Recursive Macroeconomic Theory (The MIT Press)

Great book. Very technical, but probably the most comprehensive and authoritative book on macro models and numerical techniques.

Well, I haven't read the whole book, and I doubt anyone has, but I've actively searched for several themes in it.Heterogeneity(firm and/or consumer), bubbles, financial crisis... These themes are either missing, or severely underrepresented in the book, to the point I would say that in the authors opinions they are nothing but footnotes in the macroeconomic history.This book sets the right direction for the future of macroeconomics, but at the same time it also exposes the main flaw in nowadays education of macroeconomic subject... Macroeconomists simply do not understand the tools they are using. This is starkly shown when Sargent, and others, try to condense books of functional analysis, probability theory and stochastic processes, stochastic optimal control, all into a few simple appendices, or as some inline citations... If only economists had a proper knowledge of mathematics, they would lose less time developing 'useless' à la carte models, which for policy decisions can be very easily substituted for much more manageable and simpler models - albeit of a partial equilibrium nature (don't shoot me) - and whose (DSGE) quantitative predictive performance lacks immensely, and dedicate themselves to REALLY understand the theoretical(mathematical) notions of the models being used. No, appendices of 50 pages encompassing the basic definition of measure to Pontryagin maximum principle are not even enough to give you a clue of what you do...The reason this book is so voluminous, besides the copious amounts of models, is that it follows a recipe approach. Here's model A, which is associated with technique T_A, and conclusions C_A. And then in the next, similarly. There's almost no effort to give a general framework to study these types of models. Each model, their recipe for how to solve the model. They're so busy at looking at several regions of the forest, that it seems that nobody studies the contours of the forest!Also, in a mathematical book, when we apply a theorem, we usually state them, and verify the conditions... Well in my many cases, in this book, probably mainly due to the background education in macro, the authors don't state the theorem, or when they do, the verification of the conditions for the theorem are simply brushed aside. This problem is not just in this book... Almost no book in macro, where a linearization is used ever states the Hartman–Grobman theorem, which gives some conditions where we can be sure that a linearization is a good approximations (there are cases where linearization is not a good approximation, even qualitatively speaking), and yet no one talks about it. Another example is the way that the solutions are solved in Dynare. It seems that approximations are to be used as it pleases the 'macroman', without regard for precision, and Dynare is the black box of choice. The macroman does not want to know the least how Dynare outputs a solution, as long as it doesn't show an error message, all is fine (I'm clearly exaggerating, but you get the point). Usually, when we use approximations, we would like a measure of precision regarding the true solution, or at least a magnitude on the 'distance' of solutions between iterations... Yes, let's use global methods, and just brush aside the small prints. What about the identification and specification problems when we take these models to data? Hushh child, it's not important. Let's look at the nice IRFs.The important thing is to build more and more models, with greater complexity, more rich dynamics, without almost any concern for theoretical/mathematical validity, nor statistical validity(Here's a nice article on the 'recent' state of Macro: "Publish and Perish: Creative Destruction and Macroeconomic Theory"). Criticisms like those of Canova to DSGE models have been simply ignored for most of the time.The macroman just wants to play with his toys.We could also talk about the conditions for many of the interchange of derivatives, conditional expectations and others...but these may seem minor dismisses, when we compare to the lack of rigour stated in the previous paragraphs.Sargent, if you ever read this, I love you man... it's just your book that I don't.