Relational accounting: A cultural approach

This article extends (but goes beyond) Zelier’s original concept of relational earmarking, where individuals engaged in dynamic forms of relational work categorize their spending and develop practices of resource allocation by virtue of the relationships being managed. This article pushes for a performative understanding of accounting practices by elaborating a theory of relational accounting based on the outlines proposed by Zelizer. Relational accounting begins (i) upstream where identifiable codes and structured meaning systems shape the set of options, non-options and their sense of being possible for the person engaged in action. Moral considerations embedded in these codes affect the geometric shape of the individual’s decision tree, nearly collapsing some decision branches. (ii) In mid-stream there are meaningful, ritually prescribed occasions altering accounting priorities. And (iii) downstream people are engaged in relational work where the meanings of their relationships function performatively. Deals make more or less sense by virtue of the types of relationships being managed and the meanings of the relational sites where monetary evaluations take place. Identifiable third parties sanction these relational performances. The article concludes with concrete examples of relational accounting for luxury transactions and for high cost debt.


Introduction
Since its inception in the late 1980s, Viviana Zelizer's approach to relational earmarking has been lauded and threatened across the social sciences (Bandelj, 2012;Belk and Wallendorf, 1990;Pahl, 1995;Carruthers and Espeland, 1998;Ingham, 2001;Keister, 2002;Dodd, 2005;Maurer, 2006;Steiner, 2009). In Zelizer's 1989 article in The American Journal of Sociology, she used the term 'social accounting' as the sociological counterpart to mental accounting, but it was not until her 2012 article in Politics and Society that we find a single mention of relational accounting. By then Zelizer had shown that people experience money relationally rather than mathematically as they engaged in relational work to earmark their household budgets and to evaluate financial decisions. (The earmarking came to be known as relational earmarking in order to specify that relationships rather than mental operations lay at its core.) This article builds on relational earmarking by specifying its close kin, relational accounting. I define relational accounting as the set of cultural and social processes used by individuals and households to organize, evaluate, justify and keep track of financial activities. Whereas, behavioral economists like Thaler (1999, p. 184) define mental accounting as 'a set of cognitive operations used by individuals and households to organize, evaluate, and keep track of financial activities', Zelizer (2012, p. 161) counters that the norms and practices of individuals should be examined as constituted by their relationships and the overarching cultural meanings ascribed to those relationships. Because her concept of relational earmarking does not present relational matching as a social performance but leaves it as a dynamic, interpersonal, situated event, I am developing the concept of relational accounting to make those relationships function more dramaturgically.
While dramaturgy may be absent in economic theory, norms and relationships are not. Referring to Akerlof's (2007) presidential address at the American Economic Association, Zelizer indicates that the attention Akerlof pays to norms provides the possibility for improved dialog between sociological and economic approaches to accounting. For economists to run their models without a hitch, they need to assume that any dollar that comes into a household budget is equivalent to any other dollar that arrives or that is already there (the fungibility assumption), yet Zelizer has demonstrated how cultural and social factors preclude or modify fungibility. Work by other sociologists, social and cognitive psychologists, and behavioral economists confirm her theoretical premise (Winnett and Lewis, 1995;Thaler, 1999;McGraw et al, 2003;Camerer et al, 2011;Soman and Cheema, 2011;Jolls, 2013), but too often the non-fungibility premise and the social relationships upholding it were presented as special patterns of behavior; moreover, it seemed that cultural codes and related moral concerns remained difficult for economists and psychologists to specify empirically beyond gross detection (for example, a dummy variable or the manipulation of a single topic in an experiment).
Let's take, for example, the experiments that cognitive psychologists have used to demonstrate that people take into account the context of their transaction when trying to decide which deal is the best one. These experiments had people choose between two tightly scripted scenarios. For example: Scenario 1: Imagine that you are about to purchase a jacket for $125 and a calculator for $15. The calculator salesman informs you that the calculator you wish to buy is on sale for $10 at the other branch of the store, located 20minutes drive away. Would you make a trip to the other store?
Scenario 2: Imagine that you are about to purchase a calculator for $125 and a jacket for $15. The calculator salesman informs you that the calculator you wish to buy is on sale for $120 at the other branch of the store, located 20minutes drive away. Would you make a trip to the other store? (Kahneman and Tversky, 1984, p. 347) Daniel Kahneman and Amos Tversky (1984, p. 347) described these scenarios as representing a minimal, topical, or comprehensive account. If the example were one of a minimal account, we would simply see that the decision to drive to the other branch of the store for a cost savings would represent a gain of US $5. This means that regardless of the context, an individual will evaluate a gain of $5 dollars as being the same in both scenarios (a minimal account). In reality, however, 68 per cent of the people presented with this scenario were willing to make the trip for a gain of $5 when the calculator costs $15 but only 29 per cent were willing to do so when it costs $125. (For an example of how socio-economic status affects the outcome, see Scarcity (Mullainathan and Shafir, 2013)). The dollar gain was the same, yet a cognitive bias (endowment effect) got in the way of pure mathematical thinking. This allowed the authors to demonstrate that people often frame their decisions using topical and comprehensive accounts. According to Thaler (1999, p. 186): 'A topical account relates the consequences of possible choices to a reference level that is determined by the context within which the decision arises. A comprehensive account incorporates all other factors including current wealth, future earnings, possible outcomes of other probabilistic holdings, and so on'.
Zelizer injects thick, historically relevant meanings into topical and comprehensive accounts of money: framing a purchase decision by modifying the topic of the decision is more than a cognitive bias, reliably confusing an otherwise rational calculator. She argues instead that people experience their financial decisions comprehensively, incorporating their concerns about their personal relationships as well as a range of meaningful factors derived from the past and attributed to the future (morality and cultural meanings). Properties that exist prior to and outside of the financial decision structure it and emerge as salient at the point of the decision. Indeed, external to an immediate accounting situation are the inter-subjectively understood codes that help an individual figure out what she is taking into account and how socially significant others might react to her accounting procedures (or their consequences). Even if she disagrees with these codes or cannot articulate them coherently, she acts as if she is aware of their existence by virtue of the excuses she makes for violating them or the trouble she takes to negotiate their boundaries. At the point of decision, the actor is making a payment, a purchase, a deposit, or an investment. To frame a financial decision is to bring socially significant others to the 'the top of mind' as she recognizes them as benefiting or suffering from what she has done and as she acknowledges that they sit in judgment of her actions.
Zelizer's detailed historical work finds deep moral concerns shaping budget practices (Zelizer, 1985;Zelizer, 2010). People ask whether it is morally right to give money to wives or children, whether it is godly to bet on death. And after dealing with their moral concerns and the cultural meanings they illuminate, people consider how the money was earned, by whom, and for what purpose while making decisions about how to spend (or not) differently earmarked monies. Zelizer calls this process relational earmarking and highlights the agency of individuals creatively working to understand, establish, maintain, repair, or dissolve meaningful social relationships. In short, relational earmarking and its specific manifestation in relational accounting are forms of relational work.
Moral concerns and shared cultural codes inform, energize, and constrain actors' attempts to earmark money, and these meanings act as prisms through which relationship types and relational obligations are refracted. An individual making a decision about how to spend money on her family members thinks about herself in relation to each family member by referencing relational roles and codes; she thinks about the good or harm that her actions bring to those family members as morally good, neutral, or bad. These evaluations help her make sense of what financial obligations are the most compelling ones (morally and relationally).
Relational accounting happens at the macroand micro-level, mediated by the meanings of time. At the macro-level, there are trans-situational codes for good versus bad debt, savings versus waste, family-oriented versus selfish concerns and moral versus immoral action. I emphasize that these codes are trans-situational as a reminder that their core meanings do not inhere at the level of individuals but at the group level. Likewise, these cultural codes do not depend on individuals interacting with one another and having the meanings emerge as unique to each negotiation or interaction sui generis. Although emergent understandings do arise during micro-interactions, these understandings have a meaningful character that bumps into (or refers to) pre-existing, morally charged renderings of the actions undertaken (Norton, 2014). In other words, as dynamic and creative as individuals are, they also have to manage a set of meanings that their fresh interactions did not create. Expanding the theoretical explanation to include Alexander's (2004) notion of social performance, we can argue that dramaturgical props, images and layouts can trigger further caution or escalate abandon at the point of decision. And we can see how cultural codes and meaning structures may be linked directly in a dynamic framework of relational accounting.
In my framework relational accounting begins (i) 'upstream' where identifiable codes and structured meaning systems that shape the set of options, non-options and their sense of being possible for the person engaged in action. These codes are not morally neutral but depend on the force of morality for motivation and endurance, and their moral character structures the geometry of their decision pathways. (ii) Some budgeting practices refer to meaningful time, ritually prescribed durations (mid-stream). (iii) Budgeting decisions are performed through dramaturgy, as people enact roles and accomplish meaningful scenes (downstream). These performative communications follow performance principles and constitute relational work. After explaining these three dimensions of relational accounting, I turn to concrete examples of how they work.

Upstream: Morality and Decision Structures
As people account for their monetary practices, they often talk about the why. In Why?, Tilly (2006) notes that people can assign blame (done 'by' whom/what), indicate whether an action was inadvertent or not, and can use such explanatory accounts to create, repair, modify, or terminate their social relations. Relational work relies on the moral determinations of right and wrong, intentional versus unintentional harms (blameworthiness) and worthy versus unworthy goals. The interpretive 'prisms' of morality and cultural meanings were seen as a 'sacralization effect' that constrained and directed the 'pipes' of social relationships (Podolny, 2001) in Zelizer's formulation.
While relationships offer cues about what types of spending, gifting, or withholding for what type of social tie, people come to understand their relationships as often having 'bright lines' and 'bright lights' (Hitlin and Piliavin, 2004;Vaisey and Lizardo, 2010). In other words, there are boundaries about what is appropriate given the type of relationship two individuals have with one another, and there are moments when a person's peers think that she has 'crossed the line'. At the same time, there are goals (bright lights) that people have that carry moral weight. One such goal is to nurture one's children or others entrusted to your care. Such goals are morally structured.
When individuals keep track and make sense of their spending, debt and savings decisions, they are not weighing options as if all feasible things are possible. Instead, they are sometimes shutting off some courses of action as nearly unthinkable. This is not simply a matter of what type of relationship they have with people affected by their decisions. It is also a matter of the moral stances they want to take and the corresponding effect their moral understandings have on how they structure their decisions.
Social scientists used to model economic decisions as options in a fairly symmetrical decision tree where probabilities could be assigned to each course of action (or likely payouts weighed). However, recent work from cognitive science identifies how the assignment of blame and the moral valence of an action structure the geometry of a decision (Nahmias et al, 2005;Knobe, 2010;Knobe et al, 2012). By geometry of a decision, I mean the shape of different decision points where at least two courses of action should be possible (engage or not in the action) and where the individual takes the action (or not) for a reason (the why). A decision tree represents how an actor calculates a course of action, understanding that A is done in order to achieve B and that A is done 'by' something or someone. (The 'someone' may have a significant relationship with the decider, but even among strangers general moral principles still hold.) Work by such scholars as Knobe (2010) and others reject the one-stage view of morality showing up after a decision tree is constructed. In other words, morality and culture are primary causes for how people categorize their financial options, evaluate those options and act on them. In the old, one-directional view, Knobe argues: 'People's representations of the structure of the action tree would affect their moral judgments, but their moral judgments would not have any effect on their representations of the structure of the action tree itself' (p. 556). Instead, Knobe and others find that moral deliberations happen before the decision tree forms its branches, and sometimes one branch of tree collapses onto another. When a branch collapses or contracts severely, people act as if one of the intentions of their action (a branch) cannot be considered as distinct from, or exclusive of, another intention. In short, they have trouble letting instrumental justifications serve as a legitimate basis for acting as if actions that are technically feasible can be considered or acted upon as if morality and cultural expectations could be bracketed.
Let's take the example of morally good versus morally bad conditions. And let's ask ourselves what the decision tree would look like for a symmetrical decision tree that only considers the moral valence of the decision at the point of decision (after the decisions branch) versus what the tree looks like for an asymmetrical tree that is shaped by moral evaluations before the point of decision (before the outcomes of the decision are allowed to branch as being analytically distinct from one another). I begin with the example that Knobe uses before presenting a parallel example for relational accounting.
When alternating between asking people a morally good versus a morally bad scenario, Knobe walks through a New York park gathering his sample of 43 respondents. He notes that most people think it is a moral good to help the environment and a moral bad to harm it (intentionally). The intentions of trying to earn a profit and trying to help the environment should be mutually exclusive, no matter the outcome; however, morally bad outcomes make it harder for people to see these two intentions as mutually exclusive, whereas morally good outcomes operate as expected. In the morally good condition, respondents were asked to evaluate the following scenario: The vice-president of a company went to the chairman of the board and said, 'We are thinking of starting a new program. It will help us increase profits, and it will also help the environment'. The chairman of the board answered, 'I don't care at all about helping the environment. I just want to make as much profit as I can. Let's start the new program'. They started the new program. Sure enough, the environment was helped.
They were then asked whether they agreed or disagreed with the sentence: The chairman increased profits by helping the environment.
The contrasting scenario presented a morally bad outcome (harming the environment), but all other aspects of the scenario remained the same. In the morally good condition, most respondents indicated that helping the environment and increasing profits were mutually exclusive intentions. And the forced choice nature of the experiment allowed the researchers to isolate the effects of the morally bad outcome on the decision trees of the individuals tested.
In an amoral symmetric decision tree, respondents should have come to the same conclusion about the morally bad condition. They should have responded that the environment being harmed and increasing profits were analytically distinct and that one was not done in order to bring about the other. That is not what happened. Morality mattered. When there is a moral bad, the decision tree tends to collapse. So here, implementing the policy is not a separate consideration from harming the environment. They are one and the same.
Let us now create an analogous scenario for relational accounting. When the person in charge of household budgeting is seen as acting on her own behalf rather than on behalf of others in the household (and making the family suffer for it), it is viewed as a moral bad (selfish with harm for dependents). Part of the judgment lies in the fact that parents are supposed to care for their children. (The categorical relations come with corresponding obligations). And the care parents give to their children should be given with warm affect, not cold analytics. Harming the household collapses as a separate branch for evaluation. In this new scenario, respondents are asked if enrolling in a special savings account that might require families to make some immediate consumption sacrifices can be thought of as analytically distinct from the outcome of helping or harming the family.
First, the morally good condition: A marketer went to a parent and said, "We want you to enroll in a special savings program. It will help you to insure against future emergencies, and it will also help your family".
The parent answered, "I don't care at all about what it does to my family. I just want be have as much insurance against emergencies as possible. Let's start the new program". She started the new program. Sure enough, her family was helped (for example, more stability).
Respondents were then asked whether they agreed or disagreed with the sentence: The parent insured against future emergencies by helping her family (Modeled on Knobe, 2010, p. 559).
We can model this decision using downward arrows to indicate that C (insured against future emergencies) was achieved by doing A (enrolling in special savings program); one could also argue that C was achieved by doing B (helping her family). Figure 1 represents the structure of the decision tree, indicating that A and B operate downward onto C: Insuring by enrolling; insuring by helping.
This means that it should be impermissible to say either of the following: • The parent enrolled in the special savings program by insuring against future emergencies.
Such a statement would represent a movement up rather than down the decision tree. Likewise, movements to the side are not logically permissible, ruling out the following statement: • The parent insured against future emergencies by helping her family.

Enroll in Savings
Insure against emergencies Help the Family While these rules seem to hold when there is a morally good outcome, they do not when there is a morally bad outcome. In other words, the structure of the action tree and the logically permissible attributions made in it operate by virtue of the moral evaluations brought to bear.
Let us consider this scenario with a morally bad condition: A marketer went to a parent and said, "We want you to enroll in a special savings program. It will help you to insure you against future emergencies, and it will also help your family". The parent answered, "I don't care at all about what it does to my family. I just want be have as much insurance against emergencies as possible. Let's start the new program". She started the new program. Sure enough, the family was harmed (for example, less disposable income to meet immediate needs).
Respondents were then asked whether they agreed or disagreed with the sentence: The parent insured against future emergencies by harming her family (Modeled on Knobe, 2010, p. 560).
If this scenario operates similarly to what Knobe finds in his experiments, we would expect the asymmetry to hold. Subjects in the morally good condition will tend to disagree with the statement that the parent insured against future emergencies 'by' helping her family, yet subjects in the morally bad condition will tend to agree that the parent insured against emergencies 'by' harming her family. This testable proposition suggests that people have different intuitions about 'by' (intention) as a result of their moral evaluation of a situation.
In the above scenario, one could ask whether enrolling in the savings program and helping or harming the family can be thought of as distinct (mutually exclusive) acts or whether they are nearly the same thing. Again, we would expect to see an asymmetry in how people responded to this question in the morally good versus the morally bad scenarios. In other words, for those evaluating the morally good scenario, enrolling in the savings program and helping the family are likely to be evaluated as distinct acts. By contrast, for those evaluating the morally bad scenario, enrolling in the savings program and harming the family are likely to be evaluated as the same thing Figure 2.
These experimental findings challenge our understanding of economic intuitions. Anticipated outcomes carry moral valence, and this moral valence shapes the decision tree in such a way as to render some options as being indistinguishable from one another. Rather than having an individual with enough information (bounded rationality) to make a decision among a plausible set of alternatives, the individual finds herself morally constrained in how she evaluates an option as being an option. And she operates using contrastive cases of right and wrong (binary oppositions) that switch her into rather different evaluative algorithms. While deep interpretive work can help analysts understand the meanings of these options and non-options, abductive analyses can help us back out the morally salient contrasts structuring decision trees.
Not surprisingly, morally good versus morally bad conditions often reference relationships to socially significant others. While the example of harming the environment links the evaluation to a generalized other (and perhaps, too, to the future of loved ones), the example of harming the family has direct ties to the entire family unit, people who depend on such decisions. Not only do these dependents stand in as beneficiaries of a decision, but they also appear as vulnerable to the household head's decisions (dependents). While individuals may act opportunistically without moral qualms, acting opportunistically with guile (Williamson, 1981) shifts the financial decision into qualitatively distinct territory. Likewise, acting to benefit oneself at the expense of vulnerable dependents allows the accountant to be brought to account for her decision.
Moral beliefs about right and wrong and other moral considerations operate as if in a hierarchy of higherto lower-order understandings (Battigalli and Siniscalchi, 1999). These beliefs exist prior and external to an actor engaging in a decision, and these beliefs are updated or modified over time. These beliefs are also matched to different types of relationships as the actor perceives that her behaviors will affect socially significant others who have a moral claim on the household's caretaker. Moral considerations are not only private and individually held but are also collective and inter-subjectively shared while being dependent on ongoing social relationships within and outside of the household.

Midstream: Meaningful Time
When making budget decisions that affect one's household, 'meaningful time' matters. Meaningful time does not refer to frequencies or seasonal patterns (summer versus winter spending) but rather to ritually marked moments. (For a discussion of micro-interactional sequences that are not tied to ritual events, see Wherry (2014)). Ritually marked moments include rites of passage (for example, birth baptisms, bar/bat mitzvah, quinceiñera, special birthdays, high school graduation, marriage, funerals and other coming of age ceremonies widely recognized by one's community) and highly regarded religious or religious-like events (for example, Christmas, Hanukah, Easter, Passover, Thanksgiving, New Year's). As Van Gennep (2011) notes in his classification of rites of passage, societal pressures promote ceremonies that allow people to transition from one life stage to the next. These life stages are not physiological in nature (though they often correspond to physiological changes) but depend on social meanings. I argue that Van Gennep's view of semi-civilized people, however, applies to those of us in modern, civilized, highly industrialized societies. Van Gennep writes: 'The life of an individual in any society is a series of passages from one age to another and from one occupation to another … . Among semi-civilized peoples such acts are enveloped in ceremonies, since to the semi-civilized mind no act is entirely free of the sacred. In such societies, every change in a person's life involves actions and reactions between sacred and profaneactions and reactions to be regulated and guarded so that society as a whole will suffer no discomfort or injury' (pp. 2-3). Religious events include celestial changes associated with the solstice and equinox (New Year's Day). And people making decisions about how to spend monies meant to sustain their household often include considerations of the sacred. They feel compelled to experience the transition of time meaningfully, whether that transition be a yearly ceremony or symbolic enactment of the new self crossing from one major life stage to another. People budget as if they revere their 'semi-civilized' (non-mathematically rational) minds. Merton (1957) recognizes the cultural meanings of time as temporal expected durations. The collectively prescribed durations for an event and the order of movement from one event to the next do not depend solely on a mathematical account of efficiency but rather on collective definitions of temporally constituted situations. There are widespread expectations about what time it is, what situation people are encountering, how such times should be honored and such situations performed (Halawa, 2015). Observers and direct participants express righteous indignation at non-conformers and feel compelled by internal convictions themselves. Over time these expectations can become part of standard operating procedures at work, in the tax code, at houses of worship or places of charity, conferring legitimacy on the situational moment and its requirements.
In my model time performs a semantic and syntactic function. The meaning of time depends on a cultural meaning system that helps individuals rank the importance of particular moments. Individuals do not react robotically to time but take these meanings into account as they anticipate how that moment will draw them into harmony, conflict, or separation from socially significant others. For actors to order and assemble a number of smaller time periods into a coherent (inter-subjectively recognizable) moment, they need cultural systems that align their expectations for what time it is with those of their socially significant others. These cultural systems also signal how much flexibility others are likely to give an actor in how she defines and refines her understanding of what time it is and what that means for what she needs to do with respect to it.
Let's consider the meaningful moment known as Christmas. Over time businesses have created the Christmas bonus; shopping malls, the catalog; churches, the collection; and charities, the family-oriented give-away. These practices mark time and make it difficult to pretend not to know what time it is. Although it would be more economically beneficial to wait for the day after Christmas to find the best sales, most families with children but with meager incomes still strive to be on time. They demonstrate their willingness to pay more in price and in financing to achieve their collective goal. Moreover, when parents set aside money for Christmas, they try to guard it against other more pressing concerns that may be more pressing or that may be difficult to deny in normal times.
As the heads of household plan for these important moments, they engage shared meaning systems while undertaking complex intra-household negotiations. Zelizer's work reminds us that many working class women used these savings clubs in the early to mid -1900s to send a number of complex relational messages to loved ones. These clubs enabled these women to guard their holiday monies from other household members who might have compelled early expenditure. And this was especially the case with their husbands. She writes: Indeed, the Christmas money was often reserved for domestic necessities such as a washing machine or a daughter's new coat …. the segregated Christmas money spared them from what was often considered the humiliating need to extract gift money from their wage-earning husbands. Thus, not only individual self-control but also negotiated household relations accounted for at least some of the Christmas clubs' great success in the United States. (Zelizer, 2012, p. 160) How do these savings clubs take advantage of relational accounting? Not all savings clubs are created equal because they are associated with different rituals. The Christmas Savings Club taps into an existing ritual and set of practices, formalized in a new financial instrument. If it were not Christmas or another ritual well enough regarded, the club would not have had the same power in keeping the money of wives off limits to their husbands. Christmas and its associated meanings helped these women challenge their husband's authority without threatening the relational roles of their husbands as the heads of their households. There is a chain of affirming, preparatory events associated with Christmas and a future expectation of a joyous, sacred occasion that is evaluated differently from current events and immediate problems.
Christmas happens every year, yet its frequency does not diminish its importance. Other situations happen once in a lifetime as a child comes into the world (birth) or a loved one departs (death). In the case of the former, about at least 9 months of warning allows the household to prepare financially. In the case of the latter, the transition often happens suddenly without warning. Not having enough time to prepare, however, is no excuse for not appropriately honoring the dead. Providing such honors may impose a significant strain on the family purse. Only in the most extreme situations of economic scarcity will the cheapest funeral option be viewed as an option. These expectations impose themselves on the household and both time and monetary calculation seem as if they have been suspended or at least placed in a different realm of calculation where opportunity costs are irrelevant and cheap alternatives to honoring the dead deemed defiling.
Attending to the meaningful aspects of temporality runs the risk of implying that these rituals and rites of passage remain stable and singular in their meaning or that they compel families in general and heads of household in particular to revere the dominant codes shaping the situation. We know from Lévi-Strauss (1993) that Christmas and the compulsion to spend on its behalf was a late development that met with protest and we know from Leigh Eric Schmidt that the attachment of consumer spending to Mother's Day, Valentine's Day and Christmas resulted from dynamic struggles over meaning as individuals and the groups they represented borrowed elements from different meaning systems and experimented with the possible. This pushes us to ask not simply what the meanings of the moment are but, more concretely, what the consequences are for ignoring those meanings (Fine, 1993, p. 70). One finds actors engaged with these consequences in specific sites where others may attempt to enforce the integrity of the situation. To the performance dynamics downstream we now turn.

Downstream: Relational Work and Performative Relations
To account for money is to act as if boundaries can be placed around numerical sums and that those sums can be matched to specific categories of spending. Sometimes analysts represent the process of accounting as a person figuring out her budgeting plan, where her behaviors reveal her preferences, and where those preferences reflect a sense of character (Wherry, 2008) and situated calculation (Lave, 1988). Individuals are trying to figure out how to categorize their monies and how to spend them, and they are working out how such actions affect the creation, maintenance, honoring, dishonoring, or dissolution of meaningful social ties (Bandelj, 2012;Zelizer, 2012). Their relational work leads them to perform these definitions in their accounting practices; the meanings of the relationships function performatively (Wherry, 2012). Individuals calculate value and values while 'seek[ing] to extract from others the same typifications we apply to them in the course of communication' (Alexander, 1988, p. 313). The style of communication and the execution of the relational work involved resemble a social performance (Alexander, 2004). The purchase itself and the implied budget practices that make it possible have to be dramaturgically communicated outward. The return communication inheres in the responses expected from interlocutors and socially significant others benefiting from, suffering because of, and/or standing in judgment of the purchase (or the savings/investments). The back-and-forth communication dynamics also provide opportunities for creative re-combinations, surprise, innovation, or negatively valued disruption.
Engaged in relational work, individuals match transaction media (such as money, gifts and credits) with different types of social relationships, yet this work occurs in scenes where individuals are making sense of their situations by using the 'set of sense' available from existing symbols and meanings. This matching process may be directed at dyadic ties with socially significant others or with bundles of relationships that are tied to the present situation, a reconstructed past, or an imagined future (Beckert, 2013). Parsons and Smelser's (1957) Economy and Society defined the functions of household budgeting as managing intra-household conflict and placing a household in relation to other households through the symbolic messages delivered by expenditure decisions (integrative functions). Yet they did not capture the dramaturgy of such management strategies or the performative aspects of relational work.
How is relational work enacted in contemporary household budgeting? Take, for example, an individual who comes from poor material circumstances but has managed to carve out some economic security for herself. She imagines herself in solidarity with family members she has economically left behind. And these family members imagine her to be a patron, willing to aid loved ones in dire circumstances through no apparent fault of their own. She will be more likely to receive compelling requests to co-sign for loans or to provide short-term financial assistance for friends or kin, and she will be likely to find ways to meet these requests, even if with a token gift indicating care but not paying the request in full. O'Brien develops the concept of negative social capital to account for this and measures its effects on how much households have left to save or invest. It is 'the pressure on an individual actor to incur costs by virtue of membership in social networks or other social structures' (O'Brien, 2012, p. 378). The pressure may be subtly applied or histrionically cast. And the individual placed in the role of patron looks for ways to perform that role, so that she affirms her relationships appropriately while minimizing harm (social, psychic and economic) to herself.

Relational sites
Relational performances take place in sites that differently afford sociability and meaningful claims on action. While Callon (1998) refers to such sites as the space of calculability, my emphasis on relational work takes the emphasis away from calculability back to the situated relationships being enacted. The multivalent qualities for assessment do not fully yield to market devices and are more dynamically interpreted as actors transform (partially or fully) some economic transactions into gift exchanges (Caliskan and Callon, 2009). In the relational work framework modified by my emphasis on social performance, however, meaning systems matter for interpreting a space of calculability as calculable. Indeed, meaning systems structure the logics of calculability, making elements of the site relevant for how one defines the accounting situation.
The concept of the relational site emphasizes the meaningful characteristics of the place of transaction, the dynamic interpretation of the actors making sense of the deal, and the unexpected but patterned effects on calculations occurring in places soaked in meaning. This multivalent calculation substitutes for a strict qualitative/quantitative calculation of how much a good or service might cost in a differently staged setting. The individual must relate to the site as the right place for a person such as herself (a good match). If there is a mismatch between the site's character and her own, she will be less likely to consider the goods or services being sold there, even if they are comparable to higher priced goods and services sold in the places she has deemed a good match.
Relational sites infuse the space of calculability with explicit theoretical processes found in Goffman's (1959) presentation of self and Alexander's (2004) social performance. Social life consists in the enactment of scripts as actors recognize and further define their social situations. Individuals with an easily discreditable character must pay close attention to how they perform before various audiences so that they can save face. Each individual will therefore be careful not to 'take a line' that the other actors on stage cannot respond to as if by instinct, deploying tact regarding the tact she herself deploys to maintain the integrity of the situation's meaning. These judgments about whether she has acted tastefully and will therefore be allowed to maintain face (a positive or neutral evaluation from others) hinges on whether the line she has taken 'matches' the kind of person her audience and the other actors believe her to be (as well as the kind of relationships her actions are tied to). This belief derives from group and societal level representations (Alexander, 2004(Alexander, , 2011) that exist before the initiation of the encounter and that partly steer its course. These collective representations offer paradigmatic stories about the character role she is attempting to enact, her appropriate placement within the scene, and the physical characteristics of the scene that make it resonate with the shared definition of the situation. These situational definitions affect, of course, how the individual expresses or alters the relationships she has with others. And while these performances take place using socio-technical devices (Callon et al, 2007), the actors' interpretations of the relationships they are managing (Zelizer, 2012) matter as much as the material assembly of spaces, calculative tools and other social devices that also require interpretation (Alexander, 2011, Wherry, 2012; otherwise, salience can remain low for the non-storied materials meant to shape calculation, such as the spreadsheets, financial tickers, bank text messages and cautionary letters operative in a specific site.
To define the situation, actors mark the sites of the money transfer with meaning, whether those spaces be bank teller kiosks, a pawnshop's bulletproof plastic window and reception tray (Caskey, 1994), a set of mason jars or cookie tins with taped names for their purpose (Zelizer, 1994), or a lockbox without an easily accessible key (Dupas and Robinson, 2013). These devices and spaces of calculability resemble what Callon andMuniesa (2005, p. 1231) describe as the devices that structure calculation -'An invoice, a grid, […], a trading screen, a trading room, a spreadsheet, a clearing-house, […], a shopping cart'. In my framework, these calculative devices need to be placed in a meaningful place. Where an invoice is read (the site) and the relationships (direct and implied) it marks shape the moment of calculation. The device of the trading screen being utilized in a trading room also finds itself surrounded by circuits of actors engaged in commerce who know that the rules of engagement in the room differ from those outside of it (Abolafia, 1996). The structuring of trading rooms where calculations are geographically and cognitively distributed (Buenza and Stark, 2001) also hinge on site layout, imagery and site-specific vernacular (Levin, 2001;Zaloom, 2006) along with site-specific currency that the actors openly acknowledge as 'belonging to the room'. Even in the case of virtual sites (Knorr-Cetina and Bruegger, 2002;Knorr-Cetina, 2009), there are relational tags, an interaction-order, and screen icons placing a decision to trade (or not) in a meaningful context (Preda, 2009a(Preda, , b, 2012. For more common activities such as grocery shopping, the environment of the store signals how much an individual should be willing to pay. In this way, shopping at a high-end grocery story such as Whole Foods prepares the shopper to evaluate price differently than she would if shopping at a discount grocer such as Bi-Lo. The aesthetics of each site and the carts used to carry the objects through the store affect decisions about the purchase. While the cart itself functions as a calculation device signaling how much is enough (Cochoy, 2008), it is pushed through an aesthetic space with its own imagery, in-store jargon and rules of engagement that remain site specific. The calculation device has no power to engender a specific type of calculation without the meanings of the site itself infused in the device and its user.

Third-party enforcement
The sanctioning capacity outside of the household from third parties, both informal and formal, provides a powerful constraint on budgeting decisions. Informal third parties include friends, kin and local charismatic authorities. Formal third parties include organizational agents from banks, credit counselors, debt collectors, non-profit counselors, social workers and human resources personnel. In the illicit economy, other agents such as loan sharks and enforcement personnel provide such sanctioning. In formal economic theory, sanctions appear to be merely incentives that stimulate individuals to behave favorably toward a given object or to be compelled to reject an object's allure. In social performance, however, socially significant audiences largely function as sanctioning bodies, and these audiences may be real or imagined.
Zelizer identifies third-party enforcement as a core aspect of the relational work, whereby an inappropriate budgeting decision results in having a family member or friend indicate that the budgeter 'has lost her mind' or where the budgeter herself internalizes a conversation she does not want to have in which she cannot justify her behavior in normative terms. Zelizer notes that relationally based understandings of who (a categorical but sometimes contested identity) has honored or defiled a what (a multivalent object) enable people outside of the immediate exchange to evaluate it.
For any given transaction, a different cluster of third-party enforcers will be relevant, and perceived violations of the exchange rules may activate other third parties that previously lay dormant. But these perceptions of norm violation are not automatically triggered. Rossman (2014, p. 55) writes: [Third parties] might internalize an understanding of [the exchange] as not an exchange, perceive it as an exchange but have the tact to suppress mention of it, or indignantly denounce it. We can understand these obfuscated exchanges as analogous to stigmatized people […], but with the difference that the stigma primarily applies to the interaction and only by contagious extension to the participants. Just as some stigmatized people can pass, or are allowed to pass, so too might obfuscated exchanges. Even if such an interaction is understood as reputable, this meaning can best be understood as passing, and thus its standing is tenuous and discreditable.
Third-party enforcers are trying to make sense of the implied calculations preceding an exchange. Making sense sometimes means that a budgeting decision might be construed as a non-monetary obligation. In taboo situations, a monetary gift to a romantic partner might be framed as assistance to her family (whom the giver has never met) so that both can pretend that she is not engaged in a form of commercial sex work. Likewise, a monetary payment made to a surrogate mother may be an 'extra' gift (rather than a payment), even if the function of the gift is to help the giver obtain the surrogate's services while in competition with a rival. The giver may worry that she will be accused of baby buying or of renting a womb in a competitive market. In other instances, an individual may vehemently reject the idea that she has engaged in frivolous spending by noting how much she 'saved' by purchasing an item on sale (Miller, 1998).
These justifications rely on collective beliefs to make the exchange and its calculating agents seem more straightforward (or more morally straight) than they may actually be (Bourdieu, 1977(Bourdieu, , 2000. These collective meanings provide normative 'cover' for norm violators while at the same time legitimating which accounts third-parties should enforce. Collective accounts provide more than just excuses: They let us in on the meanings of an exchange, their moral nuances, and the social relations esteemed or shamed by a budgeting decision.

Applications
I will present two applications of relational accounting to demonstrate how it works for routine purchases made by individuals and how it works for less routine budgeting decisions, such as that of acquiring a loan.

Routine purchases
In a basic accounting framework, we should assume that there is a maximum price for which an individual is willing to pay for a good or service. The individual's goal is to obtain the good at or below that price. The place where the good or service is obtained, however, affects what that maximum price will be. The situational context pushes us to evaluate a purchase decision in a specific way. We could think about the role of the situational context using a topical frame from mental accounting or a comprehensive frame from relational accounting. As we examine the differences between these two approaches, we begin to see what is at stake in our conceptual development of our concepts. First, we turn to mental accounting. Thaler (1999) explains that transaction utility emerged as a substantive critique of acquisition utility, noting that individuals do not derive a single, consistent set of utiles (satisfaction) from obtaining a good (acquisition utility). Instead, the willingness to pay for a good depends on the physical location where the good is acquired as well as the process of acquisition in the physical site (transaction utility). The concept of transaction utility tries to pare down the site where the good is obtained and has to keep all other aspects of the situation constant so that randomized experiments can be run in which people are asked how much they would be willing to pay for a Good X in a specific Place 1. This means that if the same good is being offered in Place 2, we should expect the individual to be willing to pay the same price to acquire it, all other things will have been held constant except for the characteristics of the place. If there is a difference in how much the buyer is willing to pay, it is reasonable to attribute that difference to the place of acquisition and expectations about that place, because the good itself (for example, including its quality and brand) is the same. Here's the scenario Thaler used: You are lying on the beach on a hot day. All you have to drink is ice water. For the last hour you have been thinking about how much you would enjoy a nice cold bottle of your favorite brand of beer. A companion gets up to make a phone call and offers to bring back a beer from the only nearby place where beer is sold (a fancy resort hotel) [a small, run-down grocery store]. He says that the beer might be expensive and so asks how much you are willing to pay for the beer. He says that he will buy the beer if it costs as much or less than the price you state. But if it costs more than the price you state[,] he will not buy it. You trust your friend, and there is no possibility of bargaining with the (bartender) [store owner]. What price do you tell him? (Thaler, 1999, p. 189) Thaler discovered that people were willing to pay more for the same beer purchased at a resort than at a run-down store, and he noted that it should not matter where the beer is purchased because in standard economic theory, the setting does not affect acquisition utility -'the value of the good obtained relative to its price ' (1999, p. 188). The difference in the price the consumer is willing to pay for the exact same good differs by the setting of the exchange (transaction utility) and 'the perceived value of the "deal" ' (1999, p. 189).

Upstream: Morality and decision structures
Notice how moral concerns and relationships have been largely stripped away in the mental accounting example. The only relationship we find there is the friend who will make the purchase on the individual's behalf. This appears to be a morally neutral situation, so we expect to see a decision tree that does not have a collapsed or severely contracted branch. This is the first beer of the day, and there is no hint that a moral harm will come to the individual's friend or extended family by virtue of buying a beer. In that moment, however, the individual has to wonder what signal he will send if his demand for a cold beer is too high. Would that not signal a lack of discipline, an out-of-control craving?

Midstream: Meaningful time
We have no indications that the purchase of the beer is tied to a rite of passage, nor do we see it involving highly regarded religious or religious-like events. Therefore, the social compulsion to refuse to spend more than necessary for a beer is diminished.
Downstream: Relational work as performative In this moment, the individual is managing the presentation of self before a friend and may be trying to maintain face even as he maintains his relationship with his friend. How might his friend negatively sanction a morally wayward demand: a grimace, an admonition, a suggestion that the price being offered is way too high? The individual may also have in mind how his money might have been spent otherwise. In both sites, the aesthetics of place signal lower versus higher prices, and the setting also contains clues about the kinds of people most likely to be found making purchases there. As routine as the example seems, we find a great deal of moral and relational content in it.

High cost loans
Using the relational accounting framework to analyze the decision to obtain a high cost loan would require us to begin by examining the cultural meaning systems informing the decision. Banking associations, advertisements and news editorials present a range of materials where the meanings of debt and borrowers have been made manifest. Peebles explains that saving versus borrowing were not only mathematical calculations but were also civilizational and moral attributes. Before the Industrial Revolution, the working and 'uneducated' classes 'hoarded' their monies privately, did not rely on formal banking institutions, and lingered in a meaning system where these people felt comfortable depositing their monies at the church but not at the bank. The foundation of the Savings Bank in Scotland has its own foundation myth in which the earliest banks of this kind relied on bankers who 'collected deposits after the sermon' at the Sunday Banks. Peebles (2010, pp. 243-244) writes: In a similar vein, a bank opened up by a clergyman sent out a circular to "the Industrious" when it opened. Stressing the dangers of hoarding money at home, it concluded with the statement: 'Savings Banks ensure competencycompetency inspires dignity and independence of mind. These render their possessor superior to low dishonest and groveling arts. He thus becomes a man in the strictest sense of the word. He becomes sufficient to himself, leaning on no borrowed staff, putting himself at no fellow man's disposal, soliciting no alms, nor bending submissively under a sense of obligation in the presence either of a generous or haughty benefactor'. (Horne, 1947in Peebles, 2008 To have dignity, independence and self-control required participation in Savings Banks. These are morally salient characteristics that shape how saving, hoarding, or acquiring debt will be evaluated as viable options among a set of options.

Upstream: Morality and decision structures
Taking on debt can be a moral good or a moral bad. A good debt allows a middle-class person to care for her family by ensuring that basic needs are met or by securing (and investing in) home ownership. These messages about 'good debt' contrasted sharply with those about 'bad debt'. Yet, it has been the purveyors of 'bad debt' (high cost loans) that utilized these tropes of personal responsibility and self-control to convey that poor and working-class people could borrow from themselves and thereby be responsible (moral beings); moreover, the messages came with a welcoming and understanding tone. Of course, it is easy to fall onto hard times, and PayDay lenders will not embarrass a cash-strapped individual whether or not she had a hand in her current troubles. The moral bad is not the debt but the harm that is likely to befall the family if the individual does not take on the debt.
These morally salient messages shaped the experiences of borrowing and made less salient the mathematical costs of the loan. A 400 per cent annualized percentage rate on a loan at the mainstream bank would be rebuffed or deemed unthinkable, but by virtue of being on the wrong side of a symbolic binary and needing to pay penance to get back on the right side, consumers proved themselves willing to inflict self-pain. They ignored the fact that they were paying 20-30 times as much in interest and fees at the PayDay lender for the same amount of money. And the option of not 'appropriately' providing for their families was not seen as a viable branch in the structure of their decision tree.

Midstream: Meaningful time
We have historical and contemporary indications that savings practices were tied to religious events, such as Sunday services in Scotland (Keister, 2003;Peebles, 2010;Delaney, 2012); we also have contemporary accounts of entrepreneurs being called upon to financially support saint-day festivals or individuals to make meaningful gifts for baptisms (Portes and Landolt, 2000). During such times, the likelihood that an individual defers to a request for gifting or assistance is significantly higher than in normal times.

Downstream: Relational work as performative
After analyzing the symbolic binaries and how this language is worked into mainstream depictions of lending, one can see how these messages are infused in the architecture at the banking sites where the loans are obtained. These meaningful and relational concerns include moral considerations, and in some cases, a moral imperative drove these individuals to obtain the financial product. After all, saving or extending the life of a child or an aging parent or giving one's children opportunities through education (Polletta and Tufail, 2014) are not goals that can be weighed in a standard decision tree, nor are they accounting situations where third parties do not intervene to insist that some options remain outside of the realm of good behavior. To impose a moral harm on loved ones is nearly unthinkable, and third parties help enforce that un-thinkability.

Conclusion
My extension of Zelizer's relational work and relational earmarking represents a sharp departure from individualized analyses of budgeting where interests, goals and satisfaction are brought in line with resource constraints through tinkering, cognitive shortcuts and satisficing. While Zelizer only used the term relational accounting once, I have built it into a performative theory of relational earmarking where meaning systems, moral concerns and meanings about time (as well as timing) act as structuring structures. My approach looks to internally felt morals, intuitively sensed symbolic oppositions, and ongoing rituals that shape the structure of decisions and constrain the pathways (creative and standard) for goal attainment. By the time the micro-interactions happen, the meaning systems, cultural codes, temporal expectations and relational concerns have already been in play, enabling the actors to see some types of expenditure as morally necessary (right now). The structuring structures thereby prefigure the roles available to the budgeter and to the earmarked beneficiaries, co-constituting the paths of those engaged in relational work.