A professor of mine once recalled in a lecture on modern portfolio theory, an aphorism from statistics. He told us, “You should probably know, before I move any further, all models are wrong”. To be sure, he was missing the latter half of the rather cutting observation made by British statistician George E.P Box; "All models are wrong, but some are useful". The observation is startling particularly for practitioners of finance and economics where the basis for almost all decision-making rests at least in some regard on a model or algorithm, which can only approximate the reality of things. Of course, to fully comprehend, for example, the trajectory or cumulative activity of a national economy, we would first have to address all of its individual agents (i.e households, firms, governmental entities, etc.) and anticipate their behaviour. For most economic problems that are of any consequence to practitioners and policy-makers , this exercise in developing a micro-level understanding of a system quickly becomes untenable and arguably impossible. The impossibility of ever fully comprehending a complex system, like a national economy, provides a sort of philosophical necessity for simplifying assumptions as part of these models - we would scarcely be able to understand anything otherwise.
While assumptions are wholly necessary and indeed both prudent and time-efficient, they are fundamentally distortions of reality - the more we make, the more likely we are to err, and the further our understanding will deviate from the true nature of things. It’s an interesting thought for both a student and practitioner to consider, particularly within the fields economics or finance where so much depends on modelling, accounting, and forecasting; where it is impossible to account for every possible variable, distortion or factor when building financial or socioeconomic models, which considerations are most important? And if a model must make some simplifying assumptions, are these simplifications realistic and how might they preclude use of its insights and conclusions for real-world applications ?
A recent paper, published by Professor Paul Pfleiderer of Yale University, made a bizarre claim regarding specific theoretical economic models that he had read about in academic journals; he referred to them as chameleons. Pfleiderer argues that “a model becomes a chameleon when it is built on assumptions with dubious connections to the real world but nevertheless has conclusions that are uncritically (or not critically enough) applied to understanding our economy.” Further, when the model’s critical assumptions are challenged, the model “changes colour” and resumes its original state as what Professor Pfleiderer refers to as a “bookshelf model”.
We have all, at some point, encountered these sorts of models in our economics and finance classes. In many cases, I must admit that these theoretical models are introduced and taught with a fair deal of scrutiny. One might recall, for example, the Modligliani-Miller theory of capital structure whereby the authors attempt to derive the factors which are pertinent to capital structure decisions by beginning with a “perfect world” absent any market distortions or frictions. Is a “perfect world” assumption realistic? Of course not. In the case of the Modigliani-Miller study, beginning from this Platonic condition and working backwards is a critical part of the analytical framework involved in capital structure theory – the model’s insights are far from dubious since they indicate to us clearly which market distortions are relevant to capital structure decisions . However, it would have been intellectually dishonest, according to Pfleifeder, if the authors had suggested that their “perfect world” theory (for example, the Irrelevance Theorem of the capital structure decision) had any applicability to real-world corporate decision-making on capital structure.
"As Pfleifeder notes, the trouble with “chameleon” theoretical models is how a model’s conclusions are presented and “uncritically applied” to real-world situations."
Yet as Pfleifeder notes, the trouble with “chameleon” theoretical models is how a model’s conclusions are presented and “uncritically applied” to real-world situations. Of course, this issue largely derives from the way that these theoretical models are presented in academic journals, and he makes this case with a particularly interesting example of a working paper titled “Why high leverage is optimal for banks” (2013). The authors, DeAngleo and Stulz, state that “To establish that high bank leverage is the natural (distortion-free) result of intermediation focused on liquid-claim production, the model rules out agency problems, deposit insurance, taxes, and all other distortionary factors […]and shows clearly that, if one extends the MM model to take that role into account, it is optimal for banks to have high leverage”. Pfleifeder, in a mocking way shows an equivalent conclusion derived by the same logic regarding alcohol consumption: “To establish that high intake of alcohol is the natural (distortion free) result of human liquid drink consumption, the model rules out liver disease, DUIs, health benefits, spousal abuse, job loss and all other distortionary factors […] and shows clearly that if one extends the alcohol neutral model to take that role into account, it is optimal for humans to be drinking most of their waking hours.” Each of the examples are demonstrative of what Pfleifeder refers to as chameleons, theoretical models that claim to have derived relevant insights about practical economic matters when in fact the supposed “insight” has been skewed by a careful selection of assumptions. More precisely, in each example, the omittance of factors that were critical to a proper understanding of the system at-large, whether it be a banking system exposed to high leverage or the human body exposed to high-levels of alcohol consumption.
Jokes and mockery aside – the conclusions are (or may be) entirely correct – but the presentation of the DeAngleo-Stulz argument is disingenuous at best, and only adds “noise” to an otherwise intellectual discourse on capital requirement considerations. Pfeifeder notes, importantly, that The Economist included the study’s findings in a 2018 article titled “Capital punishment: forcing banks to hold more capital may not always be wise”. In the paper, the writers argue that the DeAngelo-Stulz study shows “that it is better for banks to be highly levered even without frictions like deposit insurance and implicit guarantees.”. The use of the findings are clearly dishonest - in an attempt to make a serious argument about how capital requirements ought to be treated in the real world (where banks had failed due in part to historically high leverage), using insights derived from a perfect world. If one were to contest the findings of the article, derived from the results of the DeAngelo-Stulz study, the authors might rebut by exclaiming that all models are wrong, and based on at least a couple of simplifying assumptions or omissions. Further, they might claim that the results of the DeAngelo-Stulz study are valid until one can prove that its underlying assumptions are false or inappropriate – this is a much simpler task. In this way, the economic model morphs; the use of its conclusions are apparently not subject to the “real-world filters”, as Pfleifeder describes them; filters that would otherwise separate models which are either useful or totally impractical to use to understand “real-world” problems. The model and its assumptions do not in itself define a model as a chameleon; its use, by either its authors or others who attempt to use its findings to justify their own ends, make it so.
Seeking out chameleons in academic work and in the media is an impressively difficult task; it requires a sort of filtering system that rejects certain theoretical economic models strictly based on the realism of its assumptions. From an academic perspective, we are obliged to change the way that we think about modelling real-world economic problems. Rather than to fit our assumptions to a particular conclusion, economic models should be derived by the sort of logic that makes the Modigliani-Miller capital structure theory a foundational component of the traditional curriculum in Finance and Economics.
Some believe that this “chameleon” problem is unique to the field of modern economics, which has already in part, been addressed in other social and natural science disciplines. In his 2013 magnum opus, “Capital in the Twenty-First Century”, Tomas Piketty makes a rather blunt claim about economics as an academic discipline stating that it “has yet to get over its childish passion for mathematics and for purely theoretical and highly ideological speculation […] acquiring the appearance of scientificity without having to answer the far more complex questions posed by the world we live in”. His solution to this problem is rather simple: “start from square one, so that there is some hope of making progress”.
Far from being an exercise in abstract mathematics, economics as a discipline strives to grasp the dynamics of human interaction in the face of scarcity. Approaching economic problems in a manner which makes sturdy assumptions about the way economic agents behave in practice and presenting theoretical insights bona fide is the first key step in avoiding the sly act of crafting or presenting chameleon arguments, and in turn, being able to spot them out in the wild.