Debt and the Politics of Numbers: Hegemonic Numbers, Political Numbers, Ordinary Numbers

ABSTRACT

Numbers are both shaped by and constitutive of a certain vision of the world, and household debt is no exception. Financialized capitalism relies on hegemonic numbers that serve the economic and political interests of state government and the financial industry, which see and measure debt as a market ripe for development. In the face of this, it is crucial to build alternative numbers. The political numbers of debt conceive debt as a power relation, and quantify the degrees of financial exploitation that hegemonic numbers are blind to. Ordinary numbers seek to reflect what matters the most to ordinary people. They conceive of debt as a relationship of social interdependence, which can be a source of power, hierarchy and exploitation, but also of mutual aid, reciprocity and dignity. Far from functioning in silos, hegemonic numbers, political numbers and ordinary numbers have shifting boundaries. This article, based on twenty years of research in India conducted by a Franco-Indian team of economists and anthropologists, exposes and contributes to the politics of numbers in the field of debt.

KEYWORDS:

• Debt
• method
• anthropology
• politics of numbers
• India

JEL CODES:

• B54
• 018
• P00
• Z1

This article is referred to by:

Commentary on Guérin and Venkatasubramanian ‘Debt and the Politics of Numbers’

1. Introduction

Who decides what to count and how to count? Why do some numbers stand out and others not? Numbers are never neutral. They are both shaped by and constitutive of social and political forces and a certain vision of the world. This is what sociologists and political economists call the ‘politics of numbers’ (Alonso and Starr 1987; Desrosières 1993 [1998]; Mennicken and Espeland 2019). Worldview — both the world as it is and the desirable world — permeates the methods used and the choices made by those who produce and use numbers. These in turn often have retroactive and performative effects on the reality measured (Desrosières 2014). The sociology and political economy of quantification have highlighted the diverse ways in which some numbers can serve neoliberal or totalitarian policies, while other numbers can, on the contrary, uphold processes of emancipation and freedom (Porter 1996; Desrosières 2014). Criticisms of GDP and its inconsistencies are now well known, although it remains a central indicator for macroeconomic steering (Fitoussi, Stiglitz, and Sen 2010). Any indicator can be calculated in multiple ways, moreover, reflecting and reinforcing divergent or conflicting worldviews, and benefiting certain social categories at the expense of others. In 1996, the US Boskin report caused a scandal by revealing to the general public the methods used to calculate consumer price indices and the crucial impact on wages and pensions (Jany-Catrice 2020). Neoliberal policies depend on and promote benchmark indicators which seek to turn states, workers, citizens and non-profit organizations into calculative entities driven by performance and competition (Desrosières 2014). Similarly, and as we shall see here, financialization counts on and promotes financial indicators that transform financial speculation and market debt into normal, desirable entities.

One can regret and condemn the ‘tyranny’ of numbers (Eberstadt 1995) and the way they serve the interests of private capital or disciplinary states. One can also lament the ‘poor numbers’ that are biased, unreliable or even wrong, especially in the global south where the statistical apparatus and democratic accountability are weak (Jerven 2013). By contrast, following a long tradition of using numbers as a ‘weapon in the service of democracy’ (Desrosières 2014, p. 35), one could endorse ‘statactivism’ and adopt numbers as political tools for ‘struggle and as a means of emancipation' (Bruno, Didier, and Vitale 2014, p. 199). Numbers can contribute to the visibility of minority groups and counter ‘despotism and arbitrariness’ (Mennicken and Espeland 2019, p. 230). Numbers can serve as ‘weapon to analyse, to negotiate, and to limit the dominant players’ (Samuel 2014, p. 239). When it comes to labor and work, numbers have always been crucial for defending workers’ rights against private capital, while highlighting the many jobs and tasks that long remained invisible to official statistics, in informal and women’s labor (Benería, Berik, and Floro 2015).

Statactivism seeks to deconstruct categories that are now taken for granted, precisely because hegemonic numbers have given them life by forging specific images of the society in which they emerge. This is why ethnography is so valuable, as it enables us to capture the underlying and implicit meaning of the statistical categories used, and potentially their performative and retroactive effects. Ethno-accounting, which brings together ethnography and counting, is also crucial to understanding how people count, and counting what counts for them (Cottereau and Marzok 2012). The connections between anthropologists and statisticians have transformed over time (Desrosières 2013, chap. 8), shifting between strong complicity (in the colonial era), dialogue, mutual distrust and ignorance (the present norm). Yet as we shall see here, many current statistical innovations have stemmed from such collaborations.

The teamwork discussed here is a modest contribution to this work with a focus on household debt. With the increasing hold of debt on the everyday life and social reproduction of societies, exploring and contributing to the politics of debt numbers is urgent and necessary. This article is a reflexive analysis of our own empirical work over two decades at the Observatory of Rural Dynamics and Inequality in South India.¹ Founded and run by a Franco-Indian team of economists and anthropologists, the Observatory's primary mission is to produce different types of numbers related to rural socio-economic realities.

Numbers are symbols used to represent quantity (counting), magnitude (measuring), and position (labeling). Counting determines the number of elements in a finite set of objects, which allows to determine the size of a set. Counting also involves taking into consideration and giving value to something. Measuring implies that something already exists in a measurable form (in rupees, meters, kilos, seconds, etc.). Labeling is the process of designating, classifying and ordering. To quantify is to assign a numerical value or quantity to an entity or phenomenon so as to represent it quantitatively, whether through precise measurement or subjective evaluation. Throughout this paper, we use the term ‘number’ in its broadest sense, i.e., referring to methods for collecting, analyzing, interpreting and disseminating numeric data, and to numbers themselves.

Our reflexive work led us to pinpoint three types of numbers that can usefully shed light on the politics of debt numbers (and hopefully more broadly): ‘hegemonic’, ‘political’ and ‘ordinary’. In this paper we discuss the making of these three types. What sets them apart is first and foremost an underlying vision of the debt they reflect and produce, which expresses and shapes a specific vision of the world, both as it is and should be. Hegemonic numbers focus on financial inclusion and credit (a sum of money made available) while ignoring its other side, debt (a sum of money to be repaid); they consider credit as a desirable market ripe for development. Political numbers focus on the debt side, seen as an asymmetrical relationship to be challenged, and its consequences in terms of financial exploitation. Ordinary numbers focus on social interdependence, and approach debt as a source of social and ambivalent social ties consisting of both hierarchy and solidarity. In terms of methods, hegemonic numbers are the product of international expertise and quantitative surveys on financial inclusion as a performance indicator. Their legitimacy stems more from their communicative power than from their reliability and rigor. Political and ordinary numbers stem from our own efforts as ‘statactivists’ and draw on the iterative use of ethnography, questionnaire surveys and counting methods such as financial diaries. Our own numbers draw from a dozen villages in northeastern Tamil Nadu. They are not representative in the statistical sense: echoing statactivists’ efforts, their definition of representativeness is political, in the sense of representing a reality made invisible by hegemonic numbers (Bruno, Didier, and Vitale 2014).

Hegemonic numbers tell us that the financial inclusion of Indian households has doubled from 2011 to 2021, has tripled for women, and that the gender gap has disappeared. Meanwhile, political numbers tell us that women in our survey area are indebted at a rate of nine times their income, in contrast to three times for men, and that women bear a large part of household repayments; the interest servicing charge is 30 per cent on average (out of every 100 rupees earned, 30 are put toward repayment). Ordinary numbers tell us that Tamil villagers juggle eight loans on average, which does not necessarily mean over-indebtedness, but rather an ability to extend social relations. The share of Dalit (ex untouchables) indebtedness to non-Dalits has fallen over time, reflecting a weakening of intercaste dependency in the villages. Marriages cost four years of income, half of which is paid for by reciprocal gifts, reflecting strong social interdependency. None of these numbers are true or false, good or bad — although some, under certain circumstances, are more emancipatory than others. These three categories are standard ideals with an analytical purpose, and are therefore necessarily simplistic. The frontiers between numbers are flexible and shifting, and the making of each kind of number results from constant circulation between experts, statactivists and ordinary citizens. The fact remains, however, that building numbers that reflect the social relationships and values of a given society — what really matters to people — remains a huge ongoing challenge. We’ll come back to this in the conclusion.

2. Hegemonic Numbers: Debt and Credit as ‘Inclusive Markets’

The social and political fabric of numbers is contingent on the initial work of classification (selecting categories) and unit of analysis (e.g., individual, household, enterprise, village, nation-state). Measurement first involves commensuration, and then the choice of a unit of measurement. These different steps are far from straightforward. They involve assumptions, choices, negotiations and compromises. Moreover, numbers have a mighty performative power: they contribute to making the reality they are supposed to describe (Asad 1994; Desrosières 1993 [1998]; Espeland and Stevens 2008). Even if they do not necessarily correspond to a lived experience and have little or no concrete meaning, categories used for counting and measurement eventually take such shape. By creating categories of perception and an understanding of the world, numbers shape our perceptions and imaginations: realities that a priori did not resemble each other end up appearing identical; realities that are nevertheless fundamental end up disappearing; abstract realities become concrete.

Hegemonic numbers are not associated with a specific method, as they evolve across time and space. In the contemporary period, sociologists (and some statisticians) have widely criticized benchmarking indicators as neoliberal tools: public action is no longer driven by public accounting aggregates, employment, income and consumption surveys and consumer price indices, but by performance, comparison and incentive indicators (Bruno, Didier, and Vitale 2014; Desrosières 2014). In the global South, political economists have shown how structural adjustment plans have led to the decline of national statistical institutes and the growing use of international databases, often compiled by the International Monetary Fund and the World Bank. Despite being of dubious quality in many respects (statistical representativeness, questionable or missing data, etc.), they provide the comparisons that neoliberal policies need (Naudet 2000; Jerven 2013). Political economists have also stressed how the success of randomized trial methods has limited development policies to an extremely narrow field: the supply of private goods and incentives for behavioral change, while neglecting macroeconomic and structural policies (Bédécarrats, Guérin, and Roubaud 2020b).

To understand the making and legitimacy of financial capitalism, we need to take a close look at the fabric of financial indicators and their performative power. Here again, sociologists and political economists offer a useful critical perspective. Financial experts see stock valuation as a truth working for the efficiency of markets; yet in practice financial experts use and adapt this method according to their pre-existing vision of the world, which have above all the effect producing, maintaining and legitimizing global inequalities (Ortiz 2021). Ratios of public debt to GDP turn public spending into a loss rather than an investment, and highlight intergenerational conflicts at the expense of class conflicts (Lemoine 2016; Tinel 2016). Credit scoring techniques produce artificial classifications which in turn have substantial effects on life opportunities and social identities (Fourcade and Healy 2013). The addition of ‘financial intermediation services, indirectly measured’ (FISIM) to national accounts in 2002 transformed the cost of financial services into ‘productive’ and ‘wealth’ generating services, hiding the fact that they are potential sources of rent (Mazzucato 2018). The introduction of a wide variety of financial indicators into everyday language helped to legitimize financial solvency and profitability as acceptable and trivial requirements in the contemporary world. These for instance include the credit ratings of governments, companies or individuals, stock market indices and financial market volatility (Feher 2018).

‘Financial inclusion’ indicators are a further example. In the Global South, the rapid expansion of credit markets over the past two decades were accompanied by specific statistics that helped to legitimize a supposed demand. In India, controlling and regulating financial supply has a long history dating back to the British Empire. There are still continuities, such as the ongoing stigmatization of so-called informal (unregulated) finance, which is considered expensive and incompatible with personal freedom. But there was also a radical shift in the 2000s with the turn to credit expansion, at the expense of debt reduction.

Various types of numbers have influenced the history of financial regulation and its narratives. The first figures on debt originated from reports by British officials. A report on debt in Punjab by Malcom Darling, a member of the Indian Civil Service from 1904 to 1940, is probably the most extensive. It combines administrative data (farmers’ cooperatives), budget surveys, village monographs and detailed field observations. The report usefully describes and quantifies peasant debts and their variations according to local economies, production systems and social structures, all while lamenting the peasant mentality, driven by debt, ‘improvidence’ and ‘temptation’ rather than thrift, and considering cooperatives to be a salvation ‘to look ahead’ (Darling 1925, pp. 71; 152).² Darling’s report, and many others, supported anti usury regulation, the expansion of the cooperative model, followed by the banking model and the narrative of the villainous usurer, who was decried despite extraordinarily diverse informal lending (Gregory 1997). The obsession with regulation and control has always prevented the collection of accurate, reliable data on informal debt. Of course, from the perspective of banking law, informal lenders are a source of trouble. But for borrowers, formal credit is by no means better, whether in terms of price, borrower protection or risk sharing. From the pioneering analyses of anthropologist Chris Gregory to more contemporary observations, there have been many observed cases of unfair and predatory formal credit (seizure of assets, usurious costs, rigid repayments incompatible with irregular incomes). On the other hand, there are many examples of fair, negotiable and affordable informal credit (Shipton 2010; Guérin, Morvant-Roux, and Villarreal 2013). Both realities apply to India (Guérin, d’Espallier, and Venkatasubramanian 2013; Kar 2018).

Every ten years since 1972 (26th round of the national census), the National Sample Survey Organisation (NSSO) of India has conducted specific surveys on debt and assets (All India Debt and Investment Survey — AIDIS). Policymakers, political parties and activists have used this data for a variety of causes. These include tackling the over-indebtedness of the peasant middle class in certain regions and the successful demand for regular cancelation of peasant debts (Mohanty 2005). It includes the devastating effects of financial liberalization in the 1990s, and demands for a return to public intervention in banking provision (Shah, Rao, and Shankar 2007). More recently there has been the warning, albeit very measured, over the fall in net financial wealth, namely the debt-to-assets ratio, of the poor, particularly in urban areas (Agrawal et al. 2022).

The 2000s marked a turning point with the emergence of the microcredit narrative, followed by financial inclusion (Nair 2016; Karunakaran 2017). The stigmatization of informal finance continued, but the idea of a supposed demand for credit replaced the fear of over-indebtedness. Promoters of financial inclusion relied on microcredit organizations, banks and, increasingly, digital services to deliver financial services and products that were meant to be transparent, reliable and inexpensive. India has been very active in all these fields, supporting one of the largest microcredit markets in the world, while distributing welfare benefits through banks to boost the opening of bank accounts. It then resorted to a massive demonetization shock in 2016, to force unbanked people to bank and use digital money.

The effects of financial inclusion have been, and continue to be, the subject of much controversy. Numerous studies have shown that it can at best smooth expenditure and income, and at worst lead to over-indebtedness (Garikipati et al. 2017). Our own data from rural Tamil Nadu backs up the over-indebtedness claim, as we shall discuss further in the following section. Our focus here is not to set out this controversy in depth, but to show how hegemonic numbers helped to shape a financial industry that was blind to the other side of credit–debt, which relied on and in turn reinforced a vision of market credit as a necessary and desirable prospect.

Numerous data on financial inclusion are available, from administrative data, national public surveys and surveys by private firms. The most comprehensive and probably most widely used database is the World Bank Findex survey, which we shall as such examine here. Launched in 2011, the Findex survey was the first international survey to quantify access to financial services. It has been conducted four times since 2011 in 120–140 countries, among approximately 150,000 participants. It provides detailed data on borrowing, saving, transfers, insurance and, increasingly, digital finance, which is seen to be the future of financial inclusion. The survey’s authors present it as the ‘definitive source of data’ on financial practices (Demirgüç-Kunt et al. 2022, p. xv). The first round in 2011 concluded with an outstanding figure of 2.5 billion ‘unbanked’ (Demirgüç-Kunt and Klapper 2012, p. 2).³ In the 2021 round, the ‘unbanked’ were down to 1.4 billion, signaling ‘encouraging trends’, ‘progress’ but also ‘opportunities’ for expansion (Demirgüç-Kunt et al. 2022, pp. xi; 4). India was among the top performers: financial inclusion had more than doubled since 2011, from 35 to 78 per cent in 2021 (as opposed to 76 per cent worldwide), and for Indian women it had tripled, from 26 to 78 per cent (as opposed to 72 per cent worldwide) (Demirgüç-Kunt et al. 2022, pp. 16; 176).

Like all international databases, Findex draws its strength and legitimacy from its consistency, its global dimension and its origin — an international organization such as the World Bank (Naudet 2000). Findex data have the merit of disaggregating by gender (which is unique) and taking informality into account. In India, no nationwide statistics include both gender and informality.⁴ Like any multinational and multicultural survey however, its survey questionnaires are necessarily standardized and sometimes unsuited to the diversity of local realities (Pennell et al. 2017). The weakness of data on informal credit is a case in point, as only two possibilities are listed: saving clubs or family and friends. Yet the informal financial landscape is far more diversified, not least in India. Private lenders, pawnbrokers and local elites play a crucial role, as we shall further discuss. In fact, the inclusion of informal finance remains anecdotal, but that hasn’t deterred Findex experts from asserting that it ‘may be less safe, less reliable, and more expensive than formal methods’ (Demirgüç-Kunt et al. 2022, p. 9). Digital finance, by contrast, takes up a significant share of the questionnaire: 31 questions deal with digital issues, as opposed to 4 for borrowing. The questionnaire does not address crucial issues such as debt amounts, repayment costs and repayment constraints. Although over-indebtedness is now well documented on a global scale (albeit rarely quantified) (Guérin, Morvant-Roux, and Villarreal 2013; UNCTAD 2019), the 2022 Findex report virtually ignores it. It makes a quick allusion to it, but the questionnaire design prevents any quantification of the problem. In fact, the questionnaire is akin to a market survey, identifying existing access to financial services and diagnosing potential expansion.

Ultimately, Findex indicators help to create a void to be filled, a market to be created in order to wipe out an ‘exclusion’ deemed undesirable. Findex data translates into global maps showing different degrees of exclusion across continents and states, and consequently more or less promising market opportunities. Leading financial inclusion coalitions use Findex data to advocate for more active financial inclusion policies.⁵ The global financial inclusion industry, from investors, regulators, policy makers to credit providers, regularly use this data both to give moral meaning to their actions and to guide their targeting. Researchers also make extensive use of this data. According to the Google Scholar counting system (July 2023), the 2017 Findex survey has 2265 citations (1770 for the 2011 survey, 1480 for the 2014 survey). This data is online and accessible. The World Bank's power of communication builds trust in its numbers (Porter 1996).

3. Political Numbers: Debt as Financial Exploitation

When we started our research program in rural South-India in the 2010s, our qualitative exploratory observations revealed massive problems of over-indebtedness and financial exploitation. Decision-makers and official statistics were focusing on farmers’ debt, which had long been recognized as a public issue, but ignored casual workers and women. Rather than giving in to ‘statisticism’, i.e., the idea that statistics are necessarily false and manipulative (Espeland and Stevens 2008), we decided to follow the path of ‘statactivism’, which uses numbers’ powers of ‘denunciation and emancipation’ (Bruno, Didier, and Vitale 2014, p. 199). Our ambition was not to use statistics for ‘representativeness’, but for representation in the political sense: to make visible a reality disguised by official statistics and to consider quantification as a device to highlight an obscured reality, to give it coherence and to become its spokesperson.

In our case, with informal debt coexisting with formal debt, measuring debt is easier said than done, simply because the usual statistical categories are inadequate. This is not necessarily about manipulation or incompetence, but a reflection of one effect of the politics of numbers. When numbers ‘misrepresent society, they coordinate our misperceptions of it’ (Alonso and Starr 1987, p. 3). By definition, informal finance is unrecorded and thus difficult to capture. There is no administrative data to compare population surveys with administrative data. However, the main problem stems from the very meaning of the ‘debt’ and ‘credit’ as categories. Debt is an unstable and multiple reality, whose economic and statistical definition is just one aspect among many. We needed ethnography both to deconstruct categories and identify appropriate alternative categories and units of analysis. Our efforts simply revive a probably insufficient but long-standing tradition of collaboration between economists and anthropologists in informal and women’s work.

An article published in 1973 by the anthropologist Keith Hart was instrumental in the recognition of informal work (Hart 1973). Drawing on ethnography in Ghana, the anthropologist observed that labor statistics, by restricting labor to formal wage labor, eluded most income sources and ultimately failed to predict the economy’s productivity. Since then, the measurement of informal employment has made great progress. The ‘1-2-3’ survey is an example: it takes into account the very great diversity of working arrangements (noted by Hart) and the (partial) overlap between the ‘household’ and ‘enterprise’ units (Razafindrakoto, Roubaud, and Cling 2003; Razafindrakoto, Roubaud, and Torelli 2009). The statistician François Roubaud designed and tested this method in Mexico in the 1990s (Roubaud 2014). He first traveled around the entire country by collecting life stories from a wide range of worker profiles.

In the same vein, anthropology has actively contributed to improving statistics related to female and domestic work and wealth production (Benería and Sen 2021). These efforts have been fruitful, as most official statistical systems now include the measurement of informal and domestic work (Benería, Berik, and Floro 2015). This means that political numbers initially given voice by a minority can be transformed into hegemonic numbers, in the sense that they become part of official statistics and common sense. Such progress is the fruit of a long process of experimentation, trial and error, often involving interdisciplinary collaboration between field economists concerned with the adequacy of their tools for empirical realities, national statistical institutes, and anthropologists, or at least those open to the idea that quantification is a legitimate method for apprehending a facet of reality.

In our case, this is what we have done to highlight a reality obscured by hegemonic numbers. Any attempt to measure debt runs into various biases. Some debts are socially degrading, and people don't want to talk about them (social desirability bias). According to studies carried out in contexts as varied as the United States and South Africa, people underestimate their debt by around half (Karlan and Zinman 2008; Zinman 2009). People may forget some debts, especially when they hold three, five, or more debts (memory bias). What the statistician or field economist describes as a debt — a sum to be repaid — may be framed differently by villagers, for example as ‘financial aid’, or as ‘money taken from elsewhere’, even if it involves sums that must be repaid (categorization bias). Both men and women may tend, consciously or unconsciously, to underestimate women's debts, restricting them to small, negligible amounts earmarked solely for day-to-day expenses (gender bias). Ordinary counting practices are another challenge for the statistician. We'll look at them in more detail later. For our purpose, suffice is to say that in many circumstances, Tamil villagers rarely reason in continuous variable but rather approximately, contextually, qualitatively and comparatively (see also Appadurai 1989). For example, to the question ‘how much money did you take from this lender?’, some respondents may reply ‘Much too much’, or on the contrary ‘he gave me little’.

Ethnography was instrumental in identifying these biases and designing questionnaires that would limit them (eliminating them is unrealistic): using vernacular terms and avoiding pejorative ones, listing all expenses and asking how they were financed rather than asking respondents about their level of debt, interviewing men and women separately, conducting interviews in a conversational manner to put respondents at ease and inspire confidence; this meant sharing their social worlds and using common references related to local social, political and religious life; paying attention to their qualitative assessments of debt and then ask for the amounts (which people often know, it's just that the quality of a debt often matters more than its amount) (Guérin, Venkatasubramanian, and Kumar 2023, chap. 1). We implemented these questionnaires in 10–15 villages with 400–600 families and 1600–2000 individuals interviewed repeatedly in 2010, 2016–17 and 2020–21.⁶ Aware that memory and gender biases persist in questionnaire surveys, we also used financial diaries, a counting method designed to provide an exhaustive inventory of incoming and outgoing flows of a residential unit (Collins et al. 2009).

At the end of this long process of experimentation, improvisation and bricolage, combining intuition and trial and error, we believe we gradually arrived at a reasonable measure of debt, both at the household and individual level.⁷ This depicts a very different world from that of Findex. Virtually no families are ‘financially excluded’ thanks to a massive policy of microcredit provision and the expansion of banking. But indebtedness is massive and all the families are in debt. Formal debt is increasing over time in volume, but so is informal debt, which remains crucial for meeting urgent and unexpected needs, maintaining social ties (see next section), and also repaying formal loans. In 2020–21, informal debt still accounted for 47 per cent of total household debt. Note also the difference with official Indian data (AIDIS 2019 data for rural Tamil Nadu), thanks to our methodological precautions. While our asset estimates were similar, the percentage of indebted families was almost three times higher (99.4 per cent as opposed to 36.9 per cent) and the average amount of outstanding household debt was 2.8 times higher. There is no reason to think that indebtedness in the villages studied should be higher than elsewhere in the state.

Furthermore, financial exploitation, i.e., a substantial extraction of labor income through interest payments, is a rampant reality. Debt service, or the amount to be repaid compared to income, is huge and keeps rising. In 2010, on average, households devoted 44 per cent of their annual income to debt repayment. In 2020–21, this had risen to 68 per cent. The increase was most remarkable for Dalits and the poorest people, for whom it doubled. Another key indicator is the interest servicing charge, which is estimated by comparing the interest paid to income. On average, interest servicing charge accounts for about 30 per cent of income, with a median of 15 per cent, revealing a wide disparity of situations and extremely severe degrees of financial exploitation for the poorest people. Financially ‘inclusion’ status hardly explained such disparity. The cost of informal debt was no greater than formal debt. The real difference came from bank credits, which are cheap and regularly canceled during election periods. But they were reserved for a minority, who were mostly upper-class and upper-caste men.

Gendered data was equally instructive. Although most women are ‘financially included’, they bear a disproportionate burden of the debt. In 2016 women's income made up 22 per cent of household income on average, while their debt made up 37 per cent of household debt. Indebted women in paid employment had nine times more outstanding debt than annual income on average, as opposed to three times higher for men. In 2020–21, women’s debt share had risen to just over half (52 per cent), while the share of their income had barely increased (24 per cent). This discrepancy was greater among Dalits and the poorest households. The financial diaries method can shed further light on women’s massive share of responsibility for making repayments, which was something we missed in the questionnaire. Well beyond the category of ‘credit’, it turns out that the category of ‘repayment’ is crucial to capture gender inequalities related to debt. Hegemonic numbers have no interest in this.

4. Ordinary Numbers: Debt as Social Interdependence

Our own numbers reflect a gap in official statistical surveys. But like all numbers, they are not an objective representation of reality but represent a facet of it. Measuring debt is so challenging precisely because people have their own ways of valuing and calculating it. Economic anthropology has highlighted the diverse nature of counting modes and their associated valuation systems (Cottereau and Marzok 2012; Villarreal 2014), and the multiple ways in which people, including the innumerate, ‘inhabit’ numbers (Guyer et al. 2010, p. 37). With the invasion of numbers into everyday life, citizens have little choice but to take numbers into their ‘own heads and hands’ (Guyer et al. 2010, p. 36). Ordinary citizens are demonstrating fascinating cognitive capacities for calculation, such as converting prices and values when several currencies are in circulation (Neiburg 2016). Ordinary citizens may also use numbers as a means of resistance (Motta 2020). While we initially applied ethnography to better measure financial exploitation, it revealed modes of ordinary valuation that encouraged us to construct ‘ordinary numbers’ alongside our political ones.

We define ‘ordinary numbers’ both as ordinary methods of valuation and calculation, and as numbers that reflect these ordinary methods. As we discussed earlier, debt can be seen as a desirable market transaction (hegemonic numbers) or as an asymmetrical relationship of financial exploitation (political numbers). The way Tamil villagers count debt reveals another facet of it: villagers see debt first and foremost as a relationship of social interdependence, namely a wide range of social relations of mutual dependence. For protective reasons as well as those of identity and dignity, villagers depend on each other, are aware of this and do not seek to eliminate, but rather to negotiate it more effectively. Such a vision of debt does not exclude the market or acts of exploitation, but encompasses them.

We came to this conclusion using various ethno-accounting methods, which aimed to count how people count (Cottereau and Marzok 2012). This firstly meant gaining an understanding of how people value, count, measure and label through numbers. What counts for them, and what valuation criteria do they use? Which aspects of life are commensurable, and which are not? Once the researcher has understood the social practices of valuation and quantification, she can then try to transform this into numbers. An instructive example comes from anthropologist Jane Guyer, who used ethnography to recode national survey data and then interpreted the results (Guyer 2004, chap. 8). With the support of the renowned development economist Chris Udry, she worked on consumption data from a national survey in Ghana, the results of which defied the usual conclusions of economics and statistics: the richest people spent as much as a proportion of their income on consumption as the poor. As for debt, what economists and statisticians usually call ‘spending’ can have very different meanings. In the Ghana of the time, ordinary people tended to classify their spending according to the type of obligation it reflected. To translate this into numbers, Guyer classified spending into four categories of ‘concentric circles of social commitment’ (Guyer 2004, p. 141), and came out with two key results. Firstly, recoding produced much more intelligible statistical results; secondly, quantification broadened the ethnography by showing that the share of expenses on various sorts of social obligations was roughly the same according to social groups, confounding any assumption about the community's ability to share or redistribute.

A method tested by the sociologist and historian Alain Cottereau in collaboration with the anthropologist Mokhtar Mohatar Marzok is also inspiring (Cottereau and Marzok 2012). It looked similar to the financial diaries — tracing all the flows in and out of a residential unit — but with a few key distinctions: no a priori categorization, since the aim was to categorize according to how people themselves value the flows; the counting included in-kind flows and the time household members spent on each activity. Its striking results included the fact that the mother, classified by the census as a ‘housewife’, was in fact the most productive member of the family, thanks to her shopping tips, and that ‘fair’ remuneration for babysitting involved various potentially contradictory criteria, complicating the usual expert debates on childcare wages.

We didn't have the resources to carry out this type of extremely time-consuming survey, but we did bring ethnography into our financial diaries, using each transaction to capture the frames of reference that people use ‘on the spot’, ‘with real situations and real temporalities’ (Cottereau 2015, p. 97). We also adjusted our questionnaires in line with our ethnographic observations. Several salient results emerged through this combination of methods.

Our first step, i.e., querying how ordinary people count, revealed that debts have such multiple meanings as to be barely commensurable. The aggregate amount of their monetary debt has little meaning to many Tamil villagers. Of course, a debt of a few rupees does not have the same meaning or consequence as a 100,000 rupee debt (the equivalent of one year's average family income). But that amount is only one part of the debt experience. Debt is also, and is sometimes above all, a social relationship with a moral connotation. The social meaning of debt (its impact on the status of the debtor) and its moral meaning (whether debt is considered good or bad, fair or unfair) are shaped by many factors that are only partly connected to the economic features of the debt transaction. These factors have much more to do with the pre-existing power relationship with the creditor. This in turn depends on caste, gender, social class, and location, and its meaning hinges upon broader, constantly changing historical trajectories.

Being indebted to someone below oneself in the caste hierarchy is experienced as degrading and publicly avoided. To become indebted in one’s own neighborhood and village publicly exposes weakness, and leaves one vulnerable to rumor. Getting into debt to a prominent local man can lead to exploitation, but also provide protection and access to multiple services. For women, borrowing from a sister-in-law may create an unequal hierarchy between the two lineages. And for women, a key criteria for selecting a lender concerns the risk of sexual abuse, or just the suspicion of it. Caste relations are a notable feature of the social and moral experience of debt. Historically, high-caste domination resulted in Dalits becoming chronically, sometimes across generations, indebted to their high-caste masters. Nowadays, the ability to incur debt from multiple sources, especially for Dalits, can be a source of pride and recognition, whatever the cost or amount.

These are only a few examples of a wide range of valuation criteria and they depend as much on the profile of the individuals, their active social relations, and their aspirations. What is self-evident to the statistician or the economist, including a field economist — adding up amounts of debt that are assumed to be comparable — is not self-evident to villagers. Many cannot spontaneously state how much total debt they owe, simply because the debts are not comparable to one another, and are as such incommensurable.

To count or not to count also has a performative dimension. Counting includes both a cognitive act (apprehending a given reality with a metric) and a ritual act: the cognitive act of counting may translate into a written trace or an oral expression, may be shared publicly or kept secretly (Coquery, Menant, and Weber 2006). In various circumstances, if villagers don't count, or at least don't leave a written or verbal trace, it is to leave room for subsequent negotiation. The central feature of informal debts is their negotiability, which is crucial when coping with irregular and uncertain incomes. The very fact of not counting is a socially effective act, since it leaves the door open for negotiation.

As mentioned above, Tamil villagers are more likely to assess their debt qualitatively and contextually than numerically. Well beyond debt, villagers rarely use a metric to describe the quality of situations, objects or events. More than the price of a saree or a piece of jewelry, it is the place of purchase, more or less prestigious in terms of the quality of the weavers or the jewelers, that matters in estimating its value, as well as the type of event (a son's first birthday, a daughter's wedding, etc). By responding with qualitative valuations, people ultimately refuse numerical valuations, which they consider simplistic and distorting, and in the case of debt, incompatible with a possible negotiation with the lender.

Conversely, villagers accurately count ceremonial debts, including in written form. They are often vague about an amount owed to an informal private lender, but can state to the nearest rupee the amount they owe for ceremonial exchanges. By counting and keeping accurate account books, villagers show and reaffirm their willingness to respect their social obligations, and to track those of others. Ceremonial counting is neither new nor specific to Tamil Nadu or India. In Tamil Nadu it is an ancient practice, which was historically carried out by the few educated people in the village. Anthropologists (and some economists) have observed it in many contexts where ‘investment in people’ remains an essential form of protection and dignity (Guyer 2004, p. 9; Coquery, Menant, and Weber 2006). In other words, counting practices show that ceremonial repayments, related to social obligations, are those that matter most and from which no one cannot (and often don’t want to) escape. As economic anthropology has long shown, fulfilling one's obligations is both a matter of social pressure, personal interest — the group is a source of obligation but also protection — and dignity, as giving back allows one to hold one's rank (Mauss 1950 [1993]). Ceremonial exchanges involve considerable sums of money (on average four years of family annual income in our 2020–21 data), and the status of debtor or creditor in the cycle of reciprocal gifts is a key indicator of the financial health of families. It reflects both a position in the life cycle and a social status: good financial management means spending according to one's means and expected reciprocal gifts.

Ultimately, exploring the methods of valuing and counting debts reveals them to be chains of interdependence, of which market transactions are only a small part. Within this, villagers are both creditors and debtors, which is both shaped by and constitutive of identities, social positions, relations of power, hierarchy and exploitation, as well as mutual aid and reciprocity. Villagers do not seek to cut themselves off from these interdependency ties but to negotiate them more effectively and extract themselves from the costliest ones, financially but also and sometimes above all, socially.

Is it then possible to translate the values of social interdependence into numbers? Can we measure and quantify what matters most to Tamil villagers when it comes to debt? When we presented our results on the severity of financial exploitation, the villagers were not surprised. But this is only part of the story, they argued, as it obscures the other meanings of debt and many of the changes taking place in the local financial landscape. A few indicators may reflect what villagers value, and provide a more accurate representation of the debt experience.

In our 2020–21 survey, Dalits were indebted on average from eight different sources, reflecting their capacity to diversify their relations, which is something they strongly value (note that a classic economic view would rather use the number of loans as a proxy for over-indebtedness). Among informal loans, longitudinal analysis reveals that Dalits borrow less and less from non-Dalits (30 per cent in 2020–21, as opposed to 46 per cent in 2016–17, in volume; note that the share of debt owed by non-Dalits to Dalits remains negligible). This is a clear sign of the (relative) emancipation of Dalits on the village level, which they highly value. In our 2020–21 survey, the average wedding cost amounted to four years’ annual income, half of which was paid for by reciprocal gifts. This reflected the strength of reciprocal ties, which are also highly valued. Over the years, the percentage of families with a bank account has steadily risen, reaching almost 100 per cent in 2020–21, but the average (and median) amounts of savings remain low (500 INR, or two days’ average family income). Many villagers consider that money tied up in the bank is useless and deprives them of their wealth: what counts is that money circulates, and good management is more about circulating money quickly than storing it. By using financial diaries, the economist Jonathan Morduch and his colleagues have calculated a ‘cash flow intensity of income’ indicator that compares a family's inflows and outflows with its income (Collins et al. 2009, p. 32). This indicator reflects the velocity of money, which is particularly high for the poor, reflecting the intensity of their financial lives. In our case, it's women who show high monetary velocity.

Are ordinary numbers a better representation of reality than political numbers? Should we persist in adding up debts into monetary aggregates if the aggregate has no immediate meaning for the villagers? The ongoing challenge of measurement is to accept to value and measure different objects with a common metric. If the categories used for measurement are accepted and make sense, then the act of commensuration serves to establish differences in quantity between comparable things. Otherwise, commensuration ‘creates a relationship between objects that are not conventionally regarded as comparable’ (Espeland and Stevens 2008, p. 408). We have continued such measurements because the sum of the debts does represent a facet of the reality: it is the amount that people owe, regardless of the moral and social meaning of each debt. Despite their incommensurability, these debts will have to be repaid at some point, even if only temporarily, or non-reimbursement will have serious consequences. The repayment of these debts has a cost, which is a source of labor income extraction, and which it is therefore necessary to measure. The total amount tells us nothing about the lived experience of debt, since a small amount of debt can be much more degrading than a large one. Yet the aggregate amount tells us a lot about the person's financial situation and financial exploitation. The hegemonic numbers meanwhile demonstrate the expansion of formal services, but say absolutely nothing about the consequences for people's well-being, and social and power relations.

In other words, the plurality of numbers reflects a multifaceted reality and is a way to make the complexity of the world more intelligible and understandable. It is common for some people to have a bank account (making them ‘financially included’, which hegemonic numbers consider as progress) but they are also heavily exploited financially (according to the political numbers). Yet they may still maintain their dignity insofar as they juggle several debts (and therefore several relationships) and maintain positive ‘ceremonial balance sheet’ (ordinary numbers).

5. Conclusion

Not only are numbers social and political constructs, but they convey specific values and conceptions of a desirable world. The multiplicity of debt numbers is a clear illustration of this. Hegemonic numbers focus on financial inclusion and see debt and credit as a market potential to get rid of interpersonal relationships, which are seen as incompatible with personal freedom. Market credit, which is supposedly based on transparent relationships between equals, is the aspired-for destination. Political numbers focus on financial exploitation and view debt and credit as asymmetric power relations which include market relations. Ordinary numbers see debt as ambivalent bonds of social interdependencies.

Our Franco-Indian team of economists and socio-anthropologists was dissatisfied with hegemonic numbers and set out to construct and unveil other numbers. We have produced political numbers revealing high levels of financial exploitation unseen by hegemonic numbers, and aimed at denouncing the false promises of the financial inclusion promoters, be they from the financial industry or the state. Ethnography was crucial in adapting our questionnaires to debt and credit practices that are difficult to capture in conventional surveys. By using ethnography and ethno-accounting, we also gave voice to ordinary numbers, highlighting another facet of debt that consists of social interdependencies, and not just asymmetrical relations.

Ordinary numbers are decisive in a democratic perspective, to do justice to how ordinary people think and to their aspirations. Ordinary numbers are also crucial for broadening our horizons and imaginations, which have been narrowed by hegemonic thinking and numbers and are not always reflected in political numbers. We have sketched out a few examples of ordinary numbers here, but many others are possible, and could be a source of inspiration for other contexts. Money is only wealth when it circulates, Tamil villagers often say, and this could be translated into local money velocity indicators, probably very different from macro indicators of money velocity. Banks are useless because they immobilize money and send it outside the territory, Tamil villagers also say, and this could be translated into indicators of money leakage.

Our efforts modestly contribute to a long-running work to deconstruct dominant and misleading statistical categories, historically carried out by teams of statisticians and anthropologists, albeit unevenly across time and space. As we previously discussed for informal and domestic work, some political numbers have come to be hegemonic, and fully part of the dominant statistical methods, whether of national statistical institutes or of international organizations. The Findex surveys have the merit of including informal finance (albeit in a simplistic way that is far removed from local realities, at least in India). It is highly likely that they were inspired by the results of the financial diaries of the team of field economists led by Jonathan Murdoch and Stuart Rutherford (Collins et al. 2009). These financial diaries confirmed with numbers what economic anthropology had been observing for a long time.

What is contentious at one point in history can become conservative at another. Taking an interest in financial exclusion was contentious in the 1990s, at a time when credit organizations for the poor were reformist, using credit as part of a wide range of initiatives to boost local economies and value chains. The boundary between field economists and experts is blurred; the first statisticians of the colonial era were first and foremost field economists, as also observed in colonial Africa (Morgan 2011). Such fluidity still exists today. As far as ordinary numbers are concerned, there is an unfortunate tendency for economists to swipe up anthropological observations and integrate them into their own epistemological framework. The measurement of so-called community taxes is a good example: this indicator takes into account the redistributive efforts that weigh on people, but considers this only as an effort and an obligation, which is partly true but obscures another crucial aspect of obligations in terms of dignity and self-esteem, and what makes life worth living.⁸

Hegemonic, political and ordinary numbers circulate and interact. An urgent task is to take seriously the scale of household debt, its cost and its gendered effects. Not only do hegemonic numbers ignore debt (a sum to be repaid), but they have made credit a desirable horizon without taking into account the risks of over-indebtedness and financial exploitation that political numbers reveal. An ongoing and unresolved challenge, however, is to count, measure and quantify what really matters to people. What ordinary numbers show us in the first place is the incommensurability of debts and the relations that underpin them. The ordinary numbers we have proposed are but a pale reflection of the substance and flesh of social reality. In a world governed by numbers, it is crucial to offset the domination of hegemonic numbers and the incompleteness of political numbers with ordinary numbers, without however reducing the thickness and warmth of reality to the narrowness and coldness of numbers.

Quantifying realities hidden by hegemonic numbers runs the constant risk of transforming them into separate entities and commodities — this was the pitfall of the initially political numbers on domestic work and nature (Waring 1999, p. xix). Counting is about separating and individualizing. In highly hierarchical contexts, based on nobility or caste, recognizing people as individuals, regardless of their status, might be emancipatory. But how to count while taking into account interdependencies, both between humans, and between human and non-humans? This is certainly the most striking challenge for specialists in numbers. Statactivists believe that other numbers are possible for other worlds and other possible futures (Bruno, Didier, and Vitale 2014, pp. 213–214). It is in fact in this direction that ordinary numbers take us. The responsibility of economists and statisticians is certainly to take more seriously what the ordinary numbers are trying to tell us.

Acknowledgements

This paper is the result of the symposium organized by the Wenner-Gren Foundation and we are immensely grateful to the organizers of the symposium — Teresa Ghilarducci, Richard McGahey and Gustav Peebles — for their invitation and their constructive suggestions. The comments of all symposium participants were very helpful. This paper also owes much to the exchanges with and reviews of Florent Bédécarrats, Marion Fourcade, Deborah James, Nicolas Lainez, Jeanne Lazarus, Benjamin Lemoine, Susana Narotzky, Federico Neiburg, Horacio Ortiz, Fareen Parvez, Boris Samuel and Jing Wang. We would also like to thank the two anonymous reviewers for their constructive suggestions and the editor for his patience and meticulous proofreading.

Disclosure Statement

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

Notes

1 https://odriis.hypotheses.org/1.

2 Note that these comments are strangely reminiscent of the behaviorist morality of contemporary development economists, who see the absence of a vision of the future as an explanation for the high propensity of the poor to take on debt and their acceptability of high interest rates (Bédécarrats, Guérin, and Roubaud 2020a).

3 ‘Unbanked' are defined as opposed to ‘banked' individuals, defined as adult owners of an account with a regulated financial institution (bank, credit union, microfinance institution, post office, or mobile money service provider).

4 The All India Debt and Investment Survey focuses on households. The National Family Health Survey includes gender-disaggregated financial data, but focuses on bank accounts and microcredit. The Consumer Pyramids Household Survey, which is periodically conducted by an independent private organization, is the most exhaustive survey of consumer practices today, including finance. Here again, the informal sector is missing.

5 This includes CGAP (Consultative Group to Assist the Poor), a network of development organizations hosted by the World Bank that actively disseminates good practices in financial inclusion, Alliance for Financial Inclusion (owned and led by member central banks and financial regulatory institutions), UNSGSA (United Nations Secretary-General’s Special Advocate for Inclusive Finance for Development).

6 We stratified the sample by system ecotype (village) and by caste (household) and increased the sample size over time to avoid the aging of respondents. Full details are given on the Observatory website, along with data and a user guide (https://odriis.hypotheses.org/).

7 We summarize here the salient figures described in detail in other publications (Reboul, Guérin, and Nordman 2021; Guérin, Venkatasubramanian, and Kumar 2023).

8 There are, however, some successful examples, even within an institution like the World Bank, one of the main sources of hegemonic data on the Global South. This is the case, for example, of the economist Vijayendra Rao, who is constantly seeking to improve measurement and quantification by taking into account the robustness of social relations, whether in terms of festivals, domestic violence, or deliberations at local municipal assemblies (Rao 2002).