair amount of mathematical background that students coming to choice theory from non-economic backgrounds might not have. The book may be almost inscrutable if you are not ready to think mathematically about choice an aren't familiar with mathematic-decision speak. To that reader, I would refer you to a book called "Thinking and Deciding" by Jonathan Baron and/or the Psychology of Judgment and Decision Making by Scott Plous. For a reader with a psychology background such as myself, those were much gentler introductions, although the material in this book is worth mastering if you want to do serious work in the field of decision making research.

I have read this book more than 10 times since I came across it. Every time I read this book, I find something new. This book is based on lecture notes on axiomatic choice theory presented by David Kreps(the author) at the Stanford University. If you want to major in economics, in particular, choice theory, this book will enable you to access this area very easily.

The book seems harmless at the beginning with the mathematical proofs of some binary relations, which require a little bit of logic - and very careful writing down, if you have to turn in the problems as homework. But it gets more and more challenging - and pushes you to use some Real Analysis and Probability Theory concepts to really master the issues presented. This is not a reference book, nor a traditional textbook: it is a collection of lectures on economics of decision. The Von Neumann Morgernstern model is developed in all its magnitude, going further than any game theory textbook. You start to feel the 3 axioms and the numerical representation as second nature! But then, the Aumann/Anscombe and Savage models are introduced, as well as the limitations of the perfect VNM model. Everything you've learned seems to fall apart.
The problem sets are mainly excursions, in the very Kreps' style (if you have read his Course on Microeconomic Theory, you know what half a page problem without any equation is). In general, you don't need economics to get a good grip of the content: it is decision theory, not micro theory. Engineers and mathematicians will love it - and some economists will find it too tough!

A terrific book. If you want a comprehensible introduction to decision theory which includes all the most important stuff (von Neumann-Morgenstern, Savage, Anscombe-Aumann), look no further. You will not find anything else even close.

This book introduces the reader to the axiomatic foundations of the theory of choice. The topics covered include the von Neumann-Morgenstern axioms, the Savage axioms, risk and some of the shortcomings of the theory of choice. It is written in a very elegant style, neither too dense (mathematically) nor too shallow. The author does an excellent job in conveying the intuitive ideas behind the abstract axioms.

Nothing more need be said. If you're interested in understanding decision theory in economics, and you don't know where to start, start here.

Kreps has written an updated version of Savage's The Foundations of Statistics(1954).Like Savage,he is unable to deal with what Savage called vagueness,what Ellsberg called ambiguity and what John Maynard Keynes called uncertainty in the General Theory(1936)and weight of the evidence in the A Treatise on Probability(1921)in chapters 6 and 26 .Keynes defined an index to measure the extent of the vagueness,ambiguity,uncertainty or weight of the evidence upon which the probability estimates, calculated by the decision maker ,would be based.Letting w equal the weight of the evidence(or the vagueness,ambiguity or uncertainty of the evidence),0<=w<=1.A w=o means that the decision maker is dealing with a situation of ignorance.A w between 0 and 1(0<w<1) means that the decision maker is dealing with partial ignorance.A w=1 means that the decision maker is dealing with situations of risk,where the probabilities are further assumed to be precise,unique,exact single numbers.They can't be nonadditive(subproportional or superproportional).Kreps spends the first 13 chapters dealing with a purely mathematical construction that can only deal with the case of w=1(or rho=1 in the Ellsberg approach or the degree of evidential support=1 in the 1994-1996 Tversky approach).He then spends a total of 7 pages covering the massive amount of empirical and experimental evidence which points to the conclusion that his theoretical,axiomatic approach is a special ,limiting case that has ,practically,little or no real world application.Any reader of this review who is interested in a theoretical decision rule that deals with all of the empirical evidence briefly covered by Kreps(Allais paradox,Ellsberg paradox,certainty effect,reflection effect,translation effect,and the preference reversal effect)is advised to carefully study Keynes's conventional coefficient of risk and weight ,c.The goal of the decision maker is to maximize cA,where c=p(1/1+q)[2w/(1+w)]and A is the outcome.Risk aversion is dealt with by the decision weight(1/1+q)and not by conflating diminishing marginal utility with risk aversion.Kreps needs to rearrange the ordering of his chapters,putting the last chapter first.This would allow a potential reader to decide whether he wants to allocate his limited,finite time to reading a book that has very limited application.