Predrag Rajsic, who many of you may know through Mises Daily, wrote a short post on two different approaches to economic modelling. He describes one approach as aiming to simplify reality, to be able “to describe minute details.” This is contrasted to the approach of Friedrich Hayek and Ronald Coase, whom instead used models to “[examine] the implications of our assumptions.” I disagree that these are two distinct ways of going about economics, believing that they are really two sides of the same coin. In fact, most well-known economists make use of both methods when modeling their theories.
Rajsic starts from perfect competition, rightly pointing out that perfectly competitive models are not capable of explaining some of the most interesting aspects of the real world economy. There is no role for the firm, for information search, or, amongst many other things, for the various institutions that characterize reality. Ironically, perfectly competitive markets cannot even explain competition, because the assumptions of the model essentially describe a situation where all opportunities for entrepreneurship and competition have been arbitraged away — after all, in this model, firms cannot differentiate their products and all firms are price takers. Rajsic then notes that, by challenging the assumptions in the model of perfect competition, Coase and Hayek were able to describe aspects of the economy that would otherwise be unexplainable. But, this is not the complete story.
I too will start with perfect competition, noting that this model has a long history that precedes both Hayek’s and Coase’s involvement in economics (see, for an introduction, George Stigler, “Perfect Competition, Historically Contemplated“). To understand why such a model would even be developed, we have to ask what type of questions the developers were seeking to answer. Frank Knight, for instance, developed a rigorous theory of perfect competition to use this model as a foil for his discussion on the role of uncertainty. Earlier economists developed their own, perhaps less complete, models of perfect competition to help illustrate and explain the notion of the equalization of returns. In context, then, perfect competition does not seem so useless. Out of context and when aiming to explain other problems, perfect competition loses some of its glamor.
Believe it or not, Hayek and Coase were not the first, or only, economists to abandon perfect competition when looking to answer different questions. Thus, out of questions raised in the 1920s, the 1930s saw the development of what is called “imperfect competition.” These are markets characterized by firms with increasing returns to scale, product differentiation, market power, et cetera. These models were developed to answer questions that perfect competition cannot. Interestingly, immediately after imperfect competition was introduced to the profession, namely through Joan Robinson and Edward Chamberlin, these bundles of theory were not necessarily well received. Some, like Stigler and Milton Friedman, questioned how much value concepts like monopolistic competition really added over perfect competition. Some, such as Hayek (“Economics and Knowledge“), were more interested in developing a theory of intertemporal, or dynamic, equilibrium. Hayek also argued (“The Meaning of Competition“) that the concept of “imperfect” markets, thus holding “perfect” markets as a standard to achieve, fails to acknowledge that “perfect” markets can only ever be used as a foil to describe the actual process of competition.
Hayek’s critique makes it seem as if imperfectly competitive models were used to put into doubt the desirability of markets. This sentiment is not necessarily misplaced, given that economists like Piero Sraffa — whose 1926 “The Laws of Return Under Competitive Conditions” helped inspire the “imperfect revolution” — and Joan Robinson were sympathetic towards political ideas that Hayek was not, and it makes sense that they use their theories to support certain normative conclusions. But, this should not lead anybody to discard the usefulness of imperfect competition. Opinions on significance and meaning of the conclusions are products of interpretation, not features of the model themselves. It is only natural for economists to disagree on what the model implies.
But, when we judge imperfectly competitive models within the context of the questions being asked, we see that they do offer much of value to the economist. Thanks to these models, we can construct tractable explanations of economic processes that not only went unexplained, but were almost unacknowledged. Consider, for instance, the significance of the introduction of the concept of increasing economies of scale to the theory of trade (and the division of labor). Beforehand, traditional theories, namely comparative advantage, could not explain the real world very well. Even expansions on the original Ricardian theory were hit-and-miss, such as Hecksher’s and Ohlin’s theory of factor endowments and trade patterns. Likewise, these models help explain product diversity and markets affected by risk. More recently, and with varying success, imperfectly competitive models have also helped answer questions in the area of economic growth (see, for instance, Paul Romer, “Increasing Returns and Long-Run Growth“).
Hayek and Coase, to some extent or another, followed different routes in diverging from perfect competition. But, in the process of doing so, they created alternative models. These models may not be formal and rigorous, but they are models nonetheless. The very act of relaxing and changing assumptions is part of developing a new model. And even when economists, including Hayek and Coase, write without explicit mention of modelling, ideal types are implicit. If we could fathom the real world without simplifying, or abstracting away from reality, economics would be very different. But, because real worlds are complex and difficult to make sense of, we have to relax some of the rigor of reality when attempting to make sense of it. Coase, for example, abstracted away from concepts like vertical integration, because it would have complicated the point he was trying to make — yet, we don’t question the significance of Coase’s finding.
The point is not to place perfect and imperfect competition on a pedestal. There are as many models as there are economists. These models are useful in some cases and not useful in others. The relevance of a model is determined by the question being asked by the economist, and to judge these models outside the context of these specific questions is to leave unacknowledged their purpose. Further, when we come upon a question that cannot be answered by an existing model, we have to create alternative explanatory ideal types — we do this by relaxing and changing assumptions. Thus, these two approaches are two sides of the same coin. We cannot have one without the other.
We often disagree on how models are interpreted. I, for example, took Mark Thoma to task for his interpretation of perfect competition, efficiency, and the role of government. But, these aren’t problems with modeling, they’re problems with interpretation (I recently made the same argument in the context of the economics of knowledge). The fact is, whatever our gripe with one model or another, we cannot escape the reality of their necessity for the understanding of the real world.