Values in economics: a recent revival with a twist

ABSTRACT

This article reviews the relatively recent trend in economic methodology that consists in bringing insights from the debate in philosophy of science on values in science in order to analyse value-ladenness of economic research. The text claims that these insights from philosophy of science offer a slightly new approach to the topic of value judgments in economics that has been discussed in philosophy of economics for decades. It suggests that the perspective of philosophy of science reviewed in the article invites to rethinking analyses of feminist economists as important contributions to economic methodology.

KEYWORDS:

• Values in science
• feminist philosophy of science
• feminist economics
• values in economics
• economic methodology

JEL CODES:

• B40
• B54

Setting the stage

The topic of value has a long history in economics. The classical economists were preoccupied with understanding what economic value consists of and whether labour was its source. The marginal revolution is seen as debunking attempts to come up with a labour theory of value by equating value to the market price. Gradual abandonment of the focus on theorizing economic value was followed by proposals to eliminate all types of values from economic research: moral, social, or political ones. The view of economics as a value-free science was then solidified by the neoclassical school, with the seminal essays of Lionel Robbins (1935) and Milton Friedman (1953) that defined economics as a positive science free from any value judgements. Or so the standard story goes. As historians of economic thought teach us, debates about the notion of economic value, the role of value judgements in economics, as well as the division of economics into a positive and normative branch, were much richer than what the standard story might suggest, especially with respect to the variety of arguments and ambiguity of views (e.g. Backhouse, 2005; Blaug, 1992; Davis, 2005; Hands, 2012; Mirowski, 1990; Scarantino, 2009). Interest in contributing to these debates has never vanished among philosophers, historians, and even some economists (e.g. Anderson, 1995; Colander & Su, 2015; Dasgupta, 2005; Mireles-Flores, 2016; Mongin, 2006). However, these are not my focus on this occasion. I would like to draw readers’ attention to yet another take on values in economics that has become more present in economic methodology and philosophy of economics, mostly over the last decade.

The works I have in mind refer to (and often engage with the debate on) values in science in the philosophy of science. Recently, we have witnessed in the philosophy of economics several attempts to analyse the value-ladenness of economic research by using the terminology and philosophical approaches from this literature. I would not say that these attempts form a clear trend yet. In fact, so far, they constitute a dispersed body of texts which often do not even remain in dialogue with each other. However, I find them all a promising development in the field. In my view, the discussion on values in science in the philosophy of science offers us insights on how to identify and trace diverse normative commitments influencing economics, and on how we might possibly revisit the older debates on value-judgments in economic research by changing the analytical angle and approach. I believe that economic methodologists would benefit from paying even more attention to how philosophers of science analyse values in science. It is quite striking how few exchanges there have been until recently between philosophers of economics discussing the issue of value judgments in economics, and philosophers of science analysing how values enter science. Clearly, this situation is beginning to change.

I start with a summary of the main findings in the literature on values in science. Afterwards I report on the recent works in the philosophy of economics and economic methodology that seem to approach the issue of values in economics similarly to how it is analyzed in this literature. Then I briefly explain why I find this development promising and how I envision its direction.

Discussion on values in science in the philosophy of science

In philosophy of science we face a thriving discussion on values in science.¹ The consensus in this debate is that values, both so-called epistemic and non-epistemic ones, unavoidably enter science, and that the task for the philosophy of science is to deepen our understanding of the ways in which values impact scientific process, reasoning, and knowledge production. In addition, it is accepted in this literature that the ideal of value-freedom or value-neutrality of science is not a defensible normative option for guiding scientific practice. The ideal is seen as unattainable due to the pervasiveness of values in all fields of science and, furthermore, undesirable because it does not allow for detecting androgenic values embodied in modern science (Harding, 1991), and detrimental because it prevents scientists from considering the possible harmful (moral, or social) consequences of their claims in cases in which their hypotheses are incorrect (Douglas, 2016). Philosophers who reject this ideal claim that values entering scientific research should be made explicit and that the institutional structures of science should enable identification of value commitments and their criticism, as well as guarantee that the detrimental effects of values are controlled and neutralized, while the beneficial ones are allowed and encouraged. To this effect, they discuss which norms should guide the scientific community in order for such a criticism to be effective (Longino, 1990), or in order to ensure that non-epistemic values play a legitimate role in science (Douglas, 2009; Elliott, 2017). Some suggest democratizing science (Douglas, 2009; Kitcher, 2001, 2011), whereas others argue for the ideal of socially responsible science (Brown, 2013; Hicks, 2014; Kitcher, 2001, 2011; Kourany, 2010).

The distinction between epistemic and non-epistemic values has been introduced while discussing the ideal of value-freedom and analysing the presence of values in science. Epistemic values are normative commitments that constitute grounds for choosing one theory over another (e.g. simplicity, internal and external consistency, explanatory power, predictive accuracy, breadth of scope) (see: Kuhn, 1977; Laudan, 1984; McMullin, 1982).² Non-epistemic values are normative and emotive commitments concerning moral, social, and political life. It is mostly the role of non-epistemic values in science that is highly debated and leads to controversies.³ Yet it is perceived as being relatively uncontroversial to state that non-epistemic values can influence the choice of topics and goals of scientific research. The challenge arises with the question of whether these values have, and should have, an impact on the acceptance or rejection of hypotheses and theories, on the collection and analysis of data, on measurement, or on modelling practices. An early argument for the necessary role of non-epistemic values during hypotheses testing was offered by Rudner (1953), who argued that in the cases in which conclusion in scientific inferences does not follow deductively, scientists have to decide whether the evidence is sufficient to support the claim/hypothesis. Therefore, non-epistemic values are a necessary component of hypothesis testing. Furthermore, they are indispensable in assessing the consequences of being mistaken while making judgments about the evidential support for a hypothesis, as has been pointed out by Douglas in her revival of Rudner’s analysis (Douglas, 2000). Due to the underdetermination of theory by evidence, non-epistemic values may enter science because scientists must make some further considerations when they decide between theories, beyond assessing the available evidence (Brown, 2013; Biddle, 2013; Elliott, 2011; Intemann, 2005). Longino (1990) points out that non-epistemic values can enter scientific research via background assumptions held by researchers when they make judgements about the evidential relevance of data for hypotheses.

The focus of the literature on values in science is either on the normative proposals of how to ‘manage’ value-laden science, mentioned briefly above in the first paragraph of this section, or on the analysis of multiple forms of the value-ladenness of scientific research. When it comes to the analytical and descriptive work done in this context, philosophers often differentiate between diverse stages of scientific practice and show how values enter each of them: from framing a question and articulation of ‘a conception of the object of inquiry’ (Anderson, 2004, p. 11), through collection and analysis of data to drawing conclusions from an analysis of data (cf. Elliott’s, 2017 introduction to the topic structured around stages of scientific inquiry). Case studies in diverse fields of research focus in detail on how non-epistemic values permeate the production of scientific knowledge at each stage of inquiry. Examples of such studies include research in biology (Haraway, 1989; Hubbard, 1990; Keller, 1988; Lloyd, 2005; Okruhlik, 1994), behavioural sciences (Lacey, 2003; Longino, 1990, 2013; Małecka, in press), medicine and public health (Cohn, 2006; Katikireddi & Valles, 2015), psychology (Anderson, 2004; Intemann, 2001), anthropology (Slocum, 1975; Sterling, 2014; Zihlman, 1985), geology (Solomon, 1992), engineering (Diekmann & Peterson, 2013; Tuana et al., 2012), environmental research and climate science (Elliott, 2011; Intemann, 2015; Nordhaus, 2007), archaeology (Wylie, 2007). Somewhat surprisingly, economics is virtually non-existent in the analyses of value-ladenness in the philosophy of science.

Economic methodologists on values in economic research: bringing the perspective of philosophy of science

A philosopher who tried to initiate a dialogue between the general philosophy of science and economics over the last two decades was Hilary Putnam. In a series of publications, many of them co-authored with Vivian Walsh (Putnam & Walsh, 2012), he pointed out that economists failed to take seriously the devastating criticism of the neo-positivist philosophy of science. According to him, economics’ methodological commitments, including its view of economics as a value-free science, were strongly influenced by logical positivism – a process that started in the 1930s with the work of Robbins and has never been revisited since then by mainstream and neoclassical economists. Putnam’s work in the philosophy of science extends Williard Van Orman Quine’s arguments against logical positivism and argues for the so-called fact/value entanglement in science. Essentially, Putnam’s argument amounts to the claim that the language of science, especially in human and social sciences, contains thick concepts consisting of descriptive and normative content and that these concepts are pervasive in science and ineliminable from scientific research. However, his call for economists’ attention went largely unheard, with the exception of Amartya Sen (Sen, 2005) and, partly, of Partha Dasgupta (Dasgupta, 1993). In fact, Putnam’s point also did not spark much interest among philosophers of economics and economic methodologists (an exception being Davis, 2013; Su & Colander, 2013), even though many of them are aware of Putnam’s argument and often refer to it in passim. Yet, what paved the way towards a more substantial engagement of economic methodologists with the perspective of philosophy of science on value-ladenness of scientific research, was not a general philosophy of science and analytical philosophy of Quine and Putnam, but rather the analyses of philosophers of science oriented more towards scientific practice, mentioned in the section above.

In my overview of the attempts of economic methodologists to uncover the presence of values in economics, I will rely on the framework proposed by Julian Reiss. Building on the literature in the philosophy of science, Reiss comments on the following stages at which values enter economic research (‘positive economics’): theory formation and development of concepts, modelling, hypothesis testing and hypothesis acceptance (Reiss, 2017). Below, I adopt this framework and I add a stage of measurement to systematize the analyses of value-ladenness of economics inspired by the literature on values in science in the philosophy of science. Afterwards I mention normative proposals of how to account for values in economics raised by economic methodologists.

Theory formation

Reiss discusses the so-called theories of rational choice (ordinal choice theory, expected utility theory and game theory) in order to show how non-epistemic values sneak into economic theorizing. He points out that this happens via formal assumptions of these theories that are in fact ‘substantive normative assumptions’ (Reiss, 2017, p. 138), such as the sure thing principle that ‘tell[s] (…) individuals which features of a decision situation ought to be relevant to their preference over alternatives’ (139). Similarly, yet at a more general level, Marcel Boumans and Davis (2015) argue that the rational choice theories equate preference satisfaction with being better off and in this way they provide ‘a particular ethical interpretation of what it means to be better off’ (198). Daniel Hausmann and McPherson (2006) point out that the ethical principle of ‘minimal benevolence’ underlies these theories: a presumed value judgement stating that one should be able to make oneself better off by satisfying preferences. I refer to the values in science discussion in order to I claim that the theoretical modifications of the expected utility theory (EUT) can be better understood once we notice that decision theorists were proposing modifications of EUT by accounting for empirical evidence, as well as in the light of their commitment to the notion of rationality (‘minimal notion of rationality’ defined explicitly by decision theorists) (Małecka, 2020). This commitment is a form of value-ladenness that significantly impacted the development of EUT and contributed to, mistakenly and confusingly, treating EUT as a normative theory.

Concept formation and measurement

Many philosophers of economics draw attention to the fact that very often the concepts used in economic studies (such as poverty, inequality, unemployment, health) are thick (e.g. Alexandrova, 2018; Hausmann & McPherson, 2006). They do not limit themselves to pointing out this thickness as a form of value-ladenness, but they also analyse attempts to operationalize such terms in empirical research and then show how non-epistemic values impact construction of measures.

In her book on the sciences of well-being, Anna Alexandrova engages with the intricacies of the choice of concepts (constructs) of well-being in empirical research (Alexandrova, 2017). Even though economics is not Alexandrova’s main focus, she does engage with economics to the extent to which it concerns studies and measurement of well-being. Alexandrova analyses how normative considerations (normative preconceptions about the good life) matter in developing constructs and measurements of well-being by researchers. She shows that normative decisions in this research do not appear only at the stage of defining the object of study (which is normative par excellence in this case), but they keep reoccurring in all stages of scientific inquiry.

The value-ladenness of measurements is discussed in a detailed way in the context of measures used in health economics and public health policy. Andrew Schroeder (2019) analyses embeddedness of values in the DALYs.⁴ He points out that values may be built into measurements in two ways: in a user-assessable way, ‘which can be adjusted for after-the-fact by decision-makers’ (527), or in a fixed way when a user of a measure can’t realistically adjust it in the light of their values. Aspirational life expectancy is an example of a fixed-built value choice in DALYs: it is difficult to adjust it in light of conceptions of life expectancy that differ from the one built into a measure by an economist, as this would require a lot of aggregated data about life expectancy of different groups in a population and a relatively complex calculation.

Alessandra Cenci and Hussain (2020) discuss a novel methodology used for the measurement of welfare and health: the multidimensional first-order dominance approach (FOD) and argue that it allows avoidance of the weaknesses of other measures, such as QALYs⁵ and DALYs, that stem from their ‘subjectivist logic and underlying assumptions of standard economics analysis (e.g. cost-effectiveness, cost-utility)’ (cp., Hausman, 2015), as well as from arbitrary value judgments concerning the so-called weighting schemes in these utility-based measures. They argue that FOD as a multidimensional measure is more robust because it is based on pairwise comparisons of population distributions and at the same time it embeds the non-epistemic values of fairness-equity and redistributive justice.

Reiss discusses the consumer price index, GDP, and unemployment rate (Reiss, 2008, 2013) as examples of measurements laden with normative assumptions. According to him, consumer-price indices assume that if a consumer prefers a bundle x over an alternative y, then x is better for her than y, which is both ethically charged and controversial. Similarly, national income measures assume that countries that exchange a larger share of goods and services on markets are richer than countries in which the same goods and services are provided by the government, which too is also an ethical and controversial assumption.

Modelling

Conrad Heilmann (2017) analyses in detail the role of values in modelling in climate economics, in particular in respect to how time discounting functions are defined in these models. The assumption of time discounting – putting higher value on present outcomes than on the future ones – is problematic when climate models are applied to intergenerational decision-making, as it discounts the way in which climate change impacts future generations. Apart from claiming that time discounting is an ethical problem in itself, Heilmann also argues that the methodology of the social discount rate used in climate economics modelling ‘does not provide a framework in which facts and values can be separated’ (1341). This means that parameters of the social discount rate cannot be interpreted without referring to ethical value judgments, such as spatial and intertemporal equality.

Julie Nelson brings the perspective of feminist economics to discuss the interplay of epistemic and non-epistemic values in discussions on modelling in climate economics (Nelson, 2008), when two distinct, but interrelated value judgments are made: on moral weight put on well-being of future generations and epistemic values of what counts as rigorous scientific work in economics. They are related because defending the optimal growth model in climate economics due to its epistemic values of ‘clarity, logical rigour, precision, elegance, parsimony, and generality’ (445) comes at the cost of accepting morally problematic implications of the model for intergenerational equity. Nelson claims that celebration of these epistemic values and disregard for modifying assumptions of the optimal growth model in accordance with ethical considerations gives economists only an illusion of value-neutrality. The model leads to the moral acceptance of the status quo that is manifested in ‘performing policy evaluation purely from the baseline of the existing value income distribution’ (442), which is hardly value neutral.

Ekaterina Svetlova (2014) analyses modelling in the hybrid context of financial practices. Her analysis seeks to reinforce the argument in the philosophy of science that the boundaries between epistemic and non-epistemic values are blurred. She discusses the role of epistemic values when judgement about models’ performance is made in hybrid contexts. Svetlova shows how in the case of the Black–Scholes-Merton model, the status of the model’s simplicity as a value changed from purely epistemic to non-epistemic. In the case of capital asset pricing model,

its fruitfulness (understood as the ability of the model to give rise to a variety of successful applications in the financial industry) and scope (the ability to go beyond obvious and direct solutions) outweighed its epistemic weaknesses, thereby becoming the most important values for the judgement of the model. (94)

Hypothesis testing

Reiss (2017) remarks that the methodology of randomized controlled trials (RCTs) is used in economics to test hypotheses, in particular in the context of research that is supposed to have direct policy implications. However, he notices that ‘to demand testing by RCTs thus privileges some forms of treatment over others’ (144) because not all socially relevant hypotheses can be tested in the experimental setting that requires blinding. Thus, the methodological choice about using RCTs influences the choice of a research question and is made in the light of ‘an implicit value judgment about which questions are important to be addressed’ (145).

Donal Koshrowi (2019) further looks at the RCTs methodology often used to advance evidence-based policy (EBP). He argues that EBP is implicitly committed to the ideals of value-freedom and value-neutrality. Koshrowi notices that RCTs only identify the average treatment effect of a policy, whereas public policies invariably affect people in heterogeneous ways. Consequently, policymakers relying on RCTs cannot assess policies’ distributive consequences. One way to address this problem is a subgroup analysis, which is discarded in the methodological guidelines on EBP because it compromises the epistemic values of RCTs, such as rigour, unbiasedness and precision of the estimates and the ability to obtain causal conclusions on the basis of the estimates. Hence, Koshrowi argues that the epistemic values of RCTs are not neutral in respect to the goals that polices achieve, as this methodology is suited to propose policies that aim at achieving aggregate, or average welfare, but at the cost of ignoring their potential negative, morally, or politically contestable, distributive consequences.

Hypothesis acceptance

Reiss (2017) reminds us that economists, as all scientists, may face a moral dilemma when deciding about which hypothesis to accept and which one to reject, as has been pointed out by Rudner (1953) and further elaborated by Douglas (2000). For instance, Reiss argues that

many economists trust the natural experiments that have shown that small increases in minimum wages do not have negative employment effects. If it were in fact false, then by accepting it, many individuals will be thrown out of work following an increase in the minimum wage. But if it’s true and rejected, governments would forgo the chance of improving workers’ standards of living. (146)

Normative perspectives

Apart from analysing the ways in which values enter the knowledge production in economics, philosophers and methodologists of economics engage in the normative discussions on values in science. Alexandrova (2018) argues that she identified a form of value-ladenness not discussed in the philosophical literature on values in science yet, namely what she calls mixed claims, and then she proposes three rules that should enable the objectivity of science with mixed claims: ‘to make explicit the value presuppositions of mixed claims, to check whether the empirical claim is robust to disagreements about values, and, finally, if it isn’t robust, to expose these values to an inclusive deliberation’ (441). Both Schroeder (2019) and Koshrowi (2019) argue that the measurements and methodology they analyse are insensitive to distributive consequences due to the values they embody. Schroeder suggests that ‘health economists should seriously consider adjusting QALY- and DALY-based analyses (including cost-effectiveness analyses) to reflect egalitarian values’ (534) and Koshrowi claims that it would be legitimate in the context of EBP to rely on non-epistemic values when recommending a particular methodology over another to ‘promote the pursuit of the moral value in question, even if this proceeds on pain of sacrificing other, epistemic values’ (70).

Final remarks

There is much to be said about the above-mentioned contributions. However, a critical engagement with them is not the aim of my review. I limit myself here to mentioning why I find them a promising development in economic methodology. The discussion on values in science in the philosophy of science offers a different approach for analysing values in economic research from the older debates on value-judgements in economics, which were mostly very general and closely linked to the meta-philosophical views on a status of economics as a science. The approach I have discussed here resigns from such an abstract perspective and instead looks at how scientific knowledge is produced in economics and whether diverse normative commitments play a role in this process, and to what extent. Normative questions of how to ‘manage’ values in science, if raised, are grounded in the analytical and descriptive work, which requires engagement with the details of research and its methodology. Therefore, it is no wonder that economic methodologists have started realizing that identifying the role of values in economic research may be an indispensable part of an analysis of methodology in economics.⁶

The feminist scientists and feminist philosophers of science made important contributions, listed above, to the discussion of how to identify normative commitments in science and strongly influenced the debate on values in science. Thus, the approach to values in science from the philosophy of science brought to economic methodology offers an opportunity to rethink the legacy of feminist economists as a methodological work uncovering non-epistemic values in economics. For that reason, I include the article of Nelson in my review of the recent turn towards values in science in economic methodology. I believe that the new book by Nelson (2017) on gender and risk taking could be taken as another example of a methodological analysis of how values enter economics. The interest in values in science, with a new twist of the philosophy of science, that we observe among philosophers of economics, can also pave the way towards their greater engagement with the neglected, yet painfully actual, topic of the influence of race preconceptions on economic theorizing and research. There is much to be hoped for and awaited in the years to come.

Disclosure statement

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

Additional information

Funding

Part of the work on this project was conducted thanks to the support of the European Union’s Marie Skłodowska-Curie Fellowship EPISTEMEBEHAVIOUR.

Notes on contributors

Magdalena Małecka

Dr. Magdalena Małecka is a University Researcher at the University of Helsinki in the Practical Philosophy discipline and a Fellow of the Institute for Advanced Study in Princeton (2020–21). She has been a Marie Skłodowska-Curie Fellow at the University of Helsinki and Stanford University (2018–2020). She also held fellowships, posts and research visits at the European University Institute, Central European University, Columbia University, the New School for Social Research, University of California, Berkeley. Currently she publishes mostly in philosophy of science, especially in topics related to economics and the behavioural and social sciences.

Notes

1 It had been pointed out around seven decades ago that values may play an indispensable and irreducible role in science (see e.g., Rudner, 1953). However, over the last two decades, we have witnessed a proliferation of voices and philosophical contributions on this issue and a growing number of analyses of value-ladenness of scientific research.

2 Longino (1995, 2008) argues that values treated as epistemic by Kuhn (1977) in fact have socio-political valence, and reliance on them in scientific practice is mostly not politically neutral. Laudan (2004) proposes to understand epistemic values as only those that directly bear on the veracity of a scientific theory, such as empirical adequacy and internal consistency. For him values such as scope and explanatory power are cognitive, not epistemic, as they are valued by scientists for other, not truth indicative, reasons. Rooney (1992) argues that the distinction between epistemic and non-epistemic values should be abandoned due to the lack of clarity concerning what epistemic values are.

3 Even a moderate version of the value-freedom thesis says that epistemic values can enter the scientific process; only the strong version of the thesis states that science is objective if no values play a role in the scientific process.

4 Disability-adjusted life year. A measure of disease burden expressed as the number of years lost due to ill-health, disability or early death.

5 Quality-adjusted life year. A measure of disease burden, including both the quality and the quantity of life lived, it multiplies the utility associated with a given state of health by the years lived in that state.

6 However, the recent descriptive analyses of values in economics are still sometimes mixed, without realizing, with older discussions on value-judgements in economics, which is probably the most confusing aspect of them.