Matthew T. Clements*
St. Edward’s University
Economics is never truly value-free. The subjective aspects of economic analysis are underappreciated and should be more explicitly recognized. The presence of subjectivity need not detract from the value of economics research in understanding social phenomena.
“Like people in general, social scientists are apt to conceal valuations and conflicts between valuations by stating their positions as if they were simply logical inferences from the facts. Since, like ordinary people, they suppress valuations as valuations and give only “reasons,” their perception of reality easily becomes distorted, that is, biased.”
Gunnar Myrdal (1969, p. 50)
In Charles Ferguson’s 2010 documentary film Inside Job, the chair of a prestigious economics department makes a bold claim: that there is no need for economists to disclose any financial interests in their research, the clear implication being that the research is not influenced by the values or incentives of the researcher. It is unusual to hear such a strong statement, and perhaps few economists would take this view even if pressed. Kincaid et al (2007) argue that scientists of all kinds typically understand that there can be no sharp distinction between facts and values, that the two will both be present in their work; but that scientists still act as if the research they do is entirely value-free, treating results as if they are objective truth, to be viewed as either correct or incorrect. In this paper I argue that economic analysis involves subjectivity in a number of respects and that this aspect of economics is underappreciated; that awareness of this subjectivity should influence the way we conduct economic analysis as well as how we discuss the results; and that the presence of subjectivity need not detract from the value of economics research in understanding social phenomena.
Even when there is no overt claim to objectivity, there is often a presumption that the bulk of economic research is objective. Consider the distinction between positive and normative economics. While the former is generally defined as the analysis of how things work and is not explicitly taken to be objective, in practice it is often treated that way. The contrast to normative economics, which addresses what we should do and necessarily involves value judgment, encourages this presumption. Furthermore, economic principles are, in principle, objective:
There are facts and patterns that exist independently from our perception of them, and it is reasonable to strive for objectivity in our attempts to understand reality.
However, the results of a piece of economic research can benefit the researcher in a number of ways, and there is no reason to think that these benefits have no influence on the research itself. An economist may be more likely to pursue research that aligns with his own values or political beliefs. There may be various financial incentives, such as supporting the objectives of potential consulting clients for the researcher, or promoting government policy that benefits the researcher’s own socio-economic class. Publishing research of any kind leads to employment, tenure, and promotion, and there is thus an incentive to create publishable work.
Since the judgment of what is publishable is in the hands of other economists, themselves subject to their own biases, there is an imperfect correlation between what is publishable and what is true. On the other hand, the quest for greater understanding is itself a motivation for research and may well be the primary incentive behind most economic research, but it is unrealistic to think that economists are able to disregard all other motivations for the sake of their work.
This is not to say that the process is corrupt or that all research is advocacy. Rather, the factors noted above may influence a researcher in small, subtle ways. Consider this example: A medical doctor sometimes orders MRI scans for his patients. Together with the other doctors in his practice, he sets up a lab in the same building, where patients may go for MRI scans as well as other diagnostic tests. This is likely to be a considerable convenience for the patients, who would not need to make separate appointments in different locations to have these tests done.
However, if the doctor in general practice has a financial interest in the lab, he has incentive to order MRI scans more often than is necessary. This is not to say that the doctor consciously orders gratuitous MRIs simply to line his own pockets, but rather that the doctor cannot help
being influenced by his own benefit in the form of lab revenues. In a marginal case, where the MRI may or may not be indicated based on the medical assessment alone, the financial consideration could nudge the doctor in the direction of ordering the MRI. It is because of this kind of influence that some decision-makers, such as public officials, may be urged to avoid even the appearance of impropriety. No professional in any field can claim to be immune from such influences.
This is not a new idea—consider Myrdal (1969), quoted above—and it has sporadically reappeared in the economics literature. Putnam and Walsh (2011) have argued from a philosophical perspective that the positive-normative dichotomy is untenable. Although many may agree that economics is fundamentally subjective, the profession as a whole has never really embraced the idea. Here I focus on the practical issues: how subjectivity arises in economic analysis, and how any piece of economic research should take this into account.
Note that the presence of subjectivity does not mean that any opinion or argument is as good as any other or that advancing our understanding of social phenomena is hopeless. Even a subjective argument can be debated and critically evaluated. Whether through peer review or subsequent work, criticism of an argument can always focus on the research itself, and does not have to be concerned with the incentives of the researcher. The influence of values on research need not be an obstacle to useful research, but it should be acknowledged to a greater extent.
Methodology and mathematics
Certainly there is good reason for the extensive use of mathematics among academic economists: A mathematical argument makes a logical argument explicit and draws attention to any weaknesses; it is not difficult to make an appealing verbal argument that can be shown to be flawed when translated into mathematical terms. Use of a mathematical model illustrates multiple effects, and aids the researcher in sorting through such effects, when it may be inhumanly difficult otherwise. If an argument can be translated into mathematical terms, and there is some benefit in doing so, then it is worthwhile to perform a mathematical analysis.
Mathematical proof is just a small step away from logical argument at its purest, and as such is commonly accepted as objective. Even this notion can be challenged: That two plus two equals four relies on a set of axioms upon which there is implicit agreement. The goal of any mathematical proof is to convince someone—reader, grader, referee—that a result is true, and there can be legitimate disagreement about this, even when we agree upon axioms or assumptions.
Even if we take the math itself to be objective, there are potential problems that arise through the use of mathematics. In economic research, it is essential to recognize that the mathematical exercise is part of a larger argument, not the argument itself. One pitfall is that mathematical rigor is often taken to be synonymous with intellectual rigor, when in fact the use of math is neither a necessary nor a sufficient condition for the validity or value of an argument. There are many means of argumentation, some of which are qualitative, but the profession tends not to look favorably upon them. Graduate students and tenure-track faculty face considerable pressure to make their research mathematically complex.1 There is room for debate over what precisely constitutes intellectual rigor, but surely the use of mathematics should not be the decisive criterion. While qualitative research may have its own pitfalls, any argument can be peer-reviewed in its own right.
In some cases, a mathematical model does not provide any insight that a less formal argument would; e.g., if the results are obvious given the hypotheses. It does not necessarily follow that the issue addressed is not interesting, or that it is not important to address it. It may be that the research question is in the hypotheses, i.e. that the argument is not what the model would show us, but which is the right model to use. Other research might focus on variables that are not amenable to parameterization or measurement. Here again, this does not mean that the issue is not interesting or should not be addressed. The subject of economics is not defined by the use of a specific set of analytical tools, and it would be a mistake for the profession to ignore some issues because they are not suitable subjects for mathematical analysis. As Firebaugh (2008) says, the method should be the servant, not the master. Research that is in part driven by the desire for mathematical complexity can, at its worst, obfuscate rather than illuminate2 or lead to mathematical conclusions that are not economically meaningful.
Another potential problem is that the use of mathematics, either theoretically or empirically, lends a veneer of objectivity to the argument as a whole. This may be due in part to the analytical similarity to the natural sciences, which are even more commonly viewed as objective. However, ever since Kuhn’s Structure of Scientific Revolutions (1996), originally published in 1962, philosophers of science have widely acknowledged that even the natural sciences are laden with values.3 Whether in the realm of natural or social science, there is always the question of what mathematical analysis means in terms of a larger argument. Consider a study purporting to “prove scientifically that smoking causes cancer.” If we assume that the study correctly used appropriate methods to gather and analyze data, the only reasonably objective statement that can be made is something like this: In the study group, which is representative of the population of interest, there was a higher incidence of cancer among smokers than non-smokers, and there is only a small probability that these results could have arisen from randomness in the data. How to interpret or apply the results of the study beyond this fundamental statement is open to debate, at least to some degree. It is reasonable for a consensus to form among a community of scientists, especially after a number of studies have led to the same result; but any standard for what is required for such a consensus is necessarily subjective. For a piece of economic theory, if we again assume the soundness of the mathematical analysis itself, the reasonableness of assumptions and the validity of conclusions will always be debatable. For example, consider a signaling model in which consumers rationally infer the unobservable quality of a product. If we agree that the mathematical result is sound, we can still question whether the model is sufficiently realistic, in what contexts it is or is not applicable, or whether there are potential complications that would change the result significantly. In other words, there can be a substantial difference between a mathematical result and a real-world conclusion. The model is part of an argument that has other necessary components. I elaborate on this point in the next section.
Modeling assumptions and conclusions
Spiegler and Milberg (2009) examine economic methodology in their criticism of the application of mainstream modeling approaches to issues in institutional economics. The thrust of their argument is that the “formal models are too parsimonious to meaningfully illuminate the complex institutions they ostensibly represent” (p. 290). This kind of criticism, that a model does not apply to the phenomenon under discussion and thus the results of the analysis do not apply, is
not limited to institutional matters. Spiegler and Milberg delineate the following steps in researching economic phenomena in general (p. 294):
- Delimiting, in which the set of social phenomena under study is specified and a research question is
- Naming, in which a mathematical construct meant to be analogous to the social phenomena is
introduced, along with a ‘catalog of correspondences’ which links elements of the construct with elements of the phenomena under study.
- Solution, in which the mathematical construct is brought to a
- Interpretation, in which the mathematical solution and its implications are interpreted with respect to the research
The authors draw attention to “an important divide in the analysis—i.e. the divide between the realm of ordinary language descriptions… and mathematical language descriptions” (p. 294). Crossing this divide, i.e. proceeding from delimiting to naming and from solution to interpretation, is a necessary part of any mathematical economics research. Even if the derivation of the solution is objective, the other steps clearly are not.
Referees do tend to consider these steps, at least implicitly, in their evaluation of a paper submitted for publication. However, the reviewer’s support or criticism tends to focus on whether these steps were executed correctly, as if there were an objective answer to that question. That it is subjective does not mean that the referee need only weigh in as to agreement or disagreement. An important part of the peer-review process is the critique of the author’s thinking on these subjective issues. In some cases this warrants as much or more attention than the solution of the model.
A related issue is what we consider to be data or evidence, and what is the most appropriate way to deal with such data. Often it is possible to argue for the existence of stylized facts that cannot be measured with any precision, and where attempts at measurement would be subject to strong potential objections: for example, “The average American does not invest much time or effort in learning about political candidates before voting.” It might then be more straightforward, as well as more compelling, to build an argument upon the stylized facts. Such arguments are often dismissed out of hand, with reference to the evidence as merely anecdotal. The evidence may or may not be strong enough to justify the conclusions, but this is one of the very points of argument that peer review must consider. The bottom line of any argument is whether it is convincing.
Selection of topics
Questions of the significance of a contribution, the importance of a topic, and whether it adds significant insight to policy or previous work are all facets of the question of how interesting the research is. Readers and referees of research may willingly acknowledge that this is subjective. However, the focus of the profession as a whole, in terms of what topics are regarded as worthy of attention, is arguably too narrow, and certainly narrower than it could be. That there is even such a thing as “heterodox” economics, i.e. a considerable body of research outside the mainstream, testifies to this. Students of economics, including graduate students, are unlikely to have much exposure to anything outside the mainstream; and so even PhDs in economics might be only marginally aware of alternative approaches or ideas.
Following on to the subjectivity of what questions are interesting, I would argue that there should be more discussion of what issues we should be studying. If the profession does not take a broad view, we may miss opportunities to understand important economic phenomena, and thereby diminish our credibility. For example, there has been a great deal of research about the financial system that cannot address the financial crisis of 2008. This in itself is not a problem; there is much we can do to explore the workings of the financial system under the assumption that the system is for the most part working the way it is supposed to. But if we only focus on the workings of the system, and not the potential failings of the system, we can miss the potential for a crisis (not to mention predicting the actual crisis), and then have to scramble to understand it in retrospect.
As another example, consider the history of industrial organization. This field has offered countless insights into the functioning of imperfectly competitive markets, relying heavily on game-theoretic analysis. Prior to the 1980s, criticisms of perfect competition came primarily from heterodox schools of thought. It seems that mainstream economics did not take much interest in questions of imperfect competition until good analytic tools for dealing with them were available, i.e. until game theory had been widely integrated into economic thinking. Before this change, the model of perfect competition was more prominent in mainstream economics.
The importance of questions of imperfect competition did not change over time, but the profession’s attention to them did.
Any number of research topics began as subjects of heterodox research and have eventually moved into the mainstream of economic thought. Too much focus on topics that are currently popular can lead to marginalization of some research without regard for the value of the research itself. If for no other reason than the inherent complexity of economic phenomena, the study of economics cries out for a pluralist approach.4 At the very least, alternative viewpoints should always be welcomed as subjects for debate. Even in a case where a heterodox argument does not hold up in the face of a conflicting mainstream argument, the mainstream argument is strengthened by engaging in this conflict. Of course, assessing the value of an argument and evaluating competing arguments are subjective exercises; but if we are interested in getting at some underlying truth, more open debate is always better.
It is in making recommendations for what should be done that economists are most likely to recognize subjectivity. Still, one must realize that the “positive” conclusions upon which recommendations are made are not entirely objective. Furthermore, it can be difficult for non- economists to distinguish subjective conclusions from the research that is, in principle, objective. Referring to the “normative” part of research as that which makes use of a social welfare function, as is often done, misses an important point. The statement that a particular policy maximizes a given social welfare function is a positive rather than a normative conclusion. The real normative issue is how to think about social welfare: what is the right social welfare function to use, and why. If, hypothetically, we can all agree on a notion of social welfare, then of course we should maximize it.
Regarding many policy issues, it is easy for non-economists to get the impression that economists only value efficiency. This may well be true for some economists, but it is not in the nature of economics itself. Equity considerations are often given cursory attention in undergraduate classes, as Smith (2018) points out. We better serve our students and the rest of the world if we acknowledge that equity does matter; that there can be considerable disagreement over what is equitable, as well as the relative importance of equity and efficiency; and that, as economists, we do not have any more insight into equity than anyone else.
Consider trade as an example. Support of free trade is often based on the idea, both theoretically and empirically supported, that trade benefits everyone in the aggregate. Taking this as given does not lead directly to the conclusion that there should be free trade. The problems that trade potentially generates, such as short-term unemployment caused by the removal of trade barriers, certainly matter, and how we view the trade-off between costs and benefits of trade, and what we should do about it, is a complex and value-laden question.
Of course we will focus our research on efficiency, since our expertise speaks to this. But it is important to recognize that it is not the only thing that matters, and to acknowledge when we are stepping outside of our economic expertise in making a value judgment. Policy recommendations always involve values, and it is worth noting what values are being supported.
I have described several ways in which economics research is subjective. This subjectivity does not devalue the research, but it is something that must be acknowledged. As the author of this article, I am naturally influenced by my own values, and I have an obvious interest in publishing this piece. This does not imply that the argument I have presented here cannot be evaluated on its face. Any evaluation of a piece of research is itself subjective, but like a grading rubric or a judge’s instruction to a jury, it is reasonably straightforward to specify the criteria (which are to be subjectively evaluated) concretely. A greater recognition of the subjectivity involved in economics research would potentially influence what research is done, how it is done, and how the results are communicated. This is something for authors, reviewers, readers, and teachers to keep in mind as we do the work that we do.
Chick, V., and S.C. Dow (2001). “Formalism, Logic and Reality: a Keynesian Analysis,”
Cambridge Journal of Economics 25, 705-21. Ferguson, C. (2010). Inside Job. Sony Pictures Classics.
Firebaugh, G. (2008). Seven Rules for Social Research. Princeton: Princeton University Press.
Frank, R. H. (2007). The Economic Naturalist: In Search of Explanations for Everyday Enigmas. New York: Basic Books.
Kincaid, H., J. Dupre, and A. Wylie (eds.) (2007). Value-Free Science: Ideals and Illusions?
Oxford University Press.
Kuhn, T. (1996). The Structure of Scientific Revolutions. Chicago and London: University of Chicago Press (3rd ed).
Kurz, H.D., and N. Salvadori (2000). Understanding “Classical” Economics: Studies in Long- Period Theory. London: Routledge.
Myrdal, G. (1969). Objectivity in Social Research. London: Duckworth & Co.
Putnam, H., and V. Walsh (eds.) (2011). The End of Value-Free Economics. Oxford: Routledge.
Ratzsch, D. (2001) Nature, Design and Science: The Status of Design in Natural Science. State University of New York Press.
Smith, N. (2018). “Econ Majors Graduate With a Huge Knowledge Gap,” Bloomberg, April 13.
Spiegler, P., and W. Milberg (2009). “The Taming of Institutions in Economics: The Rise and Methodology of the ‘New New Institutionalism,’” Journal of Institutional Economics, 5(3), 289- 313.
1 Frank (2007) gives a rationale for why academic economists have too much incentive to use complex mathematics in their work: that it is a means of signaling their abilities to other economists.
2 Chick and Dow (2001) elaborate on this point.
3 See also Ratzsch (2001). Both Kuhn and Ratzsch speak to the issues in the following sections as well.
4 See also Kurz and Salvatori (2000).