Traditionally, political economy and economic sociology suggest that economic institutions and structures, such as markets, bureaucracies and class relations, operate 'identity-blind', so that any particular associations between such institutions and particular identities are contingent, that is, not a necessary consequence of the institutions and structures themselves but of other influences (Sayer 1995; 2000). However, this view has been contested by a significant amount of literature detailing the ways institutions and structures are not at all neutral with respect to identities (Smelser and Swedberg 1994; du Gay 1997; Hall 1997; Metcalf et al. 1996; Werbner 1984; Rafiq 1992; Basu 1995; Aldrich et al. 1986). This claim reflects a contemporary tendency to focus on the cultural embedding of economic and other institutions and structures (see Holton 1992 and du Gay 1997 for a general discussion).

To be sure, embedding of markets, bureaucracies and class relations can both be ‘identity-sensitive’ and ‘identity-neutral’ (Sayer 1995; 2000; O’Neill 1998; Keat 2000). So ethnicity may make a difference in terms of trust and shared meanings, norms and conventions. Markets, bureaucracies and class structures require trust, meanings, norms and conventions, and ethnicity can make a difference in deciding who to trust and belong to. Yet the social embedding of economic institutions and structures can occur for extra-ethnic reasons, and, as we shall see, as a consequence of economic and political marginality. While markets, bureaucracies and class relations are always ‘embedded’ in the cultural sphere, so that they are not just autonomous structures, they are not reducible to the cultural sphere (see Ray and Sayer 1999). It is important to retain the economic and cultural distinction in order to register how far economic and cultural problems and divisions are caused by identity-blind mechanisms, and how far they result from distinctions made by actors in the cultural sphere (Fraser 1995). As Sayer (2000) notes, it is far from progressive to ignore culturally generated problems (e.g. sexism and racism), but so too is the converse tendency of neglecting economic generated problems, or reducing the latter to the former (for example see Metcalf et al. 1996). One of the effects of the ‘cultural turn’ in social sciences is not only a marginalisation of the economic logic but also a reduction of it to the cultural logic. It is always a matter of an empirical enquiry to ascertain the extent to which ethnicity is significant in shaping economic practices. The paper examines the dynamics of petty capital, household and local authority in a multi-ethnic context by exploring retail market trading and the restaurant sector in Birmingham, UK.

While many writers (Werbner 1993; Ram 1994; Rafiq 1992; Basu 1995; Lyon and West 1995; Metcalf et al. 1996) state that ethnicity is relevant in determining petty commerce, and correlate ethnicities to petty capital, their accounts of the mechanisms remain unsatisfactory and unclear (Anthias 2001; Bradley 1996). Usually, many writers employ historical empirical research to establish how concrete microbusinesses develop and operate within identity-sensitive (or culturally-differentiated) contexts. Such studies attach significance to empirical associations, and are resistant to abstractions. They tend to assume that repeated instances of such associations indicate that the phenomena have to be related. For example, repeated transactions between buyers and suppliers of similar ethnicity is assumed to indicate greater trust (and hence competitiveness) than transactions between different ethnic groups. However, I suggest a form of analysis that seeks to make abstractions, identify casual properties, and explain the existence of social phenomena under contingent conditions (Sayer 1992; Archer 1995; Bhasker et al. 1998). Indeed, this paper seeks to penetrate beyond the simple description of phenomenal world and rhetoric of culturalism. An important consideration here, is the avoidance of new essentialist characterisations of ethnicity (Waldinger 1995; Woodward 1997; Bradley 1996).

In the next section, I shall describe the nature of the research field. The following two sections will detail some of the theoretical ideas (in particular the neutrality of markets, embedding and marginality) that inform the paper. Then, drawing upon my empirical work, I shall outline the nature of identity-neutral and sensitive mechanisms in minority ethnic commerce. Finally, I shall make some concluding remarks.

This paper on the minority ethnic petty commerce combines two distinctive studies on retail market traders and independent restaurateurs in Birmingham, UK. Interviewers were selected on the basis of the different types of role they played and according to differences of context. The restaurant research was based on detailed qualitative interviews with thirty-seven restaurants owned by migrants from five distinct ethnic origins – Bangladeshi, Pakistani and white (eight each), Indian (seven) and African-Caribbean (six). The inclusion of a variety of ethnic groups, including white, militated against the tendency of isolating one community from the rest of the small business population. A single group focus runs the risk of accentuating difference (Mulholland 1997). The independent restaurant sector was chosen because it has been a popular activity for minority ethnic businesses in Britain, and further afield (Bailey 1985; Kesteloot and Mistiaen 1997).

The owner-mangers of all 37 restaurants were interviewed. Moreover, in the overwhelming majority of cases, it was possible to interview other family members who were directly involved in the family businesses. Typically, this would be the spouse and children of the usually male owner of the restaurant. During these interviews, the role of family, the division of labour within the household, and aspirations of children were discussed. Each interview lasted around 90 minutes, and was fully transcribed. Furthermore, six regulators were selected from the Birmingham City Council who were responsible for regulating restaurants and the ‘Balti Quarter’. Also interviewed were three trade associations and two business support agencies.

In addition, eighty-six employees working in these restaurants were interviewed, focusing on the negotiated nature of employer-employee relationship. Each interview lasted approximately thirty minutes, and was fully transcribed. Subsequent analysis was based on the use of ‘Atlas Ti’ and SPSS software packages. Thirty-eight per cent of all interviews were with Bangladeshis (Pakistanis 23%, Indians 20%, African-Caribbeans 9.3%, and Anglo-Saxons and others 9.3%). In addition, only eleven per cent of the sample is female. The African-Caribbean females were two-thirds of the entire female sample. None of the female workers interviewed were South Asian in origin.

For the study of retail marketplaces, a ‘quasi-ethnographic’ approach was used, doing participant observation and unstructured interviews. This provided detailed, concrete and critical pictures of how market traders ‘made out’, and of the contexts in which they did so. Prior to formally beginning the research project, I worked for a year for several black market traders, doing participant observation as a pilot study for my research. This enabled me to get an insider’s view of market trading. The interviews were on site, in a ‘natural’ setting to maximise information flow to unearth the substantial and causal relationships and processes.

The people chosen to observe and interview were not randomly selected but were selected for likely theoretical interest; e.g. according to factors such as market traders’ ethnicity, gender and age, and product niche. However, I felt confident that I had interviewed sufficient traders in sufficient depth to have built up a representative picture of market trading. The ‘sample’ had 41 market traders, of which six were African-Caribbeans, one was West African, two were Indians, one was Pakistani, and one was East African Indian. Seven of the traders were women, of which one was African-Caribbean, one was West African and one was Indian. The overwhelming majority were white males.

Furthermore, a total of thirty regulators were selected ranging council officers from the planning and architecture, finance, environmental and commercial services departments of the Birmingham City Council to the Ministry of Agriculture, Fisheries and Food who were responsible for horticultural wholesale markets. In addition, I attended 15 markets department committee meetings and 12 market traders’ local branch meetings.

Discussions in political philosophy have long focused on the alleged neutrality of markets (Miller 1989; Offe 1984), on their tendency to operate regardless of people’s identities, or, indeed, regardless of their reasons for acting as they do. Actors’ intentions belong to the cultural sphere, but, in an important respect, markets operate independently of actors’ reasons for buying and selling. Whereas, in many other circumstances (such as family negotiation and political lobbying) actors and organisations have to justify to others what they want to do, actors and organisations do not have to do so in markets: markets are ‘reason-blind’ and judgements can be treated as mere preferences (Keat 1994). As Offe (1984:82) notes, an essential feature of the markets is that they neutralise meaning as a criterion of production and distribution.

Turning from indifference to reasons to indifference to identities, competitive markets can encourage either behaviour which is neutral with respect to identity, or behaviour which appears to be highly sensitive to it, though in the latter case it is not identity but the money or goods associated with it that count (Hirschman 1982; Simmel 1955; Sayer 1995). In the abstract, the neutrality of markets suggests that ethnicity should or will not make a difference because the market mechanism operates through unintended consequences of market choices, success may have nothing to do with merit or fairness, and much to do with luck (Steele 1992). Indeed, the logic of the market is that all that matters is what is offered for sale and its price. To be sure, the social characteristics of actors are irrelevant, since in advanced social division of labour markets operate largely under conditions of ignorance of how products are produced and sold, and who works and trades for whom. Furthermore, money is abstract and neutral - one person’s money is as good as the next’s. Moreover, as Sayer (1995) notes, under competitive conditions actors have an incentive to be neutral for fear of losing sales and bargains to others.

For instance, as Sayer (2000) notes, in the case of labour markets, competitive pressures encourage firms to employ whoever will do the job best, regardless of matters of identity, such as gender or ethnicity, and they penalise those who forego the best candidates on such grounds, though of course such incentives may be contingently overridden by racism and sexism (Peck 1996; Ram 1992; Walby 1986). At the same time, competitive pressures compel firms to take advantage of any differences in identity of groups that have economic implications as when employers, needing to cut labour costs, employ workers from a culturally-stigmatised group (e.g. immigrants, young blacks men and Asian women) that can be paid low wages (Phizacklea 1990; Jones et al, 1994). However, note the logic of this kind of selectivity: market and competitive pressures encourage employers to take them on not because of their cultural identity, but because of their economic exploitability (Sayer 2000). The latter may in turn be a consequence of stigmatisation of their identity (Ram 1992; Peck 1996), but from the point of view of market actors and market incentives, it is the economic consequences (such as costs and productivity) that matter. For example, in the sample of restaurant workers, a significant minority were migrants from South Asia, and they possessed a few recognisable skills. Only seventeen per cent of the respondents spoke English as their first language. Once stigmatised as ‘low-skilled immigrants’, they were readily exploited (i.e. long hours and poor pay and benefits) by their co-ethnic restaurateurs (see Ram et al. 2001).

Accounts of social embedding of markets, bureaucracies and class relations refer either or both to ‘necessary impurities’ (e.g. non-market, bureaucratic and informal practices, and regulation), which support the institutions and structures (see Hodgson 1988; 1999; Sayer 1995; White 1993), or to cultural meanings and identities constituting economic practices, rules and conventions (see du Gay 1997).

Markets and bureaucracies are necessarily socially embedded, and so they can never be ‘pure’ (see Hodgson 1988). Markets need non-market forms of organisation such as state regulation to make them sustainable (Christopherson 1993; Wrigley and Lowe 1996; Boyer 1997). Bureaucracies need some degree of support from other forms of organisation, particularly in terms of ad hoc action and interpersonal relations among members, if they are to function effectively. For instance, family members subsidise microbusinesses (Baines and Wheelock 1998; Song 1997). In each case, the second form of institution supports rather than undermines the first dominant form. The 'impurity principle' exposes some of the limitations of pure markets and bureaucracies.

While markets and bureaucracies may be largely identity-blind, they are embedded in the cultural sphere that is certainly not. For instance, if white Anglo consumers stop buying Indian takeaways because they prefer Italian pizzas, this is a case of identity-blind cause impacting on a culturally-differentiated context to produce effects, which are not culturally neutral. Markets and bureaucracies are embedded in the cultural sphere - the way in which market actions, class behaviour and identities develop is always in and through gender, ethnicity, age and sexuality (du Gay 1997, Slater and Tonkiss 2001). However, just because markets and bureaucracies are always culturally-embedded, it does not follow that their behaviour is wholly reducible to these cultural dimensions.

To be sure, markets and bureaucracies have a logic and a momentum of their own that go beyond the subjective experience of actors. While economic institutions depend on actions and to some extent the understandings of knowledgeable actors, markets and bureaucracies rountinise, formalise and govern actions through specific signals and rules, such as prices, money, management and administrative rules and procedures (Sayer, D., 1991). Even though these usually have to be interpreted by actors, the way the economic institutions operate is in varying degrees independent of their intentions and understandings, and disconnected from norms and values (Keat 2000; Sayer 1995; 2000; O'Neill 1998).

For instance, a labour contract requires both an exchange of property rights and an agreed price, but it is accidental whether the parties share the same ethnicity, or have similar beliefs or ideas on cultural practices. An Indian employer will not necessarily question her worker’s cultural beliefs, though she will want to know his reasons for working so as to ensure labour exploitability and compatibility, and to protect herself from labour shirking and other ‘moral hazards’. In this respect, social embedding at the workplace is driven by economic imperatives and not by cultural and moral considerations.

To sum up this section: to the extent that markets and bureaucracies are uncoupled from the culturally-differentiated contexts, they can be argued to be identity-neutral. However, it is this contention that has come under fire in recent years, particularly in relation to labour and product markets and capital structure. It is to these issues that I will address myself to next.

Recent empirical studies (such as Metacalf et al. 1996; Basu 1995; Lyon and West 1995; McLeod 1991; Ram et al. 2000a; 2001) have argued that markets and microbusinesses are in fact far from neutral with respect to ethnicity, and have uncovered specific ethnic experiences and interests marking out distinctive ethnic groups, such as Hindu Gujartis, Muslim Bangladeshis, Muslim Pakistani Kashmiris, African-Caribbeans and Indian Sikhs. This is consistent with the tendency of ‘the cultural turn’ in social sciences to de-emphasise economic structures, and to focus on the cultural agency as the dominant source of social inequalities - thereby inverting the pattern of emphasis in earlier political economic accounts (see Ray and Sayer (1999) for a general discussion).

Yet, in an important way, my studies highlight how microbusinesses are structured such that the owner’s ethnic identity is insignificant in shaping competitive advantage. Microbusinesses are located in industrial sectors that are neutral to the actors’ reasons and moral conduct. In making the claim of a cultural and ethnic style of doing business, researchers and writers (such as Werbner 1993; Rafiq 1992) may unwittingly produce new forms of cultural essentialism and hence racism (Bradley 1996; Waldinger, 1995). Such studies tend to overlook other forms of stratification, in particular class and gender.

In the case of microbusinesses, it is tempting to think of the object of study as being characterised by ‘marginality’. Yet this could refer to three different things:

• petty capital’s economic position is peripheral to the industrial sector, which is largely shaped and dominated by large capitalist organisations (see Guy 1994; Wrigley and Lowe 1996);
• owners’ income is low and insecure, and it depends on the competitiveness of their businesses and on other sources of income;
• owners are disadvantaged in a wider environment in relation to cultural and political powers. In shaping communities, identity and values are responsible for creating inclusions and exclusions.

Importantly, these sources of the inequalities and domination (i.e. competitive industrialism, market coordination, and ethnicity - see Sayer, 1995 for a general discussion) are different. Their logics can be separated into two types: identity-sensitive and identity-insensitive (Sayer 1995; see also Fraser 1995). Whereas the first two sources of inequalities (competitive capitalism and market coordination) are indifferent to identity, only the third source requires the recognition of identity and ethnic differences. Yet, how do market and ethnicity interrelate?

For the first and second sources of inequalities the social characteristics of actors are irrelevant, since in advanced economies markets operate largely under conditions of ignorance of how products are produced and sold, and who works and trades for who. Most importantly, money is abstract and neutral, and under competitive conditions, actors will be penalised for discriminating (Steele 1992).

While some black people may start up a business because of racism in the labour market and exploit racialised resources (cf. Ram 1992), this does not mean that ethnicity makes a significant difference to how, as businesspeople, they behave. This only suggests that business is affected by inequalities and marginality produced elsewhere. Once people are employed in a competitive business and operate in a market, productivity and price become crucial. Clearly, how and in what ways actors and resources come into sub-markets do not tell us how they will be actually used, since the nature of the product and type of retail outlet make a difference.

To emphasise, it is contingent how the three different kinds of domination and inequalities interconnect. These three kinds of marginality need not go together. A marginal economic position may provide an income that is not marginal for an individual who is not culturally marginalised. Thus, nature of ethnic embedding is largely accidental and contingent (see also Anthias 2001).

It is worth noting that, as Sayer (2000) points out, many writers and thinkers researching into cultural embeddedness of economic institutions and structures employ ‘associational thinking’, which attaches great significance to associations between social phenomena (say, gender and bureaucracies, or ethnicity and class). Such an approach tends to assume that repeated instances of such associations indicate that the phenomena have to be related thus, without asking counterfactual questions about the status of these relationships; say, questioning whether markets are necessarily racialised or ethnicised, or only contingently so. This inevitably leads researchers and writers to wrongly attribute concrete events to social identity that are actually produced by other social structures (see Sayer 1992 for a general discussion). It is a common and easy mistake when researchers and writers fail to pursue qualitative and counterfactual questions after undertaking empirical research surveys involving quantitative analysis and ‘associational thinking’. The latter often lead to mistaken casual inference, which ‘explain’ variations in culturalist terms. My studies point to ‘uniformity in diversity’ (Martin, 1994).

I shall draw upon my studies on retail market trading and restaurant business to detail how social structures can both be identity-neutral and identity-sensitive, and they may occur for both economic and non-economic reasons. In particular, the nature of the social relations may be a consequence of economic marginality and the nature of the product and retailing.

Recent studies on minority ethnic businesses have been keen to move away from an emphasis of black employment as an object of racism and discrimination to one in which co-ethnic ties can facilitate social networking, harmonious work relations and, in particular, access to employment opportunities. Nevertheless, market and competitive pressures encourage employers to take on workers not because of their ethnic identity, but because of their economic exploitability (cf. Phizacklea, 1990). For instance, the market traders use workers' position in the secondary labour market and their ambiguous position in relation to the welfare state to further extract unpaid labour time. Similarly, spouses are also vulnerable to economic exploitation. Often, casual workers are employed on a daily basis, working on Fridays and Saturdays and during Christmas, and are paid £15-20 for a day’s work!

The degree of which employers and workers belong to the same ethnicity in retail market trading and restaurant business is limited and largely accidental. The nature of work in market trading is rather simple and repetitive, and much is learnt by doing. Consequently, different ethnicities and languages in the workplace do not pose any serious difficulties in coordinating work. In fact, a single language (Bengali, Urdu, Hindi or English) usually dominates the workplace, though as often as not this is mixed with other languages.

In restaurants, the production of cuisine involves a particular set of shared meanings between workers, managers and customers. In complex tasks, what is required, or thought to be required, cannot be readily articulated and communicated in meaningful ways and according to a set of instructions. Indeed, tacit understanding, inarticulate information and misunderstandings are as very much part of the workplace as rules and instructions (see Fine 1994). To be sure, ethnic cuisine cannot be produced and translated into a readily set of ideas and expressions. Skills and experience are learnt ,and acquired on the job and through the guidance of senior cooks and owners (see Ram et al. 2000b). To minimise misunderstandings and to avoid costly fault lines, co-ethnic workers are sometimes in a better position to produce aesthetic and tasty works. In part, the workers and owners share the cuisine and the language from birth, so facilitating communication and understanding to accomplish the tasks. However, there are two limits to this type of co-ethnic workforce. First, while cooks, some assistant cooks and owners may share ethnicity, those not involved in the cooking (such as waiters and dishwashers) can be of different ethnicity. Second, when the production of the ethnic cuisine has become standardised and widely known, and its skills can be acquired through learning by doing on the job, different ethnicities can participate in the production and expression of ethnic cuisine. Consequently, restaurateurs are less dependent on co-ethnic relationships: often, Indian, Pakistani and Bangladeshi restaurants consist of workers of several South Asian ethnicities - in one Indian restaurant, there were workers from Nepal, Yemen, Bangladesh and India as well as of second-generation British Asians.

Significantly, many workers regard self-employment as an opportunity to escape the employment relationship (Phizacklea and Ram 1996), and provides a better alternative to causal and low-pay work (see Ram et al. 2001). In this respect, self-employment offers a measure of social prestige, income and lifestyle enjoyed by the salaried and professional middle-class. In our interviews with restaurant workers, thirty-eight per cent had at some time contemplated starting their own restaurant, or had actually started one before returning to employee status.

In market trading, the selling techniques are influenced by the unusual nature of market trading. The sales pitch and the stall display are managed in various ways that reflect some of the differences between market traders in terms of whether they have indoor or outdoor stall and its location in the marketplace, whether they are new or established traders and the nature and size of their business, and the degree of competition between traders. Invariably, market traders have to treat customers in a civilised way, and to sell reasonably fair quality to get repeat trade, so as to build up customer loyalty (Simmel, 1955). For example, Paul Cool (a white Anglo trader) commented that most of his customers were African-Caribbeans, and they were very selective, and would touch and bruise the produce whilst checking their quality. This annoyed him, but he said that it was important not to shout at them. Rather he had to put on a smile and just think of the money that he would take from them.

Waiter’s ethnicity may become a significant part of the production of ethnic cuisine for two reasons. First, their ethnicity may be important so as to mediate information between customers and cooks. Second, their ethnicity fits with the restaurateurs’ concern with the expressive and sensory qualities of the product and its production – after all, customers do usually ask themselves whether the dishes, waiters and ambience are authentically ethnic. (Of course, the sexuality of the waiting staff is also important.) However, as with the cooking staff, there are several ethnicities that can, and do, easily satisfy and perform the informational and expressive qualities of waiting. In short, as with the cooking staff, market concerns encourage employers to employ the waiting staff not because of their cultural identity, but because of their economic exploitability.

The marginal nature of retail petty trading and the unsophisticated nature of small-scale production of ethnic cuisine mean that market traders and restaurateurs can only realistically have arm’s length relationships with their suppliers. In such cases, the shared information between buyer-supplier is simple, and does not require a high level of trust for economic activity to take place. In other words, what Hodgson (1988) refers to as ‘non-market exchange’ is minimal, and economic embedding is weak (cf. Granovetter, 1973), so that transactions are influenced largely by price.

None of the traders and restaurateurs suggested that they faced open or blatant racial discrimination from their suppliers. This was partly because most of the traders had established businesses, and this meant that their trade was valuable to their suppliers to keep hold of them. For example, Mr. Patel (an East African Indian wholesaler) said that whether retailers and market traders were white or black made no difference. His only concern was to look after his customers, because there were four other major wholesalers keen to take his business. For him, people were only different in the size and type of outlet they had and the kind of quality they wanted, and the price and quality of product reflected that difference.

To sum up so far, the cultural embedding of market relations is weak and minimal, and occurs largely for economic purposes (i.e. to boost sales and reduce costs). Market transactions prioritise economic considerations, and more so in an intensely competitive environment. In such a context, there is little scope or reason for strong social and cultural embedding between the market actors (cp. Werbner, 1984; Ram, 1994). In effect, market actors are blind to each other's ethnicity.

In the next two sub-sections social embedding takes a stronger form, and sometimes for non-economic considerations.

Microbusinesses, such as those of market traders and restaurateurs, are embedded in social relations, and in particular those involving household members (Baines and Wheelock 1998; Ram et al. forthcoming). The domestic context within which they reside is important for assessing whether they make enough to satisfy their family needs. In combining the nature of household as a place of individual life course cycles and the nature of petty commerce as an economically marginal activity invariably means that market trading and restaurant business are especially vulnerable to intra-household interests, conflicts and inequalities (Baines and Wheelock 1998; Song 1997; Kay 1990; Baxter 1988). For example, some restaurateurs suggested how their private family tensions made working relationships awkward and uneasy.

How do microbusinesses manage to start up and survive in a highly competitive environment? In part, most businesses rely on family members to subsidise their business (Scase and Goffee 1987). Often, in economic marginal businesses husbands and wives work together in the restaurants, but it is not uncommon for spouses to have full or part-time employment elsewhere. Such non-restaurant employment is crucial to sustaining businesses and the household. For example, one restaurateur commented on the role of his wife,

Most microbusinesses are not a typical ‘family business’ unit since they do not absorb all the family members, who are usually involved in their own work – children are in education or ‘respectable’ employment, wives are in formal employment and/or unpaid domestic work (Mulholland 1997). Indeed, many owners hope that their children would perform well educationally and aim for ‘professional’ careers. While many petty traders and producers do not highly rate their business, as something to achieve for, they feel that it is an adequate fall-back position should the children fail to ‘break-out’ into other directions. For example, one respondent said,

The family involvement is significant though fragile, and this is a common feature for all the traders and restaurateurs. In my studies, no ethnic group had a greater access to family resources over other ethnic groups (cp. Werbner 1984; Metcalf et al. 1996). Rather, the extent of the family involvement depended on the marginal nature of the retail business, and many children were resistant to going into the business, eager instead to acquire cultural assets (cf. Bourdieu 1993; Mulholland 1997).

The nature of regulatory embeddedness of the petty commerce involves an on-going process of negotiation with others, a process that takes the form of non-market exchanges. Indeed, the regulatory system is best perceived as incomplete and partially inconsistent, in which the legal framework and regulations provide an incomplete set of instructions and incentives to those supposed to implement them, leaving considerable room for judgement and discretion (Mackintosh 1997). For sure, the concepts of intent and deliberation still remain central, yet the context and complexity of human judgement matter, and these qualify regulation in practice (Clark 1992; Wrigley and Lowe 1996). For example, there were restrictions on the numbers of different types of trade-lines (such as jeans, flowers, childrenwear, and fruit and vegetable) allowed in marketplaces so as to ensure a balance of trade-lines. Market regulators (the city council markets department) aimed to make marketplaces attractive places to shop. This usually meant there would be a quota of traders in any one trade-line. However, if the regulator felt that there was sufficient profit in a certain trade-line to allow another trader, then an entrant newcomer was allowed to trade - though this was never popular with the existing traders. In one incident, a black fruit and vegetable trader objected to a new white rival. She accused the market owner and regulator of racism and corruption. However, the owner (the council markets department) denied these charges, arguing that the trade was sufficiently robust to allow for another trader. In fact, the regulator was more interested in obtaining the extra rent that the new trader had to pay.

In part, as a consequence of the nature of the regulatory system, there was room for racism in form of promoting white businesses or crippling black commerce. Yet, I found no clear evidence of this. Rather, council regulators were governed by extra-ethnic factors. While market regulators have a dual role of disciplining regulatees and representing and protecting their interests, they also operate within their regulatory system (Clark and Stewart 1994; Mayston 1993; Ascher 1987). In other words, there are regulations that govern the regulatory body (i.e. ‘second-tier regulations’). Usually, these regulations are a mixture of contracts, competition, status and trust and professionalism (see Mackintosh 1997; Mayston 1993).

In the context where most actors are economically, and sometimes politically, marginalised, this regulatory process involves a struggle for economic survival and political legitimacy. In the case of market trading, the council markets department was in a political and social marginal position within the council hierarchy that only further undermined the traders’ capabilities to compete and survive.

To sum up this whole section: the partial autonomy of the class structure from the identity of those involved in the class formation provides an important reason for retaining a structural concept of class. Here, black and white market traders and restaurateurs as a sub-category of the petty bourgeoisie class were affected by internal and external relations as a social collectivity. What affected their earnings had little to do with their cultural identity. Rather, they occupied a structurally marginal position in the sub-retail sector, and this exposed them to intense competitive pressures (Scase and Goffee, 1987; Song, 1997). Furthermore, as a marginal sub-category of the capital class, they welded little influence in urban politics.

Hodgson (1988), Sayer and Walker (1992), Sayer (1995) and Boyer (1997) argue that markets and bureaucracies cannot create their own pre-conditions and can even undercut their own institutional foundations. Some measure of embeddedness is necessary if the market and bureaucracies are to operate, for too much price flexibility and administrative rigidity lead to instability. Economic embedding involves both ‘market’ and ‘non-market’ structures, and it can occur for economic and non-economic reasons. In some ways, markets and class relations are neutral with respect to ethnicity and other differences, while in other ways, particularly to do with hiring workers and targeting customers, they are highly sensitive to them, though for reasons of their economic exploitability (Sayer 2000).

While markets, businesses and class relations are always ‘embedded’ in the cultural sphere, they are not reducible to the cultural sphere. It is important to retain the economic and cultural distinction in order to register how far economic and cultural problems and divisions are caused by identity-blind mechanisms and how far they result from distinctions made by actors in the cultural sphere (Fraser, 1995). So ethnicity may make a difference in terms of trust and shared meanings, norms and conventions. Yet, in the case of microbusinesses, products are largely familiar items produced through learning by doing and traded at arm’s length under competitive and regulated conditions, so that trust is not necessarily imputed to social characteristics of actors, but rather to the social milieu of the market and the workplace. Nor is there need for much trust and shared meanings in uncomplicated production and simple market transactions. In addition, as a consequence of the advanced social division of labour, consumers are unlikely to know, or care, about the owners and workers’ ethnicity. Indeed, consumers are just as likely to mis-identify them.

Social embedding of some economic institutions and structures can occur for non-economic reasons (such as racism and sexism), and as a consequence of economic and political marginality. Yet, while some black people may start up a business because of racism in the labour market (Ram 1992), this does not mean that ethnicity makes a significant difference to how as business people they behave. How and in what ways actors and resources come into sub-markets does not tell us how they will be actually used since the nature of the product and type of retail outlet make a difference. As a sub-category of the capital class, the self-employeds and small employers act as a social collectivity (Bechoffer and Elliot 1981). They are especially vulnerable to the vagaries of the market and the fragility of market regulation. They are prone to failure because of family problems and inadequate domestic embedding. They have a few property assets that can be transmitted to the next generation, though the children are more likely to pursue an education.

The research project on market traders was funded by the Economic and Social Research Council (studentship award R00429334303). The research project on restaurants was funded by the Economic and Social Research Council (award no. L130241049), Birmingham City Council, Birmingham Partnership for Change, Birmingham Enterprise Link and Birmingham and Solihull Training and Enterprise Council. I am grateful to Bob Carter, Wendy Olsen and Ron Vannelli for their helpful comments on an earlier draft of this paper, also to anonymous referees for their constructive comments and advice. Many thanks to Monder Ram, Tahir Abbas and Gerald Barlow for their contribution to the study on restaurants. Finally, my gratitude to the traders, restaurateurs and regulators for their time, guidance and patience. The usual disclaimers apply.