Tag Archives: methodology

Classical Economics and Methodology

On Classical Economics

The methodological question at issues during the classical period were very similar to those that were late to agitate neoclassical and modern economists. The controversies in classical methodology included:

  1. Abstractions versus “reality
  2. Varying concepts of causation
  3. The role of mathematics
  4. The “scientific” claim of economics, and
  5. The practical relevance of classical economics

Adam Smith’s Methodology

Adam Smith’s Methodology was eclectic. The empirical, the theoretical, the institutional, the philosophical, and the dynamic were all intermingled. His definitions shifted, sometimes on the same page, and, in the course of developing his classic work, he drifted back and forth between difference conceptions of “value,” “rent,” and “real.” But, despite the numerous ambiguities in The Wealth of Nations commented on by Smith’s classical successors, as well as by the late scholars, he moved easily around the pitfalls without disaster, being sufficiently consistent during any given chain of reasoning to avoid errors in logic.

David Ricardo’s Methodology

With Ricardo economic took a major step toward abstract models, rigid an artificial definitions, syllogistic reasoning – and the direct application of the results to policy. The historical and institutional, and the empirical faded into the background, and explicit social philosophy shrank into a few passing remarks. Comparative static became the dominant – though usually implicit – approach. Ricardo declared: “I put these immediate and temporary effects quite aside, and fix my whole attention on the permanent state of things which will result from them.” Not only Ricardo, but also his disciples and popularizers, reasoned in comparative static terms – and they automatically interpreted the theories of others in comparative static terms as well.


1. Sowell, Thomas. On Classical Economics. New Haven: Yale UP, 2006. Print. 79-80.


Leave a comment

Filed under Economics, History, Notes on Economics, Uncategorized

Alfred Marshall and the Historical School

I’m currently reading Robert Skidelsky’s wonderful book, John Maynard Keynes: Hopes Betrayed, 1883-1920. I’m an enormous fan of his work, this book being no exception. However, even the best historians are sloppy at times. Writing about the great economist Alfred Marshall, Skidelsky claims that:

Marshall was also much more receptive that his predecessors had been to the use of history to discover economic laws – the so-called inductive as opposed to the deductive method – though he rejected the main conventions of German historical school (which would have destroyed economics’ claim to be a science) (Skidelsky 1983: pg 43)

This interpretation of “Alfred Marshall vs. the historical school” is fairly common among neoclassical economists, most notably Lionel Robbins, but I wouldn’t expect Skidelsky to make this error. Regardless, I suppose it’s time to set the record straight.

In fairness to Skidelsky (and others), Marshall did reject the extreme empiricism of the British and “old” historical school of economics. In his inaugural lecture, Marshall criticized the “extreme wing” of the school, claiming that their obsession with mere observations explains nothing:

Facts by themselves are silent. Observation discovers nothing directly of the actions of causes, but only of sequences in time. It may find that an event followed on, or that coincided with a certain group of other events. But this gives no guidance except for other cases, in which exactly the same set of facts occur over again, grouped in just the same way…. In economic or other social problems no event has ever been an exact precedent for another. The conditions of human life are so various: every event in the complex result of so many causes, so closely interwoven that the past can never throw a simple and direct light on the future. (Marshall 1885: pg 41)

He then went on to echo Carl Menger, saying that facts need to be “examined and interpreted by reason.” The selection and grouping of facts are subject to various assumptions that need to be carefully examined.

Marshall’s rejection of this “extreme wing” was highlighted during an exchange he had with British historicist William Cunningham. In his paper, The Perversion of Economic History, Cunningham criticized Marshall, attacking the assumptions in his economics:

The underlying assumption against which I wish to protest is never explicitly formulated by those who rely on it; but it may I think, be not unfairly expressed in some such terms as these. That the same motives have been at work in all ages, and have produced similar results, and that, therefore, it is possible to formulate economic laws which describe the action of economic causes at all times and in all places. (Cunningham 1892b: pg. 493).

Instead of concerning itself with universal economics laws, economics should be about “the actual world we live in”. The real world, according to Cunningham, is built upon empirical knowledge. It has no universality. In order to cope with the relativity of economic doctrine, economics should be concerned with classification and description. Economists should be preoccupied with taxonomy.

Obviously there are major problems with Cunningham’s argument. The taxonomic empiricism he embraced was subject to Menger’s critique of the historical school. Even taxonomy requires some universal principles before it can start classifying things. But in the grand scheme, Cunningham’s ideas aren’t particularly new for the time and in many ways they reflect the doctrines of the “old” historical school of Wilhelm Roscher. What’s more important is Marshall’s response to Cunningham. Firstly, he didn’t really go into any methodological depth. For someone who was supposedly diametrically opposed to the historical school, that’s a bit odd. Secondly, he concedes quite a bit to Cunningham and admits that historical inquiry is essential for economic theory. He accepts the idea that many economic laws may not apply to a particular historical age without modification. He even devotes a paragraph in praising the historical school and “the study of economic history.” Within the context of economic thought at the time, these statements from Marshall show that he was not an intellectual enemy of the German historical school. Far from it. In order to see why, we need to look at the “younger” historical school in Germany.

Carl Menger’s methodological attack on the historical school in 1883 stimulated an active debate in German speaking countries (also called the Methodenstreit) and the historical school’s response was lead by the economist Gustav von Schmoller. Unlike his predecessors, Schmoller recognized the limits of empiricism:

Observation and description, definition and classification are the preparatory activities. But what we desire to reach thereby is a knowledge of the interdependence of economic phenomena…. Induction and deduction are both needed for scientific thought as the right and left foot are both needed for walking. *

As opposed to empirical enquiry, Schmoller put more emphasis on institutions, how they affected individuals in an economic system, and the historical sensitivity of economic theory. Given his response to Cunningham, it’s almost certain that Marshall would have conceded much to Schmoller and the younger historical school. I’m not saying that he would agree with everything Schmoller had to say. But Marshall recognized many of the problems that the historicists emphasized and he was prepared to accept the role of history in economics. Some might even say that “Marshall himself was a product and part of the historical school tradition” (Hodgson 2005: pg 342) I’m not sure I would go that far, but there was definitely significant overlap between his ideas and the historical school’s.

It’s fairly clear, to me at least, that much of the misinformed intellectual history on the historical school of economics is the result of conflation. Rather than look at the intricacies and the development of the school, some historians of economic thought (particularly Austrians) merely conflate the younger school with the older and declare that the entire historical school was primarily concerned with some naive empiricism. Any honest look reveals that the historical school was not uniform in its methodology. Many historicists, e.g. Schmoller, Weber, and Sombart, developed economic theories and didn’t rely on the empirical methods of Roscher and others. Far from being an antagonist, Marshall should be seen as an ally, perhaps even an extension, of the historical school of economics.

*Marshall cited this exact passage from Schmoller in his Principles of Economics


1. Cunningham, W. (1892a), “The relativity of economic doctrines”, Economic Journal, Vol. 2 No. 1, pp. 1-16.

2. Cunningham, W. (1892b), “The perversion of economic history”, Economic Journal, Vol. 2 No. 3, pp. 491-506.

3. Hodgson, Geoffrey M. (2005), “Alfred Marshall versus the Historical School?” Journal of Economic Studies, 32.4, pp. 331-48.

4. Hodgson, Geoffrey Martin. How Economics Forgot History: The Problem of Historical Specificity in Social Science. London: Routledge, 2002. Print.

5. Marshall, Alfred. The Present Position of Economics. An Inaugural Lecture given in the Senate House at Cambridge, 24 February, 1885. London: Macmillan, 1885. Print.

6. Marshall, A. (1892), “A reply to ‘The perversion of economic history’ by Dr Cunningham”, Economic Journal, Vol. 2, pp. 507-19.

7. Skidelsky, Robert. John Maynard Keynes Volume One, Hopes Betrayed, 1883-1920:. New York: Viking, 1983. Print.


Filed under Economics, History, Home

Debunking Economics Part 5: Not all assumptions are created equal

It’s a cornerstone of heterodox economics to criticize the  methodology of neoclassical economics and Chapter 8 of Steve Keen’s Debunking Economics holds true to this tradition. At one point in the history of economic thought, the majority of neoclassical economists might have agreed with Milton Friedman’s classic paper The Methodology of Positive Economics. Whether that is still the case, I’m not entirely sure. However, based on the current state of economic models, I would not be entirely surprised if a majority of neoclassical economists today were at least sympathetic to Friedman’s views.

In this paper, Friedman claims that a theory should judged by its predictive power:

Viewed as a body of substantive hypotheses, theory is to be judged by its predictive power for the class of phenomena which it is intended to “explain.” Only factual evidence can show whether it is “right” or “wrong” or, better, tentatively “accepted” as valid or “rejected.” As I shall argue at greater length below, the only relevant test of the validity of a hypothesis is comparison I of its predictions with experience. The hypothesis is rejected if its predictions are contradicted (“frequently” or more often than predictions from an alternative hypothesis); it is accepted if its predictions are not contradicted; great confidence is attached to it if it has survived many opportunities for contradiction. Factual evidence can never “prove” a hypothesis; it can only fail to disprove it, which is what we generally mean when we say, somewhat inexactly, that the hypothesis has been “confirmed” by experience.

A theory should be judged by its “fruitfulness” , i.e. the precision of its predictions and how wide its implications are within a particular field, not by its realism. Friedman took this idea a step further by claiming that all important and significant theories are filled with unrealistic assumptions:

In so far as a theory can be said to have “assumptions” at all, and in so far as their “realism” can be judged independently of the validity of predictions, the relation between the significance of a theory and the “realism” of its “assumptions” is almost the opposite of that suggested by the view under criticism. Truly important and significant hypotheses will be found to have “assumptions” that are wildly inaccurate descriptive representations of reality, and, in general, the more significant the theory, the more unrealistic the assumptions (in this sense). The reason is simple. A hypothesis is important if it “explains” much by little, that is, if it abstracts the common and crucial elements from the mass of complex and detailed circumstances surrounding the phenomena to be explained and permits valid predictions on the basis of them alone. To be important, therefore, a hypothesis must be descriptively false in its assumptions; it takes account of, and accounts for, none of the many other attendant circumstances, since its very success shows them to be irrelevant for the phenomena to be explained.

To put this point less paradoxically, the relevant question to ask about the “assumptions” of a theory is not whether they are descriptively “realistic,” for they never are, but whether they are sufficiently good approximations for the purpose in hand. And this question can be answered only by seeing whether the theory works, which means whether it yields sufficiently accurate predictions. The two supposedly independent tests thus reduce to one test.

According to Friedman, the assumptions of any theory cannot possibly be thoroughly “realistic” in any descriptive sense. Any realistic theory of the market would have to include a multitude of seemingly relevant details. Any attempt to achieve this type of realism would render a theory useless. Even if we were to take a less logical extreme and assert that the assumptions of a particular theory are “too” unrealistic, there is no basis or criterion for this:

What is the criterion by which to judge whether a particular departure from realism is or is not acceptable? Why is it more “unrealistic” in analyzing business behavior to neglect the magnitude of businessmen’s costs than the color of their eyes? The obvious answer is because the first makes more difference to business behavior than the second; but there is no way of knowing that this is so simply by observing that businessmen do have costs of different magnitudes and eyes of different color. Clearly it can only be known by comparing the effect on the discrepancy between actual and predicted behavior of taking the one factor or the other into account. Even the most extreme proponents of realistic assumptions are thus necessarily driven to reject their own criterion and to accept the test by prediction when they classify alternative assumptions as more or less realistic.

While this argument for instrumentalism might seem very plausible, it suffers from several flaws. In his paper, ‘Unreal Assumptions’ in Economic Theory: The F-Twist UntwistedAlan Musgrave discusses these flaws. By distinguishing between 3 different types of assumptions, negligibility, domain, and heuristic, Musgrave shows the lack of clarity surrounding Friedman’s paper and his use of the word “assumption”.

Negligibility Assumptions:

According to Musgrave, negligibility assumptions state that certain aspects of reality have little to no effect on the topic at hand. For example, when Galileo investigated the motion of falling bodies, he assumed that air resistance had no detectable effect on these motions. So when Friedman is talking about these types of assumptions, he is correct to say that a theory containing negligibility assumptions can only be evaluated through empirical corroboration. However, he is wrong to conclude that these assumptions are “unrealistic”. Rather than asserting that specific factors are absent, these assumptions claim that these factors are “irrelevant for the phenomena to be explained” (Friedman’s claim about eye color in the paragraph above would fit under this description).

Domain Assumptions: 

When a theory is tested and fails to make accurate predictions, it may be because of a certain negligibility assumption. The particular factor that was thought to be negligible does in fact have a significant effect after all. It might be concluded that the theory will only work when this factor is absent. From this, we specify the domain of applicability of a the theory. These types of assumptions are called domain assumptions. Let’s say an economic theory assumes that the government has a balanced budget. Given such an assumption, we may conclude that any budget imbalance has negligible effects. However, we can also conclude that opposite. A budget imbalance will have a significant effect and this economic theory will only apply under conditions where this imbalance does not exist.

The switch from negligibility assumptions to domain assumptions is very important and it’s something that Friedman ignores. This switch is an ad hoc modification that replaces a stronger or more testable theory with a weaker or less testable one. As a result, some domain assumptions can render a theory untestable. If we assume that a theory only holds under a balanced budget, but the government never has a balanced budget, then this theory cannot be tested. So if we value testability (which Friedman does), we should hope that the domain assumptions in a particular theory are not always false or “unrealistic”. This is in complete contrast to Friedman’s claim that “to be important, therefore, a hypothesis must be descriptively false in its assumptions.

Heuristic Assumptions:

Heuristic assumptions are assumptions that are known to be false, but are used as stepping stones towards a more general theory. As Musgrave elaborate, “the ‘assumption’ that factor F is negligible is merely a heuristic device, a way of simplifying the logical development of the theory.” Newton’s assumption of a one planet solar system when trying to prove that planets followed elliptical orbits was a heuristic assumption. This was not a negligibility assumption, as Newton knew that planets would sometimes have detectable and meaningful gravitational effects on one another. Nor was this a domain assumptions, as Newton was not saying that his theory only applied to one planet solar systems. Newton used this assumption as a stepping stone, which was eventually used by Henri Poincaré and others to create more precise predictions and better theories on planetary behavior:

The modern theory of planetary behavior now recognizes that the stable orbits of our solar system can only have evolved – over an enormous period of time – from far less stable orbits, which must have led to collisions between proto-planets. It is now accepted that the moon, for example, was the product of a collision between another proto-planet and the early earth. (Keen 2011: 162)

So contrary to Friedman’s paper, not all assumptions are created equal and an “important” theory is not filled with unrealistic assumptions.


Assumptions, Assumption, Assumptions 


1. Keen, Steve. Debunking Economics: The Naked Emperor Dethroned? London: Zed Book, 2011. Print.

2. Friedman, Milton (1953). The Methodology of Positive Economics. Essays in Positive Economics. University of Chicago Press. 3-43.

3. Musgrave, Alan. “Unreal Assumptions In Economic Theory: The F-Twist Untwisted.” Kyklos 34.3 (1981): 377-87. Print.

1 Comment

Filed under Economics, Home