Causal Generalisations in Policy-oriented Economic Research: An Inferentialist Analysis

ABSTRACT

The most common way of analysing the meaning of causal generalisations relies on referentialist semantics. In this article, we instead develop an analysis based on inferentialist semantics. According to this approach, the meaning of a causal generalisation is constituted by the web of inferential connections in which the generalisation participates. We distinguish and discuss five classes of inferential connections that constitute the meaning of causal generalisations produced in policy-oriented economic research. The usefulness of our account is illustrated with the analysis of generalisations about unemployment put forward by the Organisation for Economic Co-operation and Development in its highly influential 1994 OECD Jobs Study. The article ends with a discussion of some crucial philosophical questions about the use of inferentialism in the analysis of causal generalisations.

Acknowledgements

The authors want to thank Kevin Hoover, Caterina Marchionni, Julian Reiss, and two anonymous referees of this journal for very helpful comments and suggestions, and Jordan Girard for research assistance. We are also grateful to the audience at the Triennial International Congress of the Société de Philosophie Analytique, Montréal, June 2015, where we presented an earlier version of this article.

Notes

1 Other labels usually given to this approach are: representationalist, truth-conditional, and extensional. Note that the dominant approach to meaning starts with a theory of reference, but does not stop there. Most thinkers then add another dimension to account for the fact that two extensionally identical expressions might intuitively have different meanings (Speaks 2017), e.g. Frege's ([1892] 1960) famous evening star and morning star.

2 There are now many labels and many species of inferentialist semantics, which include: conceptual role semantics, inferential role semantics, functional role semantics, procedural semantics, and use theory of meaning. For discussions of the different species of inferentialism, see Block (1998); Whiting (2009); and Peregrin (2012). Similar ideas can also be found in other philosophical traditions, for instance in French structuralism, starting with Saussure’s theory of language as pensée organisée (de Saussure [1913] 1995, pt. 2, ch. 4) and then influencing thinkers such as Foucault (1969) with his formations discursives. In the next section, we lay out specifications that are characteristic of what Brandom (2000, 28; 2007, 656–658) calls ‘strong inferentialism’ (which is neither ‘weak’ nor ‘hyper-inferentialism’).

3 This is the case, for example, in most versions of the currently in vogue interventionist account in which the truth conditions of causal claims are to be defined in terms of the constituent terms included in the causal claim and of their relations to their referents. This is typically done by specifying necessary and sufficient conditions for different causal concepts (see, e.g. Hausman 1998; Pearl 2000; Woodward 2003). An investigation of whether proponents of the interventionist or any other causal theory explicitly endorse referentialist semantics is a separate task which differs from the main aim of the present article.

4 The inferentialist analysis proposed here is inspired by Julian Reiss’s inferentialist theory of causal claims (Reiss 2011, 2012, 2015). Our analysis can be seen as an attempt to make his general account better suited to the semantic analysis of policy-oriented causal claims. Our classification corresponds only partially to the one he offers; for a discussion of how the two classifications relate, see Claveau (2012, 51–52).

5 For a contrast between the analysis offered in this section and the results of a referentialist analysis of the meaning of the causal generalisations in the OECD Jobs Study, the reader is referred to our previous article (Claveau and Mireles-Flores 2014) already mentioned in the introduction.

6 For the detailed argument leading to this conclusion, see Claveau (2012), sec. 1.3.

7 The OECD (1994a, pt. 1, 66) recognizes that ‘there is no direct measure of structural unemployment’ and uses three proxies to measure it indirectly: (a) the non-accelerating wage rate of unemployment (NAWRU, based on the relationship between the unemployment rate and the change in wage inflation), (b) the rate derived from the Beveridge curve (the relationship between the unemployment rate and the vacancy rate), and (c) the rate derived from the Okun curve (the relationship between the rate of unemployment and capacity use). For a detailed account of the OECD method of measurement at the time, see Elmeskov and MacFarlan (1993).

8 The ‘roughly in line’ is important here. In 1994, the evidence was fragmentary and polyphonic. We cannot do justice to the complexity here. The interested reader is referred to OECD (1994a), pt. 2, ch. 8.

9 All these terms are, for instance, used in the (critical) volume of Howell (2005). The authors also talk about an ‘orthodoxy’, and make the connection with neoliberal ideology. It seems indeed correct to say that the inflexibility claim is also inferentially connected with even wider claims about the purported ‘efficiency’ of free markets. We will not go down this road in our semantic analysis, and will leave for another time the inferentialist treatment of ‘ideology’.

10 The two recommendations most directly concerned with unemployment benefits are: ‘Restrict UI benefit entitlements in countries where they are especially long to the period when job search is intense and rapid job-finding remains likely’; and ‘Reduce after-tax replacement ratios where these are high, and review eligibility conditions where these require little previous employment history before drawing benefits’ (OECD 1994b, 48).

11 Starting with the Italian version of the OECD Economic Surveys in January 1996 (OECD 1996a), each country got its own chapter. Some countries, e.g. France, had their chapter published only in 1997. The implementation of the recommendations was further monitored at a country level in later editions of the OECD Economic Surveys, and at the cross-country level in many publications (e.g. OECD 1998, 1999).

12 A fascinating implication of the rigidity view is that two countries that seemed to have fairly good unemployment performances were treated quite differently if one appeared ‘more rigid’ than the other. This contrast is stark when comparing the reports of the United States (OECD 1996b) and of the Netherlands (OECD 1996c). These two countries had similar unemployment rates between 1993 and 1995 (averaging at 6.2%), rates which made other countries envious. The ‘flexible’ United States were offered a light medicine while the ‘rigid’ Netherlands had to act across the board: of all the recommendations about ‘labour market policy and institutions’, only 10% were applicable to the United States according to the OECD in contrast to 50% for the Netherlands (OECD 1999, 47). The United States had the smallest number of recommendations in the group of 29 countries, while the Netherlands was in fourth position—after Germany, Finland, and Norway—for the toughest medicine.

13 Note that scholars trying to defend inferentialism have also provided more principled responses to this intuition-based objection. For instance, Peregrin (2014, 11–14) maintains that the objection is a non-starter against a normative version of inferentialism such as the one developed by Brandom (1994, 2000).

14 Furthermore, the OECD recognised later that its inference from its main causal generalisation to policies emphasising deregulation of the labour market was not sound, since ‘flexibility’ might also be achieved through wise regulation instead of deregulation (OECD 2006, 19).

Additional information

Funding

The work of François Claveau on this project has been supported by the Social Sciences and Humanities Research Council of Canada (grant number 767-2009-0001) and by Canada Research Chairs (grant number 950-230644). The work of Luis Mireles-Flores on this project has been supported by TINT—Academy of Finland Centre of Excellence in the Philosophy of the Social Sciences.