On the limits of the invisible hand

17 Jul, 2016 at 13:43 | Posted in Economics | 5 Comments

It might look trivial at first sight, but what Harold Hotelling did show in his classic paper Stability in Competition (1929) was that there are cases when Adam Smith’s invisible hand doesn’t actually produce a social optimum.

With the advent of neoclassical economics at the end of the 19th century a large amount of intellectual energy was invested in trying to formalize the stringent conditions of obtaining equilibrium and showing in what way the prices and quantities of free competition constituted some kind of social optimum.

That the equilibrium reached in free competition is an optimum for each individual – given prevailing prices and income distribution – was not, however, seen by some economists as making a very strong case for a free market economy per se. It wasn’t possible to prove that free trade and competition gave a maximum of social utility. The gains made in exchange weren’t a manifestation of a maximum social utility.

wicksell2Knut Wicksell was one of those who criticized the idea of regarding the gain in utility arising from free competition as an absolute maximum. This market fundamentalist idea of harmony in a free market system didn’t live up to Wicksell’s demand for objectivity in science – and  “the harmony economists, who endeavoured to extend the doctrine so that it might become a defence of the existing distribution of wealth” were judged severely by Wicksell (Lectures 1934 (1901) p. 39).

When propounders of the new marginalist theory – especially Walras and Pareto – overstepped the strict boundaries of science and used it in ascribing to the market properties it did not possess, Wicksell had to react. To Wicksell (Lectures 1934 (1901) p. 73) it was

almost tragic that Walras … imagined that he had found the rigorous proof … merely because he clothed in mathematical formula the very arguments which he considered insufficient when they were expressed in ordinary language.

But what about the Pareto criterion? Wicksell had actually more or less anticipated it in his review (in Zeitschrift für Volkswirtschaft, Sozialpolitik und Verwaltung, 1913: 132-51) of Pareto’s Manuel, but didn’t think it really contributed anything useful. It was just the same old doctrine in a new disguise. To Wicksell the market fundamentalist doctrine of the Lausanne School obviously didn’t constitute an advance in economics.

From a methodological point of view there are also one or two lessons to learn from this history.

Models may help us to explain things by providing us with a frame/instrument for analysing and explaining the real world. However, to do that, there has to be an adequate similarity between model and reality. Otherwise they cannot function as eye openers that widen our cognitive horizon and make it possible to see and detect fundamental forces/ relations/mechanisms operating in the real world. And — most importantly — we always have to scrutinize the assumptions the models build on and so test their plausibility.

Logic, coherence, consistency, simplicity, and deductivity is not enough. Without confronting models with the real world, they are nothing but empty thought experiments — and should be treated as such.


  1. Nice post. A bit confused on social optimum part. Walrus and Pareto were assuming efficiency and not some social normative optimum. Also. You Nobel guy has been ennobled:

    I took out my GPS roamer
    and asked where is Paul Romer
    out roaming among the poor
    endogenuously helping them produce more
    Helas, I just heard he’s walking into the World Bank’s door.

  2. Once again the confusion of what constitutes micro- and macro-economics is introduced with apparently wrong happening results. We don’t expect the “invisible hand” to apply when only a few producers and kinds of products are being considered, but when the Big Picture is considered as a whole, its basic principle does become manifest. This means that for competing companies selling the same product there is a tendency for them to group together in a small region, but when we consider the whole of a large city, there is a more uniform distribution of these same kinds of shops over the whole community.

    We should be studying how big the community needs to be before it becomes worthwhile for this trend to dominate over the former one of the “all together” sales illustrated in the video clip. I think it is different for the various kinds of goods being sold. Thus food shops are more likely to be uniformly spread compared to automobile show-rooms.

  3. Models may help us to explain things by providing us with a frame/instrument for analysing and explaining the real world. However, to do that, there has to be an adequate similarity between model and reality.
    Analysis and synthesis, analytic models and operational models, concepts and measurements — these are distinct, except in the minds of economists.
    It isn’t the similarity between analytical models and reality, or lack thereof, which is the problem, but rather the false expectation that analysis is in any way descriptive or predictive.
    The problem isn’t that Walras’s tatonnement is insufficiently descriptive; the problem is in imagining that the analytical mode of thought in which Walras was engaged is ever descriptive. A geometry is not descriptive. To describe, interpret, map and manipulate the world are activities quite distinct from the practice of geometry. Geometry may aid the cartographer or navigator in her thought process, but making a map requires observation and measurement.
    What’s epistemically wrong with neoclassical economics and its complacent defense of laissez faire, is that they were arguing against observation and measurement. Theory’s analysis of what optimal means as a concept is transformed into an assertion that the world’s arrangements are optimal and no measurement is necessary or even possible. Pareto’s argument is that inquiring into the justice or efficiency of the actual distribution of wealth is beyond the pale of science.
    The economist doing a field study of actual institutions in operation is never in any danger of discovering tatonnement in operation, anymore than a surveyor will encounter any triangles lying upon the ground or trapezoids hanging from trees. But, if we are ever to learn anything about the actual economy, we must study the actual economy and its institutions and history. We cannot expect economists to do that as long as they are required to imagine that analytic models are descriptive and ceteris paribus a synonym for by-the-by.
    Whether an analytic theory lays out a coherent system in which to define optimal choice of resource allocation, price and output conceptually is a question for theorists to work out. Whatever determination is reached, it will not magically become a description of any actual market. If it is to admit an effective empiricism, economics must find methods to measure quantitatively just how efficient an actual institutional arrangement is, in operation. Against such an empiricism, neoclassical economics erected an effective blockade.
    Let’s reject Walras and Pareto and their heirs, like Lucas, for that blockade.

    • I agree. Models should not even be and cannot even be “instruments for explaining and analysing the real world”. In fact they distract us from understanding and analysing the real world. Almost certainly the most interesting and important things are the very things that cannot be explained or adequately represented by a model – sorry Romer you will not find the answers to why some economies grow and others do not using a model – you will however by going to the pre-MIT development studies literature. Unfortunately we have to go back to go forwards and basically trash the massive diversion MIT has created. Unfortunately a lot of hard won knowledge based on meticulous ground up field work has been forgotten thanks to MIT led formalisation. (MIT is actually a worse culprit in this department than Chicago.) This has not contributed to an increased understanding of why countries grow, why some do not, and they have completely ignored why some countries decline. (On a related point – there is also a difference I think between a theory and a model but for neo-classical economists they are often one and the same.) Where models may have a role, but I think as a very limited and cautiously used one, is (as one type of) empirical estimating – for example estimating the income and substitution effects of a tax cut. But they should never be a means of explaining how the world works, and most certainly should not form the centre part of that analysis or determine that analysis from square one.

  4. Bruce Wilder, you make some excellent points. However, I don’t totally agree that economics has not moved to measurement and empiricism. The problem as DeVroey shows in his recent book on Macro is that economists confuse their theorizing with actualizing. The logical constructs bring forth many good points and insights that do not necessarily have anything to do predictively or descriptively with real empirical issues. In a way, their abstract Geometry gets lost in translation.

Sorry, the comment form is closed at this time.

Blog at WordPress.com.
Entries and comments feeds.