Economics and Diversity

Carlo D'Ippoliti

• Publisher: Routledge
• Year: 2011
• eBook ISBN: 9781136718830
• Topic: Economics
• Subtopic: Economic History
• Index: Economics

The bulk of contemporary economics assumes rather than explains differences between people or groups of people. Yet, many of these differences are produced by society or they imply differing opportunities and outcomes. This book argues that economists should concern themselves with the explanation of the social causes and effects of such differences.

D'Ippoliti introduces the concept of diversity to summarise all differences that are of social origin and that a theory or model seeks to explain. This contrasts with the traditional concept of heterogeneity that instead refers to differences that are deemed to be exogenous of economic theory. In approaching this, the book ranges from the fields of methodology and history of economics to applied empirical work, as well as gender diversity which is considered in depth. The analysis of the thinking of two major economists of the past, John Stuart Mill and Gustav Schmoller, demonstrates how gender diversity exemplifies some of the fundamental issues in economics, such as the division of labour, society's capacity to reproduce itself, and the role of social institutions and their impact on individual and collective behaviour.

The book maintains that growth of GDP and of the services sector cannot be trusted to automatically bring about greater inclusion of women in the labour market. Active policy interventions are needed, spanning from the removal of discrimination to the provision of public services and the establishment of fair competition in the market, along with an improved division of social and political power between the sexes. This work will be of interest to researchers and students focusing on the history of economic thought, labour economics, social policy and gender studies.

Part I

There is Difference in Difference

Diversity and Heterogeneity

1 The Night in which all Cows are Black?

It is convenient to define two possible concepts by which economic analysis may take into account differences between people, shedding some light on the many colors of economic agents. From a substantive point of view, I define diversity as the idea that some individual differences arise from society or from the economy (that is, from people’s social status), implying certain roles in terms of the division of labor, and a group-specific set of constraints and rules of conduct (either legal or social norms) in terms of behavior. From an analytical point of view, I shall call diversity the set of inter-individual or inter-group differences that are endogenous to a certain theory; that is, those differences that the researcher seeks to explain, at least to some extent, by means of a certain social or economic story. On the other hand, I follow the mainstream use of calling heterogeneity those differences that are assumed as a datum for the subsequent analysis; that is, differences that the researcher does not seek to explain. Generally, heterogeneity does not refer to people’s roles in society, taking the form of quantitative differences between persons holding similar positions in the social division of labor. In the bulk of the economic literature, heterogeneity usually concerns agents’ endowments, preferences, expectations, information sets or cognitive abilities.

The main aim of this book is to show that recourse to diversity may at times prove useful for the study of economics, while arguing that the discipline itself – by no means the only so-called mainstream approach – currently focuses too much on the latter kind of differences, heterogeneity. In other words, contemporary economics to a large extent assumes rather than explains the existence of differences.

There are of course laudable exceptions, which I shall cover briefly in this chapter; however, they were hardly embodied within the mainstream of their respective fields or schools, being perceived instead as original and eclectic scholars, deviating from the topics and methods that all economists need to know. Sometimes these exceptions gave birth to interesting subfields of economics that nevertheless failed to enter the “core” of economic theory (in the sense of Lakatos, 1978) – that is, they did not (and do not) affect the way that all other economic issues are framed and analyzed. Two observations are necessary here.

First, heterogeneity is not a minor issue. Its assumption does not necessarily support normative backwards or conservative positions, implying that substantial inequality will necessarily remain the same. An exogenous variable is still a variable and not a constant; thus heterogeneity is a proper way to recognize the existence of some kind of difference or inequality, and possibly even to uncover its dimension. However, its adoption implies that the explanation of inequality (along a certain variable) is not the economist’s job, and consequently neither is the eradication of its roots.

Second, the assumption of certain differences as exogenously given within a model does not imply that economic science is unable to explain them: they may very well be the object of other models or studies. This is not necessarily a problem since all theories, including general equilibrium models, are anyway of the ceteris paribus kind; a theory necessarily deals with a finite number of elements while taking all others (most implicitly or hidden) as exogenous. However, whether or not certain elements can be ignored (or assumed as constant) in handling a certain problem is an issue that should at least be subject to investigation, while the parcelization of subfields risks bringing about fundamental inconsistencies in the edifice of economic theory. This is the case with heterogeneity and the representative agent, for example; two hypotheses that one stream of literature has long shown to be mutually inconsistent (see below) while another still relies heavily upon them (see, for example, Chari et al., 2009; Woodford, 2009).

There is an asymmetry between heterogeneity and diversity, in so far as the variable (or variables) over which to investigate the presence of diversity (that is, the variable that we wish to explain with a theoretical model) is by necessity a choice of the researcher. By contrast, due to the impossibility of explaining everything at one time, every theory must assume many forms of interpersonal differences to be nonexistent, or exogenous at best. In other words, due to the ceteris paribus hypothesis we must implicitly assume individuals’ heterogeneity (or even their homogeneity) over a number of variables of no immediate interest, even when we acknowledge and try to explain differences over certain other variables. For example, Part II and Part III of this book deal with gender differences while mostly ignoring differences of color, class and ethnicity. Due to our cognitive limitations we are always bound to select a limited number of topics for analysis while leaving others in the background. Thus, the consideration for diversity in some variables is compatible with the hypothesis of homogeneity and/or of heterogeneity in all other (exogenous) variables.

Concerning the theoretical status of the two concepts, notice that there is no reference in the definitions above to the logical time in which these differences are acquired (or present): to ask the extent to which differences are derivative or “original”, and to attach the concept of diversity to the former kind of difference and that of heterogeneity to the latter, is in my opinion a slippery framing of the research question. Indeed, as the major exponent of methodological individualism (Hayek, 1946) recognized, humans are originally social. That is, self-contained and mature adults do not spring out of the ground like mushrooms (to use Hobbes’ metaphor), but rather humans are socialized gradually from the very moment of birth. Thus, the distinction between the two kinds of differences is between those that are explained by the theory (that is, the endogenous variables) and those that are assumed as given (the exogenous variables). It goes without saying that neither are all differences necessarily of social origin, nor should all be explained by recourse to economic models.

As mentioned in the definition above, to some extent diversity will usually consist of inter-group differences; that is, differences between aggregates of individuals arising from their belonging to (or being considered as being part of) a group. This is because it is hardly the case that single individuals enjoy a specific and unique social status. Indeed, as will be discussed in Chapter 4, aggregates of individuals may be conveniently defined by the researcher with the very aim of highlighting diversity along some dimension(s).

The issue of using aggregative analysis (that is, to depart from the stricter versions of methodological individualism) is fundamentally intertwined with the aim of highlighting a property exhibited by some individuals and not by others, hence showing a dimension of difference. As J.S. Mill notes, if a general word is used to denote a property (for example, “slavery” to denote lack of freedom), the collection of individuals who exhibit that property (slaves) automatically emerge as an analytical class. The problem of aggregation, therefore, though not the main focus of my analysis, nonetheless constitutes a major methodological issue to be dealt with in this work.

On the contrary, heterogeneity looks at individuals’ differences independently of their belonging to any social group. However, also in this case, independence does not imply mutual exclusion, and heterogeneity may entail inter-group (average) differences, since nothing prevents the definition of two (or more) aggregates of individuals that also exhibit systematic differences of a non-social origin. Again, slavery is an example relating to the works of Schmoller and Mill. Both authors considered slaves’ behavior as different from that of free persons due to the institutional subjection to which they are bound. However, differences between the two groups of people may exist also independently of societal factors: for example, the enslavement of a specific ethnic group can imply a systematic – parametric – heterogeneity between certain free persons’ and slaves’ bodily characteristics, without this difference being produced directly by society. As we will see in Chapter 8, the extent to which differences between men and women can be related to innate and biological sources, as opposed to their social origin, has historically been a crucial analytical question for the economic analysis of gender inequality.

Finally, while the aim of this work is to advocate in favor of a greater effort on the side of economists to understand the possible social origin of many interpersonal differences, the two proposed definitions do not imply that diversity will always be necessarily greater than heterogeneity. Rather, my argument rests on two considerations: (1) that many interpersonal differences are (at least partly) of a social and/or economic origin, and therefore economics should be employed in trying to explain them, regardless of how important other differences of non-economic origin may be; (2) that differences of social and economic origin sometimes matter in the development of economic processes or in the determination of relevant economic variables, and therefore, for the sake of an analysis of these more “traditional” topics as well, they should not be ignored by the theorist.

The Economic Literature on Heterogeneity

Heterogeneity is now a founding block of mainstream economic theorizing. As of September 2010, a query for the words “economics heterogeneity” displays 403,000 results in Google Scholar, 115 in EconLit, and reaches the maximum number of results allowed in the RePEc database. While it is not feasible to summarize all this literature, it is necessary to provide a short overview highlighting the main elements to be contrasted with the concept of diversity.

In microeconomics it is generally, though implicitly, recognized that heterogeneity is the ultimate source of the exchanges in most partial and general equilibrium models. In Schmoller’s words: “the equality of men impedes the exchange” (Schmoller, 1904, p. 4).¹ Taking Edgeworth’s box as a paradigmatic example, were agents not to have different preferences or different endowments, the core of the economy (the set of possible competitive equilibria) would be empty; that is, they would have no reason to exchange anything. As subsequent literature showed, exchanges and a division of labor can still arise under non-competitive conditions; for example, in the presence of static or dynamic economies of scale (Block et al., 2007).

For reasons of simplicity, the economic literature usually refers to parametric heterogeneity; that is, heterogeneity expressed by the quantitative variation of an independent variable, be it continuous (for example, age), discrete ordinal (educational attainment) or categorical (ZIP code). However, there are streams of literature – mainly within behavioral and experimental economics – which formally model individuals who, although belonging to the same social group (thus no diversity is involved) and behaving according to the same (choice) process, do not differ parametrically, but rather in the the objective function representing their behavior (for example, to capture cognitive differences). Thus in the field of microeconometrics an emerging new stream of literature (briefly reviewed by Conte et al., 2009) is increasingly focused on the identification within a single model of three different forms of (exogenously given) systematic differences between individuals: (1) some that are correlated to observable characteristics, such as sex, age or nationality, determining a group-specific intercept or coefficient (observed heterogeneity); (2) some that take the form of individual-specific fixed or random effects (unobserved heterogeneity); (3) some that allow for individuals to be randomly or deterministically assigned to a group of subjects who follow a certain decision rule; for example, exhibiting myopia or hyperbolic discounting, that determines two or more group-specific objective functions that are not nested within a more general model (and thus cannot be reduced to a shift in the constant or in some coefficient, requiring the use of mixture models).

As colorfully explained by Harrison and Rutström (2008), this method allows one to “jointly estimate the parameters of each theory as well as the fraction of choices characterized by each. The methodology provides the wedding invitation, and the data consummates the ceremony followed by a decent funeral for the representative agent model that assumes only one type of decision process” (p. 133). However, for the aim of contrasting the nature and sources of individuals’ differences, functional heterogeneity can be considered as conceptually equivalent to parametric heterogeneity in so far as it is – in the same vein – formally exogenous to the model and substantially independent of society. Moreover, to celebrate the funeral of the representative agent, parametric heterogeneity is sufficient.

Indeed in the field of macroeconomics, (parametric) heterogeneity has been shown to undermine the idea that aggregate variables can be obtained by simple summation (or averaging, or by any linear combination) of the respective microeconomic variables. As Lippi and Forni (1997) show, with a line of reasoning not entirely dissimilar, algebraically, from the Cambridge critique of the notion of aggregate capital, if a dynamic model is obtained by the summation of microeconomic variables the parameters of a macroeconomic function – for example, consumption expenditure as a function of income – do not stand in a monotonic relation with the relevant dynamic parameters of the individual consumption functions. This holds true unless we assume very specific distributions of the relevant parameters in the population (or unless we assume agents’ homogeneity). As the authors remark in the preface, “this is hardly surprising, as aggregation as already been recognized as a major source of difficulty in many branches of economic theory and empirical economics” (p. x). Reference here is possibly made to the domain of microeconomics, where in the 1970s with the so-called Sonnenschein–Mantel–Debreu theorem (see, for example Sonnenschein, 1972) it was already accepted that aggregate market excess demand functions may be not be well behaved, despite the assumption that each individual’s demand function is.²

A simple numerical example of how heterogeneity undermines the possibility to obtain aggregate variables by linear transformation of the microeconomic variables (for example, by summation or averaging) is provided in the Appendix. This neither implies that such anomalies are the rule, nor that they are a frequent occurrence, but that when there is heterogeneity they can happen. At the same time, although anomalies of the kind shown in the Appendix can emerge, we cannot label the aggregate result as “wrong”, since even on these occasions the result is mathematically sound. Rather, it would be more correct to say that aggregate relations carry an economic logic partially separated from individual relations, due to the non-monotonic relationship between the two. The moral of the story, as mentioned, is that both aggregate analyses and micro-analyses are legitimate, though possibly only in their respective and partially separated domains.³ What is not legitimate is the illusion that it is possible to develop microfoundations by simple summation (or linear combination); that is, straightforwardly to infer macrodynamics from the exclusive consideration of individual behavior.

Outside the mainstream, heterogeneity is frequently considered within agent-based computational models and the approaches based on the concept of complexity, as well as within econophysics (see, for example, Dosi et al., 2006, 2008). According to Tusset (2010), it is the very consideration for individual heterogeneity that leads Pareto to conceive stochastic processes to be a more sound foundation for modeling aggregate trends. This example allows us to think of a particular and relevant sense in which interpersonal differences may be exogenous from social arrangements without being “original” or natural; that is, as a consequence of chance. Indeed, many authors were inclined to assume the apparently universal nature of Pareto’s law of the distribution of incomes (that was found to apply to diverse settings and societies) as proof of the existence of natural differences between persons that would ultimately underlie the distribution of income in all societies.⁴ Instead, as shown by Corsi (1995), for example, this regularity has a much simpler explanation in the emergence of a Pareto distribution in the aggregate as a consequence of random proportional variations of individual incomes (that is, by assuming that each year’s individual income is equal to the previous year’s plus a variation proportional to it, whose sign – positive or negative – is randomly determined). Indeed, if we were to develop an explicitly dynamic analysis, starting (or “original”) conditions become increasingly less relevant, while the issue of history versus chance may become central to the distinction between endogenous and exogenous sources of difference.

The Economic Literature on Diversity

Overall, heterogeneity appears to be a substantive argument for avoiding the establishment of direct and unequivocal links between microdynamics and macrodynamics, possibly implying the legitimacy of microeconomics and macroeconomics as two separate fields of enquiry for the majority of “daily” economic applications. The example of interpersonal difference that will be employed here is that of gender inequality, but in more general terms I argue that the proposed distinction between heterogeneity and diversity may also clarify some concepts in the more traditional debate about the legitimacy of the representative agent hypothesis.

For example, the argument originally put forward by Keynes to reject the equality between individual and aggregate saving does not hinge upon agents’ heterogeneity, but rather it falls in the category of diversity. It rests on the consideration of two distinct sorts of agents with two different social roles: firms (that indeed are not even persons) and households.⁵ While the former decide to invest on the basis (among other things) of the interest rate, the latter decide to save as a function of their disposable income. Thus, here the representative agent hypothesis is r...

Table of contents

• Cover
• Half Title
• Title Page
• Copyright
• Dedication
• Contents
• Preface
• Acknowledgments
• Introduction
• Part I: There is difference in difference: diversity and heterogeneity
• Part II: Difference, behavior and aggregative analysis
• Part III: On gender diversity
• Part IV: An application: gender diversity in the labor market
• Conclusions
• Appendix
• Notes
• Bibliography
• Index