Peirce’s ontology: a rejection of the dualism between the material and the ideational

Like the New Materialisms, Peirce’s semeiotics is based on an understanding of the continuity between mind and nature. The effects of our thoughts are displayed in the environment so that we cannot discern a clear boundary between the two (Short Reference Short2007, 9). Clark (Reference Clark1997) notes similarly that we tend to externalize our mind onto the world. For example, we use a pen and a piece of paper to make complicated calculations, or when we write on a computer we move chunks of text around on the screen to organize our thoughts. Our thought processes are not purely ideational, but inextricably linked to the manipulation of material objects, and by consequence those objects structure our thought processes.

According to Peirce ‘it is much more true that the thoughts of a living writer are in any copy of his book than they are in his brain’ (Peirce 1931–Reference Peirce35, 364, vol. 6).Footnote ⁵ A book is neither purely material nor purely ideational, but rather a combination of the two, even though an excessive focus on ‘symbols, which signify via arbitrary convention, has led [scholars] to treat “material qualities” of a sign, such as a word’s letters, as entirely irrelevant to its signification’ (Legg Reference Legg2008, 225). In contrast, Peirce focused on the ‘concrete instances where the specific material quality of a sign enables it to function as the precise kind of sign it is, thus in turn enabling the precise kind of reasoning it makes possible’ (Skagestad Reference Skagestad2004, 251–52).

For example, perspectival painting operated as a sign system that permitted Renaissance artists to identify new geometrical regularities that then contributed to the development of modern mapping techniques with which the territorial state could be represented (Edgerton Reference Edgerton1975; Branch Reference Branch2011). The vanishing point is the central element that orders the proportionality of all the other elements in the painting into a singular, unified, and abstract space with the help of a grid system. The exact location of the elements in the painting, their spatial relations to each other, and the distances between them all become discernible to the viewer. Perspectival painting allowed for the development of projective geometry. Out of a fascination for linear perspective, artists started to draw city views from an oblique elevation, as bird’s-eye views. Initially these oblique positions were actually existing views of the city, for example, from a hill. Once the city view became an imagined view, such as in da Vinci’s drawing of Milan, it was necessary to perform ‘an abstraction based on the principles of mathematical perspective in an effort to preserve accuracy’ (Rees Reference Rees1980, 69). From here it was only a small step to draw a city view from an entirely vertical perspective. Because such a vertical view from directly above the city did not exist at the time, painters had to extrapolate it with the help of similar geometrical calculations and methods of triangulation as the ones used for the imaginary oblique perspective (Smith Reference Smith2008). But a city view drawn from a position vertically above the city and in line with the rules of linear perspective is nothing other than a modern map. To depict territorial states it was merely necessary to extend the scope of the map. Projective geometry then allowed establishing an indexical connection between the map and the terrain, and by virtue of this indexical connection the map could resemble the terrain, it could function as an icon of the territory.

Toward a method of Peircean semeiotics

Peirce’s semeiotics permit us to decodify material and linguistic signs. We can trace how the combined effects of Peirce’s tripartite division of signs into icons, indices, and symbols establish a rich sign system with a variety of communicational patterns.Footnote ⁶ These in turn result in emotional, energetic, and logical interpretants that subsequently alter the reality that will enter the sign system.Footnote ⁷

The first step of this method is to search for signs that represent the objects that interest us by gathering qualitative data to serve as the raw material for the analysis. Any of the existing methods of data gathering can be useful, including participant observation, qualitative interviews, archival research, site visits, museum and exhibition visits, or secondary literature researches (Yanow Reference Yanow2000; Bueger Reference Bueger2014).

The next step is to identify the grounds upon which the collected signs represent their objects. The ground is the relationship that connects the sign to the object; it is the reason that a sign represents a given object. In other words, not every characteristic of the sign is important for establishing its representativeness. For example, a color-filled map of a country represents that country’s territory because it resembles the country’s shape and is causally connected to its geography by surveying techniques. But the concrete color used for filling the map is irrelevant. Furthermore, different signs can represent the same object, in each case focusing on a different ground (Short Reference Short2007). Thus, while a map is simultaneously indexically and iconically linked to the country, a flag represents a country by social convention. Peirce classified signs into icons, indices, and symbols based on the ground that connects the sign to its object. This tripartite differentiation provides the researcher with a vocabulary that directs her attention to the different relations by which signs can signify a particular meaning.

The index: It is probably easiest to first identify the indexical relations between signs and their objects because the index establishes a direct physical connection to its object (Skagestad Reference Skagestad2004) that exists independently of social convention (Atkin Reference Atkin2005). The indexicality of speech, for instance, ties spoken words to the person of the speaker and her position – a tool, which can be useful for speech act theorists.

Index and object can be physically connected in two different ways. First, the object and the index can be connected on the basis of spatial and temporal proximity (Goudge Reference Goudge1965; Liszka Reference Liszka1996, 38). The index can draw our attention to a particular object. Arrows are typical examples of indices. Yet we can broaden the notion of the index much further. The index can include memorial plaques, which highlight that a particular event occurred in a certain location. Perhaps even more politically salient, many religious sites derive their significance from being the location of a particular event, from containing holy relics, and/or from being the site of daily practices of worship. There is often a material and irreplaceable relation that ties the religious site to specific events or things, even though it is not a functional relation. How those ties are then interpreted, and which kinds of emotional reactions, thoughts, or actions they evoke, will be different for different people (Hassner Reference Hassner2003). Without a Peircean semeiotic approach it can be hard to discern how sacred space can contain a material dimension of sacredness (which is its indexicality tied to the occurrence of a particular event, the positioning of relics, etc.), while simultaneously being subjectively sacred, that is, how that indexicality is interpreted depends on the eyes of the beholder. Goddard (Reference Goddard2006), for one, did not grasp the material dimension in Hassner’s argument, which led her to claim that her own argument about the intersubjective construction of the indivisibility of sacred space is superior to Hassner’s claim of subjectively held beliefs of sacredness. Yet, the indexicality of sacred space means that there are hard limits to socially constructing sacredness, which is important for understanding the intractability of many conflicts over religious sites, as well as the search for possible solutions to those conflicts. Acknowledging this indexical dimension has the advantage of taking religion seriously and not entirely reducing it to a social construction (Gregory Reference Gregory2006).

Next to a connection based on spatial and temporal proximity, a causal relationship can exist between an index and its object. This can be based on the need for financial resources to obtain a particular sign or because of a functional link that connects the sign to its object. Much of the newly emerging status literature focuses on this dimension. For instance, China’s aircraft carriers represent China’s increasing economic wealth and development: aircraft carriers are very expensive and require high levels of technological and organizational skill to operate (Pu and Schweller Reference Pu and Schweller2014). It is not surprising that there is a correlation between the acquisition of aircraft carriers and the availability of economic resources: ‘China is not a unique case as there is always a clear correlation between economic resources and the strength of naval power’ (Pu and Schweller Reference Pu and Schweller2014, 156). The indexical relation here merely indicates that a state has the necessary resources to be able to afford aircraft carriers; it does not tell us anything about how others interpret this acquisition, whether as an unnecessary wastage, a threat, or an awe-inspiring move.

Quantitative research makes use of the index when employing ‘proxies’ as measurable variables that are causally related to potentially immeasurable phenomena. For example, number of battle deaths is frequently used as an indicator for the intensity of a war, or income can serve as a proxy for social class (Bryman and Cramer Reference Bryman and Cramer2009). It is of course not possible to simply assume the validity of proxies. Researchers need to clarify ‘the extent to which a measure assesses the construct that it is intended or supposed to measure’ (Cramer and Howitt Reference Cramer and Howitt2004, 36). Even so, proxies can be extremely useful indicators in a large variety of different fields. For clarification, it is worth noting that the causal relationship inherent in the index differs from the positivist understanding of causality: the indexical causality is merely a causality between an object and its sign. It does not necessarily establish causality between a dependent and an independent variable as two phenomena of a real and unmediated world.

We can often identify the existence of an indexical relation between an object and a sign in language usage with the help of the rhetorical device of metonymy. Metonymy is a common linguistic practice that signifies the use of ‘one entity to refer to another that is related to it’ (Lakoff and Johnson Reference Lakoff and Johnson1980, 35). ‘Metonymic concepts are grounded in our experience. In fact, the grounding of metonymic concepts is in general more obvious than is the case with metaphoric concepts, since it usually involves direct physical or causal associations’ (Lakoff and Johnson Reference Lakoff and Johnson1980, 39). Lakoff and Johnson provide numerous examples of daily usage of metonymy, such as ‘Wall Street is in a panic’ or ‘the White House isn’t saying anything’ (Reference Lakoff and Johnson1980, 35–38). We can identify from these phrases that there is a direct association of physical proximity between Wall Street and the New York stock exchange, as well as between the US presidency and the White House.

The icon: Because the indexical relationship merely establishes a material connection between a sign and its object, it is subsequently necessary to deduce specific characteristics of the object from the sign at hand. One way of doing this is to identify an iconic relationship between a sign and an object (which can at times also exist independently of an indexical relationship). An icon represents an object because of a similarity in a particular characteristic shared by both sign and object. The resemblance that relates an icon to its object can be sensory – for example, a pictorial resemblance. The portrait of a king is an icon in the sense that it resembles the king. But Peirce also considered music an icon because it incarnates musical feelings, which are its object (Short Reference Short1996).

Furthermore, the relationship between icon and object can be a structural resemblance, in which case ‘[the icon’s] parts should be related in the same way that the objects represented by those parts are themselves related’ (Legg Reference Legg2008, 207). Analogies are a typical example, because they establish a relational resemblance where the relation between A and B is the same as the relation between C and D (Liszka Reference Liszka1996). Diagrams are another category of icons that represent through structural resemblance; a typical example is a graph representing economic growth in a given country.

Icons as means of signification can be found outside of language and cognition; Peirce considered icons typical for the fine arts (Short Reference Short2007), where they can lead in particular to emotional interpretants, that is, they evoke feelings rather than thoughts. It then becomes apparent that iconic representation can be material on two accounts. On the one hand, the sign, the icon itself, can be material – an artifact, a practice, or a natural occurrence – but the icon can also be a linguistic expression, metaphors being a case in point. On the other hand, the icon is material in the sense that in its pure form the icon represents its object solely on the basis of similarity. Hence, the object can be inferred from the icon through unmediated phenomenological perception, directly by the senses and through the body, without the need for intersubjectively shared rules (Kolenda Reference Kolenda1977). The pure icon is a non-arbitrary sign, although we will see that few signs exist in this pure form.

The symbol: To determine a symbolic connection between a sign and its object, we have to identify the intersubjectively shared understanding that underlies a symbol’s signification. The symbol represents a particular object only because there is shared background knowledge that it does so. Symbols have the same characteristics as all signs in the conventional linguistic understanding of sign systems. It is therefore useful to apply the usual discourse analytical tools to the study of symbols (Doty Reference Doty1993; Milliken Reference Mick1999; Hansen Reference Hansen and Porter2006). Most, but not all, words are typical examples of symbols, but cultural artifacts or practices can equally be of a symbolic nature (Liszka Reference Liszka1996). For instance, some scholars in a growing IR literature on forms of popular culture analyze movies, board games, or computer games with the help of discourse analytical tools. Thus, Weldes edited a volume in which the authors ‘view [Science Fiction] as a generic discourse or system of meaning-producing codes’ (Reference Weldes2003, 13).

Complex signs: combinations of icons, indices, and symbols: The three categories of signs – icon, index, and symbol – are irreducible to each other; each fulfils a distinct function in semeiosis, which cannot operate if any one of them is missing. At the same time it is very hard to find a sign that is a pure icon, index, or symbol. It is rather common to find complex signs incorporating a mix of these three categories (Hookway Reference Hookway1985; Atkin Reference Atkin2005; Legg Reference Legg2008). Typical examples of signs that combine iconic and indexical components are biometrics, such as fingerprints that are increasingly used in modern border control technologies (Scheel Reference Scheel2013). Fingerprints are indexically tied to a particular person through the process of taking fingerprints, and they iconically represent the imprint of a person’s finger. Demonstrative pronouns are combinations of symbols and indexes because they are different in each language, and at the same time they incorporate an unmediated relationship to an object marked by spatial and temporal proximity (Savan Reference Savan1987; Atkin Reference Atkin2005). Lastly, metaphors are an example of signs composed of symbolic and iconic components. On the one hand, they rely on language, which is largely of a symbolic nature, on the other, they are based on similarities, on pictorial imaginations, which often seek to express otherwise inexpressible emotional states. Disentangling signs according to the different grounds based on which they represent an object can help researchers to identify the functional and symbolic connections between signs and their objects.

The combined effects of icon and index permit signs to simultaneously refer to an object, which is the indexical part, and characterize that object, which is the iconic part (Liszka Reference Liszka1996). In Peirce’s understanding the arbitrary connection between symbol and object can hinge on the non-arbitrary functions of index and icon (Sorrell Reference Sorrell2004). Over time the connection between index and object can grow into general conventions and rules (Short Reference Short2004). Thus, we might be able to establish a temporal sequence during which the sign changes its function from index to icon to symbol (Goudge Reference Goudge1965; Short Reference Short2007, 227; Legg Reference Legg2008). Process tracing such temporal progressions of changing forms of signification can be a powerful tool to understand how certain social conventions develop or how some symbolic significations emerge.

A brief sketch of the historical development of money can serve as an illustration. Money is a popular constructivist example for explaining intersubjectively shared norms (see e.g. Searle Reference Searle1995; Adler Reference Adler1997). According to the conventional constructivist understanding money can only function as a medium of exchange if it is accepted and recognized as such within a given community. Constructivists highlight the symbolic nature of money when they rightly claim that whales’ teeth are unlikely to be accepted as a medium of payment in a North American supermarket, even though they might still be accepted on the Fijian islands.

While I do not want to deny the symbolic nature of money, a more historical perspective suggests that indexical and iconic features played a crucial role in the emergence of particular forms of money. Contra linguistic turn approaches, the materiality of money shaped the meaning it evoked and the uses it served. Especially, in its earlier stages of development, money had to fulfil certain material and functional characteristics – specifically, the objects serving as money had to be transportable, relatively rare, quite hard to counterfeit, and easily divisible. Precious metals were one type of material that fulfilled these criteria. A cursory glance at monetary development in Western Europe demonstrates that during the Middle Ages the nominal value of a coin was initially equivalent to the value of the metal out of which the coin was made. An indexical and an iconic relationship existed between a coin’s nominal and intrinsic metallic value. But over time, in order to overcome budgetary difficulties, the princes who had the right to coinage began to debase the currency. So-called ‘token money’ emerged. The nominal value exceeded its metallic content, although the coins looked the same. The indexical relation between the coin and its value disappeared, whereas the iconic relation remained intact. Although this maneuver was initially considered fraud, in the 16th and 17th centuries economists started to suggest that it was necessary to have token money in circulation for the smooth functioning of the economy, as long as the supply of coins did not exceed the demand for transactions (Cipolla Reference Cipolla1956). Thus, a social convention emerged, and a symbolic character was added to the relation between coins and their value. Toward the end of the 17th century banknotes came into circulation (Braudel Reference Braudel1979, 414–18, vol. I). By then the relationship between money and its value ceased to be an iconic one, and became a symbolic relation based on a shared rule of accepting those pieces of paper as money.

We can identify a very similar temporal progression of changing forms of signification in an entirely different case, namely the establishment of the principle of sovereign equality on the basis of diplomatic forms of representation. The precondition for this development was the emergence of a clear indexical relation that tied a diplomat to his sovereign (instead of having a diplomat serve numerous sovereigns simultaneously), which emerged by the end of the 17th century and was materially manifested in the diplomat’s wage, his letters of credence, and his diplomatic instructions (Queller Reference Queller1967). By virtue of the indexical relationship between the diplomat and his sovereign, the diplomat could also operate as an icon that represented the sovereign through a relationship of structural similarity. Perhaps the most immediate indicator of the rank of a particular sovereign in the European hierarchy was the embodied practices of diplomatic precedence, in particular the spatial positioning of diplomats at social occasions. Where a diplomat was located in a procession indicated the social rank of his sovereign. As a result disputes over precedence were very common. In order to avoid these disputes rulers increasingly resorted to diplomatic practices that would express the equal standing between rulers, such as entering a room simultaneously on the sound of a horn (Hamilton and Langhorne Reference Hamilton and Langhorne2011). Over time the social convention of sovereign equality could emerge out of these iconic diplomatic practices. At the Congress of Vienna in 1815 rulers agreed that diplomatic precedence seized to play its iconic role. The norm emerged that among equally ranked diplomats the person who is on her post for the longer time period takes precedence (Wight Reference Wight1977). The new diplomatic practices symbolized sovereign equality by virtue of the agreed social norm; they did not represent it any longer iconically.

The interpretant: The last step of the analysis is to identify the interpretant. The interpretant is the potential for interpretation or misinterpretation that is inherent in a sign (Skagestad Reference Skagestad2004; Sorrell Reference Sorrell2004), even if that potential does not always become an actuality, or does not always become an actuality in the same way. In Peirce’s understanding signs are usually interpreted in relation to other signs (Gottdiener Reference Gottdiener1995). The context of a given situation matters for completing the meaning of a sign, as do the historical experiences and knowledge of the interpreter, ‘the collateral information’ (Metro-Roland Reference Metro-Roland2009, 274). Depending on context, a sign can stand for different interpretants (Short Reference Short2007). Thus, the palace of Versailles with its adjacent gardens evoked a sense of glory and prestige in Louis XIV, while it provoked a fear of supremacy in other rulers, which led them into a balancing alliance against the Sun King (Mukerji Reference Mukerji1997).Footnote ⁸

The researcher’s task is to trace the connections between sign and interpretant and identify how particular components of a sign, given a specific context and collateral information, lead the actors to feel (emotional interpretant), act (energetic interpretant), or think (logical interpretant) in certain ways. A first approach can be to draw ‘on the researcher–analyst’s participative experiences as proxy for others’ behaviors and actions: through those firsthand, immediate experiences the analyst gains entry into understanding others’ responses’ (Yanow Reference Yanow2000, 64). Nonetheless, these experiences of the self should be further backed up with interviews or documentary analysis in order to identify other people’s responses to a particular sign in a given context and relate those to their historical experiences and knowledge (Yanow Reference Yanow2000).

Peircean semeiotics provides one way to highlight the profound interconnectivity of the material and the ideational. At the same time, other methodological approaches have different advantages and it is possible to combine Peircean semeiotics with other tools. One typical example would be to combine Peircean semeiotics with discourse analysis. One could also combine Peircean semeiotics with abduction – the process whereby we intelligently guess the most likely explanation for a particular phenomenon, usually by moving back and forth between theory and empirics (Peirce 1931–Reference Peirce35, 530, vol. 6; Hookway Reference Hookway1985, 224). Peirce would not have considered his semeiotics incompatible with abduction, a concept he himself invented. One of several ways that he related abduction and semeiotics was to suggest that the semeiotic apparatus permits the analysis of signs that provide the empirical material from which abduction can develop.

GDP’s indexical features

The emergence of money, accounting, and numeration have been historically linked (Roover Reference Roover1963; Schmandt-Besserat Reference Schmandt-Besserat1984; Latour Reference Latour1990; Robson Reference Robson1992). Scholars have argued that commerce and monetary exchange were at the origin of index numbers in general (Klein Reference Klein1997, 76). The exchange that occurs on markets in monetary terms allows accountants to observe the monetary, and therefore numerical, values that are attributed to the goods traded on markets. Accountants can thus aggregate the monetary values of a large number of items, which otherwise do not share any other characteristic in common. ‘A fundamental merit of market transactions is that they provide “objective” prices that serve to value quantities of goods and services’ (Stiglitz, Sen, and Fitoussi Reference Stiglitz, Sen and Fitoussi2009, 86).

The concrete ways in which national accountants observe the monetary values of traded items, that then become aggregated to a GDP figure, include tax returns, import and export documents, the public accounts of the government, and the institutions it finances, ‘published financial accounts of private enterprises such as listed public companies, producer boards, and nonprofit organizations’ (Waring Reference Waltz1989, 116), data gathered in connection with other governmental regulatory activities, data collected by trade bodies, large-scale censuses, periodical questionnaires, and sample surveys (Fogel et al. Reference Fogel, Fogel, Guglielmo and Grotte2013). One specific method developed by Wassily Leontief, namely input–output tables, traces the movement of intermediate goods through the economy to compute the value added in production (Coyle Reference Coyle2014). The sum of all the values added in the economy gives us the GDP measured from the production side of the economy.

GDP’s symbolic features

Despite the clear indexical connection between GDP and the economy, many symbolic features enter into the calculation of GDP. GDP is an aggregate indicator, and therefore depends on categorizations which regulate what is included within its purview and what is excluded. ‘Distinct normative purposes determine what is to be measured’ (Perlman and Marietta Reference Perlman and Marietta2005, 227),Footnote ¹¹ but at the same time it is necessary to take into account what can be measured. In the case of GDP, there is a connection between the socially agreed upon rules about which items are included in the measurement of GDP, and the possibility of directly observing the monetary values of those items. That is to say, there is some connection between the symbolic and the indexical features of GDP. GDP is mainly, but not exclusively, a measure of market production (Stiglitz, Sen, and Fitoussi Reference Stiglitz, Sen and Fitoussi2009). It requires the Peircean semeiotic apparatus to be able to identify this connection between the symbolic and the indexical dimension. A discourse analytical approach cannot account for the fact that certain components are excluded from GDP because they lack the indexicality that would allow to attribute a numerical value to a certain object, as discourse analysis does not consider the possibility of a direct link between a signifier and a signified. Old materialists, in turn, do not consider the role of socially agreed conventions that play a role for including certain components in the calculation of GDP while excluding others.

When items are not exchanged on the market, it is impossible to observe their monetary value and identify a recorded exchange (Anderson Reference Anderson1991). We have to work with imputed values, which are highly assumption-driven data – data, which are based on social conventions and theoretical commitments, and therefore of a symbolic nature. In such cases the indexical link to the real economy is very dubious; the data become less reliable (Stiglitz, Sen, and Fitoussi Reference Stiglitz, Sen and Fitoussi2009). Yet, some items are included in GDP calculations although they are not recorded market exchanges, among them measures for the home production of goods, imputed rents of homeowners, or the shadow economy. Other items, such as leisure, or the home production of services like cooking, cleaning, and raising children, have not been included in GDP (Waring Reference Waltz1989). Given that the amount of time dedicated to these tasks varies among countries, an inclusion of such measures would affect comparisons between countries. For example, it is estimated that the production of household services is at the level of 35% of GDP in France, while it is 30% of GDP in the United States (Stiglitz, Sen, and Fitoussi Reference Stiglitz, Sen and Fitoussi2009, 130). An additional inclusion of leisure as a good that people enjoy, might add another estimated 80% to GDP (Stiglitz, Sen, and Fitoussi Reference Stiglitz, Sen and Fitoussi2009, 90). However, the question how to attribute market prices to leisure is quite daunting. The unreliability of imputed data also haunts alternative aggregate economic indicators, such as Net Domestic Product (which accounts for the depreciation of capital goods), or environmentally adjusted GDP. These alternative indicators place a higher premium on sustainability, but national accountants resist their implementation, because they rely so heavily on imputed data.

Also the ways in which GDP is calculated depend significantly on social norms. Many of the rules according to which GDP is computed are derived from economic theory. In particular, Keynesianism was at the origin of GDP’s invention in the 1940s. Following Keynesian theory, fundamental identities underlie the calculation of GDP. For example, if we want to calculate GDP from the demand side of the economy, we have to follow the equation Y=C+I+G+(X−M), that is income=consumption+investment+government expenditure+(exports−imports). Economists contested these identities in the 1940s (Suzuki Reference Suzuki2003). For instance, while Kuznets wanted to subtract government spending from GDP rather than add it to it, Roosevelt was interested in accounting for government spending as an addition to GDP, because he did not wish for the war effort to result in an ostensible reduction of output (Carson Reference Carson1975; Coyle Reference Coyle2014).

Following another underlying assumption for the measurement of GDP, we can calculate GDP in one of three ways: by adding all the incomes in the economy, all the expenditures, or all the output. In line with equilibrium theory and the technique of double entry bookkeeping, these three methods have to yield the same results. By definition the income side has to match the expense side. These theoretical commitments were so strong that Stone, one of the leading economists in the implementation of the UN System of National Accounts, remarked ‘it has always seemed to me that in national income and expenditure work where the true values of the variables must satisfy certain definitional equations it would be desirable to adjust the observations actually obtained so that these equations were exactly satisfied’ (Stone, 3 March 1949a, in Suzuki Reference Suzuki2003, 485). So in this case the indexical and the symbolic features of GDP run against each other, with the symbolic features gaining the upper hand.

While GDP measures are occasionally adapted to be in line with theoretical commitments, at other times it is impossible to obtain specific components of GDP by direct observation and measurement. It can, for example, be the case that a state lacks the capacity to collect all the relevant data, as is the case in many developing countries. Under such circumstances mathematical equations that depend on specific assumptions and theoretical arguments help to determine the missing numbers. Even Keynes’s notations of the calculation of GDP contained such footnotes as ‘obtained by subtracting’ (Suzuki Reference Suzuki2003, 497). Nowadays many GDP data of developing countries are not based on observed measures, but on social conventions about how to calculate specific variables. Thus, the UN reports GDP data for 47 sub-Saharan countries for the time period between 1991 and 2004, even though it obtained measures for less than 50% of these observations. In cases of missing data, the World Bank Statistics Manual uses ‘a method for filling the data gap, which is based on the assumption that the growth of the variable from a period of which data exists has been the same as the average growth for those other countries in the same regional or income grouping, where data exists for both periods’ (Jerven Reference Jerven2013, 22). The World Bank transforms social conventions into statistical numbers. GDP’s symbolic features fill the gap left by missing indexical features.

Even in cases with good data availability, national accountants have to rely on some theoretical assumptions because it is tricky to observe certain phenomena directly. For example, inflation interferes with the stability of GDP numbers (Robson Reference Robson1992). In order to make GDP data comparable from one time period to another, we want to evaluate which portion of the change in GDP is due to a mere increase in prices, and which portion of the change is due to changes in the real economy. Unfortunately, measuring the GDP deflator with the help of which we can identify inflation is very complicated.Footnote ¹² One particularly daunting issue is the question of how to differentiate between a product’s price increase that is due to the product’s qualitative improvement, and one that is merely the result of inflation. In the United States, the Boskin Commission focused on this very issue, and created a hedonic price index. The hedonic price index estimates how the price of each feature of a product is linked to specific quality improvements, such as an increase in the size of the memory of computers. Any residual price increase not linked to qualitative improvements is attributed to inflation. However, the method is complex, contested, and depends on many social conventions (Stiglitz, Sen, and Fitoussi Reference Stiglitz, Sen and Fitoussi2009). Even so, the Commission reached the conclusion that not taking into account qualitative improvements of products led to an overstatement of the inflation rate by 1.3% and a corresponding understatement of real GDP growth. The adoption of the hedonic price index resulted in increased US GDP and its growth over several years, which has been linked to the Silicon Valley boom in the late 1990s and the early 2000s. Other countries started to catch up with the US economy once they included similar measures in their calculations of GDP (Coyle Reference Coyle2014, 88–90). The episode highlights the importance of social conventions, but these conventions evolve around efforts to establish the appropriate indexicality of GDP as a sign of the economy. Available methodological approaches are unable to capture this complex interplay between symbolism on the one hand and indexicality on the other.

GDP’s iconic features

The economy and economic growth are either ‘invisible to the naked eye’ (Latour Reference Latour1990, 42), or so complex and confusing that they cannot be deciphered without signs. Only after a lengthy process of extracting data and combining them in various ways ‘can the economy be made visible inside piles of charts and lists. Even this is still too confusing, so that redrawing and extracting is necessary to provide a few neat diagrams that show the Gross National Product’ (Latour Reference Latour1990, 38). The figure of GDP makes the economy visible; it makes it readable (Speich Reference Speich2011). Some would even say it brings it into existence (Mitchell Reference Mitchell2002). The specific GDP number represents the economy by similitude. This is GDP’s iconic feature.

The iconic character of GDP as a number is an important, but often overlooked, material feature of the sign. Existing methodological approaches lack the tools needed to understand the role a sign’s material qualities play in the process of signification. Without these material characteristics, we lose important dimensions of signification, equating all signs to linguistic signs, or omitting the characteristics of signs all together.

As a number, GDP portrays certain useful characteristics that go beyond its rhetoric purposes. GDP is significantly more mobile than most other forms of signs, be they material artifacts, or even words, which can be more easily reinterpreted in different contexts. The GDP number is light, but maintains its consistency and can thus easily travel across a variety of different settings, through computer systems and networks, without being corrupted (Latour Reference Latour1990). In addition, the number can be recombined with other numbers with the help of mathematical procedures, including multiplications, divisions, and additions, as well as statistical methods, such as regression analyses, to yield new interesting findings in ways that are precluded to linguistic signs (Robson Reference Robson1992; Hansen and Porter Reference Hansen2012). With the help of GDP numbers we can calculate an average GDP per capita, ranking countries according to economic performance while their diverging characteristics disappear from view behind the unified numerical notation (Speich Reference Speich2011).

The decimal character of the Arabic numerical system helps to perform these mathematical functions, which could not be performed so easily if numbers were written out in words, or even in the Roman numerical system, given their heterogeneous bases. ‘Computing the sum LXI (61) plus III (3) is manageable without writing, but the point is that the adjacent columnar ordering of these number-symbols is arbitrary because of the operation of inconsistent bases. The use of Arabic numerals assists calculation of complex sums by allowing columns of numbers to be ordered by their decimal relation. A sign for zero supplies a written symbol for what would otherwise be an empty space’ (Robson Reference Robson1992, 696).

Furthermore, GDP’s representation in quantitative form creates an air of objectivity. Despite the many imprecisions and social norms associated with measuring GDP, the overwhelming majority of policy makers and scholars treat the sign as if it provided a factual measure of the size of the economy. I would suggest that this is due to GDP’s numerical notation. Numbers reduce the ambiguity that is typical for qualitative statements, as quantification establishes an unequivocal identity (rather than merely a similarity) between the objects that are counted toward the aggregate indicator. By virtue of being aggregated those objects have to be identical in some regard (Robson Reference Robson1992). In the case of GDP, actually observed product prices, are aggregated with assumption-driven imputed data and government provided services, the latter of which are still measured according to the costs that go into their production rather than according to their (non-existent) market value. These differences in data accuracy and signification disappear behind a singular GDP number.

Additionally, all the doubts and assumptions associated with measuring GDP are swept away in their official publication – the material and iconic form in which they appear. Instead of publishing GDP numbers as a bandwidth, or adding comments about data reliability, GDP has been published as a singular number since its invention in the 1940s – despite the fact that GDP data have always been subject to revision and uncertainty. In one particular instance Keynes remarked ‘“you know as well as I do that the change in Commonwealth balances (or whatever it was) cannot have been zero last year, it must have been several hundred millions; what do you think it was?” “We really do not know”, was the reply, “but probably between three and four hundred million.” “Then put it down at £350 million” said Keynes “and try to get some accurate information in the future for by your own admission it is very important”’ (Stone, 21 June 1976b, in Suzuki Reference Suzuki2003, 497). The missing indexicality in the data were not noted in their final publication. Even today the World Bank does not publish the data it uses to calculate GDP. In response to Jerven’s query about GDP metadata, the World Bank noted that ‘raw data provided by the National Statistics Agencies are not available for external users and only a handful of people at the World Bank have access to it’ (Jerven Reference Jerven2013, 97). The absence of qualifying information concerning the reliability of the data increases the apparent objectivity of GDP as a measure of the economy.

GDP’s interpretants

Interpretants are the possibilities of interpretation that emerge from a sign. They are the effects a sign can have on the emotions, actions, and thoughts of human beings. According to Peirce’s tripartite division of object, sign, and interpretant, there can be no interpretant without a sign. This means that prior to the invention of GDP in the 1940s, GDP could not have influenced leaders’ policy decisions, such as balance of power considerations. Thus at the Congress of Vienna, the great powers established a balance of power on the continent purely on the basis of population numbers. Although they realized that wealthier inhabitants might be bigger assets than poorer inhabitants, they did not know how to systematically measure the income of the people (Webster Reference Webster1963). Yet, materialist scholars who focus on the relation between economic strength and great power status routinely make use of reconstructed historical GDP data (such as Maddison’s data set which contains GDP figures up to the Roman Empire) to back up their claims (Kennedy Reference Kennedy1987; Brooks and Wohlforth 2000–0Reference Brooks and Wohlforth1; Krasner Reference Krasner2000; Maddison Reference Maddison2007). The point is not merely that GDP might not be the right measure to evaluate economic strength in different historical and geographical contexts (Speich Reference Speich2011; Coyle Reference Coyle2014), but also that it is unclear how statesmen and diplomats could have acted upon a measure which did not yet exist. The fact that GDP is not a purely indexical sign, but contains important symbolic elements that are subject to change over time, makes matters worse.

Black (Reference Black2008) has argued that instead of focusing on GDP to identify great power status, scholars should rather rely on what people at the time thought was a great power. The present approach certainly agrees with this perspective, but it would add that whom statesmen considered a great power at a particular time depended upon the signs that portrayed great power status. Thus in the late 17th and the early 18th century, consumption was considered the ‘foundation of all wealth’ (Boisguilbert in Kendrick Reference Kendrick1970, 292), and Louis XIV’s lavishly decorated palace of Versailles rose concerns among other rulers, who partly for that reason formed a balancing alliance against the Sun King (Mukerji Reference Mukerji1997). From a Peircean perspective the existence of specific signs is a necessary condition for particular interpretants to emerge. Actions, thoughts, and emotions can only result from particular objects, if those objects are visible – that is, if they have a sign dimension. In the following, I highlight some of the interpretants that have resulted from GDP as a sign of the economy.

Emotional interpretants: Emotional interpretants are the emotions that result from GDP as a sign. It is a commonplace that reduced GDP growth signifies an economic downturn and evokes fears and anxieties, while high GDP growth arouses euphoria. Beyond a sense that individuals’ employment prospects and prosperity depend on the GDP indicator, GDP has become a sign of sovereignty for many developing countries, and as such it generates a sense of national pride. Like a seat in the United Nations, a national flag, or an anthem, the GDP figure is a representation of the state. For most countries high GDP and economic growth levels engender strong national self-esteem (Kendrick Reference Kendrick1970). As a result, those activities that enhance GDP are highly valued, whereas activities that are not included in GDP evoke indifference. Foremost among those are housework and leisure (Waring Reference Waltz1989; Stiglitz, Sen, and Fitoussi Reference Stiglitz, Sen and Fitoussi2009).

Energetic interpretants: One set of energetic interpretants that results from GDP is due to the fact that GDP has served as an input for policy decisions. Policy makers have used GDP figures to reduce the effects of recessions with the help of Keynesian demand management. When economic growth is low governments can increase public spending in order to increase aggregate demand. Or central banks can decrease interest rates to encourage investments. Next to influencing government actions, GDP also has effects on the behavior of a whole range of other actors. Private investors decide whether a sovereign debt is sustainable partly on the basis of debt/GDP ratios, and accordingly buy or sell government bonds. Additionally, GDP measures influence action in international organizations. For example, countries with low levels of GDP get access to comparatively inexpensive loans from the World Bank. Too high debt/GDP ratios can lead to emergency lending from the International Monetary Fund, complete with conditionality measures to ensure repayment.

Next to informing actions, GDP is also often the target of particular activities. Governments can decide to support certain undertakings rather than others in order to increase GDP. For example, they can encourage mining at the expense of a protection of the environment, because environmental degradation does not directly affect GDP, and is therefore less visible (Waring Reference Waltz1989). Similarly, as a result of enhanced productivity societies could produce the same amount of goods in less time and thus increase their leisure, but usually there tends to be a bias in favor of producing more goods instead, perhaps because more leisure time does not increase GDP (Anderson Reference Anderson1991). For these reasons Stiglitz, Sen, and Fitoussi have proposed a range of other indicators to supplement GDP in the hope that ‘the report and its implementation may have a significant impact on the way in which our societies looks (sic) at themselves and, therefore, on the way in which policies are designed, implemented and assessed’ (Reference Stiglitz, Sen and Fitoussi2009, 9).

Logical interpretants: Logical interpretants are the thoughts that emerge from specific signs. Econometricians use GDP figures in their analyses, generating a wide range of theories about business cycles and long-term economic development that inform economic thinking. With the introduction of GDP and other aggregate economic measures, economics gradually transformed into a discipline based on quantitative analyses (Suzuki Reference Suzuki2003); the apparently precise nature of GDP as a number might have provided a greater sense of scientificity than qualitative indicators. From its origins among the other social science disciplines, economics grew from the 1960s ever closer to statistics and acquired a highly quantitative focus (McCloskey Reference McCloskey1998; Fourcade Reference Fourcade2009). Quantitative measures like GDP extract the economic from its social, political, and natural surroundings when they aggregate the purely monetary dimension of particular items and provide a value for the size of the economy. In this sense figures like GDP might have contributed to the imagination of a separate economic sphere (Mitchell Reference Mitchell2002).

In the policy world GDP measures are not only ‘a basis of calculation for management’, but also bare a ‘significant impact upon government accountability, reporting and consensus on matters of the national economy’ (Suzuki Reference Suzuki2003, 502). Low economic growth is interpreted as government failure, high economic growth affirms success. Over time our societies have become used to interpret GDP not merely as a measure of economic market production, but rather as a measure of overall well-being. The phenomenon is not linked to the sign of GDP itself, but rather emerges from the social context in which it is used.