DeLong, Summers & Krugman on models

2 January, 2016 at 20:09 | Posted in Economics | 10 Comments

Larry Summers, Brad DeLong, and Paul Kugman are having an extended discussion on the role of models in economics on their blogs this week.

That’s good. Since the model is the message in economics today, that is actually the most important discussion possible to have in economics.


Krugman is arguing that models are not ‘always the right guide for policy, but still necessary for disciplining our policy preferences.’ According to Krugman, ‘the discipline of thinking things through in terms of models is really important.’

This emphasis on the value of modeling should come as no surprise. Paul Krugman has always — although sometimes admitting that economists have a tendency to use  ‘excessive math’ and ‘equate hard math with quality’ — vehemently defended  the formalization and mathematization that comes with the insistence of using a model building strategy in economics.

But if these math-is-the-message-modelers aren’t able to show that the mechanisms or causes that they isolate and handle in their mathematically formalized macromodels are stable in the sense that they do not change when we ‘export’ them to our ‘target systems,’ these mathematical models do only hold under ceteris paribus conditions and are consequently of limited value to our understandings, explanations or predictions of real economic systems. Building models only to show Krugmanian  ‘self-dicipline’ is setting the aspiration level too low.

According to Keynes, science should help us penetrate to ‘the true process of causation lying behind current events’ and disclose ‘the causal forces behind the apparent facts.’  We should look out for causal relations. But models — mathematical, econometric, or what have you — can never be more than a starting point in that endeavour. There is always the possibility that there are other (non-quantifiable) variables – of vital importance and although perhaps unobservable and non-additive not necessarily epistemologically inaccessible – that were not considered for the formalized mathematical model.

The kinds of laws and relations that ‘modern’ economics has established, are laws and relations about mathematically formalized entities in models that presuppose causal mechanisms being atomistic and additive. When causal mechanisms operate in real world social target systems they only do it in ever-changing and unstable combinations where the whole is more than a mechanical sum of parts. If economic regularities obtain they do it (as a rule) only because we engineered them for that purpose. Outside man-made mathematical-statistical ‘nomological machines’ they are rare, or even non-existant. Unfortunately that also makes most of contemporary mainstream endeavours of mathematical economic modeling rather useless. And that also goes for Krugman and the rest of the ‘New Keynesian’ family.

The DeLong-Summers-Krugman discussion is certainly a question of methodology. And it shows the danger of neglecting methodological issues — issues mainstream economists regularly have almost put an honour in neglecting.

Being able to model a ‘disciplined’ credible world, a world that somehow could be considered real or similar to the real world, is not the same as investigating the real world. Even though all theories are false, since they simplify, they may still possibly serve our pursuit of truth. But then they cannot be unrealistic or false in any way. The falsehood or unrealisticness has to be qualified (in terms of resemblance, relevance, etc.). At the very least, the minimalist demand on models in terms of credibility has to give away to a stronger epistemic demand of appropriate similarity and plausibility. One could of course also ask for a sensitivity or robustness analysis, but the credible world, even after having tested it for sensitivity and robustness, can still be a far way from reality – and unfortunately often in ways we know are important. Robustness of claims in a model does not per se give a warrant for exporting the claims to real world target systems.

In his plaidoyer for disciplining thought by the ‘reasoning tools’ that we call models, Krugman puts too much emphasis on modelling as an epistemic genre. Even if epistemology is important and interesting in itself, it ought never to be anything but secondary in science. The primary questions asked have to be ontological. First after having asked questions about ontology can we start thinking about what and how we can know anything about the world. If we do that, I think it is more or less necessary also to be more critical of the reasoning by modelling that has come to be considered the only and right way to reason in mainstream economics for more than sixty years now.

On the all-important question of ‘external validity’ of economic models, Krugman obviously halts at stressing the heuristic epistemological value of disciplining thought by modeling it:

What, after all, are economic models for? They are definitely not Truth. They are, however, a way to make sure that the stories you tell hang together, that they involve some plausible combination of individual behavior and interaction of those plausibly behaving individuals.

Read literally this is, of course, an absolutely absurd standpoint. If we can’t warrant that the premises (assumptions) on which our model conclusions build are true, then what’s the value of the logically correct deductions we are supposed make with our models? From false assumptions anything logically follows!

Krugman, as most other mainstream economists, subscribes (although not very consciously or explicitly) to a deductive-nomological view on scientific explanation and prediction (an explanation of an event being nothing but a prediction of  its occurrence), in which prediction and explanation  are things to deduce from law-like hypotheses and a set of antecedent/initial conditions.  But — and on this both Hempel and Popper were very explicit — to count as adequate/sound, the explanans must be true. Explanation and prediction in the models we construct and use is not only a question of logical form.  Models that intend to say something about the real world — and the policy models that DeLong, Summers and Krugman discuss certainly do — can’t escape dealing with ‘Truth’!

Model reasoning as an ‘object to enquire’ into activities, is anyway not, from a scientific point of view, on a par with the much more important question if these models really have export-certificates to the real world or not.

Questions of external validity are important more specifically also when it comes to microfounded policy models. It can never be enough that these models somehow are regarded as internally consistent. One always also has to pose questions of consistency with the data. Internal consistency without external validity is worth nothing.

Yours truly has for many years been urging economists to pay attention to the ontological foundations of their assumptions and models. Sad to say, economists have not paid much attention — and so modern economics has become increasingly irrelevant to the understanding of the real world.

I have spent a considerable part of my life building economic models, and examining the models that other economists have built. I believe that I am making reasonably good use of my talents in an attempt to understand the social world.

I have no fellow-feeling with those economic theorists who, off the record at seminars and conferences, admit that they are only playing a game with other theorists. If their models are not intended seriously, I want to say (and do say when I feel sufficiently combative), why do they expect me to spend my time listening to their expositions? Count me out of the game.

Robert Sugden



RSS feed for comments on this post. TrackBack URI

  1. Excellent

    Models are for the entertainment of children

    If there were any “models” in the General Theory, I didn’t see them as such


    I guess I was just too gobsmacked by the depth of the thinking – about how the world actually works

  2. “not the same as investigating the real world. ”
    The other problem is that the real world has been subject to policies that are derived from the imaginary models for some considerable time now. So much so that it can be said that we have a Procrustean Economy – where all outcomes are crushed into a shape that fits the models. So we can have no idea what a 5ft or 7ft economy looks like in the real world since all of them have now been stretched or crushed into the 6ft Neo-Keynesian bed. There are no 5ft or 7ft real economies to test variant models against.

  3. A very important discussion. What we need here are some historians and philosophers (and it seems mathematicians as well) who need to explain and provide some good examples of why the use of models, contrary to what Krugman asserts, are in fact a very, very, bad idea.

    At the very best they are useless because they do not tell us what the causation mechanisms are. At the very worse they are can lead to serious damage if they are intellectual basis of policy formulation because people have not properly investigated what the causation mechanisms are (ie what the causes are of the problem we are trying to deal with). There is absolutely nothing in Krugman’s or anyone else’s model that tells us what we need to know: ie what is the actual cause of the so-called ‘ liquidity trap’. Keynes does give one possibility: uncertainty and animal spirits. To be sure though that that is the cause, you need primary (which includes non-quantifiable) evidence from borrowers, lenders, corporations, governments, central banks and much else. You might find out that this is the answer. It might not be. Most likely it will be part of the answer but there are other factors at play. All of this you will need to know to get the right policy prescription.

    Your model is not miraculously going to substitute for the hard field work of actually finding out what is going on.

    • Krugman should be given credit where it is deserved. He says:

      “What model of the inflation process do you have in which an expansion of the Fed’s balance sheet translates into inflation without causing an overheating of the labor market first? I’m not saying that there is no possible story along those lines, but spell it out so we can see how plausible it is.”

      I must say I often seen studies which show money supply growth and inflation and they say that one is the cause of the other without spelling out the mechanism. But really making up a story to get a model is not the answer. Again it is not about the model, it is about finding out the real cause. To understand the causal mechanism you have to find evidence that shows that behind the inflation is money supply growth. A lot of interviews, a lot of primary documented material…. I’m not saying it is easy, it is a hell of a lot of time-consuming work ( and involves a lot of resources). But that is what a serious investigation involves. And it is the only way.

  4. The future or economics: why you will probably not be admitted to it, and why this is a good thing
    Comment on ‘DeLong, Summers & Krugman on models’
    “… economics is a big omnibus which contains many passengers of incommensurable interests and abilities.” (Schumpeter, 1994, p. 827)
    Economics is a scientific failure. Being stranded in the middle of nowhere, evidently, no one other is responsible than the confused drivers/passengers of the big omnibus themselves. These can be roughly divided into four sects: Walrasians, Keynesians, Marxians, and Austrians. What all have in common is substandard scientific abilities. Generally speaking, the four approaches are built upon unacceptable premises and therefore violate Aristotle’s first principle of science: “When the premises are certain, true, and primary, and the conclusion formally follows from them, this is demonstration, and produces scientific knowledge of a thing.” (Posterior Analytics)
    What are the premises that are accepted by the majority of economists? Krugman put it thus “most of what I and many others do is sorta-kinda neoclassical because it takes the maximization-and-equilibrium world as a starting point …”. This starting point has to be abandoned because this premises are by no stretch of the imagination certain, true, and primary.
    At this critical juncture, the economist has to make up his mind: either to defend the indefensible beliefs of one of the four sects or to replace the foundational assumptions and to begin in earnest with the overdue reconstruction of the whole theoretical superstructure of economics. Based on the history of economic thought it is a fair bet that the representative economist will mess up things. However, this has to be proved, so here is the challenge.
    The most elementary economic configuration is the pure consumption economy which consists of the household and the business sector which in turn consists initially of one giant fully integrated firm. This minimalist configuration is defined for one period by three equations.
    (i) Yw=WL wage income Yw is equal to wage rate W times working hours L,

    (ii) O=RL output O is equal to productivity R times working hours L,

    (iii) C=PX consumption expenditure C is equal to price P times quantity bought/sold X.
    If you cannot understand or accept these almost self-evident equations, which hold for the world economy as a whole and every closed national economy, you are out of economics. These premises are certain, true, and primary, or stated in relative terms, obviously superior to the neo-Walrasian axioms (Weintraub, 1985, p. 147) or to Keynes’s defective formal basis (1973, p. 63).
    For the graphical representation of the three equations see here At any given level of employment L, the wage income Yw that is generated in the consolidated business sector follows by multiplication with the wage rate W. On the real side, output O follows by multiplication with the productivity R. Finally, the price P follows as the dependent variable under the conditions of budget balancing, i.e. C=Yw, and market clearing, i.e. X=O. Note that the ray in the southeastern quadrant is NOT a linear production function; the ray tracks ANY underlying production function. Note also that the wage rate W is an AVERAGE if the individual wage rates are different among the employees, which is the general case.
    Under the conditions of market clearing and budget balancing in each period the price follows from the three equations as P=W/R (1), i.e. the market clearing price is always equal to unit wage costs. To repeat, the price is here taken as the dependent variable, of course, it can be treated as an independent variable (2015). Also to mention is that money as transaction medium is left out here for brevity; for the full picture see (2015).
    The elementary consumption economy works as follows. If the wage rate W is lowered, the market clearing price P falls. If the number of working hours L is increased the price remains constant, provided productivity R does not change. If productivity decreases the price P rises. If productivity increases the price falls. In any case, labor gets the whole product, the real wage W/P is invariably equal to the productivity R according to (1), and profit for the business sector as a whole is zero. All changes in the system are fully reflected by the market clearing price P. The pure consumption economy is reproducible for an indefinite number of periods.
    The changes from period to period are formally given by:

    (iv) Wt=Wt-1(1+wt) The wage rate in period Wt is given by the wage rate in the previous period Wt-1 and the rate of change for the current period wt.

    (v) Rt=Rt-1(1+rt) Analogous for the productivity.

    (vi) Lt=Lt-1(1+lt) Analogous for labor input.
    The rates of change for future periods wt, rt, lt are random variables with an a priori unknown distribution function. Because of this we cannot predict the price in period t=10 but we can test it in period t=10 or any other future period. As a matter of principle, (1) is a testable proposition.
    Given the enumerated premises and conditions, the market clearing price performs a random walk which is determined in turn by the random walks of wage rate and productivity. Equation (1) in combination with (iv) to (vi) replaces the ridiculous supply-demand-equilibrium model of econ101.
    From the premises and conditions follows for a start.

    — The elementary economy constitutes itself through the interaction of real AND nominal variables. There is no such thing as a ‘real’ economy, in other words, ALL ‘real’ models are a priori false.
    — There is no such thing as an equilibrium. The product market is cleared and the budget is balanced but the economy is not moved by an Invisible Hand towards some equilibrium, e.g. full employment. In other words, ALL equilibrium models are a priori false.
    — The commonplace quantity theory does not hold. Inflation/deflation depends initially on the development of wage rate and productivity and nothing else. If the period changes of the wage rate are exactly equal to the changes of the productivity i.e. wt=rt, absolute price stability prevails over all future periods despite the random variations of productivity and employment. Hence, price stability/inflation/deflation is not a monetary phenomenon.
    — The marginal principle, which ultimately derives from the unacceptable behavioral assumption of utility maximization, does not apply and plays no role at all for the price determination. ALL marginalist models are a priori false.
    — Supply and demand functions are nonentities. Well-behaved production functions and decreasing returns do not exist. These premises are neither required nor admissible.
    — Neither the income distribution nor the distribution of the real product depends on the marginal principle. ALL marginalist distribution models are a priori false.
    In the next analytical steps the number of firms is increased, which leads to the determination of relative prices, and the conditions of market clearing and budget balancing are lifted, which gives rise to the phenomena of inventory changes, profit/loss*, and the increase/decrease of the stocks of money and debt. It is pretty obvious that all economic phenomena can successively be derived from the three premises (i) to (iii). Stock-flow consistency is guaranteed ab initio.
    In order to develop the first economic theory since Adam Smith that satisfies the scientific criteria of formal and material consistency the unacceptable foundational propositions of Walrasianism, Keynesianism, Marxism, and Austrianism have to be abandoned. Scientists would immediately perform the necessary paradigm shift but economists are no scientists. Their acceptance of the maximization-and-equilibrium world for more than 100 years is forever disqualifying. Because of this, the still confused sorta-kinda economists are kindly asked to leave the big omnibus now.
    Egmont Kakarot-Handtke
    Kakarot-Handtke, E. (2015). Major Defects of the Market Economy. SSRN Working
    Paper Series, 2624350: 1–40. URL
    Keynes, J. M. (1973). The General Theory of Employment Interest and Money.
    The Collected Writings of John Maynard Keynes Vol. VII. London, Basingstoke:
    Schumpeter, J. A. (1994). History of Economic Analysis. New York, NY: Oxford
    University Press.
    Weintraub, E. R. (1985). Joan Robinson’s Critique of Equilibrium: An Appraisal.
    American Economic Review, Papers and Proceedings, 75(2): 146–149. URL
    * See ‘Profit and the collective failure of economists’

    • “…Keynes’s defective formal basis (1973, p. 63).”

      Would it be possible for you to specify this defective formal basis?

      • Answer to Bragi
        This is the corpus delicti from the General Theory “Income = value of output = consumption + investment. Saving = income – consumption. Therefore saving = investment.” (Keynes, 1973, p. 63)
        This two-liner is conceptually and logically defective because Keynes did not come to grips with profit theory.
        “His Collected Writings show that he wrestled to solve the Profit Puzzle up till the semi-final versions of his GT but in the end he gave up and discarded the draft chapter dealing with it.” (Tómasson et al., 2010, pp. 12-13, 16)
        Because profit is ill-defined the whole theoretical superstructure of Keynesianism is false, in particular all I=S and IS-LM models.
        For the formal proof see ‘Why Post Keynesianism Is Not Yet a Science’
        Keynesians will not make it into the future of economics because of proven logical incompetence over more than 80 years.
        Egmont Kakarot-Handtke
        Keynes, J. M. (1973). The General Theory of Employment Interest and Money. The Collected Writings of John Maynard Keynes Vol. VII. London, Basingstoke: Macmillan.
        Tómasson, G., and Bezemer, D. J. (2010). What is the Source of Profit and
        Interest? A Classical Conundrum Reconsidered. MPRA Paper, 20557: 1–34.

  5. This is indeed a very necessary reflection, but I have my doubts if one can bring down a single mainstream economic model by such a general discussion. In order to achieve this, it is necessary to complement the general critique with a specific one, and also to show an alternative to the criticised model.
    In the case of the so-called Ricardian trade model – arguably the most famous international trade model –, I have tried to accomplished it here:
    The proposed alternative is quite surprising, namely Ricardo’s original numerical example in the Principles.

  6. Larry Summers referenced work by Olivier Blanchard et alia.
    It is interesting to read that text as an example of the empiricism of the model thinkers.
    Blanchard doesn’t “export” his analytic model to the world. What he does is stylize the facts so he can import them into the analytic model. For example, his model has a policy rate, so he stylizes a fact, a selected interest rate on one type of bond, and identifies it as the policy rate.
    This method — stylizing facts so they can be “imported” into an analytical model — results in a highly stilted and awkward way of talking or writing. The points Blanchard and his colleagues are making about policy dilemmas are not without interest. But, without reference to an institutional context, they cannot be politically informative as to the stakes in an interested contest over policy.

  7. Summary on ‘Musings on Whether We Consciously Know More or Less than What Is in Our Models…’
    “In order to tell the politicians and practitioners something about causes and best means, the economist needs the true theory or else he has not much more to offer than educated common sense or his personal opinion.” (Stigum, 1991, p. 30)
    Economists do not have the true theory, they merely have a heap of incoherent models. These models are stand-alone constructs that are not held together by an overarching theory. To begin with, the relationship between theory (e.g. gravitation) and model (e.g. solar system) is poorly understood. In economics one has models but no theory.
    The models in turn are built on premises that are methodologically unacceptable. As a rule, economist do not know what is in their models, see ‘The future of economics: why you will probably not be admitted to it, and why this is a good thing’
    The common logical blunder of Walrasians, Keynesians, Marxians, and Austrians is that the elementary concepts profit and income are incorrectly defined. Yet, profit is the pivotal concept for the analysis of how the economy works. Without a correct profit theory economics is vacuous. Conventional profit theory is logically indefensible and because of this the familiar economic models fail to capture the essence of the market economy. There are many opinions but no scientific understanding of how the economy works.
    In general it holds that the representative economist is at best dimly aware of the crucial distinction between opinion and knowledge, which constitutes science since Aristotle: “There cannot be both opinion and knowledge of the same thing at the same time.” (Posterior Analytics, Wikipedia)
    The discussion of Paul Krugman, Larry Summers, and Brad DeLong about methodology is roughly on the level of old witch hunters who exchange their opinions about whether the incubus or the succubus is a greater threat to humanity.
    The scientific embarrassment of this sitcom consists in the plain fact that economists have no idea of what is in their logically defective, i.e. provably false, models.
    Egmont Kakarot-Handtke
    Stigum, B. P. (1991). Toward a Formal Science of Economics: The Axiomatic
    Method in Economics and Econometrics. Cambridge, MA: MIT Press.

Leave a Reply

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

You are commenting using your 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 )

Google+ photo

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

Connecting to %s

Blog at
Entries and comments feeds.