Home > Uncategorized > Where modern macroeconomics went wrong

Where modern macroeconomics went wrong

from Lars Syll

DSGE models seem to take it as a religious tenet that consumption should be explained by a model of a representative agent maximizing his utility over an infinite lifetime without borrowing constraints. Doing so is called micro-foundingthe model. But economics is a behavioral science. If Keynes was right that individuals saved a constant fraction of their income, an aggregate model based on that assumption is micro-founded.FRANCE-US-ECONOMY-NOBEL-STIGLITZOf course, the economy consists of individuals who are different, but all of whom have a finite life and most of whom are credit constrained, and who do adjust their consumption behavior, if slowly, in response to changes in their economic environment. Thus, we also know that individuals do not save a constant fraction of their income, come what may. So both stories, the DSGE and the old-fashioned Keynesian, are simplifications. When they are incorporated into a simple macro-model, one is saying the economy acts as if… And then the question is, which provides a better description; a better set of prescriptions; and a better basis for future elaboration of the model. The answer is not obvious. The criticism of DSGE is thus not that it involves simplification: all models do. It is that it has made the wrong modelling choices, choosing complexity in areas where the core story of macroeconomic fluctuations could be told using simpler hypotheses, but simplifying in areas where much of the macroeconomic action takes place.

Joseph Stiglitz

Stiglitz is, of course, absolutely right.  

DSGE models are worse than useless — and still, mainstream economists seem to be impressed by the ‘rigour’ brought to macroeconomics by New-Classical-New-Keynesian DSGE models and its rational expectations and microfoundations!

It is difficult to see why.

Take the rational expectations assumption. Rational expectations in the mainstream economists’ world imply that relevant distributions have to be time independent. This amounts to assuming that an economy is like a closed system with known stochastic probability distributions for all different events. In reality, it is straining one’s beliefs to try to represent economies as outcomes of stochastic processes. An existing economy is a single realization tout court, and hardly conceivable as one realization out of an ensemble of economy-worlds since an economy can hardly be conceived as being completely replicated over time. It is — to say the least — very difficult to see any similarity between these modelling assumptions and the expectations of real persons. In the world of the rational expectations hypothesis, we are never disappointed in any other way than as when we lose at the roulette wheels. But real life is not an urn or a roulette wheel. And that’s also the reason why allowing for cases where agents make ‘predictable errors’ in DSGE models doesn’t take us any closer to a relevant and realist depiction of actual economic decisions and behaviours. If we really want to have anything of interest to say on real economies, financial crisis and the decisions and choices real people make we have to replace the rational expectations hypothesis with more relevant and realistic assumptions concerning economic agents and their expectations than childish roulette and urn analogies.

Or take the consumption model built into the DSGE models that Stiglitz criticises. There, people are basically portrayed as treating time as a dichotomous phenomenon – today and the future — when contemplating making decisions and acting. How much should one consume today and how much in the future? Facing an intertemporal budget constraint of the form

ct + cf/(1+r) = ft + yt + yf/(1+r),

where ct is consumption today, cf is consumption in the future, ft is holdings of financial assets today, yt is labour incomes today, yf is labour incomes in the future, and r is the real interest rate, and having a lifetime utility function of the form

U = u(ct) + au(cf),

where a is the time discounting parameter, the representative agent (consumer) maximizes his utility when

u´(ct) = a(1+r)u´(cf).

This expression – the Euler equation – implies that the representative agent (consumer) is indifferent between consuming one more unit today or instead consuming it tomorrow. Typically using a logarithmic function form – u(c) = log c – which gives u´(c) = 1/c, the Euler equation can be rewritten as

1/ct = a(1+r)(1/cf),


cf/ct = a(1+r).

This importantly implies that according to the neoclassical consumption model that changes in the (real) interest rate and the ratio between future and present consumption move in the same direction.

So good, so far. But how about the real world? Is the neoclassical consumption as described in this kind of models in tune with the empirical facts? Hardly — the data and models are as a rule inconsistent!

In the Euler equation, we only have one interest rate, equated to the money market rate as set by the central bank. The crux is that — given almost any specification of the utility function – the two rates are actually often found to be strongly negatively correlated in the empirical literature. The data on returns and aggregate consumption simply are inconsistent with the DSGE models.

Although yours truly shares a lot of Stiglitz’ critique of DSGE modelling — “the standard DSGE model provides a poor basis for policy, and tweaks to it are unlikely to be helpful” —  it has to be said that his more general description of the history and state of modern macroeconomics is less convincing.  Stiglitz notices that some of the greatest deficiencies in DSGE models “relates to the treatment of uncertainty,” but doesn’t really follow up on that core difference between Keynesian ‘genuine uncertainty’ economics and neoclassical ‘stochastic risk’ economics. DSGE models are only the latest outgrow of neoclassical general equilibrium (Arrow-Debreu) economics. And that theory has never, and will never, be a good starting point for constructing good macroeconomic theory and models. When the foundation of the house you build is weak, it will never be somewhere you want to live, no matter how many new — and in Stiglitz’ view better — varieties of ‘micro-foundations’ you add. As Lance Taylor writes (personal communication):

Aside from consistent accounting (Walras’s Law), the Arrow-Debreu model is useless for practical macroeconomic purposes. Its “agents” could never carry through their assigned optimization exercises, a feat beyond the capacity of a universal Turing machine, let alone feeble human brains. The Sonnenschein-Mantel-Debreu theorem, moreover, shows that microeconomic rationality assumptions have no significant macroeconomic implications …

taylor_press_lowresThe phalanx of models marshalled by Stiglitz uses “expectation” to mean the first moment of some objective probability distribution, shared by some or all players. In contrast, Keynes adopted a non-frequentist (and non-Bayesian) interpretation, whereby we cannot make a probabilistic assessment about whether some future event(s) will occur. “Fundamental uncertainty” is a kissing cousin of Donald Rumsfeld’s “unknown unknowns.” Moments computed from numbers of the past are irrelevant. Plugging them into algebraic machines generates no useful information.

As with all great bodies of thought, many strands of what Keynes has to say appear to be contradictory. Strict macro accounting on one hand and belief in the fundamental instability of capitalism make up just one example. Bastard Keynesian is just an attempt to sidestep the contradictions. Whether Keynes’s own ideas will help us disentangle the unknowns of the future is unclear. It is clear that Stiglitz’s enormous tool bag will not be very helpful.

  1. csissoko
    September 18, 2017 at 1:14 am

    A more important critique of Arrow Debreu has always seemed to me to be that the equilibrium concept isn’t incentive compatible unless the agents are infinitesimal in size. Given that in the real world at least one side of every transaction usually has some bargaining power, the flaw here is that either the equilibrium concept isn’t really economics (because the concept is not incentive compatible) or you assume infinitesimal agents and find that your results have nothing to do with the real world.

    Basically I prefer the modern micro approach (at least as I learned it in grad school where game theory ruled): competitive equilibrium can be an interesting and useful reference point, but in and of itself is not really a good tool for economic analysis.

  2. September 18, 2017 at 5:06 pm

    What I find bewildering in this discussion, not having studied economics formally, is that people are debating the coefficients of an equation. Is the discounting factor in the model consistent with macro measurements in the real world. Or they’re debating philosophically whether to model macro behavior by aggregating micro agents is ever legitimate. Usually a criticism stats as “micro-foundations are in principle unsound” and then leaps to “the model doesn’t agree with observations”.

    To anyone familiar with complex systems, a simple algebraic agent model is patently unsound, not because the coefficients are wrong but because they don’t model complexity. If all the micro agents are simple and the same that will *for sure* generate wrong results, no matter how close the coefficients to some statistical average. There are no single correct coefficients to set in a simple micro agent model, because in the real world people vary widely in the form and coefficients and the stochastic evaluation of their preference functions. This complex system produces emergent results that are substantially different from a simple system where all the agents are simple and the same, no matter how well calibrated.

    Building macro economic models by simulating millions of micro agents, with realistic variance and complexity, sounds like a lot of fun. It might even evolve into a powerful prediction engine. This is not what most economists are doing, because I don’t see many economists competing with climate scientists to run their large scale sims on supercomputers, or sharing their simulation results. What economists seem to do is try to guess algebraic forms that shortcut the complexity and the emergent phenomena that they should be modelling in the first place.

    More economists need to up their tech and develop micro-agent models with the right amount of complexity and variance, the right scale, and the right discipline for fitting to empirical data, before there can be any kind of scientific conversation as to whether these models, done properly, have useful predictive power. Until then we’re like philosophers squabbling over Platonic conceptions that have no bearing on the reality of a topic.

  3. September 22, 2017 at 4:26 am

    I highly respect Josef Stiglitz. He is the most reliable economist among the mainstream economists. By his personality he is almost dissident in the neoclassical economics and yet he retains influential status among leading-edge economists and policy makers.

    I have read his working paper: “Where modern macroeconomics went wrong” and I was disappointed. He may be addressing wider audience who still believe in the validity of macroeconomics based on DSGE model. As such, his paper may be effective. It may induce some doubts and create a chance to let those believers to convert from the present research program to another. I admit that it is a necessary work, but for a more concerned economists on the very basis not only of macroeconomics but also of microeconomics.
    Stilgitz explains how and why DSGE models are wrong and missing the point. But, as Lars Syll remarks,

    “Stiglitz acknowledges that his approach ‘and that of DSGE models begins with the same starting point: the competitive equilibrium model of Arrow and Debreu.’ ” (first paragraph of Syll’s post “Stiglitz and the full force of Sonnenschein-Mantel-Debreu.”)

    Stiglitz’s enormous tool bag (Lance Taylor) is obstructing him to descend on the depth of trouble of whole economics. Stiglitz is talking on the assumption that a good macroeconomic model is necessary and it is possible. From this assumption, he recommends various amendments to be incorporated in the new macro model. However, his assumption is quite doubtful.

    I willingly admit that a good macroeconomic model is necessary. If there is none, somebody would invent a substitute and it will be used. Macroeconomic model is a kind of astrology. Without sufficient knowledge of the world, princes need something to direct them. Astrology was one of them. Catholic Church was in principle against the use of astrology, but could not prevent it to be used in the end of Medieval Era. Macroeconomics is present day astrology. It does not matter if the prediction is true or not, or in other words, accurate or not. From necessity point of view, macroeconomics is necessary. For such a purpose, there is no need that it is a science. Astrology is not a science or a pseudo-science and yet it played a big important function in some societies. However, as for the possibility of macroeconomics, it is highly questionable if macroeconomics is a science at all.

    The geocentric system of Ptolemy was an “exact science.” When Kepler started to reconsider the orbit of planets, Ptolemaic system was exact up to 8 angular minutes. In other words, the maximum error of Ptolemaic system was only 0.14 degree. It is sure that it was much more accurate than any of macroeconomic models. (I am talking figuratively. Comparison of two predictive systems is not easy.) Despite its accuracy, Ptolemaic system was a wrong system. Further development of astronomy depended on the overthrow of the system. Economics today requires a Copernican revolution. Interestingly, in one place of his working paper, Stiglitz talks about “Ptolemaic attempt.” (p.2) But he feels no such necessity. He might be a guru of Ptolemaic system and may contribute still a bit for its improvements. However, the limit of further development is already visible.

    What we need is a prospectus for a Copernican revolution. Stiglitz provides no hints on this direction. The most urgent role of a methodologist like Lars Syll is, in my opinion, to contribute to this revolution. Even if his role is only to give a rough suggestion, it may be a big contribution to the future of economics.

    • September 22, 2017 at 10:13 am

      See my comment on “Break this!”

  4. September 22, 2017 at 12:03 pm

    Re flow structure in my comment and Lars’ quotation of Lance Taylor: Keynes’ focus on “liquidity preference” is hinting at something which flows. This cannot be measured accurately and added but can be referred to as a whole, hence his making a distinction between ‘macro’ and ‘micro’. Quantification here is not numerical but logical, capable only of distinguishing All, Some, One and None, or judging Everything, Enough, Not Enough and No Longer Any (but also that judgements of direction of flow can be Right or Wrong). Keynes’ argument was about too much unemployment meaning too little purchasing power, not predicting that x amount of government expenditure would generate y (useful) jobs. C.f. the familiar paraphrasing of his General Theory ch.3: “Better to be roughly right than precisely wrong”. I see the GT not so much wrong as still anticipating the truth: the start of a seminal “work in progress”. The lack of appreciation and misrepresentation of it by jealous rivals I find frankly disgusting.

  5. September 22, 2017 at 3:33 pm

    Although the paper of Joseph Stiglitz is disappointing, the question that he pose is important. Every economist concerned with the future of economics should ask “Where did modern macroeconomics go wrong?” Only getting a correct answer, we ca restart in a right direction.

    As Lance Taylor’s comment hints it, there will be no easy answer. But the question is still important, because the answer decides the basic strategy for the reconstruction of economics.

    The first point is to question whether criticizing DSGE models is sufficient. Stiglitz mainly claimed how it is wrong. It seems that Stiglitz is thinking it is. Syll makes no mention on this point. Robert Gordon (2009) once claimed that macroeconomics was sound until around 1978.

    Many others believe that Hicksian macroeconomics (ISLM) miss-oriented Keynesian economics and it is necessary to return to Keynes’s General Theory. Syll seems to be close to this position.

    Meir Kohn (1986) was much more radical, because he claimed that macroeconomics went wrong before Keynes’s General Theory. He hinted that we should go back to sequence analysis which was developed in the 1920-30’s. Keynes once subscribed to this method (e.g. Treatise on Money, 1930) but adopted equilibrium framework in his General Theory. See my question in the ResearchGate “Return to Keynes, is it sufficient?”.

    My personal opinion is summarized in two points:

    (1) We should abandon equilibrium analysis as the framework of analysis and adopt instead sequence analysis (or process analysis). See Shiozawa (2016a).

    (2) We should abandon demand and supply framework (i.e. the most fundamental framework of neoclassical economics) and restart from the classical theory of value (of course modernized as the 21 century economics.) See Shiozawa (2016b; 2017).

    Gordon, Robert (2009) Is Modern Macro or 1978-era Macro More Relevant to the Understanding of the Current Economic Crisis?

    Kohn, Meir (1986) Monetary Analysis, the Equilibrium Method, and Keynes’s “General Theory”. Journal of Political Economy 94(6): 1191-1224.

    Shiozawa, Yoshinori (2016a) Microfoundations of Evolutionary Economics

    Shiozawa, Yoshinori (2016b) The Revival of Classical Theory of Values. In Nobuharu Yokokawa, Kiichiro Yagi, Hiroyasu Uemura and Richard Westra (Eds.) The Rejuvenation of Political Economy, May 2016, Oxon and New York: Routledge. Chapter 8, pp.151-172.

    Shiozawa, Yoshinori (2017) The New Theory of International Values: An Overview. In Shiozawa, Y., T. Oka, and T. Tabuchi (Eds.) A New Construction of Ricardian Theory of International Values: Analytical and Historical Approach, Singapore, Springer Nature. Chapter 1, pp.3-75.

    My question in ResearchGate: Return to Keynes, is it sufficient?

  1. No trackbacks yet.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.