International Network for Economic Method (INEM)

Modality and How-Possibly Explanations in Economics

Short summary (180)

There is a growing interest concerning the modal features of modelling across the sciences, with

economics being no exception. Economic modelling practices do not fit the procrustean bed of

full how-actually explanations and the demanding criteria on models that come with it. Instead of

dismissing these practices as epistemically inert, more effort must be invested in investigating

the role they play in establishing modal claims and providing how-possibly explanations.

However, we still have a limited understanding of how how-possibly explanations are formed

and used in practice and of their relationship with how-actually explanations. We also lack

methodological criteria that would help us to assess the epistemic and heuristic value of how-

possibly explanations.

This session aims to make three chief contributions. First, we show that the practice of

establishing modal claims is well entrenched in the practice of economists, even those with an

eye on policy. Second, we illuminate how how-possibly explanations relate to economic

modelling practices and how they are used to acquire how-actually explanations of phenomena

of interest. Third, we provide methodological resources to appraise how-possibly explanations in

economics.

Topic description (991)

How we can learn about the world with highly idealized models has been a vexing philosophical

issue. These models often do not faithfully describe the world as it is, yet it also seems they are

not epistemically vacuous. Philosophers have started to draw attention to the modal features of

scientific modelling (e.g. Betz 2015; Bokulich 2014; Knuuttila and Koskinen 2020; Massimi

2019; Rice 2016; Shech and Gelfert 2019; Weisberg 2013). According to that growing body of

literature, scientists sometimes seek to establish possibility claims. For instance, biologists

synthesize potential organisms (Elowitz and Lim 2010) and physicists model impossible

perpetual motion machines (Feynman, Leighton, and Sands 2010) to further our understanding of

the world. They pursue the epistemic goals of science by inquiring into the possible and

impossible.

We observe a similar interest concerning economic models. They are sometimes said to describe

credible worlds (Sugden 2000, 2009), establish possibility claims to rebut impossibility ones

(Grüne-Yanoff 2009), or provide how-possibly explanations (Aydinonat 2007, 2008; Ylikoski

and Aydinonat 2014). All this draws attention to the modal features of economic modelling

(Hausman 1984; Rappaport 1986, 1989, 1998): economics pursues not only its aims by

investigating what is actual, but also what is possible. Moreover, while some philosophers of

economics are pessimistic about the capacities of economic models to accurately represent the

actual world and explain (Alexandrova and Northcott 2013), others have suggested that they

might fare better as representations of genuine possibilities (Grüne-Yanoff 2013; Verreault-

Julien 2017; Ylikoski and Aydinonat 2014), thus offering an alternative and nuanced vindication

of current economic modelling practices.

The methodology of economics needs this nuance. Prediction and explanation are commonly

considered as the most important goals of economic modelling. Yet, typical accounts of

economic modelling that rely on representational adequacy and robustness as criteria for success

suggest that most economic models fail to achieve these goals (Aydinonat and Köksal 2019). It is

true that economic models often do not meet the decomposability assumptions needed for

minimalist accounts of idealization (Cartwright 2009; Grüne-Yanoff 2011) and that most of them

are not robust to many kinds of de-idealizations (Knuuttila and Morgan 2019). Thus, they are

unlikely candidates for providing good predictions or explanations. But it does not follow from

this that most economic modelling is deficient. The reason is that the dominant accounts of

economic modelling do not do much justice to practice of economic modelling and particularly

to its modal features. Before dismissing these practices as epistemically inert, one must invest

more effort in investigating the role models play in establishing modal claims and providing

how-possibly explanations.

However, we currently have a limited understanding of modal modelling. No systematic survey

of modal modelling practices in economics exists, and few proposals have been offered how to

categorize and systematize these practices. Nor are there clear analyses of the epistemic

contributions of economic modal modelling. Do they consist in how-possibly explanations that

have epistemic value in their own right, or do they indirectly contribute to how-actually

explanations, and only through that acquire value? Furthermore, we also lack methodological

criteria for assessing the quality of specific how-possibly explanations.

This session focusses on those modelling practices that relate to how-possibly explanations. It

consists of three papers that examine distinct features of how-possibly explanations in

economics. Each paper documents some specific modal modelling practices in economics,

ranging from central bank modelling through trade theory to modelling cooperation. Each paper

then analyses how these practices make a substantial epistemic contribution, and concludes by

identifying criteria for assessing the quality of these modal modelling practices.

The first paper focusses on models considered potential candidates for providing causal

explanations, as one finds them, e.g., in central bank multi-model platforms. When considered on

their own, these models merely provide 'thin' how-possibly explanations of little epistemic

value. Yet, in practice, these models are crucial for acquiring how-actually explanations. The

paper illustrates such practices by showing how the Bank of England integrates multiple how-

possibly models in its COMPASS platform, aiming at how-actually explanation of real-world 

phenomena, and it investigates the success conditions of such how-actually explanations in terms

of how well they facilitate reasoning about reliable intervention.

The second paper emphasizes that economists often make use of multiple models along with

other sources of information in their quest for explaining actual phenomena. Therefore,

individual models rarely function as explanantia. Rather, economists typically use a set of

models (each of them highly idealized) to construct a 'menu' of possible explanations,

viz. explanations whose truth isn't yet established. When doing so, they rely on a diversity of

models and sources of evidence. Determining the actual explanation is then a matter of

eliminating the possible candidate explanations and establishing the truth of the relevant

explanantia.

The third paper argues that the practice of providing 'mere' how-possibly explanations does not

entitle economists to formulate any just-so story they wish. It proposes a conceptual framework

that allows assessing 'good' from 'bad' how-possibly explanations. This conceptual framework

relies on two key distinctions: the first between how-actually and how-possibly explanations, the

second between epistemic and objective how-possibly explanations. It argues that both of these

latter categories afford their own epistemically relevant results, and both have the conceptual

resources to distinguish between good and bad models. The paper illustrates these arguments

with examples from trade theory, economic geography, and social cooperation.

With these three presentations, the session aims to further our understanding of how-possibly

explanations and modal modelling in economics. In particular, we pursue three goals. First, we

show that the practice of establishing modal claims is well entrenched in the modelling strategies

of economists, even those with an eye on policy. Second, we illuminate how how-possibly

explanations make substantial epistemic contributions, either by establishing relevant free-

standing modal claims or by facilitating, in conjunction with other sources of information, the

development of how-actually explanations of phenomena of interest. Third, we provide

methodological guidelines for assessing the quality of how-possibly explanations, thus showing

that they are distinct from methodologically naive just-so stories.