The first, and to my mind the most compelling, has been urged with great force by Herbert Simon (among others): people simply are not perfectly rational, perfectly prescient utility maximizers, and they couldn't be even if they wanted to. Second, the kind of economy assumed by neo-classical theory is a very special and recent phenomenon, originating in particular countries in particular epochs, and while it has since spread from there and seems likely to conquer the globe, the neo-classical theory is restricted to the places where economic life is conducted in markets with unrestricted, alienable private property, free labor, etc., so it is not a general theory of economic behavior, something deeply desirable. Third, even in the countries with the appropriate legal-institutional framework, perfect competition is exceedingly rare, very important parts of the economy are manifestly oligopolies, and certain kinds of competition are even legally prohibited by e.g. intellectual property laws. (The market for Friday night fish-fries in Madison, Wisconsin is very nearly perfectly competitive; the global market for commercial jets is not, and if you try to compete with Microsoft in providing Windows operating system programs, or Time-Warner in recordings of Madonna, you'll go to jail.)
The matter of imperfect competition has begun to yield, in the last few decades, to hard work and mathematical trickery. This development, springing ultimately from work by Edward Chamberlin and Joan Robinson in the 1930s, merely involves, so to speak, breaking a few mathematical bones so they will set in a better position. The objections on the grounds of our "bounded rationality" would seem to call for much more radical procedures, on the level of major surgery. But what about the middle group of objections, those about institutions --- do they call for bone-setting, or for a bone-marrow transplant?
There is a a long-standing American tradition of "institutionalism" which has decided that desperate measures are called for: among its leaders one may name Thorstein Veblen, John Kenneth Galbraith and Madison's own John R. Commons. This sort of institutionalism was long on facts, short on general theory, and almost free of mathematics, which is to say it was very poorly adapted to the environment of graduate schools and research universities. More math or more theory or both would be nice in their own right, and provide fine subjects for graduate seminars. Currently there are two substantial schools each proposing to bring theory and mathematics to the study of economic institutions, known, just to be confusing, as neo-institutional economics and new institutional economics. Thráinn Eggertsson's book is an exposition of the first-named school, which was essentially founded by Douglass North, and is of the bone-setting variety.
Neo-instititional economics, that is to say, accepts Homo economicus quite unreconstructed: he is the same greedy optimizer he always was, able to solve NP-complete problems in a single bound. But H. ec. no longer inhabits that idealization of nineteenth century Britain and post-bellum America, that world of frictionless competitive markets, where a man is free to sell whatever he damn well pleases, and buy anything he has the money for. Instead he finds himself, variously, a share-cropper in the South, a farmer in the Iceland of the sagas, a Roman slave-owner or slave, an African peasant, the director of a non-profit organization, and the ruler of a state. He now must pay the cost of conducting transactions, of gathering information, of negotiating contracts, of enforcing contracts. His property rights, so far from taking a simple, absolute and fixed form, can be arbitrarily convoluted and restricted, and are subject to change from within the economic system itself (not to mention what the state is able to both decree and enforce). The presentation is skillful and even, sporadically, entertaining, though it suffers from the textbook's "we will show that --- we are now showing that --- we have shown that" syndrome, and a barbarous system of citation. Eggertsson starts with situations close to the neo-classical ideal and gradually weakens its assumptions about markets, institutions and property rights.
Fairly early on we get an account of "the firm as a nexus of contracts," which is an explanation of why firms exist at all, rather a puzzle from the neo-classical viewpoint. The answer turns on cost of conducting transactions, in particular the cost of negotiating and enforcing a contract. Think for a moment of Adam Smith's pin factory, with its eighteen separate pin-making operations, its suppliers of raw materials, its distributors, its providers of workspace and equipment, et cetera. For all the persons involved in this production process to negotiate a universally satisfactory contract with each other could well take (as those who have been subjected to consensus decision-making can well imagine) from now until Doomsday, if it ever got done at all; and then there is the problem of enforcement. These contracting costs are drastically reduced if instead they all enter into their own bilateral contract with the firm. This of course gives an advantage to those who control the firm, which (being rational self-aggrandizers) they will exploit, but at least now the work gets done, and those involved get the benefits from that, whereas before the transaction costs were so high they got nothing. (In other words, the bourgeoisie are catalysts.) A natural corollary is that the degree to which the production process is integrated by firms will depend on the cost of negotiating and enforcing the various relevant contracts, and that when these transaction costs are low enough, we will see "commerce substituting for management". (There is an obvious role here for labor unions, as they lower the cost of negotiating the labor contract, and lower the cost to each worker of enforcing it; but unions are not even mentioned in Eggertsson's index.)
By the end of the book, while the modeling style hasn't changed at all, the original laissez-faire economy with a night-watchman state has been left completely behind, and we are contemplating competing forms of property rights in stateless societies (Iceland; comparison with the Atlas Berbers or the Pashtuns would be most interesting) and the neo-institutional economic theory of the state itself, considering such points as why rulers may decree and enforce systems of property rights which reduce the total taxable wealth, and how the changing costs of weaponry affects the distribution of political power. The theory of the state is strongly suggestive of Ibn Khaldûn, which is a point in its favor, and I eagerly look forward to a neo-institutionalist account of the dynastic cycle. Where there's empirical evidence bearing on the various models, Eggertsson reports on it, and is willing to admit that certain models are unsupported. Naturally, though, most of the evidence is favorable.
Empirical support for a neat body of theory is normally a Good Thing, but here it's slightly embarrassing. All the models rely on assuming that economic actors are so many members of the species Homo economicus, but that beast simply can't exist; human beings are not equipped to do the necessary calculations, period. Evolutionary arguments à la Alchain and Friedman don't work: monitor lizards can evolve into Komodo dragons, but no selection pressure, however intense, will squeeze them into the winged and fire-breathing kind. (Hoarding treasure and maidens I'm not so sure about.) The only resolution I can see is that sometimes Komodo dragons are good enough dragons, that boundedly-rational searchers will find and be satisfied with arrangements pretty close to the optima settled upon by perfectly rational agents. When this is not so, when there is a big difference between what can be done by way of outsize lizards and the fire-breathing ideal, then neo-classical and neo-institutional economics will no longer serve as good approximations to the True Economics. (To revert to our earlier metaphor: at that point we must dismiss the model-building bone-setters, and call in the conceptual surgeons.)
Other caveats could be added almost indefinitely: for instance, unmeasurable transaction costs make my operationalist thumbs prick. On balance, however, Economic Behavior and Institutions is a fine point of entry into an active and even valuable field of research which, owing to its expository structure, should be accessible to anyone with a basic grasp of conventional microeconomics.