OUR THEORY AND ITS CONTRIBUTION TO EXISTING RESEARCH

Existing research on social capital in Political Science can be classified into two different categories, based on the unit of analysis for examining the concept: the group or the individual. Group-level research posits that repeated interactions within associative organisations increase mutual trust, norms of reciprocity and altruistic tendencies that in turn lead to social goods such as political cooperation and government responsiveness (Putnam Reference Putnam2001; Varshney Reference Varshney2003: 362–98). Individual-level measures of social trust, often captured in survey data, focus on the orientations of individuals to broad social categories of ‘others’ and their expressed levels of trust (Stolle Reference Stolle2002; Freitag & Bauer Reference Freitag and Bauer2013).

Group-level research is founded on studying the constellation of associational life, and the relationship between groups in a given social context. Putnam (Reference Putnam1994, Reference Putnam2001), in studying associative organisations, posits that social capital arises through the interaction between members in stable relations. This conception of trust is then broken down into different types of capital, like bonding versus bridging Footnote ¹ or bracing Footnote ² which all have differential effects on aggregate trust, with bridging capital increasing overall trust and bonding capital decreasing it. The major difference in impact between these ‘types’ of capital are in the composition of the groups in which the trust is fostered (see e.g. Pena & Lindo-Fuentes Reference Pena and Lindo-Fuentes1998; Varshney Reference Varshney2001, Reference Varshney2003; Gacitua-Mario & Wodon Reference Gacitua-Mario and Wodon2002; Larsen et al. Reference Larsen, Harlan, Bolin, Hackett, Hope, Kirby, Nelson, Rex and Wolf2004).

Individual-level research, based on survey data, also makes distinctions among different kinds of trust, mostly focusing on particularised versus generalised Footnote ³ trust (see e.g. Freitag & Traunmüller Reference Freitag and Traunmüller2009; Newton & Zmerli Reference Newton and Zmerli2011; Uslaner Reference Uslaner2002). As with bonding and bridging group-level distinctions, there seems to be an inverse relationship in different kinds of individual level trust. Particularised social trust seems to undermine (or at least be correlated with lower levels of) generalised social trust at the individual level (Uslaner & Conley Reference Uslaner and Conley2003). Studies examining individual-level social trust largely depend on survey instruments, asking either directly or indirectly about how much respondents trust people in their society. While direct questions are subject to significant heterogeneity in respondent interpretation based on the wording of the question (Sturgis & Smith Reference Sturgis and Smith2010), indirect questions often confuse the indicators of social trust (such as tax compliance or rule-following) and the causes of social trust, in terms of individual orientation (Knack Reference Knack2002). Additionally, survey instruments’ questions often evoke responses mirroring the generalised trust environment that people live in, rather than trust as an experienced phenomenon or a personal trait, which leads to high levels of volatility in responses, and disconnection from other posited measures of social trust such as membership in voluntary organisations (Newton Reference Newton2001).

Both group-level and individual-level research in Political Science share three common weaknesses in understanding trust: the contextual sensitivity required to interpret the critical distinctions between different forms of trust, the lack of sensitivity to individual diversity in experienced trust, and the need for static relationships within which to measure trust. In comparing particularised trust versus generalised trust, or bonding versus bridging organisations, context determines the differences. In other words, a bonding organisation in one society could be a bridging organisation in another, contingent on the salient social groupings (tribal vs. religious) and cleavages (ethnic vs. economic) (Narayan & Cassidy Reference Narayan and Cassidy2001). At the level of survey instruments, the ability to discern and compare different kinds of trust is obstructed by the necessity to translate specific terms from one language to another – such as ‘trust’ and ‘stranger’ – or even account for variations in cultural contexts, wherein respondents could implicitly assume that they share or differ in religious and ethnic identities with said strangers (Adam Reference Adam2008; Torpe & Lolle Reference Torpe and Lolle2011). The problem of individual-level survey instruments is further compounded by co-ethnicity effects and social desirability of responses to questions, depending on the perceived identities of researchers collecting data (Adida et al. Reference Adida, Ferree, Posner and Robinson2016).

Even in conditions when social characteristics are similar, the interactions through which generalised trust develops are measured as occurring within stable and mutually recognised dyads of individuals, groups or entities like the state. For example, Putnam's argument that the relationship between increasing diversity and generalised trust resembles a normal curve (Putnam Reference Putnam2007) assumes that repeated interactions and friendships occur between members of in-groups and/or between members of out-groups. Uslaner's counterargument based on American and British societies that inequality and residential segregation, not diversity, strengthens ‘in-group loyalty that works against trusting people who are different from yourself’, also assumes that repeated interactions occur between individuals of in-groups or out-groups (Uslaner Reference Uslaner2010). Both assume a stable social network of identities, and a single salient social cleavage.

The last major problem in the Political Science literature, the assumption of homogeneity of experienced trust, has been partially dealt with by scholars in neighbouring disciplines, especially in Sociology. Work from scholars such as Briggs (Reference Briggs1998), Lin (Reference Lin2000) and Ferragina (Reference Ferragina2013) has examined the ways in which social inequalities affect the level of experienced social capital in both theoretical and empirical ways. Yet, the literature in Political Science has been largely silent on this measure.

While our method draws insights from both sociology and economics, this protocol seeks to inform the ways in which social trust has been used as a monolith primarily in political science. In the vein of Putnam, we are seeking to understand social trust stemming from the individual and manifesting in institutions without succumbing to the problems of either survey or group-based research as they stand in political science. By providing a less context-specific measurement than associative organisations, our theory provides a more general strategy for observing social capital and one that implies how social capital can be measured via individual-level interactions. We test our theory through data collected in South Africa, though it could be applied in almost any context, and compared by using institutionally similar circumstances.

WHAT IS A CHECK-POINT?

While moving through a regulated space, whether in commercial exchange or interaction with state institutions, there are times and places in which individuals are asked to demonstrate with documentation or proof that they have adhered, or are currently adhering, to the rules. Proof, constituted through individual or combined verbal signalling, written documents, or physical actions whose legitimacy are recognised by the institution, is the basis on which access is granted. Such proof may or may not be created by these institutions.

From a practical standpoint, such interactions can encompass a wide variety of exchanges, including everything from demonstrating proof of payment to get coffee from the barista, to showing documentation at a border in order to gain access to a country. Proof can be in the form of receipts of sale, passports or visas, or verbal affirmations to demonstrate familiarity. But such exchanges also frequently happen to gain entrance to membership spaces, such as gated communities or country clubs, accessing universities through security perimeters, going through roadblocks, or through airports which demand proof that you are a paying customer with an impending departure in order to get through layers of security. Proof is demanded in order to move forward. Check-points, in their essence, are an exchange of proof for access.

If one imagines, for example, the act of buying a coffee at a shop, then there are multiple stages through which you will progress. When you approach the till and communicate your order, two different processes are set in motion. The cashier calculates the total that you owe, and the barista begins to prepare your order. Between the transfer of money from the customer to the cashier and the transfer of the beverage from barista to customer, how often are you asked to demonstrate the rightness of your place within the shop? Are you asked to give proof of your order or payment to remain in the café, or in order to pick up your drink? Is such proof required of all customers, or only some?

WHAT IS A TRUST-POINT?

Trust-points are, in many ways, the opposite of check-points. They are the roadblocks that do not exist, the policemen not asking for citizenship or residency proofs at a routine traffic stop, or the fact of picking up a coffee without showing a receipt or identification. Within a given institutional space, a trust-point is when a check-point could exist, but does not: in other words, when individuals are permitted to move into the institutional space without documentation. In individual contexts, it is exceedingly difficult to identify trust-points as such, since they are characterised by their absence. They are visible mostly in comparison of similar spaces in different contexts, or with different people in a given space.

In the example of a coffee shop, what are the differences in experience ordering a coffee from the same chain in an affluent neighbourhood versus an underprivileged one? Or, ordering coffee in a town-centre versus an airport? There are routinely places where, despite an overall similarity of governing rules, different contexts provide the opportunity for the creation of trust-points, such as not showing proof of purchase to pick up goods, like dry cleaning or coffee.

Such trust-points, as noted above, are largely visible through comparison. For example, in the USA, we implicitly trust restaurant staff to correctly charge our credit cards out of our sight. However, in South Africa, because of rampant credit card fraud, regulations were enacted which dictate that machines must be brought to customers so that such cards never leave their direct line of sight, at the recommendation of both foreign governments and major financial institutions (Deloitte 2014; Overseas Security Advisory Council 2019). Although not clear by only examining the experience of dining in the USA, it becomes clear through comparison that an enormous bond of (financial) trust is depended upon in the course of the average experience of eating at a restaurant. The absence of this trust imposes costs – in terms of restaurant internet infrastructure, staff training and equipment, as well as time – on the establishments who adopt such innovations.

WHAT IS A FAST-LANE?

Several questions can be asked about how check-points and trust-points operate. How often are individuals asked to demonstrate adherence to rules, and how often is proof not required? Who is asked to demonstrate proof, and are the demands for proof applied regardless of the individuals’ identity characteristics? These points of inquiry are the basis on which check-points and trust-points can be used to demonstrate the presence of social capital in a given society.

There are some check-points that are only checks for some individuals, but points of trust for others, based on their ascriptive identities. In other words, individuals’ interactions with the established – ostensibly neutral – check-points and trust-points are contingent on their race, ethnicity and language, among other identities. Compared with check-points and trust-points that measure aggregate social capital, the presence of fast-lanes helps to ascertain their differential impact on distinct social groups. As a result, societies characterised by trust-points rather than check-points could also have many fast-lanes: revealing that although such societies have high social trust, certain groups continue to face barriers and experience observably lower levels of social trust. Conversely, in societies characterised by check-points, the prevalence of fast-lanes would reveal that certain groups enjoy economic and social advantages denied to others. In this way, the presence of fast-lanes is indicative of the very real costs imposed by discrimination and the ways in which prejudice can be manifested within the context of interactions with the state and with private institutions. It is certainly true that the selective application of surveillance is often rooted in institutional and individual prejudice. By understanding it in terms of social trust, and the costs imposed by its absence, we are trying to illustrate how social trust environments are conditioned by prejudice and identity, in ways previously overlooked in empirical studies.

The proliferation of ‘Show Me Your Papers’ laws in states such as Arizona (SB 1070) in 2010, and Utah (HB 497) in 2011, or the application of stop and frisk laws in places such as New York City, provide good examples of the ways in which check-points can be selectively applied, based on individuals’ ascriptive identity. When challenged in the court, critics of such laws suggested that the only people from whom authorities were demanding proof of residency or citizenship were those perceived as Latino, or as threats, by law enforcement officials.

Our anecdotal observations regarding the University of Pretoria in South Africa also demonstrate how race continues to play a major role in access to the campus. Frequently students forget to bring their identification cards that must be swiped at the turnstile gates, in order to enter the campus on foot. The same needs to be done by drivers of vehicles at the proximate traffic booms manned by (black African) guards in a small guardhouse. Students who have forgotten their identification cards usually go up to the guardhouse and negotiate access into the campus. The official method is to register oneself in a logbook, including contact details, National Identification number and particulars of purpose.

However, in reality, race, class and gender determine the ease of access into campus under such conditions. For white women, such access is easy, and always with a sense of deference. The guards usually smile and wave them through. Similar conditions exist for black African women, but accompanied by light banter and cheerful warnings about carrying their cards. For white men, however, the experience can be dichotomous. There are instances when they are treated similar to white women. If they choose to engage in conversation, then a warm exchange is undertaken, and cheerful warnings about carrying cards are also delivered by the guards. At other times, if they choose to be belligerent or are perceived as such, then verbal confrontations arise. The result, however, is nearly always that the white men are permitted inside. In terms of black African men, the guards are the sternest. Detailed questions are asked and some are turned away. There are some other social variations within the category of African men that affect such interaction. On the one hand, the guards frequently take pleasure, as evidenced by smiles and jibes, in turning away young African men in expensive cars. This may signify class animosities between the relatively lower-income guards and the young affluent students. On the other hand, older African men dressed in conservative western clothes are allowed in, which may reflect the still hierarchical and patriarchal nature of South African society.

CHECK-POINTS, TRUST-POINTS AND FAST LANES IN PRACTICE

INTERVIEW FINDINGS

Different interview participants did report some check-points that were largely consistent across all different demographic considerations. For example, the vast majority of respondents reported being stopped at least twice a month while exiting a grocery or chain general store, and being asked to present proof of purchase, generally in the form of a receipt, to private security guards before exiting. No respondents reported that the interaction was threatening, and many of them explained the stop as procedural, non-targeted, or by explaining that security guards were ‘doing their jobs’. About 20% of respondents attributed such stops to high levels of petty crime in South Africa, which is borne out by official statistics of crime in the country that are significantly higher than comparable countries in the region or globally (Statistics South Africa 2015).

Despite the consistency of reporting such stops, however, there were vast differences in the levels of time and intrusion that were reported. While two different young, white men reported that the interaction took less than a minute,Footnote ⁴ others, for example a young black womanFootnote ⁵ and a young black man,Footnote ⁶ reported that the stop regularly took between 5 and 15 minutes each time. Thus, even though this checkpoint seems to be consistently applied, regardless of the ascriptive characteristics of the person being stopped, the application of scrutiny varies significantly, in ways that are broadly connected with both race and class.

Other stops were more obviously fast-lanes, where certain individuals reported being stopped frequently by the police while driving or conducting daily activities. While race and class did seem to condition the frequency and intensity of fast-lane stops, the respondent's status as either a citizen or a foreign resident was most often reported as the reason for suspicion by authorities. This condition seemed to apply almost exclusively to those people perceived to be, or who were actually from, other African countries. Actual or perceived foreignness was an important reason for explaining stops by police and in public places by both menFootnote ⁷ and women.Footnote ⁸ Those South African citizens who reported being ‘seen as foreign’ reported that they stressed their accentsFootnote ⁹ or facility with local languages,Footnote ¹⁰ as ways to avoid escalating confrontations with officials, and part of their repertoire of evidence that allowed them to proceed through the interaction. Foreign residents of South Africa from other African countries often reported that their status as foreign nationals made them immediately suspicious to authorities,Footnote ¹¹ and often made the interactions more contentious or expensive in the form of necessitating the payment of bribes.Footnote ¹²

Perceived class identity was another justification for the fast-lanes that were reported by participants. While many respondents reported that their external indicators of affluence allowed them to pass through barriers unscrutinised,Footnote ¹³ the opposite was also true.Footnote ¹⁴ Indicators of class, in the form of car models or sense of dress were the most second most frequently offered justifications for stops that were targeted at particular segments of the population.Footnote ¹⁵

Although interviews did not include questions about the effects of co-ethnicity, some interview participants did note that shared racial identities could promote trust, or increase distrust,Footnote ¹⁶ with officials. While some black South African respondents noted that it was often more difficult to bribe white police officers when seeking to avoid censure,Footnote ¹⁷ others noted that trust was greater with black officials.Footnote ¹⁸ Whether this constitutes the creation of fast-lanes because of co-ethnicity – insofar as black officials create opportunities for some people to avoid censure based on racial identity – or whether this is a stop which only people read as black by the officers have to abide, is unclear from the data.

While co-ethnicity was not often cited as a conditioning factor by white South African interview participants dealing with white officials, there were several who brought up the idea that there existed tensions in cross-racial interactions. Where referenced, however, such effects were often seen as minimally important, or peripheral.Footnote ¹⁹ Interview participants who identified themselves as white reported that race was only a significant contributor in the context that their racial identity promoted trust with officials.Footnote ²⁰

The last major category of fast-lanes conditioned by identity were created by differences in gender. Women interview participants regularly noted that their gender helped to create fast-lanes with male officials, both because of inherent sympathies and because the women could draw on conditions – like pregnancy or menstruation, whether actual or fictionalised – to avoid being detained or showing identification.Footnote ²¹ In one particularly telling set of interviews, two Nigerian respondents, both in their mid-20s, but of different genders, tell vastly different stories of their levels of experienced trust with officials. The Nigerian woman reported that she did not regularly face problems with the male police officers and was able to use her gender to avoid checkpoints.

Her male counterpart reported that, not only was he was regularly stopped, but that the only way to avoid such detentions was to pay bribes to officials.

Such differences in experience, both from foreign nationals from the same country, of the same age belie the strength of gender as a conditioning factor in determining the differences between fast-lanes and checkpoints.

What is evident from the range of responses about the presence of fast-lanes and check-points, however, is that such ascriptive characteristics are intersectional in their effects. Nationality, class, gender, race and other identities compound to create the experienced trust landscape of individual respondents.

LIMITATIONS OF THE STUDY

The limitations of this study fall into three broad categories – case selection, interviewer effects and sample size – and should be addressed before proceeding to a discussion of the findings. The authors recognise that a single province of South Africa is not necessarily indicative of either the country as a whole, or indeed other contexts. However, in this study, we were seeking to explore a new measurement and implementation to study social capital, and while the sample size of the study and the geographic reach of it were small, the results remain compelling as a way to demonstrate the variation of social trust within a given context, and the costs of its absence.

Structurally, we did attempt to limit the within-study interviewer effects (Adida et al. Reference Adida, Ferree, Posner and Robinson2016), to maximise internal validity and proof of concept. The research assistants themselves were a mix of South African citizens and foreign nationals, from multiple (self-identified) racial backgrounds, in order to try to minimise interviewer effects. While all interviewed participants were from both within and outside of their own status groups, the majority of the interviews were within shared-status groups. Additionally, there were limited social desirability effects because of the lack of overtly partisan or political content in the interview questions (Adida et al., Reference Adida, Ferree, Posner and Robinson2016). Additionally, our observational data, which confirm many of the findings from the interviews, overcomes the challenges of interviewer effects because of the relatively more passive nature of data collection. However, future and larger studies will need to build in additional protections against distortions introduced by co-ethnicity, including larger research teams, larger samples, and more attention to the shared and dissonant identities of interviewers and participants.

Lastly, as noted above, this project was largely exploratory in nature, and given the relatively small sample size, the data themselves are not very amenable to quantitative content analysis or coding. We hope that future, larger studies will allow for more splicing of the data into categories, and quantitative content analysis of responses, including coding and word frequency counts. We anticipate these approaches being especially fruitful in analysing interview data.

EXPLORING THE TRANSPORTABILITY OF THIS APPROACH: ACCRA, GHANA AND DELHI, INDIA

The proposed measures for social capital appear to receive support – after contextualisation – in other urban settings within developing societies. Exploratory observations by the authors in 2015 and 2016 provide anecdotal evidence that Accra, Ghana and Delhi, India are sites where similar trust-points, check-points and fast-lanes exist. Although in Accra the trust-points and check-points function akin to those of Gauteng, they are notable by their absence, thus indicating high levels of social trust. In contrast, check-points and trust-points in Delhi indicate variance in social expectations of trust across income groups, indicating low levels of overall social trust.

Accra's coffee shops and informal bars do not ask for the presentation of receipts before serving drinks. More strikingly, Accra's roadside restaurants and bars seldom ask for pre-payment. For example, behind a shopping centre in the Osu area of Accra, an informal restaurant set up in the evenings served Ivorian cuisine to clients seated at plastic tables placed on the unpaved road. The diners bought their drinks from a liquor shop across the lane, in what appeared to be one-half of a garage.

There was an intricate process of ordering the meal. Customers asked for the prices of the tilapia fish entrée, which depended on the size of fish, and then ordered the fish and side dishes such as Ivorian couscous and avocado salad. Subsequently, customers went to the liquor shop and ordered their drinks, which extended from local beer to Scotch whisky. The drinks and dinner were served, and customers paid in cash afterwards. There was no method or means of restricting customers’ exit from either the liquor store or the restaurant: even using the facilities required going to a nearby lane. The authors’ observations in similar establishments in other urban settings in Ghana, specifically Cape Coast and Tamale, suggest that such trust-points are widespread. In the University of Ghana in Accra, although a security infrastructure akin to that of South Africa's University of Pretoria and University of Witwatersrand exists, identification cards or registration in a log-book were not required and entry and exit were generally unhindered. The presence of security guards within the campus was also minimal.

In a more complex instantiation of the check-point, our cursory observations in Delhi, India revealed that the treatment of customers by employees is contingent on the income segments of the populations that the coffee shops target. Nevertheless, the number of check-points is similar across companies serving the same or distinct income groups.

In one popular coffee chain, where a standard cup of coffee cost approximately US$1: customers ordered a meal or beverage from a server, while seated at a table; were served the beverages at the table; the customer then asked for a bill, which was brought by the server; and, the customer paid the server. Thus, check-points where customers could be asked for their capacity to purchase beverages did not exist. Customers paid after consuming the products. Also, most transactions were in cash because the amounts were deemed too low for credit cards.

However, the transaction process differed in the middle segment of cafes where coffee cost approximately US$2. In a partially self-service system, customers ordered at the counter; they paid while ordering; customers then proceeded to occupy a table; and were subsequently served at their table either by face recognition, or internal memos indicating table numbers for servers, or even by signalling to servers. Here, a significant checkpoint – paying at the till – existed prior to the consumption of the beverage, but no proof of purchase was required to acquire the product. Also, unlike its cheaper counterpart, credit and debit cards were used more often in these cafes.

The higher-end coffee shops represented by popular international coffee chains, where a cup of coffee cost approximately US$3, followed a procedure similar to its self-service counterpart in the USA. Customers ordered at the counter; they paid while ordering; their names were announced; and they picked up their drinks. Patrons were not asked to produce a receipt. A significant number of customers used credit cards. Consequently, whereas customers were not treated equally in the different types of coffee shops, the numbers of check-points did not differ significantly. Customers at the full service, cheaper cafes were not asked to present proof of purchase before being served, but neither were their more expensive partial and self-service counterparts.

The benefits of this approach to studying social capital, as outlined above in the literature review, are that these contexts are comparable, within capital cities in developing countries, without the need for translation or for significant contextual knowledge of associational culture. The contexts of Accra and Delhi, as large urban areas in developing countries, prove fruitful comparative sights, in part because of the fact that they reveal the texture of social trust landscapes, the manifestations of institutional and individual prejudices, and the ways in which social trust varies. While not infinitely transportable, this approach to studying and measuring social trust allows for more comparison both within and between contexts.