Concept and Causation: Issues in the Modelling of Institutions

Abstract

The article analyses the conditions of the accurate modelling of ‘institutions’, including more realistic models such as computational ones. In particular, modelling is challenged by the composite nature of the concept of institution and the associated causations. Collective behaviour undoubtedly exhibits regularities, which are clarified by models. Yet, the concept of institution is not a unitary entity that has a stable reference across time and space and is a term in unambiguous causalities; therefore it may not be a cause or outcome that is computable from agents’ attributes and behavioural rules. Also, agents behave according to the rules that have been assigned to them by the modeller. Yet in the ‘real world’, individuals may depart from social norms without any ‘reason’ (‘rationality’): models under-address the cascade of cognitive processes that underlies the emergence of institutions and the context-dependence of their relevance for individuals.

Keywords:

• institutions
• computational models
• causation
• methodology

Notes

¹ As shown by, for example, the theoretical insights put forward by J. M. Keynes, G. Myrdal or A. O. Hirschman.

² The author is very grateful to Paolo Ramazzotti for having suggested this reference.

³ Since their birth in ancient Greece, thought experiments have been widely used by physics or philosophy, well-known examples being, among many others, Erwin Schrödinger’s cat, John Rawls’ ‘veil of ignorance’, John Searle’s ‘Chinese room’, or Hilary Putnam’s ‘brain in a vat’ (for a textbook, Tittle, 2005).

⁴ Among many others, as is well-known, ‘old institutionalism’, neo-institutionalism, themselves divided along the lines of, e.g. evolutionary economics, behavioural economics, law and economics, or economics of organisations (see, e.g. Samuels, Medema, & Schmid, 1997; Samuels, Schmid, & Shaffer, 1994).

⁵ The distinction between representations, explorations, conceptualisations of the world that would be ‘true’, ‘credible’, ‘plausible’, ‘adequate’, goes beyond this article, as it is a founding question of philosophy.

⁶ e.g. ‘institutions’ in developing countries have been proxied by the protection of colonisers’ settlement, itself proxied by settlers mortality rates (Acemoglu, Johnson, & Robinson, 2001); urbanisation and population density have been used as proxies for per capita income (Acemoglu, Johnson, & Robinson, 2002); social norms have been proxied by ‘cultural values’ such as ‘generalised trust’, itself proxied by ‘family ties’ (Alesina & Giuliano, 2013).

⁷ e.g. an individual being herself an immigrant may with time change her behaviour towards more recent immigrants.

⁸ Yet the absence of ex ante theorisation may be explicit, as in the case of ‘big data’ analysis.

⁹ The author is grateful to Fabrice Tricou for having underscored this crucial point.

¹⁰ Indices of institutional phenomena may be examples of such confusions, as an index ‘is’ not the concept itself and moreover these indices gather together attributes that belong to heterogeneous levels of reality (in addition to the problem of their weights). For example, the World Bank indicators of ‘governance’ thus gather together six aggregate indicators as heterogeneous as ‘voice and accountability’, ‘political stability and absence of violence’, ‘government effectiveness’, ‘regulatory quality’, ‘rule of law’ and ‘control of corruption’ (Kaufmann, Kraay, & Mastruzzi, 2010).

¹¹ More generally, it may be argued that the analysis of causal relationships is inherently confronted with limitations when social complex phenomena are involved (Krauss, 2016).

¹² Among many others, Loasby (1999), Mantzavinos (2001); for a textbook from an ‘institutions and organisations’ perspective, Scott (1995).

¹³ The concept of ‘institution’ may be heuristically viewed as referring more to ‘written rules’, and that of ‘norm’ to ‘unwritten rules’. Yet from the point of view of the individual mind, what is pertinent are her ‘mental representations’, their deontic force, relevance, capacity to disseminate, etc., and some ‘unwritten rules’ may have a stronger obligation force than ‘written’ ones (Sindzingre, 2006).

¹⁴ As obviously shown by, e.g. political disagreement, dissidence, rebellion.

¹⁵ A related point has been made by Dow (2012): models claim to be more rigorous and precise, but the very formalisation of a ‘verbal theory’ (of an argument) can change the latter’s meaning.

¹⁶ This argument is a modality of Albert (1963/2012) and Kapeller (2013) critiques of the ‘ceteris-paribus’ clause that is used by neoclassical models. In another domain, epidemiologists are aware of this non-repeatability of contexts or of ‘events’, modelling future pandemics from past ones being very uncertain (as was shown, e.g. by the prediction errors that have affected the models of the 2009 influenza).

¹⁷ ‘Unexpected’ behaviour (unexpected by models) differs from deviant behaviour, or from defecting behaviour, which are well-analysed in game theory and behavioural economics: such qualifications imply that the rule vis-à-vis which there is a deviation or defection is ex ante known and identified by the observer.