Local Adaptation of Work Practices: The Case of BancoEstado’s “CajaVecina” Correspondent Banking System

This article contributes to the discussion of everyday domestic finance technologies by looking at CajaVecina, a correspondent banking network coordinated by BancoEstado, a leading Chilean financial institution. Differences in perceptions between actual users and designers of ICT for development projects (ICT2D) emerged from structured interviews with executives of financial intermediaries, customers, and shopkeepers. The extent to which independent merchants operating CajaVecina’s bespoke terminals confront and solve the “design-actuality gap” questioned whether the CajaVecina system enabled neighborhood retail stores to act as a de facto bank branch. Empirical results suggested that was not the case. Instead of following strict contractual behavior, participants in the correspondent banking network addressed a design gap through social interaction and leveraging relationships with repeat customers. This behavior builds on information emanating from what they called “operating quotas.” Operating quotas enabled BancoEstado to diversify risk, document financial services habits, and forecast the performance of merchants (particularly small, independent retail shops). Merchants used trends in operating quotas to tailor services offered through the CajaVecina terminal while aiming to increase the loyalty of trusted customers. These results further the understanding of correspondent banking services aiming to increase financial inclusion by providing evidence of a previously unexplored aspect of these networks, where social dimensions take precedence over economic, financial, and technological aspects.


Introduction
The research documented in this paper explores a financial technology initiative in which non-financial providers work in tandem with a well-established financial institution while aiming to improve the cash cycle journey. The technological, geographical, and policy objectives of this system, which sought to increase financial inclusion in a South American country, positioned the research within the overall umbrella of contributions exploring information technology projects for development (ICT4D). Fieldwork took place in Chile-a country where, at the time of the research, banknotes, and coins remained an important medium to settle on-the-spot retail transactions. The technical system at the heart of this research consists of a rudimentary, easy-to-install package featuring a point-of-sale (POS) terminal with a keypad as its only accessory. As is the case in Diniz et al. (2012Diniz et al. ( , 2013Diniz et al. ( , 2014 and Jayo et al. (2012), the organizational form that is the subject of the research is known as correspondent banking, where a dedicated POS terminal enables local small businesspeople (namely independent grocery stores) to serve as a representative of a financial institution, handling simple transactions while increasing access to cash by low-income Chileans. But as will be evident below, confronting "the social problem required more than a transactional exchange would allow" (Diniz et al., 2013, p. 6).
Empirical results suggested a gap or difference in perceptions between design and actuality (Diniz et al., 2013(Diniz et al., , 2014. Participants solved this gap through the social system around which the Chilean correspondent banking network is built. This active and creative role of the user of the technology has also been termed "appropriation" (Eglash, 2004). There are thus alternative terms to describe uses of a technology that were not part of the original design by engineers. In the case documented in this article, this alternative use builds on information emanating from what participants called "operating quotas." A quota is the level of credit that the bank gives to its correspondents. The bank could alter such a quota, based on the longevity of the independent retailer within the network, the volume of business, and the grocer's credit behavior. The information emanating from the operating quotas was relevant in at least two ways. First, the flow of information from the operating quotas enabled the financial institution to diversify risk and learn about the financial services habits of merchants (particularly about economic agents who seldom leave a footprint within organized financial markets and institutions such as small, independent retail shops). A second relevant aspect was that information emanating from operating quotas allowed merchants to tailor services offered through the POS terminal to fit the behavior of trusted customers and thus transform a correspondent banking service into a new and dissimilar type of solution, one different from that of the established bank, and which leveraged relationships with repeat customers.
The possible emergence of a "new payment space" is very much at the center of the research (Maurer, 2008(Maurer, , 2012. Such new space is characterized by redefining the interactions of payment providers with users, and at the same time minimizing the involvement of or even excluding financial institutions from the transaction. Indeed, Bailey et al. (2018) and Diniz et al. (2014) offer evidence suggesting the emergence of a new payment space amongst correspondent banking in Brazil. Diniz et al. (2014, pp. 22-24) note how users of correspondent banking services expect that their demands for financial services are met by the correspondent in the same manner as a bank employee or bank branch, while envisioning correspondents as mere transaction processors. In other words, Diniz et al. (2014) document evidence in which users expect the neighborhood retail store to act as a de facto bank branch.
Research in this article departs from the conceptual framework of the previous fieldwork documenting the use of correspondent banking to deepen financial inclusion in Latin America, by following the economic anthropology of payment media (Maurer, 2008;Nelms et al., 2018;Swartz, 2020). This alternative conceptualization enables us to further explore how neighborhood grocers modify the social system to expand their role in a correspondent banking network. Specifically, we question assumptions by Maurer's (2008) "new payment systems" and by Diniz et al. (2014), by asking whether the neighborhood retail stores act as a de facto bank branch. In other words, by looking at the operations of Chilean grocers and studying their views regarding the correspondent bank system, we question whether the behavior of independent grocers was solely guided by financial and administrative guidelines and, therefore, whether the retailer was a mere transaction processor.
We find evidence of how correspondents modified the social system to expand their role. This was in line with empirical evidence from Brazil documented by Bailey et al. (2018), in which correspondents redefined the social space not as a result of purposeful design or financial incentives, but as an unintentional result of the system designers' structural conditions for the development of such behavior.
Research results thus question the extent to which small, independent grocery stores strictly behaved within the confines of the contractual relationship with the financial institution in the creation of a new payment space. Research results also highlighted how trust and customer loyalty take precedence over efficiency and effectiveness within a new payment space. Such redefinition shows an apparent disruption in the social and financial relationship between the bank and the correspondent. This disruption involves the grocery store that operates the system going beyond the terms of the agreement and the design of the system, by leveraging trust to distinguish and categorize users. Such redefinition of the relationship is part of the new payment space because in maintaining the loyalty of trusted customers, grocers eliminate the financial institution while classifying customers in monetary terms.
In summary, we expand previous empirical insights that show differences in perceptions between design and actuality of the correspondent system (Bailey et al., 2018;Diniz et al., 2013Diniz et al., , 2014. Research in this article offers a detailed analysis and documents how correspondents modified the social system to expand their role within a correspondent banking network based on ideas emerging from economic anthropology. We question whether the retail store acts solely as a transaction processor for the bank. In answering this question and documenting the calculation modes used by small independent grocers in their daily operations and lives, research results also contribute to the new sociology of domestic finance and payments (Ossandón, 2017).
The remainder of this paper proceeds as follows. The next segment offers further details on the motivation for the research and frames the discussion by detailing the conceptual underpinnings as well as the administrative workings of correspondent banking networks. The subsequent section offers the settings and data collection methods of the fieldwork. The fourth section presents the empirical evidence, while the fifth and last section provides a discussion and tentative conclusions.

Research Context and Motivation
The organizational form that is the subject of the research emerged to support international trade in early modern capitalism. Correspondent banking agreements envision a contract through which one financial intermediary provides services on behalf of another for a fee (charged to the originating bank, the customer, or both) and in which the service provider (correspondent) will typically keep sight deposits from the nominating bank to facilitate customer transactions. In the absence of electronic fund transfers or global financial institutions, correspondent banking relationships allowed many banks to grow their balance sheet substantially. For instance, London-based Midland Bank (today HSBC), was able to claim the accolade of "biggest bank in the world" in terms of assets in the 1940s and 1950s thanks to correspondent banking agreements with banks across the world (Holmes & Cooper, 1986). In subsequent decades, the global payment system based on cross-border, inter-bank relations had to deal with the growth of globalization, the liberalization of payments, disruptions to the international monetary system, a series of high-profile bank failures, and the advent of ICT-led innovation (Shenk, 2021). More recently, a study by the International Monetary Fund (Liu, 2017), established that correspondent banking continued to facilitate overseas interchange businesses, global exchange in commodities, and cross-border payments. Correspondent banking thus continues to play a significant although substantially less important role within financial markets in the 21st century.
Throughout its history, inter-bank relations at the heart of correspondent banking as an organizational form evolved in two important ways. First, during the 20th century in the USA, correspondent banking enabled small and middlesized banks to deliver business services bypassing regulatory constraints to the territorial expansion of their branch networks (e.g., Osterberg & Thomson 1999;Richardson, 2007). But as was the case of its international predecessor, correspondent banking in the USA form chiefly involved business-to-business relationships between financial institutions.
According to Diniz et al. (2012), a second organizational transformation took place in Brazil starting in 1973, with the biggest changes taking place between 1999 and 2003. This new organizational form built upon ICT-enabled points of service installed by banks at small retail shops and groceries, pharmacies, post offices, etc., to offer services otherwise provided at retail branches of financial institutions in lowincome areas (Contel, 2020;Diniz et al., 2012;Loureiro et al., 2011). 1 The aim of recruiting retailers as bank correspondents were primarily to help distribute federal government monies in remote locations otherwise underserved by traditional banking. Diniz et al. (2013) highlight the pivotal role of Henrique Constabile in building this system. These three organizational forms suggest how correspondent banking networks are a fundamental part of wholesale and retail financial markets, while their chief aim is to answer the dilemma of physical detachment between financial institutions and their customers. In this paper, we use the term "correspondent bank-retailer" to denote and differentiate the Brazilian innovation from constellations in which agreements take place exclusively between financial service organizations.
The novelty of "correspondent bank-retailer" networks dispersed through the Americas and the Caribbean thanks to the efforts of supra-national organizations, non-governmental organizations (NGO), national and local governments as well as financial technology (FinTech) entrepreneurs (Felt et al., 2017;Financial Stability Board, 2017;Huilca Pérez & Jiménez Jaramillo, 2016). The correspondent bank-retailer model in Latin America aimed to reduce distance as well as to address harsh and challenging geography, high-level criminality rates in urban centers, and insufficient financial and telecommunications structures. Nearly all of these programs sought to reduce financial exclusion and increase financial inclusion by allowing low-cost admission to essential financial services for individuals residing in areas where it was too expensive to manage retail bank offices and automated teller machines (ATM). Indeed, the possibilities of low-cost delivery and enhanced financial inclusion led BancoEstado, a state-owned but business-oriented deposit accepting financial institution, to deploy a correspondent bank-retailer network by reaching out to Chilean merchants, corner shops, and groceries while aiming to introduce a correspondent bank-retailer system in 2005 called CajaVecina. 2 Following Diniz et al. (2012Diniz et al. ( , 2013Diniz et al. ( , 2014, Jayo et al. (2012), and Maurer (2008) research documented in this paper provides an understanding of the reconfiguration that took place at the point of sale with the advent of the CajaVecina in Chilean urban spaces. This while we looked at aspects related to changes in the physical and social infrastructure of connections between point of sale (POS) terminals, bank computer infrastructure, and changes in cash flow and (potentially) the working capital of small, independent grocery stores that adopted CajaVecina. Empirical research documents how issues of security impinged on the everyday operation of the CajaVecina POS terminal by grocery stores, the regulatory and administrative infrastructure offered by both BancoEstado and the Chilean state for the operation, and conflict resolution within the CajaVecina network (Espinosa-Cristia & Alarcón, 2016).

The Bank and Its Correspondents
In a broad sense, it is not difficult to see the introduction of corresponding banking-retailing as fostering the processes of marketization (Callon, 2016) and financialization (Chiapello, 2015;Martin 2002). Marketization processes are defined in contested forms. On the one hand, strategies to reduce financial exclusion are part of the modernity of the market and maneuver that, in tandem with the process of increasing financial inclusion, destroy the "social fabric" of those that get bancarized. On the other hand, the introduction of a correspondent banking-retailing network could be a civilizing strategy for entering retail banking services. Callon (2016) explains that the cited marketization understandings apparently look quite different. However, both notions of marketization do not problematize the very notion of the market.
Callon calls for a detailed look at the market conceptualization. Markets, remarks the author, are complex and depend mainly on innovation strategies. Such is the case of BancoEstado effectively producing "market agencements" that are "the expression of [market] competition itself. . . where political and moral reflection is at the heart of markets and pushed out to their fringes" (Callon, 2016, p. 1). Such an idea of market agencement opens the possibility to see the relation between small grocers in a correspondent bankretailer network and the conventional bank as one where both parties transform once they establish this relationship.
Alongside issues and the nature of any potential conflict of interest and the responsibility of the established bank and its correspondent, there arises the level of confidence a customer of the nominating bank may have in an operation she performs with a correspondent. Trust is central to retail financial services, due in large part to the fiduciary nature of many relationships within the sector (Devlin et al., 2015). Indeed, an important self-reported barrier to financial services is the distrust of financial institutions (Allen et al., 2016), while trust has been identified as an integral element in fostering greater customer support, and engagement in bank operations (Devlin et al., 2015;Ennew & Sekhon, 2007). However, extant literature is unclear as to whether trust is transportable throughout a correspondent banking agreement, particularly when dealing with small-sized, independent grocers.
To address the gap in extant literature, we draw on the economic anthropology of payment media (Maurer, 2012(Maurer, , 2015Nelms et al., 2018;Swartz, 2020). In his research, Maurer (2008) argues that technological and regulatory innovations in the late 20th and early 21st centuries evolved to deliver a new payment space. This encompasses how electronically mediated retail payment systems helped to redefine the user experience, create new services, and reduce "frictions" for money to be exchanged, transferred, and used to settle accounts. New retail payment systems are of two types: those that directly rely upon established financial institutions and others that use information technology to minimize the involvement or even exclude financial institutions from the transaction. This novelty, however, often assumes that technology works appropriately and seldom questions the proclivity of users to embrace the new solution. It seldom allows for the interaction between user and technology to develop in ways other than those for which the technology was designed by engineers.
Research in this paper questions the assumptions around Maurer's "new payment systems," by asking: does the neighborhood retail store acts as a de facto bank branch? Considering this problem situates our research within contributions that document the sociology of domestic finance (Ossandón, 2017), specifically, the modes of calculation used by small independent grocers in their daily lives, and to a lesser extent, individual customers who interact through and around the Chilean correspondent bank-retailer network called CajaVecina.

The Sociology of Everyday Finance
Sociological research exploring the role of finance and technology in daily life analyzes the behavior and events that contextualize small value, repeated, market transactions often paid in cash. This approach is in sharp contrast with others that aim to establish the economic determinants of a market transaction. It also differs from that of critical sociology, which describes the role of retail financial services in everyday life as a form of systematic socioeconomic training that promotes capital accumulation by imposing forced financial thriftiness and frugality (Baudrillard, 2016). Wilkis (2017), for instance, is among those pointing beyond the "distant banking logic" implicit in the dominant view of critical sociology while highlighting the importance of considering social ties within commerce and retail finance. Along similar lines, Ossandón (2017) argues that an analysis of people's "daily" finances that considers social ties enables a better understanding of the links between financial inclusion and exclusion.
Daily finance-also known as the finances of everyday life-has an essential antecedent in the sociology of personal finance and the exchanges that economic actors carry out in commercial circuits (Zelizer, 2010). For this purpose, relational sociology provides the concept of "special monies" (Zelizer, 1994). The concept of "special money" aims to show communicative capabilities that are critically connected with communities and collectives. As Moor (2018) explains, this relates to the archaic relationship between the concepts: "communication" and "communion." Moor (2018) also considers that currency and prices send messages.
As forms of money and payment evolve, it looks likely that they will face many of the same questions that can be traced throughout monetary history,-that is, "the tension between state or corporate power on the one hand, the autonomy and privacy of ordinary people on the other, and the tension between money's power to generate collectivity and its power to divide and exclude." (Moor, 2018, p. 580).
Moor's analysis led to problematize any local or personal agency when we work with symbolic generalized means (Luhmann, 1998). Therefore, concepts like Zelizer's special monies need to be used carefully and we must not forget that money and payments evolve in a continuous tension expressed in its history. In our case, this tension is at the core of the CajaVecina operation, where there is a local agency and at the same time, the bank and the whole monetary system operating in tandem.
Zelizer's special monies concept is defined as the classification that users make when identifying, organizing, saving, and categorizing money in its daily use. In a similar vein, Science and Technology Studies, or STS (Felt et al., 2017), offers a line of inquiry looking at how people classify and maintain the order of the world around them. Classifications are generative (Bowker & Star, 2000), which is to say that they produce a reality, a reality that has to do with, for example, differences amongst the ordering and classification that operators of CajaVecina's POS terminals carry out in their shop. These classifications and orderings are relevant because small, independent shops operating the CajaVecina's POS terminal in their store must accommodate such transactions within minimal working capital and liquidity. As a corollary, classifications in commercial circuits and special monies are generative in the emergence of systems like CajaVecina. Zelizer (1994Zelizer ( , 2010 provides a framework to understand the complexity of the social links forged around a cash transaction. This is evident, for instance, when considering that the source and form of a transaction can result in a small, independent shopkeeper displaying distinctive preferences (and therefore an ordering) between selling $10 in groceries, mobile top-up airtime, or cigarettes. In its simplest form, the ordering of these potential transactions may be associated with different profitability. But many other factors may also come into play in determining the shopkeeper's preference and the ordering of transactions as different sales categories will apply according to whichever is the dominant ordering rationale.
However, the shopkeeper's scope of action may be limited when involved in a correspondent bank-retailer network as the contractual agreement will dictate that she acts as a sub-agent of the customer's bank, in this case, BancoEstado. Research in this paper thus aims to ascertain how shopkeepers display their preference for different sales categories, the ordering of these preferences, as well as the effect those choices have on the classification of users of cash and noncash transactions through a dedicated correspondent bankretailer POS terminal. Answering a different but related question, Cholez and Trompette (2016) claim that nearby networks serve as reductions of market uncertainty. Cholez and Trompette (2016) also claim that the more formal the economic exchange, the less kinship network necessary to give life to the economic circuit. Classifications can serve then as an uncertainty reduction device.
All in all, exploring the ordering and classifying of users of cash and non-cash transactions in a correspondent bankretailer POS terminal is also important from a techno-social perspective. Indeed, a retrospective analysis of Zelizer's (1994) "The Social Meaning of Money" places paymentand correspondent banking-technology as experiencing an important "moment" of change in today's world (Bandelj et al., 2016). Furthermore, new technologies are viewed as a central phenomenon that deserves analysis, since such technologies would entail facilitating transactions for users rather than blocking or differentiating them.
Technologies are not just an ensemble of nuts, bolts, chips, and programing, but also those solutions based on information technology that change the relations between actors that are part of a market agencement (Callon, 2016). Based on empirical evidence from Chile, Ossandón (2017) documents how under certain conditions commercial credit practices have generated "quota economies," economic circuits working under the same or very similar principles of solidarity as savings cooperatives. A dual process of collective learning can be observed in the quota economy, says Ossandón (2017): one relates to how individual consumers engage with credit infrastructures, while the other relates to the construction and development of trust and distrust across networks and the accumulation of social capital. In these calculability exercises, individual actors operationalize not only the common good, but also seek to maintain and even increase quotas.
Growth in quotas depends on results from systematic ontime repayment (i.e., no delinquency) as well as from benefits emanating from marketing and institutional propaganda devices (such as gifts, coupons, or redeemable points). Interestingly, Ossandón (2014) also provides theoretical support from his analysis in the consumer credit market, where the credit client is "sowed" and "grown" in a process where the credit issuer retailer learns about its customer and uses their new knowledge to manage their clients' quota amounts.
Considering the above, we contend that the quota depends on the systematic analysis of the correspondent bank behavior. Trust builds through repeated interactions between the established bank and the correspondent grocery store. Further, our working hypothesis entailed operationalizing the concept of trust as "ontological security" at the center of daily transactions around the services of CajaVecina. Giddens defines "ontological security" as the one that relates: ". . .and applies to all cultures, pre-modern and modern. . . ontological security is one form, but a very important form, of feelings of security in the wide sense. . .[Ontological security] refers to the confidence that most human beings have in the continuity of their self-identity and in the constancy of the surrounding social and material environments of action. A sense of the reliability of persons and things, so central to the notion of trust, is basic to feelings of ontological security; hence the two are psychologically closely related". (Giddens, 2013, p. 92). Ontological security is at the heart of both the growing quota process and the classification and ordering of monies by the correspondent grocery.

Settings and Methods
As noted above, the BancoEstado's mandate to increase financial inclusion was the original motivation to create a bank-retailer correspondent banking network in 2005. BancoEstado defined CajaVecina network as: "A service channel for customers and non-customers of BancoEstado operating in commercial premises during business hours. With store terminals, banking transactions and financial services performed using BancoEstado ATM cards or RUT Account debit cards." (Gobierno de Chile, 2013, p. 41).
As suggested by the statement above, the CajaVecina system envisaged the implementation of point-of-sale (POS) terminals at small retail grocers in urban and countryside locations to offer low-income Chileans access to retail financial services. This as the CajaVecina POS terminal communicates with BancoEstado's central computer system, which acts as an automated clearinghouse.
More than 30% of the overall inhabitants use CajaVecina POS terminals as they find an access point throughout each and every borough in Chile. The CajaVecina terminal offers the most essential services of a bank branch. When individuals go to their neighborhood grocery store, they can make withdrawals and deposits from BancoEstado's current account (the so-called RUT or DNI account 3 ), balance inquiries, pay bills, purchase airtime for mobile phones, or transfer money (between RUT accounts), just as they would if they were going to a regular retail branch or ATM of BancoEstado. Stake (2006) identifies three types of case study: intrinsic, instrumental, and collective. As is the case of correspondent bank-retailer networks explored in Diniz et al. (2012Diniz et al. ( , 2013Diniz et al. ( , 2014, our case is both intrinsic and instrumental, this as CajaVecina had a network of 31,000 correspondents or almost 70% of the 45,000 in operation in Chile at the end of 2019 (Diario Financiero, 2020). BancoEstado was the first to launch a correspondent bank network in 2005. During the following 15 years, CajaVecina continued to grow in number of POS and transaction volume.
Broadly speaking, the CajaVecina ecosystem can be seen as yet another step toward a "cashless economy" (Bátiz-Lazo, 2018;Bátiz-Lazo, Haigh, et al., 2014;Bátiz-Lazo, Karlsonn, et al., 2014;Bátiz-Lazo & Smith, 2016), a change that is often aligned to reducing the number of individuals otherwise excluded from the banking system (The World Bank, 2019a, 2019b). Through an inquiry into the characteristics of CajaVecina, we confront the operational complexity of the correspondent bank-retailer network as it mediates between financial and non-financial parties, a network developed and built by BancoEstado to facilitate the inclusion of large numbers of low-income Chileans into retail banking markets. Studying a network of grocery stores acting as correspondent bankers enables greater understanding of unbanked users, payment processes, and financial operations in general (Elyachar, 2010;Langley, 2014;Maurer, 2012Maurer, , 2015. Results thus shed light on how relationships build in the retail payment space, while considering the general way in which personal finances occur today when payment technology finds itself in a crucial "moment" of change in the world (Bandelj et al., 2016).
Following among others Maurer (2015) and Zelizer (2010Zelizer ( , 1994, fieldwork built upon the ethnomethodology approach (Garfinkel, 1967), since we examined social communications through an analysis of human activities, emphasizing the methods and strategies employed by individuals to construct, make sense, and give significance to their daily social practices. In this process, particular attention was paid to purposeful practices on which the operation at the point of sale of the CajaVecina network stabilizes itself.
We collected data mainly through semi-structured interviews (Brinkmann, 2014). Semi-structured interviews incorporated three elements during data-collection: (i) a script that accorded to the variables we wanted to explore; (ii) allowing objectives to remain close to the research question that had been proposed as well as to the dimensions covered in the theoretical framework; (iii) modifying the script according to the rhythm of the conversation, allowing the researcher/research subject to converge in an environment that may give rise to complications and/or dissonances.
The sample of informants from 24 retail stores was selected to mimic the geographical distribution of the CajaVecina network. For this reason, a unique requirement of adequacy was made (Lynch, 2005): (2) Business Profile (d) The sample considered only retail stores run and managed as small family businesses, where a single family member could perform various functions (e.g., purveyance, serving customers, cleaning, among others). (e) Stores had been established for no less than 1 year in the neighborhood. (f) Stores had no less than 1 year of experience in the running of the CajaVecina service. This criterion aimed to find retailers that were well acquainted with the diversity of the products offered by CajaVecina. (g) For most stores in our sample only one member of staff managed the CajaVecina terminal. The profile of these operators was chiefly middleaged females (9 men and 15 women, between the ages of 40 and 60 years old). 4 (3) Adjustments During Fieldwork (h) As the sampling progresses and the importance of "operating quotas" became evident, the sample considered different levels of quotas for banking transactions through CajaVecina. Note that retailers' "operating quotas" fluctuate according to two criteria: the length of time the store has been operating CajaVecina's terminal and the average number of banking transactions carried out through CajaVecina per month. As time and transactions increase, the total quota increases, either at the request of the stores or as an offer from BancoEstado. Thus, it was possible to observe retail stores that had quotas of between $400 and $1,400 US dollars.
Fieldwork took place between January and June of 2017. We followed a qualitative approach to sample the grocers. A total of 24 interviews with grocery stores were collected in two rounds-where questions in the second round of interviews were amended after a partial analysis of transcripts from the first round. About 12 informants were located in the metropolitan region of Santiago and 12 in Valparaíso. Three additional interviews with managers and mid-level administrators of the CajaVecina system at BancoEstado enabled us to "triangulate" responses from grocers. Our methodology sought concordances within different grocers' system operations. The basis for our discussion sought to find phenomena that were common within the interviews. Coding was emergent, rendering a global reading of the interviews. Following Moser and Korstjens (2018), who recommend that each analytical unit is immediately compared with the others as it emerges, data analysis was carried out by generating a combination of inductive category coding with a simultaneous comparison of all units. This latter comparison was done to factor the analysis and reach significant data reconstruction.
Looking at concordances in the use of the correspondent bank operations and more specifically, quota mechanisms, we inductively found similitudes within interactions between correspondent operators and bank customers. Such similarities allowed the team to find broader codes that could contain such repetitive interactions and explanations about the functioning of the system.
Taking care of ethical aspects, interaction with interviewees followed a strict protocol based on the Bioethics Committee of the Universidad de Chile. As a result, individual grocers and executives from CajaVecina signed consent forms before an explanatory interview. This form ensured that generally accepted ethical procedures were followed to guarantee that all human subjects choose to participate of their own free will and that they had were fully informed regarding the nature of the research project, any potential risks, as well as issues of confidentiality and anonymity of their responses.

Results: Agency in the Operation of the CajaVecina Correspondent Bank-Retailer Network
Empirical results suggested the possibility of a central mechanism to explain the operation of CajaVecina as a correspondent bank-retailer network. This mechanism was similar to that proposed by Ossandón (2014). BancoEstado implemented a quota-based mechanism from its very beginning. As an former senior executive told the research team: "Quotas it is the mechanism that BancoEstado uses to know its customers better, delivering low fees initially and increasing based on customer knowledge." (Interview with former senior executive from CajaVecina at BancoEstado).
Results based on observation and grocers' interviews suggest that the "sowing of credit," that is, the extent to which BancoEstado advances credit to merchants operating CajaVecina POS terminals, is performative. This is because grocers in CajaVecina's network made distinctions and ordering regarding their users in line with the amount of credit advanced by BancoEstado.
The following main trends within the interviews supported the possibility of that central mechanism: (A) The "operating quota," or credit advanced in the form of an overdraft facility by BancoEstado, was the most significant issue defining the interaction between bank and grocer. (B) The operating quota enabled the grocer to record transactions through the POS terminal. But at the same time, it became a "sowing" mechanism as it enabled BancoEstado to collect information on the grocer's financial and economic behavior. (C) The sowing mechanism was not limited to vertical learning by the operator of the correspondent bankretailer network (in this case BancoEstado). The sowing mechanism also enabled the grocer to learn about customer behavior as part of his/her economic and mercantile flows. This was possible because the grocer leveraged customer trust while establishing distinctions and mechanisms to reduce uncertainty around the daily operation of CajaVecina (including the management of the operating quota).
The trends above emerged from the systematic analysis of interview data where grocers recounted how initial contact with BancoEstado arose either from bank staff offering a correspondent banking agreement or from a direct request by the store owner to the bank. Regardless of the initial approach, staff at the bank's headquarters performed "due diligence" in the form of a credit check and commercial evaluation of the store and the store owner while assessing the potential default risks posed by its owners. A positive evaluation resulting in the owner being deemed as low default risk was followed by BancoEstado advancing a line of credit that became CajaVecina's "operating quota." Quota rather than credit or overdraft was the most popular term used during interviews with owners of grocery shops who operated within the correspondent bank-retailer network. Henceforward we use the terms operating quota, quota and overdraft interchangeably. We also use interchangeably customer, client, and user. And we use interchangeably the concepts of grocer, retailer, and shop owner.
As will be evident below, references to the quota had a central place in the store owners' discourse. Indeed, the quota could be described as a core mechanism for understanding the operation of the correspondent bank-retailer agreement. We will return to discuss the quota after shedding some light on the bank's perception of the operation of the correspondent bank-retailer agreement.
For bank staff, the correspondent bank-retailer network is primarily constituted as a technology that systematizes and dehumanizes exchange and then as a business model that facilitates low-value transactions while increasing their volume. The following statement illustrates the first distinction: "The first thing is the machine. . . The initial operating model considered that when you go to a business there will be people making transactions, and this is how the business operated since the time when cash emerged, it was like the predominant [ The manager's comment that the "system is a machine" suggests bank staff conceives the nature of CajaVecina as an abstract, dehumanized, robotic computer ensemble. This vision reduces the grocer to a human ATM or a button-pushing appendix of the POS device. The same viewpoint disregards the social interactions between grocer, customer, and bank while reducing the relevant element in the correspondent banking operation of CajaVecina to the collection of POS devices deployed across grocers' counters.
The deployment of the POS terminal does not seem to pose high implementation or adoption challenges for grocers beyond those related to reliable telephone connectivity. This was evident as bank staff reported there were no special measures to deploy CajaVecina's POS terminals in remote locations such as Easter Island or even Antarctica. They all achieved connectivity.
As has been documented elsewhere (Bátiz-Lazo, 2018;Bátiz-Lazo & Woldensenbet, 2006), it could then be said that BancoEstado implemented the CajaVecina business model using "tried and tested" ICT applications, that is, applications that were known, reliable, and robust. This approach rewarded incremental change as it offered a sound and reliable system. This is in sharp contrast to the uncertainty associated with reliability and other risks linked to potentially disruptive innovation.
CajaVecina managers at BancoEstado, however, explained that in their opinion the "business model" around which the network operated was more important than the technology. The design of the business model predated the technology and according to interviewees, it was based on a self-sustaining and self-managing network. Explaining CajaVecina's development since 2006, a senior executive at BancoEstado said: "Security transport vehicles no longer enter this system. Instead, the business runs on its own, it's a self-sustaining and selfmanaging network. So how did we achieve that? As suggested by the statement above, the business model of CajaVecina aims to guarantee the security and reliability of the operation, as a withdrawal by a customer will credit the grocer's overdraft and a deposit will debit that overdraft. The operational model assures the technological apparatus of the correspondent bank-retailer network delivers as expected. The fact that the device is a POS terminal is beside the point. The interface could be chipbased, web-based, satellite telephony, or any other. This would have no greater effect than the operational business model built by the bank with which the grocers engage in their daily routine. The credit line is what inaugurates the possibility of operating the commercial establishment as a sub-agent of the correspondent bank. Moreover, a large number of relatively small value overdrafts to grocers gives the bank a degree of operational security and diversification of financial risk.
According to bank staff, the analysis of a grocer's potential credit risk is no different from the way in which a line of credit is granted to any potential bank customer. This is clearly illustrated from a grocer's perspective in the following statement: "I thank 100% the person who installed my [CajaVecina or CV], because initially they didn't want to give me the CV because I'm [currently] married but was [previously] separated. Because my [previous] partner was an alcoholic and, in the end, he [didn't] want to show [them any records but] finally they gave it to me. That's why I say it's a thing. . . even me, they didn't want to give me a current account for the same reason, but they did in the end because I've shown things are very different. My partner is one thing, and my things are mine. But I haven't had problems with the CV." (Interview n° 12, Santiago Metropolitan Region) This grocer's statement illustrates the cumbersome process some retailers face. In the case above, she had to demonstrate her financial and operational independence before being able to secure the CajaVecina POS terminal. The quote further illustrates some of the tensions between the system and the stores operating the CajaVecina terminal. In this relationship, the device is expected to deliver high levels of security in the transactions for the bank but, as the quote suggests, this is not always the case as presumably others have "had problems with the CV." Meaning that, for reasons discussed below, the system is not foolproof as technology and the business model is imperfect. The following statement illustrates the relative importance of the concept of quota for the grocers: "The thing is that they give you a quota here, not all of the [grocers] have the same quota. Some have one million pesos [approximately US$1,500 dollars], others two million." (Interview n° 11,Valparaíso Region) This statement tells how the credit assessment places individual grocers on different footings because there will be more possibilities for business the greater the quota. Interestingly, the size of the overdraft (i.e., quota) can be increased by a combination of the age of the grocer's agreement with the pattern and volume of transactions that the store generates within the existing line of credit. So that regardless of the size of the shop, a larger overdraft will be granted the older the agreement with the bank and the greater the transaction volume.
The grocer, who considers the quota as a business opportunity, will need to avoid "having problems" with the "working capital" 5 that the bank has given her through the overdraft. Interviews suggested that the grocers realize that the line dividing the in-credit/in-debt threshold must be actively managed. The longevity of individual grocers in the correspondent bank-retail network depends on it.
Interviews also suggested that the grocers are aware there is a mechanism that gradually increases the quota. The overdraft will increase (up to an undisclosed limit) to the extent that the terminal is operational. Operating time is thus the chief element to increase the quota. Other fundamental variables to increase the overdraft are transaction volume and "good behavior" by the grocer. It is unclear what the latter exactly entails but it results from BancoEstado monitoring the grocer's performance.
It is worth noting that the behavior to increase a grocer's overdraft in CajaVecina is strikingly similar to the one developed in the credit system of department stores. They use it to "sow" credit and grow the number of customers as well as the number of transactions by individual customers (Ossandón, 2014).
As mentioned, managing the in-credit/in-debt overdraft threshold is critical for the grocer. Since a withdrawal by a customer will credit the grocer's overdraft and a deposit will debit that overdraft, grocers continuously recalculate their overdraft, in some instances, even after each transaction. This is the case when the grocer relies on the fees from CajaVecina as their main or major source of income. The quota or overdraft thus frames the experiences of those who engage every day with the processes of the correspondent bank-retailer relationship.
A closer look at the management of the overdraft suggests that its operation is related to the category of "managing transactions," for grocers create strategies to manage their float by deciding which customers and how the grocer will privilege. During interviews, it became evident that grocers work around CajaVecina while aiming to increase the loyalty of regular customers while discriminating against casual users. This can be seen in the following statement, where a grocer describes how she manages periods of low liquidity or little elbow room in the overdraft: "Of course, you need to learn how to manage your cash. If you know [that a] strong month's end is coming, and you have a 1 million quota [US$1,500 dollars overdraft]. And a little girl comes and pays 300,000 [pesos or US$450 dollars], and the gentleman from the liquor store comes and pays 200,000 [pesos or US$300 dollars]. You're out, you know? I don't make payments or major movements" (Interview n° 4, Metropolitan Region) As the statement suggests, this grocer refuses to accommodate requests for what she considers to be large value transactions. The relative value of these transactions is determined in comparison to the size of the available overdraft.
Discretionary management of transactions thus represents an active way of operating the correspondent bank-retailer agreement. Note that such social behavior disrupts the continuous delivery of information to the bank on the retailer's economic performance as the bank is unable to ascertain the value and number of "large" transactions that failed to take place. Nonetheless, the bank will estimate a series of metrics while aiming to get to know its correspondent bank-retailer network better. These metrics and learning that takes place around them are the basis to increase or maintain the size of the overdraft.
Social behavior that strictly speaking disrupts the correspondent bank-retailer agreement takes many other forms. For instance, another form of deviant behavior connects to safety measures. Stores with POS are frequently situated in elevated delinquency areas or countryside zones, that is, places where the bank remained very eager to boost financial inclusion and not often attended by bank offices or any automated teller machine. But the store proprietor and fellow citizens are perfectly mindful that servicing the CajaVecina system might increase cash placed at the store, hence expanding the incentives to a possible theft. In response, the store proprietor embraces tactics that discourages casual transactions and recompense frequent clients or, alternative, help to deal with periods of high demand and customer "foot traffic." Other forms of this deviant behavior result from grocers becoming true quota-management experts to secure steady access to their overdraft facility. Their actions include moving resources between the POS deposit box and the grocer's main cash register, calling on the grocer's credit and/or debit accounts (as opposed to the business' account), or travel to the nearest BancoEstado branch to withdraw and/or deposit cash. These actions show how the management of CajaVecina requires the POS operator to be active and attentive to the quota available throughout the day. The following statement by a grocer provides further evidence: "Simple, it's like a wheel. If you haven't paid, the wheel won't turn, and you can't withdraw. Active management of the overdraft facility goes far beyond allowing the quota to enable or close-off possibilities of servicing casual users and recurrent clients. The grocer's statement above also includes evidence of "marking" (Zelizer, 1994). This process happens when the interviewee states that the "CajaVecina was created for our neighbors, for those 5, 10, 15 buck transactions." This is interesting because it suggests a limit for individual transactions, limits that were not included in the bank's guidelines. Active management thus also depends on the grocer's "understanding" of what is and what is not an appropriate transaction.
The statement above also helps to at least partially answer the complex question proposed by Maurer (2015) regarding who owns the "payment space." It should be clear that the bank's sowing mechanism allows a degree of control to minimize the bank's operational risk. Yet the reported management of the overdraft facility described some transactions and behaviors that take place based on individual criteria by the grocers that, in turn, question whether the financial intermediary is the actual "owner" of the payment space. The said marking and classifications of what is or not a valid transaction for the correspondent operation at the grocery dispute the ownership of the new payment space that CajaVecina creates. More importantly, the CajaVecina operator has plenty of "space" to accommodate their operations and select who is the beneficiary of the transaction based on their own interests and their idea of what CajaVecina is created for.
Active management of the overdraft by the grocer also relies on trust mechanisms and the potential classification of customers. This results in a space of financial operations that can be unique to the grocer and distinct from the one expected by the bank. Such a new retail payment system directly relies upon established financial institutions.
However, as the evidence from the interviews suggests, grocers can minimize the involvement or even exclude financial institutions from complete control of the transaction. The financial institution is not fully disintermediated because BancoEstado retains control of the information flux and limits the volume of potential transactions by changing the level of operational quota. Nevertheless, the new payment space allows for the interaction between user and technology to develop in ways other than those for which the technology was designed by engineers. This is possible as the credit sowing mechanism on top of the overdraft is not just a form of vertical learning-that is to say, by the bank that owns the POS terminal and manages the correspondent bank-retailer network. As a result of active management of the overdraft by grocers, the sowing mechanism opens the possibility for the grocer to leverage trust with individual clients and as a result establish distinctions and mechanisms to reduce uncertainty for the daily operation of CajaVecina.
The following statement from a grocer suggests how there is a certain leeway in the operation of the POS terminal. This leeway enabled distinctions to be that which relate to the level of trust that the grocer or operator of the POS terminal share with the casual user or recurrent customer: "Many come with [cards other than BancoEstado], but in general people that do come here with other cards are the moms, the children. I already know who these people are, and if another person comes from elsewhere, I have no idea if it's the same person as the one who has the card or not" (Interview n°5, Valparaiso Region).
The statement above illustrates how the level of trust between the POS operator and the user is active in the administration of the grocer's overdraft facility. The statement suggests that the grocer is ready to engage in a transaction regardless of the payment card when she has confidence in the identity of the user but will be reluctant to engage in a similar transaction or be flexible with a casual user. The grocer has thus made a classification of POS users based on trust rather than technical, legal, or administrative criteria.
The level of trust the customer has in the grocer also redefines the payment space. For instance, reported behavior included instances where a repeat customer would leave behind her account details and even her Personal Identification Number (PIN) so that the grocer could finalize a transaction at a more convenient time for the grocer or whenever the management of the overdraft facility allowed it. The following statement captures this behavior: "Yes, they leave me their account numbers and bank cards here and I pay them to the extent that my quota increases. A lot of people do it that way." (Interview n° 23, Valparaíso Region).
Note how in the statement above a casual client who has no prior knowledge of a grocer is highly unlikely to leave her details or electronic password. Instead, she will remain in the shop until the transaction is recorded or continue her journey if unable to carry out the transaction. Indeed, the grocer's agreement with the bank is that the transaction should always be completed with "client present" and the use of the correct PIN will be taken to certify that. Yet it is the trust between the repeat customer and the grocer that bends the rules enabling the grocer to take ownership of the payment space. That is how this payment space is reformulated, and its ownership is disputed by the correspondent and the bank.

Discussion and Conclusions
The research documented in this paper delved into the daily practices of operating a correspondent bank-retailer network. To this end, we examined research on special monies, social studies on finance, as well as some research from science and technology studies. Empirical evidence suggested how CajaVecina builds upon a system that mediates and participates in the distribution of trust between users and grocers. However, this element cannot be completely understood by merely considering grocer-client interactions. Research results expand our understanding of the difference in perceptions between design and actuality in correspondent bankretailer networks as documented among others in Diniz et al. (2013Diniz et al. ( , 2014. Research results suggest how grocers modify the social system by pivoting their role as correspondents. The discussion is framed by literature on the economic anthropology of payment media (Maurer, 2008;Nelms et al., 2018;Swartz, 2020). This conceptual framework allows us to understand CajaVecina as a technological and regulatory innovation of the late 20th and early 21st century that created a new payment space for their users and store operators, raising the question: does the neighborhood retail store act as a de facto bank branch?
The article advances and documents a detailed analysis of how correspondents modify the social system to expand the payment space. Fieldwork in this article shows that small independent grocery stores behave within the confines of the calculation modes used in their daily operations and within the contractual relationship with the financial institution. However, and as expected by the domestic finance and payments literature, the grocer operating the system goes beyond the terms of the agreement and the bank-designed technology by leveraging trust to distinguish and categorize users. This insight resonates strongly with some of the findings in the so-called "new sociology of domestic finance" (Ossandón, 2017). Empirical results thus suggest that loyalty takes precedence over efficiency and effectiveness within a new payment space. Such redefinition shows an apparent disruption in the social and financial relationship between the bank and the correspondent.
Empirical evidence on the behavior of banks and grocers also corroborated previous research on the "sowing of credit" by Ossandón (2017). This finding emerged as the quota is granted without a strict or widely available credit scoring system but an individual assessment of the grocer like that of opening a bank account. We argue that this mechanism is very similar to the one developed in the credit system of department stores, which "sow" and grow their clients, a mechanism that has been documented by Ossandón (2017). This author explains that the "sowing" generated by credit systems allows individuals to create performative events on that sowing, to adapt to the rules established by the systems, be this credit in the case of credit cards and vulnerable populations, or banking in the case of the CajaVecina's operations.
The credit risk assessment which BancoEstado performs on its clients is key to the notion of CajaVecina as a selfsustaining and self-managing network. This assessment opens the network to an individual grocer while BancoEstado diversifies some of the risks in its lending portfolio and uses information collected from the POS to learn about grocers' banking behavior.
Meanwhile, the shop owner engages in the management of the overdraft which, in turn, justifies the development of techniques on the part of grocers to ensure the continuous functioning of the CajaVecina POS terminal. However, note that it is information and trust in the customer that allows a shop owner to transform the correspondent bank-retailer agreement into a different kind of solution and to appropriate the payment space. This solution is independent of the bank while the shop owner takes benefit of the concentrations of confidence cultivated with her loyal clients.
Grocers realize, not always very clearly, that the overdraft should be managed and that the long-term success of the CajaVecina in her shop depends on it. They are also aware that there is a mechanism that could gradually increase the size of the overdraft facility and that being active, and the duration of the agreement, will positively influence that process.
As a result, the paper contributes to advancing a more complex view of the correspondent bank-retailer model, suggesting that store owners appropriate their correspondent work, helping the system to succeed even when they are navigating their own interests and needs within the correspondent banking system. Such appropriation is based on their use of the overdraft mechanism (quota) and how individual grocers "play" the system. In this manner, research in this article builds on the contribution by Bailey et al. (2018) and more broadly from extant literature on correspondent banking systems (e.g., Diniz et al., 2013Diniz et al., , 2014. Expanding these previous findings, research in this article focuses attention on the socio-economic implications of the quota mechanism as the basis for the reformulation of the payment space.
More broadly, CajaVecina case analysis provides some insight into the integration of "daily finances" and "high finances," that is, the integration between low value, high volume transactions and large value, low volume transactions in capital and investment markets. This as the quota enables us to draw a direct line from a small value cash transaction in a grocery shop to the diversification of financial risk in the portfolio of a large financial institution such as BancoEstado.
The case of CajaVecina also shows the mechanisms with which the on-site operation of the correspondent bankretailer agreement is organized and configured into a specific payment space. In this sense, Wilkis (2017) correctly argues that observing those spaces allows us to better understand the convergence between those everyday spaces that are interconnected via mechanisms like correspondent banks with the bank's operation in capital markets.
Leveraging trust to distinguish between potential users suggests a reformulation of the payment space as defined by Maurer (2008Maurer ( , 2015. This reformulation has its conceptual support in Zelizer's (2010) ideas of special monies, that is, the fact that grocers distinguish clients by classifying them in monetary terms. As noted by Bowker and Star (2000) imposing a classification, in this case by discriminating users' access to the POS terminal, allows the grocer to impose aspects related to the general operation of her business over those detailed by the correspondent bank-retailer agreement, this as the grocer is willing to bend administrative rules when dealing with repeat users but not when dealing with casual users. The grocer is protecting her relationship with repeat users as they purchase regularly from her overall business and not only transactions related to CajaVecina.
The main point here is that the classification made from special monies and the management of the overdraft facility relates to the level of trust between grocer and user. Paraphrasing Giddens (2013), CajaVecina is a mechanism that permits the strengthening of both the client's and grocer's identity, both of whom are connected, in a consistent way, with their surrounding social environment and material of actions. This grants them reliability and the power to operate in the world in such a way that trust is at the center of their actions.
This ontological security allows the POS terminal to be operated in an environment where many of its transactions go beyond the temporal here and now (the present). In other words, they surpass the moment in time when grocer, client, and POS terminal exist in the same location. For instance, it is based on trust whether or not the grocer will accept a transaction when her overdraft is almost full. Turning down a casual user has no additional consequences when a repeat customer might be offended and cease purchases of other items at the grocery store (on the same or future date). This is one of the reported mechanisms used by grocers to manage risk based on the trust history she has generated with repeat customers. The grocer is in effect using the services from CajaVecina to increase customer loyalty.
Furthermore, the CajaVecina case permits the study of those tensions that appear when authors adopt concepts like special monies (Zelizer, 1994) and money as a symbolic generalized means (Luhmann, 1998). These conceptualizations are part of the continuum, where local agency and the whole monetary systems are operating together in the day-by-day payment and financial realm. These conceptualizations also help to highlight the importance of social issues that live side by side considerations of effectiveness and efficiency that dominate other studies.
Finally, the case of CajaVecina suggests that one of the tasks in researching domestic finance and payments entails searching for new payment spaces, how these are created and appropriated by different actions, as well as scenarios where they differ across national payment systems. This is important to better understand the extent to which retail payment systems are regulated by institutional and legal considerations and the extent to which the regulation of these systems is maintained through social practices (i.e., self-regulation). Practices that are considered unmonetized, such as trust or retribution, or by others perhaps emerging from innovations created by those who use and operate those systems.
Banco Central do Brasil (ND), Understanding the National Monetary Council, available at http://www+.3.bcb.gov.br/ pre/cmn/entenda_o_cmn_en.asp (accessed 11-Jul-2019). 2. BancoEstado, Caja Vecina, el banco más cerca de todos, available at http://bech-version3-dev.win4.kiteknology.com/sala-de-prensa/ noticias-bancoestado/noticias2012/2015/09/04/cajavecina-elbanco-m%C3%A1s-cerca-de-todos (accessed 11-Jul-2019) 3. More details on the RUT account can be found at the following link: https://www.bancoestado.cl/imagenes/_personas/productos/cuentas/cuenta-rut-home.asp 4. It seems BancoEstado offers training to only one person per retail store. The lack of training sometimes created service issues. For instance, services through the CajaVecina system could only be offered if the trained person was within the store premises. The operator was also the only person able to solved problems with clients, helping them with their transactions (financial socialization), and was also in charge of initiating conversations with the BancoEstado representatives regarding any issue or problem dealing with the daily operation of the system. The operator also is the only one who receives news and new training opportunities to deal with CajaVecina's new services and features. 5. Here we use working capital loosely as per the statement above rather than a strict definition of the difference between a grocer's current assets and current liabilities.