Patriarchal Investments: Marriage, Dowry and the Political Economy of Development in Bangladesh

ABSTRACT This article explores why dowry inflation persists in Bangladesh, despite the country being widely heralded as a development success, especially with regard to gender. The article asks three questions. Does rural Bangladesh show changing patterns of marriage similar to those reported elsewhere in South Asia and more broadly? What might explain the persistent spread and inflation of dowry payments? How might changes in marriage and the inflation of dowry be related to the broader political economy of development in Bangladesh? Analysis of primary data from rural Bangladesh affirms shifts in norms of marriage arrangement and conjugal relationships, but also emphasises considerable continuity. The political economy of development in Bangladesh is briefly described. Dowry is argued to function not as compensation for perceived weakness in women’s economic contribution, but to bolster men’s. In mobilising “additional” resources, dowries help sustain the economic system and indicate ongoing commitment to cultural idioms of masculine provision and protection, against a background of widespread corruption and political and gender violence.

It has long been recognised that economic modernisation is associated with changes in family and household structures and the ways that personal identity is construed (see, for example, Engels 2004Engels [1884 ;Stone 1990;Carsten 2004). In accordance with this, contemporary studies across the world identify apparently similar tendencies towards more nucleated household structures; growing stress on individual agency and personal choice in narratives of how marriages occur; and greater emphasis on and expectations of the conjugal relationship. 1 Interpretations vary as to how radical these changes really are, and whether they represent liberation for women or enrolment in "new structures of domination" (Abu Lughod 1990, 52). As seen below, these issues have been extensively discussed in South Asia and especially India. What is unusual in the case of South Asia however, is that this relatively upbeat, agencyfocused literature on affinity exists in parallel to a shadow-side literature on dowry which stresses immiserisation, coercive demands and many forms of violence against women.
Conventional modernist scripts would tend to explain this duality in terms of a tension between modern trends towards increased agency and conjugality in marriage and "traditional" inertia in relinquishing dowry practice. In fact, however, the rapid expansion and inflation of dowry gifting is itself associated with modernisation. Traditionally dowry was practised only amongst high caste Hindus, but it is now found throughout India, Pakistan and Bangladesh, across almost all communities and social groups. With one example provocatively termed "bridegroom-price" by Caplan (1984), contemporary forms of dowry are typically distinguished by the high levels of payment involved, the fact that the amount to be given is integral to marriage negotiations, and the fact that the transfer goes directly to the groom or his family. While the picture varies somewhat by geography and community, it is clear that this is largely a post-colonial phenomenon, emergent in the 1950s and 1960s but gathering pace since the 1970s and 1980s. An early commentator noted how the consumer goods transferred symbolised modern masculinity, with the motorbike as an emblematic example (Lindenbaum 1981). Materially, dowries have now become increasingly central to young men's ability to establish themselves in the modern economy, providing capital for business or the cash or connections required for career advancement, including through international migration (Bannerjee 2002;Tenhunen 2007;Rozario 2009;Rao 2012).
This article discusses developments in marriage in rural Bangladesh against the background of these apparently contradictory trends. First, and most straightforwardly, it reviews the evidence on changing patterns of marriage in South Asia and explores the extent to which similar tendencies are shown in rural Bangladesh. Second, more ambitiously, it asks what might explain the persistent spread and inflation of dowry payments. Third, in particular it seeks to explore how changes in marriage and the inflation of dowry may be related to the broader political economy of development in Bangladesh.
The field research was conducted in 2006 in two villages, one in northwest Bangladesh (Dinajpur district) and the other more central (Manikganj district). The locations were chosen as part of a much larger study of 1,500 households, which sought to explore differences in levels of well-being across six rural and urban sites. 2 This larger project provided important background material on the two villages, and also meant that the research team was already familiar to local people. However, the small numbers involved in the sub-project reported here make it dubious to generalise about differences by site. I therefore differentiate respondents by religion, economic status, gender and age, but not by location. 3 The sub-project involved individual interviews with 70 respondents and 16 focus group discussions. Of the 70 respondents, 58 were from couples in which husband and wife were interviewed separately; ten were elderly people who provided individual life histories; and two were focused on religion and family life. Checklists of key issues were drawn up, which local researchers used in conducting a first round of interviews. They produced transcripts in Bengali and English. I undertook content analysis on these, and sent the team my analysis and reflections. The research team then met to discuss the findings. This discussion produced a further checklist for a second round of interviews. While most people were interviewed once, 20 of the couples were revisited with the more targeted set of questions.
Changing Forms of Marriage in South Asia Giddens (1992) provides a mythic frame for studies of marriage and modernity with his idea of the "pure relationship" of late modernity, in which the old compulsions to stay together because of economic dependence or fear of scandal are no more. The purity he talks of is pure utility, where the only binding is the satisfaction of the partners based on their mutual self-disclosure. When the relationship is no longer providing what they seek, it can be dissolved and new more fruitful partnerships formed. As a depiction of the reality of relationships this is easily critiqued (see Jamieson 1999). Its interest lies rather in its ideological character, conjuring a relationships market, engaged in by individual consumer/agents seeking to make and re-make "the reflexive project of the self" (Giddens 1991, 5). Twenty years on, the rise of internet dating and mobile applications such as "Tinder" offer a material instantiation of just such a vision. This does not, of course, make it any less ideological, but indicates its power to beguile.
A critical counterpoint is presented by Collier's (1997) reflection on her fieldwork in Andalucía (Spain) across 20 years. Collier acknowledges marked change in the ways people describe their actions and relationships. Whereas in the 1960s people tended to imply "a disjunction between internal desire and outward action," in the 1980s, "people tended to portray actions as reflecting the actors' intentions" (Collier 1997, 6). While patterns of courtship and marriage had indeed shifted, Collier doubts whether these in reality reflected the radical expansion of personal agency that the rhetoric would suggest. Instead she links the narratives to the broader socio-economic context, in which all her respondents "knewat least on some levelthat market processes systematically disadvantaged people like them" (Collier 1997, 52). Narratives of greater agency in their personal lives mirrored a growing stress on individual merit and personal choice in their explanations of economic inequality. These reflected the dominant ideologies which legitimated unequal accumulation and diverted respondents from conscious recognition of the structural constraints they faced in engaging in the changing economy.
Social anthropology on marriage in South Asia occupies the mid-space between Giddens and Collier. It recognises, though tends not to investigate in depth, the association of economic change with shifts in marriage discourse and practice, and in most cases remarks growing stress on the agency of young people and expectations of marital intimacy. For some communities which have historically been marginal, however, changing practices may mean less room for manoeuvre for young people rather than more. Kapila (2004, 382) describes how the Gaddis, a pastoral people in Himachal Pradesh (North India), are indeed increasingly stressing conjugality, but that this means more restrictions on sexual expression, especially for women). The changes in part reflect state legislation, and in part social aspirations which involve becoming more mainstream, more middle class, and more Hindu. Parry (2001) makes some similar observations of industrial workers in Chhattisgarh. While he believes that there is nothing new in hopes of intimacy in marriage, especially amongst women, what seems to him to have changed is that amongst the better off working class these have moved from an alternative to an authorised discourse, to which men also increasingly subscribe (Parry 2001, 816). Marriages for this group of workers seem more stable than in the past, with less scope for the secondary relationships in which both men and women historically enjoyed considerable flexibility. The stress on romance does not then lead to Giddens' pure pragmatism, but to a belief in the "indissolubility of young love" (Parry 2001, 817). This point of Parry's is important, because it shows that modern Indian myths of marriage as once-for-all romance are very different to the mix of idealism and utility that Giddens puts forward in his idea of "the pure relationship." Abeyasekera (2013) reinforces Parry's comment, in her observation that while middle-class young people in Sri Lanka are encouraged to find their own marriage partner, there is great pressure on them to "choose right," first time and swiftly, as especially girls' reputations will suffer if they are seen to have "got friendly" with more than one boy. Marriage videos constructed on a motif of Bollywood romance are laced with implications of sexual interest between the couple that would have been unthinkable a generation ago. This may indeed appear a kind of liberation where young people's sexual behaviourafter as well as before marriagehas historically been significantly controlled by the older generation (see Fuller and Narasimhan 2008). It is, however, at least as deeply inscribed with gender difference as the earlier models of proper relationship, and very far indeed from any feminist-inspired vision of gender equality. It is also important to recognise, as the "offstage" songs and stories reported by Raheja and Gold (1994) attest, that Indian women's sexual awareness and expression are not new creations of the Bollywood age. What has changed is that they may now be more openly displayed, if generally in somewhat coy form. In both cases, of course, as Abu Lughod (1988) argues in her study of women's songs and poetry in Bedouin society, these are cultural productions, not simple revelations of women's "true feelings." And the stage on which they are produced significantly shapes the form that they take. The frank vulgarity of village women's everyday joking in India and Bangladesh as well as some of the songs and stories that Raheja and Gold present, is very different from the authorised, sanitised desiring for one's true love depicted in the wedding video.
The most obvious challenge to marriage as "pure relationship" in South Asia, however, is its importance to family strategies for class and/or caste advancement. This is a central theme in the Osella and Osella's (2000) study of the low-caste Hindu Izhava community in Kerala. Large dowries and elaborate wedding celebrations provide a means of both demonstrating and achieving the status to which they aspire. In later publications they note also the changing form of weddings in Kerala from the couple as bearers of a ritual into a "filmi" spectacle or show (Osella 2012, see also Kendall 1996, on Korea, andAbeyasekera 2013, on Sri Lanka). Unusually in this literature, the Osellas draw attention to the ways that the new emphasis on romance and conjugal relationships is crowding out established patterns of same-sex sociality and queer sexualities.
Marriage for migration offers a striking example of its strategic use for collective advancement. Gardner (1995Gardner ( , 2012 demonstrates how significant are alliances with Bangladeshis in the UK in terms of the social and economic status of individual families and the development of Sylhet as a district. 4 For British Bangladeshis, however, having a marriage arranged with a villager from Bangladesh may be viewed with some alarm. Rozario (2012) interviewed young, urban, middle class and educated Muslim Bangladeshis in Dhaka and London and found in both places that wives laid great stress on "respect" in terms of their expectations of husbands' attitudes to them. She also notes the importance of security, in terms of having a reliable and trustworthy partner, which young men and women believe to be more likely if they marry someone who is serious about Islam. Seeking such a partner was also one means that young UK citizens could use to evade a cross-national marriage without rupturing relationships with the older generation.
Changing norms of marriage arrangement, which occupy a pivotal place in this literature, clearly demonstrate the continued collective investment in marriage. Ahearn (2001) analyses love letters in Nepal, and considers how literacy combines with changing discourses of development and nationalism to give young people new notions of agency, including in love. This is nonetheless still cast in patriarchal terms, in which young women's consent must be wrested from them, but, Ahearn says, at least their consent is required (Ahearn 2001, 250). Many authors contest the opposition of "love marriage" and "arranged marriage," pointing instead to similarities or hybrid forms between them. Donner (2002) is one of these, writing of middle-class Bengali households in Calcutta. She stresses that marriages of all kinds are major sources of anxiety (Donner 2002, 85) and that whatever their form "the perfect match never seems to materialise" (Donner 2002, 94). Most women, she says, consider the best way of managing love marriages is by treating them as arranged marriages. This is only a problem if the match crosses the accepted rules of appropriate partnering (Donner 2002, 88). Grover (2009) is an exception in this predominantly middle-class literature, as she researched people living in slums in Delhi. Inverting the conventional wisdom, she finds that for this population the lack of kin support amongst people who have made love marriages actually makes their marriages more stable. Where marriages are arranged in the neighbourhood, women can easily take refuge with their natal families if things get tough and so may never develop a strong presence in their marital household. If nearby, the woman's natal family is also more likely to get involved if there are disputes in the marital family, which can mean the difficulties escalate. 5 Caplan (1998) discusses Anglo-Indians who have historically been identified by themselves and others as choosing their own marriage partners. Even so, he states, it was important that the match be suitable and pre-nuptial contact was supervised by elders (Caplan 1998, 22). Indeed, the involvement of elders in marriage arrangement may be becoming greater, he believes, due to a combination of dependence on elders' networks with diaspora settlement and the financial pressures of the "competitive wedding scene" (Caplan 1998, 24).
For South Indian Brahmans, Fuller and Narasimhan (2008) find similarly that companionate marriage is the "modern ideal," and that this includes emotional satisfaction. However, they point out that this is very different from "love marriage" as it involves collaboration between parents and children and a strong belief that "matching" of backgroundsincluding ideally caste endogamyis most likely to produce marital happiness. The result, they emphasise, is very important in the reproduction not only of caste, but also of the (still largely upper caste) middle class (Fuller and Narasimhan 2008, 752). While acknowledging changes in expectations of marriage and love, Osella (2012, 242) stresses the considerable continuity between past and present in "the many connections which people do clearly expect, tacitly or explicitly, between 'economic well-being' and a 'happy family life'. She also argues the need to go beyond the love marriage/arranged marriage distinction not only empirically, but also analytically, "since all marriages across all social classes involve a mix of practical-pragmatic, economic and affectual-passionate considerations and forces" (Osella 2012, 244). As described below, this resonates strongly with the evidence from rural Bangladesh.

Dowry in South Asia
Where dowry appears in the marriage literature, it tends to be represented in relatively positive and voluntaristic terms. Osella and Osella (2000, 97) for example, describe poor families as marrying with minimal dowry, but large amounts being given willingly by those with high aspirations in a "fierce hypergamous marriage market in which opportunities to display and augment prestige are maximised." Amongst middle-class and elite families in Bangladesh, as elsewhere in South Asia, gifting at marriage is seen as the willing demonstration of love for one's daughter. This may, of course, be a euphemism, as dowry is not seen as something to be proud of. However, a higher dowry may also be seen as a fair price to pay to secure a desirable groom of good family and employment status who will be able to provide his bride with a comfortable life (Kishwar 2005). Representation of dowry in the poverty and gender literature tends to be entirely negative. There are three main aspects to this. First, is the ruinous impact on household economic status of having many daughters. Davis (2007, 3) notes that dowry was rated first amongst causes of household economic decline by 116 focus groups across 11 districts in Bangladesh in 2006, including 62% of poor female and 55% of poor male groups. Tomalin (2009, 2) states that the economic pressure of dowries has exacerbated tendencies to son preference and resulted in a "well documented" impact on adverse sex ratios in many regions of India, due to sex selective abortion and female infanticide. Second, dowry is presented as an unfree exchange, as brides' families are forced to give more than they can afford or face their daughter remaining unmarried as they compete with other families to "buy" a groom (Rozario 2009). In Bangladesh dowries are thus commonly referred to by the English term "demand." Third is the strong association of dowry with violence, that women may be threatened or beaten either when their parents have failed to pay what they promised, or as a means to extort more money after marriage, leading in the worst cases to "dowry deaths." Identification of dowry as a social evil has led to both legislation and social activism. Dowry was outlawed in India in 1961 and in 1980 in Bangladesh. In neither case was the ban effective. From the 1980s, publicity surrounding "dowry deaths" and other forms of violence against women led to demands by feminists in India for a "dowry boycott" and for new legislation to strengthen protection for women against dowry-linked abuse (Kishwar 2005). Subsequent amendments to the Dowry Act in India have hardened its provisions, to the point that the Law Commission has suggested that new amendments are now required to protect grooms' families against abuse (Times of India, January 20, 2012). A similar shift is evident amongst activists and in the academic literature. More recent commentators want to problematise the widespread identification of dowry with "dowry deaths" and "bride burning" (see Tomalin 2009;Wyatt and Masood 2010). Wyatt and Masood (2010) describe how closer investigation of "dowry deaths" reveals that there are often much more complex stories of unhappy marriages behind them. Tomalin (2009) states both that there are other bases of gender violence than dowry, and that dowry is not necessarily associated with violence. Kishwar (2005) retracts her earlier total rejection of dowry, to state that activism on dowries should target compulsion rather than those that are freely given, resist all forms of violence against women and campaign for women's inheritance rights.
Attempts to explain the rise of dowry seek to ground the social in the material. Its early identification as women's "pre-mortem inheritance" (Goody and Tambiah 1973) has been rejected, both on the grounds that it goes to the groom not the bride and that it constitutes in many cases a far larger amount than could be expected as a share in ancestral property (Osella and Osella 2000, 100). The "economic compensation theory" (Sharma 1980) takes various forms. At its simplest, it holds that the groom's family needs to be compensated for taking on an unproductive member. This sits rather uneasily with the fact that, at least in poorer households, the need for women's household work is one of the main factors motivating grooms' households to seek a new bride (Grover 2009). More sophisticated versions suggest that the economic value of women's work is falling relative to men's; or that the need for cash income makes households newly dependent on male earning capacity (Sharma 1984(Sharma /1993. Other theories look to population. A simpler version holds that the population growth rate combined with the fact that brides are generally several years younger than grooms means that there will always be more girls seeking a smaller absolute number of boys (Rozario 2009). If this were correct one might expect that dowry competition would have eased in recent years, as the rate of population growth has fallen sharply. A more nuanced version links this to modernisation, holding that it is the relative scarcity of grooms with the desirable quality of salaried employment which engenders competition (Lindenbaum 1981).
Whether the explanations look to economics or population they seem to share two fundamental assumptions. The first is the attempt to derive price (the dowry) from some notion of real value, rather than seeing notions of relative value as themselves socially determined. The second is that dowry-gifting is seen as the response to some kind of deficit (economic or numerical) on the women's side and advantage on the men's. Evidence presented later in the article on dowry in Bangladesh suggests that both of these need reconsideration.
The widespread practice of dowry reinforces the evidence reviewed above that there remains strong family involvement in the arrangement of marriages in South Asia. For the parents of daughters the threat of economic ruin can make love marriages appear in a more positive light, as the only chance to escape paying a hefty dowry, other than marrying their daughter to someone whom age, poverty, health status, character or an existing wife make clearly undesirable. Overwhelmingly, however, dowry gifting indicates that marriage in South Asia is not predominantly about the romance of young love, but a collective investment in advancement. The emphasis shifts from the social or emotional strongly onto the economic. The recurrence of references to marriage "markets" or the need to "buy" a groom is indicative. Bringing together the bodies of literature on marriage and dowry underscores the persistence of patriarchal structures despite some relaxation of gender differentials in intimate relationships. It also tempers discourses of personal choice, romance and agency with darker motifs of competition and compulsion.

Political Economy of Development in Bangladesh
Having long been a by-word for "patriarchy and poverty" (Greeley 1983) Bangladesh is now widely heralded a development success. A World Bank commissioned report of 2007 thus opens by declaring "Bangladesh stands out as the shining new example in South Asia of a poor country achieving impressive gains in gender equality" (World Bank 2007, 3). It goes on to list as major achievements: the halving of fertility rates between 1971 and 2003 6 ; that in much of the country more girls are attending secondary school than boys 7 ; that there is no longer excess mortality amongst girl infants 8 ; that women's solidarity and earning potential has been raised by "the micro-credit revolution"; and that "vast numbers" of young women are moving from village to town to work in garment factories where previously "young women were rarely seen outside their homes" (World Bank 2007, 3). Interestingly, marriage and dowry constitute one area in which the confident tone falters. Confounding the generally positive picture, the report notes that dowry has "spiralled" over the past 30 years and states that it is associated with poverty, violence and less voice for women in the household (World Bank 2007, 13). Moreover, marriage is considered the "boundary" that mainly determines a woman's access to resources and her scope for agency (World Bank 2007, 12). It is also seen as largely beyond the reach of direct policy intervention (World Bank 2007, 13).
In reflecting on this report it is important to acknowledge that the changes in fertility, equalising of the sex ratio, increased levels of education and rates of women's economic activity are indeed remarkable. The demographic shift in particular is a game changer, where "over-population" was the problem emphasised in all development writings about Bangladesh throughout the 1970s and 1980s. On the other hand, this report is a classic development text, conjuring a very particular view of the world. Agency is attributed to programmes and projects, key actors are the government and the non-governmental organisations (NGOs). Domesticity is identified with invisibility and restriction, women working outside the home with liberation. Peopleespecially womengenerally appear eager for change but held back by cultural and institutional inertia. While admitting at some points that "there may also be wage and hiring discrimination against women," 9 overwhelmingly the report exudes confidence both that change is happening and that its direction is progressive.
What this suppresses, or acknowledges only in sideward glances (particularly towards the rise of violence against women) is the rather different realities which are other faces of development in Bangladesh. At independence in 1971 Bangladesh was a largely peasant economy, with rain-fed, low-technology agriculture and significant inequalities in landholding. Industrial production was limited and concentrated on raw materials processing, in the shape of textile, jute and sugar mills. State capacity was restricted both by the inexperience of its leaders and functionaries, many of the key positions in the bureaucracy had been held by West Pakistanis (Van Schendel 2009, 176), and by the dominance of party-political concerns over decision-making. The ruling party "engaged in an abundant politics of patronage that continues to plague the Bangladesh state machinery today" (Van Schendel 2009, 178). The early years of independence saw economic productivity fall well below pre-war levels and in 1974 there was widespread famine. In 1975 the President, Sheikh Mujibur Rahman, was assassinated in a coup and military rule was established, which continued under different regimes until 1990.
In place of state-centred, Indian-inspired ideas about development, the post-1975 economic modelsupported by Western donorsfavoured export-oriented growth, de-nationalisation of industry and input subsidies to increase agricultural production (Van Schendel 2009, 193-194). The 1980s saw a "green revolution" style intensification of agriculture with the widespread expansion of irrigation (mainly through the extraction of ground water through tubewells), adoption of high yielding variety seeds and application of chemical pesticides and fertiliser. While this has been an undoubted success in terms of increasing rice production, it has not raised the living standards of rice producers and there are serious concerns about longer-term food security and environmental impact (Van Schendel 2009, 235). Large amounts of Western aid secured the Bangladesh government's compliance as an early adopter of "structural adjustment" through the 1980s and the ongoing pursuit of economic liberalisation through the 1990s. It also enabled governments to remain in power and continue to provide some welfare programmes despite low levels of domestic resource mobilisation. Aid further fostered a burgeoning NGO sector. Lewis (2011, 110) quotes figures of 206,000 "not for profit" organisations accounting for 8% of annual gross domestic product (GDP) in 2005. In addition, aid provided a whole range of privileged opportunities for accumulation that have helped produce the highly polarised society that is Bangladesh today, ranging from an extraordinarily wealthy section of the transnational elite to significant numbers of people suffering absolute destitution.
The 1990s saw the financial importance of aid eclipsed by export earnings from garment factories and remittances from migrant workers abroad (Lewis 2011, 136) with earnings from intensive shrimp cultivation coming to comprise the third largest source of export revenue (Lewis 2011, 151). In 2014 GDP growth was 6.1% compared with 2.1% in 1982. While absolute levels of poverty have declined, 10 inequality has worsened (Lewis 2011, 136). Urbanisation has been rapidin 1980 85% of the population were in rural areas, while just 66% were in 2014 (World Bank 2015). The capital city, Dhaka, has a fast-growing population, with high levels of pollution, over-crowded and poor quality housing, and permanently jammed traffic. Access to power is extremely unreliable, with even well-served areas suffering frequent cuts. Both materially and socially, connectivity is highly unequal, such that those living in the area which produces a large proportion of the gas that powers Bangladesh gain neither energy nor employment from its production (Gardner 2012).
Despite the prevalence of social initiatives pursued by the government and other agencies, the underlying model of development in Bangladesh is red in tooth and claw. It is highly exploitative both of workersas witness the long hours and poor conditions in garment factoriesand of the environmentas indicated by the destruction of trees, crops and increases of soil salinity, let alone the forced expropriation of land, caused by shrimp cultivation (Van Schendel 2009, 236). It is tempting to see this as running counter to the aid industry, but this may be to fall for development propaganda. The development tropes noted in the World Bank report cited above are indicative. Two images predominate on gender, that of "making visible," and "bringing out." In both cases these relate, directly or indirectly, to bringing women into the market. But they may also be seen more broadly as what development, or modernisation, or the transition to capitalism, is about. As the construction of "what counts" in economics makes clear it is all about the market (Elson 1991). Women working in their own homes or subsistence production "don't count" in classical economic terms; their labour is only accorded value once it is exchanged in the market. Feminist writers have of course contested this, pointing to the many ways in which "domestic labour" is vital to sustaining the system of production (see Edholm, Harris and Young 1977). From the viewpoint of capital this can easily be concededwomen's entry into the wage labour market does not typically mean they relinquish responsibilities for housework, they are simply enrolled in a dual form of exploitation. What is true of women/labour is also true of other kinds of assets. Gold handed down as dowry from one generation to the next may have intrinsic value for its holders but it does not become productive unless it enters market exchange. Land passed on through inheritance is similarly "unproductive" in economic terms unless it is capitalised such as to produce sufficient surplus for exchange. Read this way, the urge to "bring women in" to the market which is emblematic of gender and development, might at the structural level be less about women's needs to be emancipated, than about the "needs" of global capitalism. These require the "discovery," as McCarthy and Feldman (1983) argued long ago in relation to women's employment in food for works programmes in Bangladesh, of "new sources of capital and labour." In Marxism, this is the process theorised as "primitive accumulation." Along with patronage, political violence has become endemic in Bangladesh, seriously undermining both the credibility and the workability of the political system (Moniruzzaman 2009). Devine, Basu and Wood (forthcoming) observe that power is concentrated in increasingly few hands, and warn that deterioration in the "quality of political life" is "real and current." The space for critical comment and debate has severely narrowed. Deeply implicated in both the politics and the violence, and similarly feeding off both development assistance and economic liberalisation, is a parallel network of touts and gangsters, who control access to virtually all the public and private goods of modernity, and demand extra payment if any business is to get done (see Devine 2007;Van Schendel 2009;Wood 2015). In development-speak this tends to appear as "governance issues," or less politely, corruption. While both labels may convey something of the truth, sociologically what this network reflects is the acute competition for scarce resources, such that "playing fair" will never be good enough. To succeed in the education system, for example, children not only need to work extremely hard at school, but also have a private tutor at home. And this private tutor may even be their schoolteacher, who then also favours his or her private pupils when it comes to school-based work. Whether it is the Bangladesh state relying on international aid or remittances from workers operating in very different economic contexts, or an electricity technician requiring a "tip" before agreeing to make a connection, there is the sense that the economic system itself is never sufficient, never self-sustaining. To operate it has constantly to be fuelled by additional resources from outside.

Marriage and the Order of Things in Bangladesh
Marriage is virilocal and kinship patrilineal. The culturally approved model is for a bride to join the household of her father-in-law, and this is still what happens for at least a period in most cases. 11 Overall, the proportion of nuclear family households has increased, but less rapidly than the expectations of many villagers and scholars. Village studies show percentages of nuclear households in the low 50s in the 1960s and 1970s to the low 60s in the 1990s. 12 Polygyny is allowed amongst Muslims but it is relatively uncommon, being seen as both economically and socially undesirable. 13 There was only one case in our sub-sample of 58 marriages and only 16 (1%) in the wider survey of 1,500 households. 14 Divorce is much more common amongst Muslims than Hindus, with village studies recording rates between 15% and 30%. Hindu widows are not allowed to re-marry. Young Muslim widows and divorcees may re-marry, but do so on very unfavourable termsoften as a second wife or to a much poorer and/or older man. Long years of co-residence and the shared underlying folk culture mean that religious boundaries have been relatively porous, with most authors emphasising the syncretism and malleability of Islam in Bengal (see Ahmed 1981;Banu 1992). More recently, however, a number of reformist movements have sought to introduce a much stricter Islamic line, with gender and family relations an important focus (Huq 2008;Samuel and Rozario 2010;White 2010).
The influence of the state is uneven. From low levels of formal registration historically, Shehabuddin (2008, 83) found that 72% of her informants' marriages were registered. 15 Despite laws prohibiting child marriage, UNICEF reports around one-third of Bangladeshi women aged 20-24 were married by the age of 15 (UNICEF 2011, 33). Our research found that age of marriage was rising in both locations, but some people noted contradictory pressures to marry girls younger. These derive on the one hand from fears that higher dowries will have to be given for older girls; and on the other from a sense that society is becoming increasingly lawless, jeopardising their daughters' honour.
In our Dinajpur study village which was near the administrative centre, marriage disputes were frequently brought to local government (union parishad) for adjudication, but in most villages they are still mainly heard by the customary gathering of male village elders (shalish). Issues of dowry, maintenance, sexual misdemeanours and domestic violence constitute around half of all disputes that go before the shalish (Siddiqi 2004, 58). Unlike a court of law, the people Siddiqi spoke to saw the primary orientation of the shalish as to restore harmony and secure reconciliation (Siddiqi 2004, 50). The power they wield is grounded in the dominance of wealth, age and gender, and the settlements that they deliver reflect this.
The idiom of guardianship dominates both family and community, and these are closely intertwined. 16 Members of the shalish are guardians of the community, as the husband and father is the guardian of the home. Traditionally, at least, women in Bangladesh should always be under male guardianship, of their fathers in childhood, their husbands and fathers-in-law in marriage, and their sons in widowhood. The guardian is responsible to his wife and children for provision and protection. A guardian is also responsible for his wife and children to the social and religious community, somaj (White 2007). This entails rights to discipline through violence. If they misbehave, he will be held at least partly to blame: his honour, and the honour of their lineage, depends on their compliance. It would be mistaken, however, to suggest that the responsibilities for guardianship lie all in one direction. Wives may also have duties as moral guardians. Razia Begum, a middle-income Muslim woman, described this as follows: If a woman wants to be an ideal wife, she has to be aware where the husband is going, what he is doing outside, whether he is going to any bad place, and whether he is coming into contact with any bad company. If the husband takes money from the wife, or even if he spends his own money, the wife has to know the purpose of spending that money and whether he's telling the truth or not regarding this. Therese Blanchet (pers. comm.) confirms that the expectation that a bride will be able to reform a young man who is going off the rails belongs to a well established cultural narrative.
In Bangladesh, therefore, marriage is not only concerned with individuals, or even couples or families. It is a core social institution, which both materialises and symbolises the underlying moral order, which configures people being in the right position and living in right relationship with one another, each following his or her proper path in life. 17

Affinity, Conjugality and Choice
Confirming the broader pattern in South Asia, there is clear evidence in rural Bangladesh of shifts in the norms governing the arrangement of marriage. While marriages should still be arranged through family negotiations, pre-marital relationships are becoming increasingly common, and considerably more open, than they were 30 years ago. This is particularly evident in towns and university campuses, where young unmarried couples can be seen openly sitting or walking together. Such inappropriate mixing, especially in educational contexts, transgressing class or religious boundaries, was the stuff of scandalous stories in our research villages. Such stories are not anything new -Hartmann and Boyce (1983, 110) report from their 1975 fieldwork how scandalous stories of Muslim girls eloping with Hindu boy classmates were used to justify keeping girls out of school. In fact, the only case of "love marriage" in our sample was well-matched in terms of class, religion and education. 18 They tried to get their families to agree to the marriage, and only when this was refused did they decide to elope. Even then, they worked hard thereafter to re-build relationships, and were eventually accepted into the man's parents' house six years later. In general, young people conducting pre-marital relationships still had to keep them hidden, with girls seen as "spoiled" if it became known. In our focus groups especially, "love marriages" based on sexual attraction rather than parental wisdom were evoked as figures of disorder incarnate. This reflects conflicting ideas about sexuality. On the one hand, sexual compatibility is an important part of the force of attraction that people value in a marriage. On the other hand, sexual attraction and the "love" that it evokes is seen as treacherous, short-lived and non-dependable, putting girls in particular in positions of great vulnerability. Instead of instantiating the moral order, marriage on such grounds constitutes a threat to it: a decline in family authority; the basis of inappropriate alliances; and resulting in households adrift, without the social support they will need to sustain them.
More common than open challenge are moves to accommodate new patterns of relationship within existing idioms. The norm has shifted to involving greater consultation with the couple themselves, backed up by a legal requirement that the girl's consent should be given. There is also some evidence of young people deciding whom they want to marry, and then getting their parents to arrange it. There are several reasons that parents may be only too happy to accept their children's increased involvement. As mentioned above, a love marriage is perceived by many parents as their main chance of avoiding having to pay a dowry. Parents also say that since marriages are now happening when the couple are more mature, and especially when they are educated, then they have more of a right to a say. Finally, parents also express the fear that if things go wrong in the marriage, they will bear all the blame. To involve the couple themselves in the choice offers some protection against this. Again, it is important not to exaggerate the change. There are cases amongst the grandparents' generation of marriage based on the couple's desire, as well as cases of minimal contact amongst the young.
Confirming the wider patterns noted above, we found that people talked openly about the importance of intimacy in marriage, in terms of a force of attraction (moner tan) or like-mindedness (moner mil) between husband and wife. Women especially spoke of how their husbands were everything to them, and widows about the pain and loss of colour in life when their husbands died. While material sufficiency was valued, the quality of the marital relationship was also crucial. A poor Muslim widow, Anjumon Bewa, states this clearly: "In a good marriage there is no fighting, no arguing between husband and wife and they don't lack anything they need. Someone who can provide wealth, but no peace is not a good husband." A middle-income Hindu woman, Shanti, puts this more positively: "If your husband is good you can be happy even living in the forest!" Aside from this, people spoke of understanding without need for words, being able to say to one's spouse what you could not say to anyone else, thoughtfulness about what the other might like, bringing gifts and food, looking out for each other, care shown by the other to one's parents or children, or of the spouse as a treasure which cannot be replaced. References to mutual support and commitment, indexed by the pairs of terms good and bad, advantage and problem, security and danger (bhalo/mondo, subidha/ osubidha, apod/bipod) are common. An elderly middle-income Muslim man, Aynuddin Serkar, sums it up simply: "I am there in her bad times, she is there in mine, we live side by side." Such statements seem compatible with the wider literature which finds a growing emphasis on the marriage relationship. However, without similar data reported in previous studies it is hard to be sure of how much this represents a real change, or is it simply that attention is now being paid where formerly such comments would be overlooked? It is notable that many of the ways people talk about intimacy are still embodied in the material provision of resources or care, rather than abstracted in purely emotional terms. Also, from previous fieldwork in 1985-86 I feel doubtful that the substance here is newmarital intimacy then also seemed critical to people's sense of well-being. However, it may be that people now feel freer to talk in such ways than they did before, and there are certainly some shifts in form. Nazma, a wealthy Muslim woman in her mid-30s, for example, told how she and her husband give each other gifts on their wedding anniversary, which is not a traditional practice. Older people talk about shifts in style of address, that in their day husband and wife never used each other's names, whereas this is common practice now. In fact, however, only Nazma said that she had called her husband by name. Three woman in their 20s and 30s kept their own names on marriage: they all were of middle wealth with at least undergraduate level education, and two were teachers. However, all of the women who had children, these included, had come to be generally known as (their child's name)'s Ma. While this practice is interpreted in the gender literature as part of women's subordination, a denial of their individuality, the women themselves universally welcomed it, seeing it as a celebration of the identity of which they were proudest. Many of the men similarly welcomed being known as their child's father, and said their greatest happiness was to hear their children calling them Dad. Or, as Hasina Bewa, a Muslim middle-income widow, put it: "What could be nicer than hearing my children calling me 'ma' with their mouths full!" Sometimes intimacy is expressed in an elliptical way, by a more distant form of address. Thus little girls are affectionately called "old lady." Ayesha Begum, a middle-aged and middle-income Muslim woman declared: "Whoever calls me whatever, I'm happy with it. Yet when my husband calls me 'Begum,' I like it most. In this name there is affection, and at the same time it expresses that I am the head woman of this household." 19

Guardianship: Provision and Protection
As introduced above, in Bangladesh the dominant idiom for marriage is guardianship. This involves the twin aspects of male provision and protection. The dominance of development concerns in Bangladeshi social science means that work and property-related dimensions of marriage have received the most attention. This follows the assumption that increasing women's economic activity and so reducing their dependence on men constitutes the predominant way to advance gender equality. It is not surprising, therefore, that discussion of changes in marital relationships in Bangladesh is mainly found in studies of women factory workers (see Ahmed and Bould 2004;Dannecker 2002;Kabeer, 1997Kabeer, , 2002; women's use of microfinance (see Kabeer 2001;Goetz and Sen Gupta 1996); and migrant workers (see Blanchet 2010;Rao 2012).
Rather than emphasising women's dependence, Siddiqi (2004, 22) casts it in terms of rights, remarking: "the right to maintenance is at the heart of the marital relationship." This is at once symbolic and substantive. One of the commonest ways to describe marriage is to say a woman "eats her husband's rice" (swamir bhat khai). Marriage breakdown is signalled similarly: "she won't eat her husband's rice" or "he wouldn't give her rice." Such statements conjure male provisioning and female consumption, and this is an important motif in Bangladeshi narratives of marriage. To be able to provide adequately for one's wife and family is a significant aspect of male honour, which receives cultural emphasis in the purdah prohibition on women working outside of the house. What this suppresses is the complement to male provision of resourcesfemale nurture and stewardship. Thus in the classic peasant household, while men are to provide the resources through cultivation or purchase, it is women's thrift and skills in storage, processing and preparation that transform these into food to sustain the family.
The figure of the house is central to the imagination of male-female relations. Repeating the same pattern, house construction is men's work, but the day-to-day care and maintenance of the house belongs to women. "Ghor-songsar kora," to run a house and family, is another common way of describing married life. This applies to both men and women. Husbands should be family-centred: hard-working; of good character; free of harmful addictions (to drugs or alcohol) and extra-marital affairs. For women, however, the associations with the house are far more elaborate and explicit. The following quote from middle-income Abdul Karim gives a typical example: The ideal mother will raise the children according to the tenets of Islam, and she won't give value to what outside people say. The ideal mother should be kept busy with the children, family, cooking and general cleaning in the house. The ideal mother won't go outside to gossip. Nowadays, mothers are going out to work because of shortage, this isn't anything bad…but they have to maintain purdah. This is an Islamic rule.
This shows the fusion of the spatial and the gender orderwomen are both the centre of and (properly) centred (and contained) in the house. It links together the values of hard work, self-sacrifice, right speech, care and right example. Although this speaker refers to Islam, these values are closely associated with the figure of Lakshmi, the Hindu goddess of wealth. A strong presence in the common cultural fund of imagery in Bengal, Lakshmi's qualities include deification of thrift; order; and the use of nature's gifts; and that she symbolises wealth, property, fertility and new kinship ties (Fruzzetti 1990, 123-124). Known locally as "Lokkhi," she has also been incorporated as an adjective which is used freely by Hindus and Muslims to describe one who is particularly good, fortunate or blessed (usually a woman, or a young person of either sex). The association between a wife and Lakshmi, and a house without a wife being like one without wealth, goes back to the ninth century (Fruzzetti 1990, 123). Two women, the first Muslim and the second Hindu, show how this has become a common saying, which is enlisted in the contemporary debates about women's work: Women are the Lakshmi of the home. If they go outside, there will be no Lakshmi in the home.
Since my younger daughter has gone out for work the Lakshmi of her house has also left. That's why they can't get out of deficit even with both their incomes.
It may not be coincidental that both of these women are poor, and so are likely to be thinking of manual, degrading "outside" work.
With increasing landlessness; high yielding varieties of rice which require irrigation, fertiliser and pesticides; rice mills in place of manual husking; and increased urbanisation, "traditional" gender complementarities have been replaced by a greater demand for cash. 20 This has combined with growing numbers of women having to take paid work because of poverty and increased employment opportunities for women as the economy has developed. Purdah prohibitions on women doing work "outside" the household have accordingly shifted. While some of our respondents still maintain that women should not "go out," for most people the major issues now are the kind of work, what and where it is done, and how women conduct themselves within it. Office work or teaching, for example, are seen as enhancing status, fusing as they do the symbolic capital of salaried employment in public service; educational achievement which serves as an index of moral intelligence as well as formal qualifications; a genteel modern occupation involving non-manual labour; and an inside location. For women to work in a rice mill or road construction is, by contrast, widely disapproved of. Here the symbolism is reversed, with heavy, manual labour attracting low rates of pay and carried out under the hot sun in full public view. These criteria of value are obviously not limited to women; they are more general indices of class-based status in Bangladesh. For women, however, there are two additional dimensions. First, teaching in particular links to ideas of women as mothers, passing on knowledge and guidance to the new generation, and so is easily assimilated as gender appropriate. Conversely, women working on the roads or in construction constitutes a direct affront to gender norms. Second, norms of female modesty are easily accommodated within office work, including the wearing of a burkah if desired. 21 This is compounded by the view that educated women know how to conduct themselves with honour. By contrast, physical labour outside is seen as immodest, as a woman will not be able to keep her body properly covered. Again, this is intensified by the cultural association between lower class status and greater proclivity towards sexual misconduct.
There is a clear parallel here with the analysis of marriage arrangement presented above. The shift in discourses around women's work clearly reflects an attempt to accommodate the "new" circumstances within the "old" status markers. But perhaps more significant is the intertwining once more of material and symbolic, gender and class or social status. The new economy requires women's labour and their families need the income it brings to defend or advance their economic position. Notions of respectability which bear class and gender markings are re-deployed to ensure that this augments, rather than compromises, broader differentiation in families' social status.
While women's outside work is increasingly accepted, this does not mean that the ideology of male provisioning has necessarily weakened. Confounding expectations in the gender and development literature, women's status often depends on their hiding the extent of their earnings, and colluding in conventional narratives of female dependence. Kabeer (1997, 271) describes how husbands of garment workers in Dhaka feared their wives' reduced economic dependence could undermine their own dominance at home. By contrast, women went to considerable lengths to say their earnings had no effect, and tended to avoid any overt challenge to their husbands' roles as breadwinner. For some women, being able to support themselves through their wages meant they could leave unhappy marriages and/or avoid having to re-marry Kabeer (1997, 293-295). Ahmed and Bould (2004) also found that factory-working mothers showed little evidence of (the traditional) son preference, and asserted instead the (at least) equal value of daughters. In general, however, both Kabeer (1997, 298) and Dannecker (2002, 198-199) report that women felt the fact they were working in itself signalled a major battle won, without need for any broader challenge to male authority. Blanchet (2010) reports similar findings amongst women who had migrated unaccompanied from a village in Bangladesh to work in the bars in Bombay. She recounts an example of a daughter-in-law heralded by her marital family as "lokkhi bou" 22 after sending large amounts of earnings home from (sex) work in Malaysia, because she colluded in the fiction that it was her husband who had earned the money. Blanchet also describes how once the opportunities for bar work declined and thus the flow of money dwindled the voices of conventional morality in the village were able to reassert themselves. Some of the women who had worked in bars were brought before the shalish and required to undergo the ritual cleansing and humiliation usually reserved for prostitutes. Some resisted this, however, and bar work continues. Taken as a whole the article emphasises the territoriality of morality in Bangladesh, and the way that "border crossings" enable the shift from one moral regime into another. Despite all the difficulties, Blanchet states, the women experienced the employment as liberating. Their status within the village, however, required the women not to display their earnings, but rather to conceal them so as to bolster the fiction that their husbands were competent providers. Kabeer (1997, 296) locates the reason for women's collusion with ideologies of male provision in the other aspect of guardianship: protection. She states: In the context of Bangladesh, a striking feature of gender subordination is the extent to which women rely on male protection as much as they rely on male provision. Consequently, there is a social as well as an economic dimension to female dependence. This notion of male protection is a core but under-researched aspect of the construction of marriageand indeed patriarchyin contemporary Bangladesh. The growth in numbers of women and girls going out to work or study has been accompanied by an increase insometimes violentsexual harassment (Siddiqi 2004;Chowdhury 2005). One of Kabeer's informants captures this beautifully as she talks about "the fear women have," which makes them value even the equivocal "protection" of a violent husband (Kabeer 1997, 297). Constructions of male protection rely heavily on the inside/outside axis noted above. Violence against women in the home is rife. The 2007 Bangladesh Demographic and Health Survey (NIPORT et al. 2009, 201) states that more than half of all women reported having experienced physical or sexual violence in marriage, and 18% in the year before the survey. The importance accorded to having a male guardian reflects the extent to which the honour-sexuality nexus constitutes extreme jeopardy for women, rather than any guarantee of female safety or security (see also Chowdhury 2005;Feldman 2010).
The social necessity of male protection constitutes a significant brake on the capacity of women's independent earnings to deliver the "empowerment" that liberal/development discourses envisage. In our research women in particular repeatedly emphasised that a woman without a guardian is highly vulnerable to unwelcome attention or abuse from other men and the gossip that surrounds the suspicion of sexual misbehaviour. In one of many such statements by our informants, Maya Rani, a poor Hindu woman, put it like this: My husband is the power, like the pillar of the house. If I go out with my husband no one dares to say anything to me. Even if I go out without my husband, since I am married no one troubles me because I have a husband. If I do anything bad, it is no one but my husband's duty to account for that.

The Logic of Dowry
As introduced above, the increase in value, spread and compulsory character of dowries is tightly tied, symbolically and materially, both to economic modernisation and to masculinity. Lindenbaum (1981, 396), one of the earliest commentators on this, notes how the Japanese and American origins of dowry goods also signified Bangladesh's role as consumer in the global marketplace. It is a sign of the times that in our research the only case of successful resistance to dowry took a rather different symbolic form. Wealthy pharmacy owner Mahbubul recounted how both he and his brother-in-law have managed to marry their daughters without any dowry. In explanation he exclaimed, "There is God!" The girls were kept in strict purdah, but both studiedhis daughter at school to grade 10 (16 years old), his niece tutored by his wife at home to MA level. Both grooms had studied, and one taught, in an Islamic institution, and were similarly strict in their observance of reformist Islam. The son-in-law's conduct in the pre-marriage negotiations was also distinctive. When he came to see the bride he asked her name and nothing else. When her married sister said he could ask more he demurred: girls are not cattle in a market that need to be checked over thoroughly. This re-doubled the approval he won in his future mother-in-law's eyes.
While the heightened profile of Islam is certainly a feature of contemporary Bangladesh, the tide of accumulation still constitutes a powerful undertow. Tellingly, in research between 1963 and 1974, Lindenbaum already found some cases in which brides' parents were paying for grooms' higher education, to the point of commenting, "In many respects, the groom payment makes the man" (Lindenbaum 1981, 398). This trend has only intensified since. Since the mid-1980s, Ahmed and Naher (1987, 160), Rozario (1992) and Rao (2012) found jobs for the groom in Dhaka or the Middle East forming a part of marriage settlements, and even being demanded after marriage, against threat of divorce. In our own research, dowry similarly featured significant amounts of cash, gold, or the promise of a job. Grooms were commonly using the payment that they gained on marriage to establish themselves in business. Rozario (2009, 48) sums the situation up: "for many grooms' families, access to modern goods, modern salaried jobs…a new business venture, or expansion of an existing business is only possible through a large dabi [claim] from the brides' families." What has been the effect on dowry of the two most famous economic developments for women, the rise of the garment and microfinance sectors? Kabeer (1997, 291) reports cases of women working in garment factories who were able to marry without dowry, but none of these was amongst her own respondents, so the information might not be reliable. Her informants did, however, include mothers who were working to help pay daughters' dowries, and unmarried women who were aiming to help earn their own (289-290). Dannecker (2002, 188) by contrast reports that few of her respondents had been able to save much for their dowries. A garment factory worker who had managed to marry without dowry had an ongoing responsibility to send her wages to her parents-in-law (Dannecker 2002, 162). More evidence is thus required on whether work in garment factories is a means to offset heavy dowry demands, or to fund them. Rozario (2009) reports that the major suppliers of microcredit to women in Bangladesh, the Grameen Bank and the huge NGO, BRAC, both have anti-dowry policies. In practice, however, their loans are frequently used for dowries. Even more worryingly, she states that families of grooms use the availability of such credit to argue for even higher dowry payments (Rozario 2009, 48).
Far from women's contribution being in decline, the data suggest the need to bolster men's. The material security of former days, land to be inherited from the father, is no longer available, or no longer sufficient, to guarantee the male provisioning role. The desirable alternative (or supplement) of salaried employment is difficult to achieve, with high levels of unemployment especially amongst educated young men (J. Devine, pers. comm.). Read this way, one might say that demand dowry does not constitute as Goody and Tambiah (1973) have argued a form of female inheritance, but rather male, through the affinal line. Transfer from older to younger men is now bi-partite, following the ties of birth and ties of marriage, a form of collective (though not free) investment in the male provisioning role. If we follow the logic of practice, it seems that the symbolics of demand dowry may be not a veil for economics, but in a sense the economics themselves.
To understand this further I draw inspiration from the analysis of Collier (1997), and seek to connect this pattern of collective investment in the male provisioning role to the broader political economy analysis sketched out above. This suggested four broad points. First, that the predominant economic model in Bangladesh continues to depend on the exploitation of its "raw materials," in the form, for example, of the labour of women in garment factories and migrant workers, the harvesting of shrimps, the intensive use of chemicals in agriculture, or the extraction of natural gas. Second, that rather than celebrating aid intervention as an emancipatory enterprise, we need to recognise how it serves as the midwife to capitalism, in supporting the erosion of the subsistence sector and the "discovery" of "new forms of capital and labour" (McCarthy and Feldman 1983). Third, that the social and economic structure is marked by intense competition for relatively few development "goods," which are thus allocated through a mix of market and extra-market mechanisms such as patronage and extortion. Fourth, that the economic system has thus failed to become self-sustaining, relying for its fuel on continually drawing in "additional" resources from outside.
Considered in these terms, dowry comes to appear consistent with the development of capitalist modernity, rather than an anomaly within it. As attested by numerous accounts of the hardship occasioned by dowry amongst the poor, it represents a form of primitive accumulation whereby resources are in effect forcibly extracted from the parents of daughters. This contributes to further shrinking the subsistence sector as land or gold is sold to finance dowries, and also provides the "outside" fuel which the system needs to sustain itself. At the family level, the dowry provides a means to generate resources that will enable a higher level of market engagement. The uneven development of capitalism in Bangladesh promotes maximum competition in a context of extremely limited resources, with the result, as noted above, that the forced mobilisation of "additional" resources is the norm in virtually every transaction. Since marriage is a key moment in social mobility, it would perhaps be surprising if the marriage "market" were the exception to the rule. The social pressure to get girls married for fear of "dishonour" clearly puts families of girls at a disadvantage. Meanwhile, in economic terms, investment in meneven if he is your son-in -law rather than sonstill makes sense in terms of family advancement strategies, since men's mobility and earning power is so much greater than women's.
This structural bias towards male employment reminds us that gender identities are affirmed rather than undermined through "modern" employment. This is indeed inscribed in the repeated stress on "women's work" (most obviously in the emblematic garment factories). It is not feminist commitments that make garment factories overwhelmingly recruit women workers. Rather gender difference is incorporated, not dissolved, in the production of employee subjects who are both relatively cheap and relatively docile, in part because of the patriarchal authority to which they are still subject at home and in wider society (Elson and Pearson 1984).
While gender differentiation is thus utilised for capitalist accumulation, it cannot be explained entirely in economic terms. As shown by the idea of marriage as instantiating the moral order, it is important to recognise an independent cultural or ideological structure of patriarchy. When families invest through dowries in male provisioning, they are not only making a "rational" economic decision given men's greater earning power. There is also a perceived patriarchal dividend in buttressing male capacity to provide. This is also evident in individual motivations. Liberal feminist discourses of women's emancipation through economic activity notwithstanding, the predominant impetus for women in Bangladesh to enter the wage labour market or to undertake small enterprises funded through micro-finance has been to address their families' needs for more income in the face of increasing landlessness and the growth of the market economy. Interviews with working women repeatedly emphasise this desire to support their husbands, to make up for the fact that they can no longer provide alone sufficiently for the families' needs (see Ahmed and Bould 2004). This is not to deny that having an income may give women more say in the household, or the scope to manage independently if a relationship breaks down. Both are clearly the case, and from a feminist viewpoint, important. But for the vast majority of women this is not what primarily motivates them. The reason that women hide the extent of their earnings and collude in fictions of male provision is that they are also committed to the overall patriarchal order. They may want to re-shape this so as to gain rather more room for manoeuvreas much from the older generation as from their husbands. Overall, however, womenlike their parentstend to believe in a moral order which codifies male provisionthe problem is that the one they have is no longer able to deliver.
Buttressing this commitment to male provisioning is the continuing stress on male protection and its ambivalent combination of threat and security for women. Almost all commentators agree that violence against women is increasing, clearly indicating the threat to women if they do appear to step out of the patriarchal line. This is reconfirmed by the more general context of political violence, clientelist state and gangster rule introduced above. Its (anti)heroes are the live fast, die young maastans (godfathers) who codify a modern, violent masculinity which offers both practical and symbolic challenge to the cultural norms of family values and age-based deference. At a symbolic level this seems the home context of demand dowry: something forced and without normative legitimacy, yet an unavoidable fact of social practice. More materially, it re-emphasises the old identification of public space as quintessentially male, which women enter at their peril. The cultural investment in the priority of males is therefore further reinforced, as the politicisation of resource access depends on the mobilisation of a man among men.
The rise of demand dowry represents the most significant challenge to the association of marriage and the moral order in both practical and symbolic terms. Against a vision of the moral order as grounded in virtuous exchange, this stands like a cultural vortex, an ironic caricature of the rampant growth of the market; the rewards it brings to some and the ruin to others; and its erosion of moral relationship. Practically, it exposes women and girls to additional violence, and can bring economic disaster to families with more daughters than sons. Symbolically, it rejects the virtues of living in one's right position and right relationships, for an absolute priority to financial gain. Whether this represents a new moral order based on the primacy of competitive advantage over social responsibility, or a choice for personal economic gain against the moral virtues of relational well-being, is a matter for future research.

Conclusion
This article posed three questions. First, does rural Bangladesh show changing patterns of marriage similar to those reported elsewhere in South Asia and more broadly? Second, what might explain the persistent spread and inflation of dowry payments? Third, how might changes in marriage and the inflation of dowry be related to the broader political economy of development in Bangladesh?
With regard to the first question, the article suggests that in rural Bangladesh, as elsewhere, shifts in the structure of economy and polity are affecting the ways people practise and talk about marriage. It cautions, however, that change may be less radical, and continuities more marked, than either local people or wider scholarship would sometimes imply. Despite a shift in symbolism to romance and coupledom, families remain heavily involved in South Asian marriages, and a critical aim in marriage is to secure or advance the family's status in class and/or caste terms. The practice of dowry offers a powerful instantiation of this collective investment in marriage. While there are commonalities in this across contexts, the symbolics at least seem to be bifurcated by class. Where the rich extend conspicuous dowries "for love and in happiness," amongst cash-strapped parents of daughters they are experienced as demand and plunder. In addition, while a stress on affinity, conjugality and choice is evident in discussions of marriage in rural Bangladesh, the structural idioms of male provision and protection are still of far greater importance than is suggested by much of the literature on India. This may in part reflect the difference in populations studied, poorer and more rural in the case of Bangladesh, more urban, middle class and upper caste in the case of India. It may additionally reflect the greater degree to which "maastanocracy" (Van Schendel 2009, 252-253) dominates in Bangladesh.
In reviewing possible explanations for the persistence and inflation of dowries, it is suggested here that demand dowry is better understood as a response not to the perceived weakness of women's economic contribution, but rather to bolster men's. Dowries have become a vital source of capital for young men seeking to enter business or employment in an economy characterised both by intense competition and by endemic reliance on "additional" resources to secure access to the majority of public and private goods. Social pressure to get girls married for fear of dishonour notwithstanding, gender wage differentials mean that family interests in economic advancement may also be served by this investment in males. This is reinforced both by cultural ideology and by the political context of endemic violence and gangsterism. In sum, therefore, dowries help oil the wheels of the economic system and represent a symbolic and material investment in the patriarchal idioms of male provision and protection which stands in stark contrast to modernist ideologies of growing gender equality.
Finally, the article suggests that political economy should be brought more strongly into discussion of "social" issues such as changing marriage practices. Where the literature has tended to emphasise changing "structures of feeling" (Williams 1978) at family and individual level, these can also be used as a vantage point to provide insight into the political economy of social and economic transformation. In South Asia it helps to pursue such a perspective if dowry is brought into the centre of discussions of changes in marriage, rather than kept, as is largely the case at present, in a somewhat separate literature. For Bangladesh, reflecting on marriage and dowry together brings into sharp relief the contradictory character of capitalist development, which combines extreme competition with the reliance on social mobilisation of resources to support and extend market engagement. This implies that the extensive investigation of the impact of socio-economic change on marriage should be complemented by greater attention to the role of marriage in promoting socio-economic change. Far from being a matter simply of personal happiness, marriage in Bangladesh represents an important opportunity for family advancement or decline. Viewed more structurally, it also constitutes a vital means to equip especially male economic subjects with the qualities required to promote the reproduction and expansion of capitalist modernity.

Notes
1. Examples of studies include an edited collection by Hirsch and Wardlow (2006); for Africa a collection by Cole and Thomas (2009), plus Abu Lughod (1990), Egypt; for Europe, Giddens (1992), UK, and Collier (1997), Spain;Illouz (1997), USA; and for Asia Kendall (1996), Korea, Ahearn (2001), Nepal, Osella and Osella (2000), India. 2. Further information about this project can be found at http://www.welldev.org.uk. 3. The respondent profile is as follows: Manikganj total 33: 7 Hindu, 26 Muslim; 7 rich, 13 middle and 13 poor; Dinajpur: total 37: 5 Hindu, 30 Muslim, 2 Santal (Adivasi); 4 rich, 26 middle and 7 poor. The economic classifications were taken from the wider Wellbeing in Developing Countries (WeD) survey, in which household heads were asked to identify themselves as "richer," "average" or "poorer" compared with others in the same community. The sample was chosen to provide a range of experience, not to be representative of the villages as a whole. All names have been changed. 4. A high proportion of the families of Bangladeshis in the UK originate from Sylhet district, which has a long history of maritime involvement. 5. This latter point is echoed by White (1992, 108-109) for mid-1980s rural Bangladesh, and Osella and Osella (2000, 91) for Kerala in the early 1990s. 6. Fertility rates declined from an average of 6.4 births per woman in 1980, to 2.2 in 2013 (World Bank 2015). 7. The female literacy rate was 18% in 1981, rising to 55% in 2012. Overall literacy has increased from 29% in 1981 to 59% in 2012 (World Bank 2015). 8. The sex ratio, an absolute marker of gender equality in indexing the right to life, has increased from 0.94:1 females to males in 1990 (Sen 1990) to being now almost at parity (CIA 2015). 9. Interestingly, this point is made in discussion of agricultural employment. By contrast, the employment of women in the garment factories is naturalised: "since sewing has been a skill traditionally learnt by women and girls in Bangladesh…" though there is also a glancing mention of the fact that young women "provided cheap labor to help keep these industries competitive" (World Bank, 2007, 6). In fact, tailoring has historically been a primarily male occupation in Bangladesh. 10. The figures on this are a matter of some debate, see Sen (2000). 11. Ghar jamai (literally, son-in-law of the house) marriages, where the groom goes to live at the bride's family home, are relatively rare, usually occurring when the bride's family has some land but no sons. 12. These figures are necessarily tentative, since different authors classify household structures in somewhat different ways. Studies are done in different parts of the country with different areas of primary focus and so clearly cannot be read as providing a simple illustration of change over time. However, they do provide some longitudinal perspective on broad social trends. I have not been able to find a village study since the 1990s. 13. I know of no case where polygyny openly involved more than two wives.