The need for methodology in the construction of economic theories (with Joan Miranda)

Originally published in La Rotonda (in Spanish). Co-authored with Joan Miranda.

Although it is impossible to deny Economics its status as a science, it becomes necessary to ask what the methodological requirements upon which it rests actually are — questions such as: What is the logical structure of a theory? What properties characterize the set of propositions that constitute a theory? What are the common fallacies in theory construction? How are causal relationships established? And when can a theory be said to be invalid?

Painting: Wassily Kandinsky

I. Introduction

In their daily work, economists draw on a body of theoretical knowledge that allows them to grasp — in an ideal sense — the movement of social reality. Simply put, this body of knowledge is called “theory,” while the “grasp” it enables is nothing other than the instrument of the scientist, serving both to “explain” and to “predict.” As will become clear throughout this article, the methodological structure across the empirical sciences — including Economics — is analogous¹; on this point, however, it is worth noting the following: it is the very nature of the object under study that imposes real limits on a given science, thereby defining its field of inquiry, as well as the manner of investigation. Thus, while natural scientists study relations between material objects, social scientists study relations between people within a given society (Figueroa, 2008). This demarcation does not imply, of course, a juxtaposition of different branches of knowledge²; rather, it concerns the study of a specific form of knowledge as part of a broader genus of scientific explanation (Blaug, 1985 [1980]).

Although it is impossible to deny Economics its status as a science, it becomes necessary to ask what the methodological requirements upon which it rests actually are — questions such as: What is the logical structure of a theory? What properties characterize the set of propositions that constitute a theory? What are the common fallacies in theory construction? How are causal relationships established? And when can a theory be said to be invalid? The exposition that follows is organized around these questions.

II. The Propositions of an Economic Theory

In general, economic theories are systems of statements or logical propositions. Their structures are a complex mixture: on one hand, they are systematic, hierarchical, and ordered languages of reasoning; on the other hand, they are statements that abstract essential features of reality (Friedman, 1967 [1953a]). That is, propositions must comply with certain rules that protect them from errors in the logical relations they establish with one another. Nevertheless, the rules of knowledge — as a point of support — come from outside science itself: from epistemology (Figueroa, 2008).

Economic propositions are not of just any kind: they are empirical. This requires the following distinction. The propositions of the formal sciences differ from those of the empirical sciences: the former require only that the principle of non-contradiction be satisfied in order to determine the truth of propositions, whereas the latter additionally require a confrontation with reality in order to be “true.” For this, empirical propositions must be susceptible to testing, since it may happen that, however empirical they may be, they are never refuted³.

Consider the following case (Figueroa, 2008):

“A shaman tells his client: if you have faith, my medicine will heal you.” This proposition is irrefutable. For if it does not come true, the shaman can always reply that the client lacked faith.

III. Common Fallacies in the Propositions of an Economic Theory

As mentioned earlier, one condition a theory must meet is that it must be logically correct. However, scientists are not free from committing fallacies — that is, arguments that appear to be correct but are not logically so.

Our science is made up of diverse theories that share a single objective: to explain the functioning of human society (Figueroa, 2016). A scientific theory uses the logical device of initially analyzing the behavior of a representative individual⁴, and then proceeding to explain aggregate behavior. The scientist must not ignore, however, the interactions that exist between individuals when moving from one level of analysis to the other — whether from individual to aggregate or vice versa — since otherwise one risks committing a fallacy of composition in either direction. This fallacy takes, respectively, two forms: i) the fallacy of aggregation and ii) the fallacy of division.

Consider the following proposition:

“If a borrower cannot repay their debt to a bank, he or she is in trouble. If all borrowers cannot repay, then all borrowers are in trouble.”

In fact, if all borrowers cannot repay, it is the bank that faces a serious problem, not the borrowers. This is an example of a fallacy of composition — specifically a fallacy of aggregation, in which an individual-level observation generates a false conclusion at the aggregate level.

On the other hand, an example of a fallacy of division would be: “If national income increases, then the income of all individuals will also increase.” A society can find itself in a situation of macroeconomic boom and microeconomic distress (Schuldt, 2004), meaning that the virtues of macroeconomic growth do not necessarily translate into improvements in individual welfare.

The most important product of a theory is to reveal causal relationships, which the scientist must test through statistical methods. This testing provides provisional grounds for maintaining the theory as valid. However, in some cases one falls into the fallacy of statistical causality — that is, when the scientist claims that a causal relationship derives from the correlation between two variables. First, the scientist must bear in mind that correlation is a statistical description showing the degree of association between two variable samples. Second, this association does not provide the information needed to distinguish which is the dependent and which is the independent variable. And finally, association does not mean explanation; the latter comes from theory alone. In econometrics there is a well-known and widely used statistical test: the Granger Causality Test. This test analyzes the sequence of two variables and, on the premise that the future cannot affect the past, concludes that the lagged variable is the Granger cause of the future values of the other variable (Montero, 2013). This test, however, has been criticized for being atheoretical, purely instrumentalist in the strictest sense, and for abandoning any pretension of integrating econometrics with economic theory (Arestis & TAIFA, 2004). This leads us to reflect on the importance of analyzing data in the light of theory.

It is often believed that the ultimate purpose of economics is to forecast variables; if it fails in its forecasts, it should not be considered a science. The social sciences use information from the past to falsify their theories, since this procedure can only be carried out with observed data. A theory that has successfully passed the falsification process will not be able to forecast the exact value a variable will take in the future, but it will be able to identify which variables determine that value.

The causal propositions derived from a theory can be expressed in a simplified form. Suppose:

Y = f( X₁(+), X₂(+), X₃(−) )

Variable Y depends directly on X₁ and X₂ but inversely on X₃. The interaction of these three variables in the model can predict the behavior of Y but cannot forecast its exact value. An economic theory cannot determine the value of the endogenous variable in the future because the theory itself does not determine the values of the exogenous variables; rather, these draw on information from the past. In many cases, the challenge for econometricians will be to draw on a theory that illuminates the most relevant and determinant variables to incorporate into their model. Once again, the role that theory plays becomes all the more evident.

So far, this article has developed some premises that approximate a theory of economic theories — that is, a methodology of scientific knowledge that demands the “logical steps” necessary for constructing economic theories. These “logical steps” entail the following negative requirement: an economic theory must be open to refutation by experience. In other words, empirical testing can only yield “positive decisions” that temporarily validate a theory, because “negative decisions” can always overturn it (Popper, 1962). Economic theories are in perpetual flux.

References

Arestis, P., & TAIFA, S. d. (2004). Crítica a la economía ortodoxa. Barcelona: Universidad Autónoma de Barcelona.

Blaug, M. (1985 [1980]). La metodología de la economía o cómo explican los economistas. Madrid: Alianza Editorial.

Figueroa, A. (2008). Nuestro mundo social: introducción a la ciencia económica. Lima: Libros PUCP.

Figueroa, A. (2016). Rules for scientific research in economics: The alpha-beta method. Springer.

Friedman, M. (1967 [1953a]). La metodología de la economía positiva en: Ensayos de economía positiva. Madrid: Gredos.

Georgescu-Roegen, N. (2007). Ensayos bioeconómicos. Madrid: Los Libros de la Catarata.

Jevons, W. S. (1998 [1871]). La teoría de la economía política. Madrid: Ediciones Pirámide.

Montero, R. (2013). Test de causalidad. Documentos de Trabajo en Economía Aplicada. Universidad de Granada.

Popper, K. (1962). La lógica de la investigación científica. Madrid: Editorial Tecnos.

Schuldt, J. (2004). Bonanza macroeconómica y malestar microeconómico (Vol. I). Lima: Fondo Editorial, Universidad del Pacífico.