The neo-classical “Microfoundations” are themselves not microfounded…
Two arguments missing from this debate (Wren-Lewis comes close to the first):
A. So called micro foundations are not derived from ‘first principles’ and are not micro at all. So called micro-founded macro models are, contrary to what the articles state, not based upon the behavior of individual people. They are (among other things) based upon the ad-hoc imposition, without any kind of discussion, of the neo-classical individual homo economics choice model of upon the entire sector households, supposedly as an ‘emergent property’ of the macro-behavior of the sector households (an ‘emergent property’ is, for instance, the ‘swimability’ of water. This is not a property of individual water molecules but it is a property of large amount of water molecules). One can be critical, like Dorman (and me) of the neo-classical modelling of human behavior. But such criticism does not a-priori exclude that, as an emergent property, the entire sector households behaves in a ‘neo-classical rational way’. However. Nobody ever tries to explain on logical or empirical grounds that this is the case. Nobody. Ever. To the contrary. The Arrow-paradox states that aggregating individual preferences runs into serious ‘transitivity’problems. This is never discussed in any of the models (at least, not in the models I read – I don’t do this anymore as I consider it to be a waste of time). The ad-hoc imposition is therewith not derived from what microfundamentalists call ‘first principles’. They do not prove or try to prove (with good reason!) that adding individual indifference curves yields a ‘household sector’ curve with the same properties. Sideline – when a single ‘representative consumer’ is used to model the sector households, ‘unemployment’ is by definition excluded from the model (the representative consumer chooses (!) to work a little less or a little more). Hmmm.
B. In a monetary economy, accounting identities are much less prone to the ‘Lucas critique’ than econometric estimates. The reason to model micro foundations is supposedly the idea that people learn and the world changes. And indeed – people do learn. An example is the change in the behavior of interest rates after the forties and fifties, when they hardly and only slowly reacted to changes in inflation. Nowadays – they do react faster and more, which is surely caused by , among other things, the development of better and ‘quicker’ metrics of the price level (even economic science proceeds: during the last years, the economic statisticians have crafted much better statistics of house prices, for instance, though the people at the ECB still have to come to grips with this)! As people learn (and as other things change, too) we can’t rely on econometric models, which are based upon the past, to predict the future, according to Lucas. Before him, Keynes had, as early as 1939, of course voiced comparable criticisms of econometric models though his stance lacked the ‘predestination’ streak somewhat inherent in the work of people like Lucas and emphasized fundamental unpredictability, (I did not read the Tinbergen/Keynes discussion but I did consult this useful tract about this discussion by Hugo Keuzenkamp). So much about econometrics. But as long as we are living in a monetary, the quadruple (and sometimes even eightfold) entry accounting identities of the National accounts do hold. See Koo, who uses these identities to analyse balance sheet recessions. Or see Bos, who explains the mechanics of this kind of eightfold entry accounting. A penny spent (by you), is a penny earned (by somebody else) – which makes for quadruple entry accounting. This is, when you think about it, part of the essence of monetary transactions. It’s a model which exists of multiple players whose actions by definition influence the incomes and balance sheets of other players and which is estimated by aggregating the interrelated monetary consequences of the actions of all the individual humans and companies. Which has as an undeniable consequence the fact the ‘sector households’ is not the master of its own fate – which again may be the psychological reason why people like Lucas despise these kind of economic models and tried to replace them with this idea of a hyper-rational sector households, which forges its own future. Reality catched up with this endeavour: “Oh, what a tangled monetary sub-prime housing bubble web we weave, not knowing who we really deceive…”. Keynes, by the way, did contribute to the development of National Accounting and was much more favorable about this model than about the econometric ones.
Fun fact about rational expectations: the Keuzenkamp paper states that (29 years before Muth):
“Tinbergen’s 1932 paper, Ein Problem der Dynamik (A problem of dynamics), is the first that explicitly uses rational expectations in an economic model(Keuzenkamp, 1991). Tinbergen (1932, p. 172) assumes that expectations of economic agents are rational (‘vernünftig’) in the sense that they should be consistent with the relevant economic model. In addition, Tinbergen (1932, p. 172) makes the crucial assumption that these subjective expectations coincide with the objective expectations”