Phil 345 Essay 1

Essay 1 §

Friedman argues that a theory should be accepted if it makes well confirmed predictions for its “intended domain.” Do economic models such as Akerlofs or Schelling’s support or undermine Friedman’s “methodology of positive economics?”

I claim that neither Akerlof’s nor Schelling’s models comply with Friedman’s methodology of positive economics, but they should nevertheless be considered useful for their potential. I invoke Sugden’s conception of credible worlds (CW) to argue that while both models cannot be empirically verified, they represent parallel worlds that could become credible with more refinement and thus should be considered useful.

The scientific process is often simplified as a linear chain of hypothesis, modeling, testing, and accepting or rejecting. This view is oversimplified not only because it assumes linear theory generation without providing a method for generating hypotheses, but also because it presumes the entire process is conducted by one person or a small team over a short period. In reality, a falsifiable scientific hypothesis is generated through a tortuous, winding path, encountering dead ends and following beliefs, myths, and hunches, often involving decades or even centuries of random search by scientists worldwide. From this viewpoint, Friedman’s methodology is both:

1. Valid and consistent with existing paradigms of the philosophy of science.
2. Utterly useless for contemporary positive economics because much of economics is still in its infancy as a science

Friedman’s argument that “theory is to be judged by its predictive power for the class of phenomena which it is intended to explain” aligns with concepts like falsifiability, instrumentalism, and robustness, as well as methods in the hard sciences like physics. (Friedman, 1953) Within the scientific process, this corresponds to the “verification/falsification” step. However, inventing hypotheses ripe for such verification is often the more difficult step, especially for economists, considering that (i) economics rests on human psychology, and despite J.S. Mill’s hopes, we lack a precise theory for it, and (ii) there are too many variables, as highlighted by Rosenberg. (Friedman, 21; Rosenberg, 300) Friedman glosses over this: ”[…] besides [empirical evidence’s] obvious value in suggesting new hypotheses […],” despite it being the most crucial inductive step, often referred to as creativity or, in Sugden’s Schema 3: Abduction, the inductive leap from real-world evidence to abstraction in a credible, parallel world. (Friedman, 7; Sugen, 20) Induction from the real to the CW is one type of induction; the induction from the credible to the real is the popular one—namely, empiricism—with the former having taken a backseat.¹ While Friedman offers a good process for the latter, he fails to address the more crucial former step: the generation of hypotheses, or, in Sugden’s conception, the induction of CWs.

As Sugden argues, the importance of Akerlof’s and Schelling’s papers is that they “give structure to a statement that is often made about the real world,” i.e., they induce a credible world from real-world phenomena. (Sugden, 3) A market for used cars is a “how-possibly” explanation for a theory of asymmetric information, as is the checkerboard argument for segregation. Their significance lies not in their findings but in their inductive leap from evidence (e.g., segregation exists) into a CW. While not yet fully fleshed out to take the second inductive step of prediction and being “abstract and unrealistic,” they have “introduced to economics the concept of asymmetric information,” sparking off a whole branch of economics: the economics of information. (Sugden, 2) The value of Akerlof’s paper is in abstracting from real-world phenomena to conceptualize asymmetric information. Refining, extending, and verifying is the work of subsequent research, and Akerlof is not required to provide a full outline; his role was to generate or inspire hypotheses regarding asymmetric information.²

Economics is, compared to other subjects, in its infancy; it must be given freedom to explore, conceptualize, formalize, and extend without immediate regard for testability—not because it is unimportant, but because the universe is too complex for one theory or concept to emerge fully formed from a few papers. Only after centuries of barely testable macroeconomic policies have we gained enough tools (statistics, surveys, computing power) to consider verifying classical economics. There is hope that game theory, a relatively recent branch of economics, can generate testable hypotheses in fields such as auctions or signaling games—a significant achievement and a stepping stone. Possibly, in the future, we may have enough concepts and tools to generate an economic equivalent of general relativity, with great predictive power and generality, as Friedman envisioned. Right now, however, Schelling’s and Akerlof’s models solidly contribute to the process of scientific discovery as nascent seeds of hypotheses that may, in the future, generate “fruitful theories,” yielding precise predictions over a wide area.

Essay 2 §

Defend or Criticize: The most reasonable conclusion to draw from Reiss’s “Explanation Paradox” is that models in economics do not provide scientific explanations, but that they are nevertheless accepted by economics as explanatory because ‘explanation’ in this discipline isn’t scientific, it’s just plausible story telling.

I claim that Reiss’s characterization of the explanation paradox and its last arm—only true accounts explain—to be an unreasonable imposition on causal explanations and scientific theories in general. This is because:

1. Storytelling has been an indispensable part of science
2. Science includes systematic classification between good and bad stories—not true or false
3. Scientific stories are not necessarily monist and reductionist

I. For Storytelling

In the discussion of the third arm of the explanation paradox, Reiss doubts economists’ intuitions of credibility, stating that “the ‘credibility’ of an account of a phenomenon […] is not per se a reason to accept it as an explanation.” (Reiss, 56) However, the history of science is replete with scientists seeking credible accounts—a good story. From Archimedes’ insight into density in a bath, Kekulé’s ouruborus dream leading to the benzene ring, to Descartes’ coordinate system inspired by a fly on the wall, storytelling has been central. These compelling stories were neither fully evidenced nor objective from the outset, yet they inspired causal explanations. We then filter out the unfalsifiable and empirically test the remainder; only then do we consider them scientific theories or true-enough accounts.

From this perspective, a scientist essentially performs a random search for CWs, relying solely on skillful storytelling, and eliminating those that fail falsifiability tests, and finally evidencing them through experiments. In such conception of science we need not reject, unlike Reiss’s suggestion, the importance of’aesthetic sensibility’ and intuitive recognition of theory as elegant or beautiful by a subject expert in determining which CWs are good explanations.³ For instance, contrary to Reiss’s interpretation, the Galilean conception that the universe is pythagorean is aesthetically pleasing, and evidenced extensively, but still rigorously unjustified according to his criteria.⁴ (Reiss, 56)

Indeed, the scientific revolution was not necessarily a deductive proof of the scientific process but instead a method of sorting the evidentiary and useful out of a plural set of stories, an identification of the way of knowing that produces, historically, useful and good knowledge. What sets apart human science from simply an automatic theorem prover is that we need induction—creativity—, like Galileo and Newton’s idea to use mathematics to describe the physical world. Friedman even acknowledges that empiricism “can never reduce [a multitude of possible theories] to a single possibility […] consistent with the finite evidence,” and that “the choice among alternative hypotheses […] must [be] arbitrary, though there is general agreement [of] relevant considerations”—as we have selected mathematical beauty for physics, we aim to find a useful heuristic for economics, and we are doing just that, as outlined below. (Friedman, 5)

II. For Pluralism

It is also false that storytelling is a monistic endeavor; economics especially prides in its pluralist family of theories. Reiss claims, via Kitchen’s notion of argument patterns, that models contain generalized abstract argument structures, and good scientific argument patterns are simpler and more fruitful, i.e. reductionist and monistic. Reiss posits that “The most important inference rule in economics is ‘Solve the model using an equilibrium concept’ […] but of course there are many equilibrium concepts [and] no clear rules which among a set of Nash equilibria to select […] the justification for using [one over another] is very thin.” (Reiss, 58) This is, also, a mischaracterization. Scientific fields often comprise multiple stories: in physics, of general relativity and quantum mechanics, or psychology, of behaviorism and cognitivism.

Ultimately, the CWs of science cannot at all simultaneously be reduced into simpler or abstractly general statements from the outset; a grand unified theory of economics is impossible without first resorting to the “higher-level” storytelling. The strictly reductionist view also manifests in Reiss’s repeated characterization of “contemporary economics,” referring only to neoclassical economics while in reality, there is no one unifying story of contemporary economics: computing and math in algorithmic game theory, dialectical methods in Marx’s historical materialism, psychology’s methods in behavioral economics, or statistics in quantitative finance all are equivalently powerful; to argue for reductionist stories alone is akin to ontological dictatorship.

Much of economics is simply CWs, some verified, others instrumentally useful, and the remainder neither verified nor useful—that we hold on to with a certain trust in the aesthetic sensibility of a scientist⁵, that it may be stepping stones in building more verifiable and useful theories. Science has never been a grand unifying endeavor nor has it aimed to be; it simply searches for credible explanations, respecting the universe’s complexity and emergence, while giving us a method for distinguishing between the credible and good, and the credible but bad.

Bibliography §

Reiss, Julian. “The Explanation Paradox.” Journal of Economic Methodology 19, no. 1 (2012): 43–62. https://doi.org/10.1080/1350178X.2012.661069.

Friedman, Milton. Essays in Positive Economics. Chicago: University of Chicago Press, 1953.

Sugden, Robert. “Credible Worlds: The Status of Theoretical Models in Economics.” Journal of Economic Methodology 7, no. 1 (2000): 1–31. https://doi.org/10.1080/135017800362220.

Akerlof, George A. “The Market for ‘Lemons’: Quality Uncertainty and the Market Mechanism.” The Quarterly Journal of Economics 84, no. 3 (1970): 488–500. https://doi.org/10.2307/1879431.

Schelling, Thomas C. Dynamic Models of Segregation. 1978.

Footnotes §

1. The reason for this may be that hard sciences have simpler laws and thus the first induction step is easier, while data is hard to gather, leading to a focus on experiments and testing; on the other hand economics has much more complex laws and more numerous variables, making the first inductive step harder ↩

2. This is remarkably similar to how physics progressed in the beginning of the scientific revolution. It took decades to centuries to conceptualize location (via cartesian coordinates), velocity, acceleration, or momentum; decades more to formalize it, and some more to finally result in the empirically testable hypothesis of Galilean relativity. Equivalently it took years of observations by Tico Brahe to verify the planets, years more by Kepler and his conception of “orbits” lacking mechanism, until Newton’s testable hypothesis of gravitation. ↩

3. Reiss lapses into deferring to such sensitiblites quoting Friendman: a good model is “‘simpler’ the less the initial knowledge needed to make a prediction […] more ‘fruitful’ the more precise the resulting prediction.” ↩

4. It even surprises the hard scientists; the “Unreasonable Effectiveness of Mathematics in the Natural Sciences” is a phrase coined by physicist Eugene Wigner in his influential 1960 essay, reflecting on the surprising correspondence between mathematical concepts (credible world)—often developed without any consideration for physical application—remarkably correspond to and predict natural phenomena (real world) ↩

5. While Reiss claims that “the credibility of an account of a phenomenon of interest to an individual or group of researchers is not per se a reason to accept it as an explanation of the phenomenon,” I consider this to be putting science on an undeserving pedestal of objectivity. Science comprises not more than few researchers, with individual biases, “upbringing and educational background,” creating good stories based upon their derived nature and nurture, via the limited tools we have of falsification and empiricism, believe will hold—as we did believe for centuries of the “irrefutable truth” of Newtonian mechanics, the Central Dogma in biology or now, general equilibrium theory, all of which has been overturned or minimized. Our search for truth is indeed a process of iterative improvement, from a subjective story (of the credible world) to an objective one (of the real world), a process of telling better stories, and without either the creativity and rigor of our bards of science, we couldn’t have gotten this far. ↩