Normative and behavioural economics: a historical and methodological review

Abstract

Behavioural economics has not only posed serious challenges for the empirical adequacy of rational choice, but also for its normative status. Since the 1990s, a large body of work has proposed various normative approaches that account for inconsistent choices. Focusing on the relevant approaches, history, methods, and limits, this article offers a literature review of the relationship between normative and behavioural economics.

Keywords:

• Choice
• opportunity
• preference
• rationality
• welfare

JEL CODES:

• B21
• B41
• D60
• D90

Acknowledgement

A preliminary draft of this paper was presented at the 2019 conference on Public Economic Theory in Strasbourg and at the 2019 Happiness Economics and Interpersonal Relations conference in Rome. I thank the audience for comments. I am grateful to Francesco Guala, Cyril Hédoin and Robert Sugden for helpful remarks, and most particularly to Wade Hands for insightful discussions that considerably helped me improving the ideas presented in this paper. Finally, I thank Pierre Van Zyl for proofreading and two anonymous referees for their extremely valuable remarks. All mistakes remain my own.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Rabin (1998) and DellaVigna (2009) offer extensive reviews. For historical reviews of behavioural economics, see Heukelom (2014) and Kao and Velupillai (2015).

2 See Sugden (1991) for a survey of rational choice from a normative perspective—i.e., how individuals ought to choose.

3 On the historical relationship between psychology and economics, see Hands (2010). On the historical debate around the measurement of utility—from early neoclassical economics to behavioural economics—see Moscati (2018).

4 The axiom was later reformulated by Houthakker (1950) into the Strong Axiom of Revealed Preference (SARP) and by Afriat (1967) into the General Axiom of Revealed Preference (GARP).

5 For a detailed panorama of how different branches of revealed preference theory evolved over time, see Hands (2013).

6 See in particular the historical interpretation of Hands (2010) about the role of psychology in consumer choice theory, from the early utilitarian stage of neoclassical economics to the ordinalist revolution.

7 This point was acknowledged by Paul Samuelson himself in a correspondence with Hendrik Houthakker regarding SARP: “I imagine that you are right that many individuals looking at this paper will be induced to believe that there is after all very little, and very little of interest, in the modern theory of consumer’s behavior. However, if this is indeed the truth, we should not try to keep it a secret. By all means let us make clear how little and how much the existing theories of economics contain.” (Paul Samuelson to Hendrik Houthakker on the 31st of July 1952). Few decades later, goodness-of-fit methods were proposed to measure how far individuals deviate from GARP (Afriat 1973; Houtman and Maks 1985; Varian 1990).

8 We can also include a third assumption, according to which (iii) individuals’ subjective beliefs are updated using Bayes’ rule. Yet in the contemporary approach of revealed preference I here discuss, such an assumption is (to my knowledge) absent of the theory.

9 A famous example in choice under uncertainty is Savage’s (1954) reconsideration of the status of expected utility theory after he violated the independence axiom himself (Allais 1953). Savage’s response to Allais was that the theory should not be seen as descriptive (how individuals actually choose) but as normative (how individuals should choose).

10 Empirically, measuring social welfare in the standard framework is based on households’ revealed preferences, with several restrictions that are well summarised by Slesnick (1998)—among which households are supposed to constantly maximise their utility.

11 Note, however, that language plays a decisive role in Kahneman and Tversky’s experiments, which makes it not so straightforward to interpret individuals’ responses as mistakes of logical reasoning. See Jullien (2016).

12 One may check the English definition of a “mistake” to consider the various meanings the word can take. According to Oxford Languages, a mistake is an “an act or judgement that is misguided or wrong”. This raises the obvious question, wrong according to what?

13 In the nudging literature—to be seen as the practical application of libertarian paternalism in various domains (e.g., health, education, environment, among many others)—the distinction between mistakes in terms of logical reasoning and welfare became, however, blurred. For example, eating healthier over time is a moral/ethical principle, while choosing an apple over a cake in intertemporal choice is a time-consistency principle, i.e., a rational principle. Yet a nudger would not necessarily be clear about which meaning of “mistake” he/she holds (either in terms of logical reasoning or welfare, or both, or perhaps yet something else).

14 There are at least two terminologies than can be found in the literature to designate the “normative” turn of behavioural economics: “normative behavioural economics”—coined by Berg (2003)—and “behavioural normative economics”—used for example by Dold and Schubert (2018). These terminologies should not be assimilated with “behavioural welfare economics,” which is a subset of (either) “normative behavioural economics” or “behavioural normative economics”. This is because suggesting a normative approach that accounts for inconsistent choices does not necessarily imply that it is based on a welfare criterion. Sugden (2004, 2018) takes this non-welfare approach (to be reviewed in Section 3).

15 Knowing that the “grand hedonistic tradition” dominated early neoclassical economics for its (roughly) first fifty years, this approach—explicitly presented as a “back-to-Bentham” approach—is just a special case of such a hedonistic tradition. For a subtle distinction of some interpretations of the Benthamian utilitarian principle and their implications for welfare economics, see Baujard (2010). Note also that Bentham may not always be seen as a primitive predecessor of rational choice theory and welfare economics (Danchev 2016).

16 See in particular Kahneman (1999), who discusses a consequent body of empirical studies in psychology on how human sensory experience works, which (he argues) provides arguments for the possibility to make interpersonal comparisons of utilities.

17 In particular, the distinction between the happiness and well-being interpretations of preference satisfaction is that the former defines “what makes individuals better off” in terms of hedonic experience, while the latter does not take any position on defining “what makes individuals better off” and prefers to let this information private to individuals. In other words, the well-being interpretation does not refer to something in particular as what makes individuals better off. This poses serious challenges for policy applications, which practically require to know what makes individuals better off in order to increase social welfare (to be discussed in Section 3).

18 What follows synthesises Sugden’s self-acknowledgement of the influence of these authors over his own work, which is therefore also based on his own reading of these authors. For a critical review of Sugden (2018), see Mitrouchev (2019).

19 Commenting Sugden’s approach more extensively would lead us to further specifications that are outside the scope of the present article. For the respective influences of James Buchanan, J. S. Mill and Adam Smith over Sugden’s normative theory, see Sugden (2018, Ch. 1, 2, 3, 6, 11).

20 Bhatt, Ogaki, and Yaguchi (2015) and Ogaki and Tanaka (2017, Ch. 11) proposed a “virtue ethics” criterion, according to which it is good to satisfy individuals’ meta-preferences for what is judged to be desirable for society. There is also the “meaning” approach proposed by Loewenstein (1999), Karlsson, Loewenstein, and McCafferty (2004) and Dold and Stanton (2021), according to which it is good to realise individuals’ expectations about living a meaningful life.

21 The literature on measuring experienced utility includes Kahneman and Snell (1990, 1992), Kahneman and Varey (1991), Varey and Kahneman (1992), Kahneman et al. (1993), Fredrickson and Kahneman (1993), Kahneman (1994, 1999, 2000, 2011 [Part V]), Redelmeier and Kahneman (1996), Kahneman, Wakker, and Sarin (1997), Schreiber and Kahneman (2000), Redelmeier, Katz, and Kahneman (2003), Kahneman et al. (2004), Kahneman and Sugden (2005), Kahneman and Krueger (2006), Kahneman and Thaler (2006) and Dolan and Kahneman (2008).

22 For an extensive analysis of the axiomatic foundations of the normative theory of experienced utility and its limits, see Mitrouchev (2023).

23 See in particular Varey and Kahneman (1992, p. 169), Kahneman (1994, p. 21), Kahneman, Wakker, and Sarin (1997, p. 377) and Kahneman and Sugden (2005, p. 176), who make that point explicit.

24 During the production of this article, Daniel Kahneman passed away. The author wishes to express his warm thoughts to his family, and acknowledges the enormous intellectual debt owed to him, as an outstanding contributor to behavioral research.

25 The literature proposing true preference as a standard of well-being includes Bleichrodt, Pinto, and Wakker (2001), Madrian and Shea (2001), Benartzi and Thaler (2002), Camerer et al. (2003), Thaler and Sunstein (2003, 2009), Thaler and Benartzi (2004), Beshears et al. (2008), Loewenstein and Haisley (2008), Dalton and Ghosal (2011), Rubinstein and Salant (2012), Pinto-Prades and Abellan-Perpiñan (2012), Thaler (2018) and Sunstein (2019). Note that in the 1970s, pioneering experimental studies aimed to capture individuals’ mistakes by letting individuals reconsider their choices (MaCrimmon 1968; Moskowitz 1974; Slovic and Tversky 1974; MacCrimmon and Larsson 1979). In choice under risk, there is a recent interest of measuring individuals’ preferences over principles of rational choice (Benjamin, Fontana, and Kimball 2020; Nielsen and Rehbeck 2022; Breig and Feldman 2023).

26 Perhaps the most eminent and influential policy application is the nudge (Thaler and Sunstein 2009; Halpern 2015). Note that before a large number of behavioural economists got interested in the true preference criterion, some authors had already given considerable support for it. The concept of true preference follows that of Harsanyi (1977, 29–30) in his defence of utilitarianism. Fine (1995) aimed at distinguishing the two concepts of true preference and observed choice from a social choice perspective. From a philosophical perspective, some authors had already defended the satisfaction of self-interested “informed,” “rational,” or “laundered” preferences as to what constitutes goodness (Gauthier 1986, Ch. 2; Arneson 1990; Goodin 1992).

27 Further experimental research could nonetheless provide elements for justifying or rejecting the hypothesis of true preference. See in particular Benjamin, Fontana, and Kimball (2020), Nielsen and Rehbeck (2022) and Breig and Feldman (2023), who propose different experimental designs for measuring true preferences.

28 See Berg (2003, 431), Berg and Gigerenzer (2010, 147–148), Hands (2014, 398), Whitman and Rizzo (2015), Lecouteux (2016) and Dold and Schubert (2018).

29 Simply put, paternalism is here justified under individuals’ consent. For empirical surveys about Europeans’ acceptance of nudges, see e.g., Reisch and Sunstein (2016), Reisch, Sunstein, and Gwozdz (2017) and Sunstein, Reisch, and Kaiser (2019).

30 See in particular Mitchell (2005), Rizzo and Whitman (2009, 2019), Welch and Hausman (2010), Grüne-Yanoff (2012), Rebonato (2012), Hédoin (2015, 2017), Sugden (2017b) and Scoccia (2019).

31 The literature proposing the choice-basis criterion as a standard of well-being includes Bernheim and Rangel (2007, 2009), Köszegi and Rabin (2007, 2008), Salant and Rubinstein (2008), Loewenstein and Ubel (2008), Bernheim (2009), Dalton and Ghosal (2012), Rubinstein and Salant (2012), Manzini and Mariotti (2014) and Bernheim (2016). There is also a related literature, which aims to extend measures of rationality deviations (Afriat 1973; Houtman and Maks 1985; Varian 1990) with behavioural foundations. This literature includes Apesteguia and Ballester (2015) and Dziewulski (2020). Very generally, one can say that the choice-basis approach is merely bound to the old problem of recovering preferences from choices (Samuelson 1938; Mas-Colell 1977, 1978).

32 See e.g., Bernheim and Rangel (2008, 156), Manzini and Mariotti (2014, 343–344) and Bernheim (2016, 24–25), who advance the standard “ordinalist” argument, according to which welfare economists should evaluate individuals’ situations based on individuals’ own conception of goodness (not happiness nor true preference). They also argue that choice is a far less obscure concept because there is more available data on choice than on mental states.

33 The literature on the opportunity criterion includes Sugden (2003, 2004, 2007, 2008, 2010, 2017a, 2018).

34 What follows applies to Sugden’s individual opportunity criterion, not to his interactive opportunity criterion. See Sugden (2018) for the different versions that the opportunity criterion can take.

35 This point actually refers to a complex debate in social choice theory on how to measure opportunity, and whether opportunity is measurable at all. I come back to this point further on.

36 The literature on choice overload includes Iyengar and Lepper (2000), Hutchinson (2005), Botti and Iyengar (2006), among others. For meta-analyses providing mixed results, see Scheibehenne, Greifeneder, and Todd (2010) and Chernev, Böckenholt, and Goodman (2015).

37 The literature on self-constraint includes Elster ([1979] 1998, [1983] 2016, 2000). See also Thaler (1980), who discusses situations where individuals voluntarily restrict their choices, deliberately not choosing so as to avoid psychic costs that the choices might induce.

38 See Pattanaik and Xu (1990), Sen (1991) and Sugden (1998, 2003, 2010) for a debate.

39 The literature on elaborating a diversity metric of opportunity includes Pattanaik and Xu (2000), Bossert, Pattanaik, and Xu (2003) and van Hees (2004), among others.

40 For more about the lack of psychological substance of the opportunity criterion, see Schubert (2015).

41 I deliberately do not use the usual vocabulary of “welfare-relevant domain” nor “welfare relation,” because a comparison is not necessarily based on well-being (or welfare, or individual utility).

Additional information

Funding

This research is supported by a grant from the Métropole Européenne de Lille (MEL).