Child Money and Food Stamps: A comparative analysis of Mongolian welfare programmes in the Ger Districts of Ulaanbaatar


 Until recently, the Mongolian welfare system was entirely category based. However, a new food stamps programme funded by loans from the Asian Development Bank, which targets aid according to proxy means testing, has been introduced as part of the bank’s aim to push Mongolia towards a fiscally sustainable welfare model. The food stamps programme is presented as efficient and responsible in contrast to Mongolia’s universal child money programme. Based on long-term participant observation research in the ger districts of Ulaanbaatar, areas inhabited by many rural-urban migrants living in poverty, this paper compares the two programmes, interweaving street-level accounts of the experiences of residents and bureaucrats alike with the respective histories and funding sources of the two programmes. Doing so provides a multi-level analysis of the emergent welfare state in Mongolia, unpicking the ‘system’ that ger district residents encounter, linking the relative influence of international financial institutions to democratic and economic cycles, and offering a critique of the supposed efficiency of targeted welfare programmes.


Introduction
For the hundreds of migrant families from the Mongolian steppe that lost or sold every animal they once had and turned their sights on Ulaanbaatar in hopes of finding work to either build a new life or save enough money to return to herding (Save the Children 2013; UNESCAP 2009; Upton 2010), the reality of life in the city's 'ger district' outskirts has often been harsh and unforgiving. Already accustomed to a lifestyle that was far from easy, the infrastructural situation in these city outskirts has nevertheless proved challenging for Mongolia's internal migrants (IOM 2018;Terbish/Rawsthorne 2016). Homes in the ger district are not connected to the city centre's water and waste supply lines. Families therefore collect water by hand from local kiosk delivery stations, left to find their own way to move heavy drums along the unpaved roads and up mountainsides (Kamata et al. 2010: 25). Ger district dwellers are also left to construct their own long-drop latrines, leading to issues regarding sanitation and contamination of ground water (Kamata et al. 2010: 27). Most importantly, however, in this, the coldest capital city in the world, where winter temperatures regularly drop to -30 degrees centigrade, ger districts are not connected to the city's central heating network. In the countryside, gers 1 and baishin 2 are usually heated by burning dried dung (argul) collected from herd animals, or wood, but in the city, without herds and with a steady supply of relatively cheap coal, ger districts have become an incredible source of air pollution. Outdoor air quality in Ulaanbaatar remains at levels designated by the UN as dangerous or hazardous for almost the entire winter, and the fine particulate matter produced by home stoves exposes ger district residents to extremely poor-quality air indoors as well (UNICEF 2016(UNICEF , 2017. 3 Equally challenging for former countryside dwellers is the loss of purpose and cultural prestige that can accompany the material losses they have experienced. To be a herder in Mongolia is a hard but highly respected way of life (Sneath 2006); to be one of the countless faces of the ger districts is not only to struggle, but to suffer the distain of the urban elites who consider the smokeblackened gers that ring their apartments a blight on their city (Fox 2019a: 95-96). Recent rural-urban migrants struggle to find paid employment that can sustain households incorporating multiple generations, even as kindergarten and school classes overflow, limiting the future opportunities of ger district children (IOM 2018: 22). And yet, whether left with little option due to the collapse of rural livelihoods, or drawn to the opportunities for education and employment that others seem to somehow access in the city (Bruun 2006), over the last 30 years rural families have continued to make the transition to urban life.
Following the end of socialist-era restrictions on internal migration, ruralurban migration, especially to the capital, was inadvertently incentivised by a series of laws that privatised urban land for residential and family use (Kamata et al. 2010: 15), and eventually, in 2002, entitled all citizens to a 0.07 hectare plot in the capital (Byambadorj et al. 2011: 170;Miller 2017). As Mongolians already have a portable form of dwelling -the ger -many families sold any remaining animals and moved their gers to the city. Establishing new fenced homesteads (hashaa) on the unclaimed land on the outskirts of the city and negotiating the complex but flexible legal property regimes for registering their homes, either before or after their occupation, rural-urban migrants have expanded the city's borders at an exponential rate (Byambadorj et al. 2011). Named for their distinctive architectural form, the ger districts now stretch out across Ulaanbaatar's valleys and mountainsides, encircling the city centre to the west, north and east (Park et al. 2019).
Although referred to as 'informal settlements' by some academics and development professionals, ger districts are differentiated from most UN-defined informal settlements by their cultural history and legal status (Byambadorj et al. 2011: 166-167;Fox 2019a;Terbish/Rawsthorne 2016: 93). Most of the migrant residents in Ulaanbaatar have rights to their homes and land and many have constructed wooden or brick houses (baishin) alongside their gers. Therefore, unlike in the case of other non-legal settlements, the government of Mongolia has had little legal recourse to slow their expansion. Instead, as many ger district residents feel, successive governments have developed unwritten policies of neglect towards these areas, as they are not provided with the urban infrastructure of the city centre. 4 Furthermore, from 2017 the government decided to stop issuing residency permits to new arrivals, nominally banning migration to Ulaanbaatar until 2020(IOM 2018. This policy does not directly prevent families from moving to Ulaanbaatar but does restrict them from formalising their migration. Unregistered migrant households are "nearly 20 percentage points more likely to report having difficulties at destination compared to registered households (72% to 54%)" (IOM 2018: 37) as, without a registered Ulaanbaatar address, migrants are barred from accessing vital local services, including certain residency-based social welfare services. The ban on registra-|| tion does not stop migration, instead incorporating welfare into an assemblage of subtle barriers aimed at making urban migration a less attractive prospect (Lindskog 2014: 884;Terbish et al. 2020: 8). 2 Social Welfare in Migratory Contexts: The Mongolian Case Arguments regarding economic sustainability become especially heightened in migratory contexts where the boundaries and correspondent limits of state responsibility come under challenge (Ben-Yehoyada 2015) and where "economic and moral distinctions become difficult to separate" (Morris 2019: 80). Studies of the intersection of welfare and migration highlight issues around the real and perceived effects of migration on the welfare state (Brochmann/Dølvik 2018;Handler 2009: 87), such as anxieties over boundary-making (Freeman 1986) and the emergence of "welfare chauvinism" (Schumacher/Kersbergen 2016). Restricting access to welfare is a well-worn technique of "hostile hospitality" -"hostipitality" in Derrida's terms (2000) -applied to unwanted migrant populations (Khosravi 2009: 53). Passive 'non-policies' of neglect, "ranging from a reduction in economic support to a complete exclusion from the welfare system", are used in complement with active measures, such as detention and forced removals in order to encourage and enforce return (Valenta/Thorshaug 2011: 6). Access to welfare may even be mobilized as an 'inducement' made conditional upon migrants' cooperation with return procedures (Rosenberger/Koppes 2018). When the migration under study is international, citizenship becomes a key status that indexes a person's access to services (Joppke 1999(Joppke , 2010. And yet, even in international migration contexts, binary citizenship frameworks can fail to account for positions of partial membership. Similarly, research on welfare and social protection systems invariably reveals the social stratifications that structure access to services among citizens and the moral discourses according to which such lines are drawn, including, for example, that of the deserving and undeserving poor (Morgen/Maskovsky 2003;Schneider 2001).
Studies of migration and welfare highlight boundary-making practices that take place within nation states, and their intersection provides a unique anthropological viewpoint on a complex and overlooked social context: Ulaanbaatar's ger districts. This study therefore examines two social welfare programmes in Mongolia and the initiation of a targeted means-tested programme to Mongo-lia's welfare landscape. The programmes in the spotlight are not specific to Ulaanbaatar -one is a universal allowance offered to all Mongolian children (CMP), the other is a food aid programme aimed at the poorest of the poor -but the paper focuses on the experiences of ger district residents, Mongolia's ruralurban migrants and the local bureaucrats that administer these social benefits. As only a very small number of studies of Mongolian social welfare have been published thus far (Dugarova 2019;Lindskog 2014;Terbish/Rawsthorne 2016;Terbish et al. 2020;Smith 2015) and ethnographic studies of low-level bureaucrats in Ulaanbaatar are also limited (Fox 2019a(Fox , 2019bPlueckhahn/Terbish 2018), this study offers a glimpse into a distinctive, post-socialist welfare system of a unique democracy, wedged between Russia and China.
The ethnographic discussion presented in this article is drawn from six months of participant-observational fieldwork carried out in the welfare office of Aglag, an eastern ger district of Ulaanbaatar, conducted as a part of twenty months of residential fieldwork in the area between 2014 and 2017. 5 Between February and July 2016, I conducted research in the district's local government office (horoo), located on the first floor of a building located next to the bus terminus that contained the small police station, district health clinic, and citizen's meeting hall (all of which had separate entrances). During this period, I shadowed Ganbold, the care administrator (halamjin mergejilten) as an intern. I was assigned a vacant desk in the small office that housed the district's three social welfare officers: Ganbold, the district social worker (niigemiin ajiltan), and the labour exchange manager (hödölmöriin mergejilten) whose remit covered both Aglag and the neighbouring district. 6 Other offices in this area of the building included one administrating services for the elderly, one administering the in-and out-migrations for the area and providing 'proof of residency' certificates, a room for Ganbold's team of twenty 'sector workers' (hesegiin ahlagchid), a room for the cleaning staff and the private office of the local managing director.
Ganbold's office is the first point of call for Aglag residents with questions regarding their welfare status and benefits, and the office was also used as a meeting room for horoo staff for both formal discussions, such as those regard-|| 5 All names including that of the district are pseudonyms. 6 When I first arrived, the desk was occupied by a young woman, an intern and Social Work student. However, she was moved to another position in the Bayanzurkh düüreg soon after I began. The labour exchange manager moved between the two horoo-s on a weekly basis. Sometimes Ganbold himself was moved on a temporary basis to other horoo-s to help cover their workload. During those times, he was not replaced at Aglag and citizens were advised to come back when he returned. ing cutting households from the food stamps programme, and informal gatherings over tea and lunch breaks.
Taking inspiration from anthropological studies of "street-level bureaucrats" (Lipsky 2010F;Dubois 2010;Herzfeld 1992), I examine the everyday experiences of care administrators and their clients in Ulaanbaatar's ger districts. On the level of the street, the paper describes the techniques of governance, including household surveys and home visits of Ganbold and his team, that shape the daily functioning of the system. It also provides the first analysis of the specific issues that have arisen at the level of the household and the district office following the introduction of the food stamps programme (FSP), Mongolia's flagship means-tested welfare payment. Exclusion from the FSP leads to conflicts between ger district residents and welfare officers, particularly when the officers are seen to 'pass the buck' and deny responsibility for the programme's implementation. Seeking to avoid simple critiques of 'the system', however, this study provides a comparative analysis of the development of the two programmes in question (the CMP and FSP), charting their histories and the domestic and international forces that have influenced their design and implementation. As such, the paper traces the funding backgrounds of each programme and uses this to interrogate the role and influence of international financial institutions on Mongolia's welfare system.
Attending to welfare in migratory contexts, such as Ulaanbaatar's ger districts, demands an awareness of welfare as a transnational, globalised phenomenon. As Bob Deacon and Michelle Hulse have noted: globalisation introduces a new field of enquiry into the subject -the supranationalisation or globalisation of social policy instruments, policy and provision. This field of enquiry embraces not only the advice offered to national governments by supranational agencies, but also the capacity of these agencies to regulate economic activity in the interests of social protection (Deacon/Hulse 1997: 44). In the case of the Mongolia, there is clearly an ongoing power struggle between international financial institutions (IFIs) and successive Mongolian governments, as each seeks to control welfare system's present and future development. This paper demonstrates that Mongolia's democracy plays a pivotal role in this struggle, as politicians court voters with promises of welfare reform or expansion -while, on the other side, IFIs render loans and debts as metaphorical carrots and sticks to push Mongolia towards greater welfare 'efficiency'.
The effects of such power struggles at the level of international organisations and domestic politics on the everyday life of citizens, however, are often overlooked. As Paul Stubbs has argued, "literature on globalisation and social policy (cf. Deacon et al. 1997) has focused little on the interconnections between the different levels at which policy is made; much less the impacts on, and interactions with, welfare subjects" (Stubbs 2002: 362). This paper therefore constitutes an answer to calls to integrate street-level ethnographic perspectives with the effects of social policy globalisation in the Mongolian case, examining the FSP and the CMP programmes across the scales of international activity, domestic politics, and local interactions. This broad perspective demonstrates the extent to which bureaucratic actors, infrastructures and discourses are interlinked and offers an ethnographic critique of the concept of efficiency integral to mechanisms such as proxy means testing (PMT).

Discourses of Corruption, Dependency and Discontent
Mongolia's seventy years as a socialist satellite of the Soviet Union remain a part of people's living memory. During those years, employment provided the foundation for the country's 'welfare' system, with workers assigned jobs in the city in factories, state enterprises, and cultural establishments, or roles in rural agricultural and farming collectives (Gochoosuren 2013). In the present era, known as "the age of the market" (zah zeeliin üye) (Wheeler 2004), the welfare system consists almost exclusively of a combination of pensions and benefits that are awarded according to categories of merit or vulnerability. For many in the ger districts, these benefits are a vital and significant portion of household income; Mongolia's welfare system is recognised as being generally pro-poor: the poorest 20 percent of the population receive 34 percent of total social welfare transfers (World Bank 2015: 12). However, while popular support for certain merit-based benefits -such as that awarded to the "Elderly with State Merits"continues, as the country's economic growth has slowed over recent years, the migrants of the ger district have become increasingly viewed as undeserving recipients who have become dependent on "hand-outs" (Yeung and Howes 2015: 23). Furthermore, as a series of scandals regarding the misappropriation of public funds -such as the infamous small and medium enterprise (SME) affair that revealed elected officials to have shamelessly channelled almost 1.3 million USD in government loans to family and friends (The Diplomat 2018) -have continued to erode public confidence in government accountability, welfare becomes a site of suspicion: how can the population be sure that only the neediest are receiving aid when examples of corruption pervade elsewhere in state activity? Thus, partly in response to the pressure to reduce 'dependency', and partly in an attempt to respond to calls to make sure Mongolia's welfare system is reaching those who need it most, a single needs-targeted food stamps programme has been introduced. A targeted benefit requires, first and foremost, a system of evaluation. In Mongolia's case, needs are determined according to proxy means testing (PMT), a system recommended by institutions such as the World Bank for use in contexts where income data is unreliable or difficult to access (World Bank n.d.). Rather than using financial data, PMT relies on making an "informed guess" regarding a household's income and consumption based on the household's characteristics (World Bank n.d.: 2). In Mongolia, data on each household is collected by street-level bureaucrats who visit people's homes and fill out a standardised survey that covers the household aspects that are then used to determine eligibility.
As the care administrator for the area, Ganbold sits on the fourth level of the governance structure of Ulaanbaatar: Democratic government in Mongolia has five levels: national government (1 house, 76 elected members); city administration (28 elected members, with a powerful city mayor); district administration (25 elected members); Khoroo (1 elected manager, paid staff, including social welfare and civil registration officer); and units (sub-groups of Khoroo, with unit leaders) (Terbish/Rawsthorne 2016: 89-90). 7 The city of Ulaanbaatar is composed of nine administrative districts (düüreg) divided into 141 sub-districts (horoo). Ganbold is responsible for the area of Aglag, one of the twenty-eight administrative sub-districts (horoo) of the Bayanzurkh district (düüreg). Not all who live in Ulaanbaatar's ger districts can be considered rural-urban migrants. Indeed, the ger district is a longstanding and widespread urban form in Mongolia, one that long-predated the Soviet-era city centre (Campi 2006;Fox 2019a). However, Aglag is an area at the eastern edge of the ever-expanding city that only became residentially inhabited following the end of socialism. Aglag's registered population now fluctuates around 15,600 (divided into ~4,000 households), growing from 9,885 in 2007, when the first population census of the area was conducted (http://www.ubstat.mn/StatTable=221). It is known as an area with a high percentage of families living in poverty. This information comes from the Household Survey.
|| 7 There is no consensus for the transliteration of Mongolian into English. Thus, horoo is often written as khoroo.
While the survey is meant to be conducted every three years, various circumstances can affect this. For example, the second round of surveying that was meant to be carried out in 2016 was delayed by the parliamentary election that took place that summer. As people's circumstances also fluctuate over the years, the sector workers and the care administrator may carry out spontaneous visits to households that request a reassessment of their level.
During my time at the horoo, I accompanied Ganbold on his evaluative visits to households in the district. On these visits, Ganbold bears witness to the living situation of the household in question. Accepting a cup of tea or simply hot water as he sits to complete the survey, he is obliged, however briefly, to enter the lifeworld of that family (Dubois 2014). A resident of this ger district since his childhood, the little wood and brick houses (baishin) or circular gers he evaluates are not unfamiliar territory for Ganbold, although the poverty he regularly encounters is not his personal experience thanks to his government job and his wife's work as a doctor. Nevertheless, his role as translator -translating the living reality of a household into its paper version (Jacob 2007: 250)forces him to carry out a standard practice of bureaucratic dematerialisation (Hull 2012a;2012b). As Ganbold completes the survey, through the series of marks he makes on the pre-printed paper, the household whole is rendered into parts, parts legible to the bureaucratic form (Scott 1998: 22), transformed invisibly into proxy values for the family's capacity to afford food.
To the designers of the survey and proponents of PMT, Ganbold's work represents good, transparent practice. However, for families living in the ger district, the dematerialisation experience is far from transparent nor legible. As one woman exclaimed after describing how she had been turned down for food stamps, despite being the sole breadwinner with five children to feed, "They come to your house and they check all the things you have. Then they tell you, you don't need help, you have a freezer. But what are they talking about? When there's no food, can you eat the freezer?"

Mongolian Social Welfare
Although much has been written in anthropology about the political and economic reforms that took place in former socialist countries following the collapse of the Soviet Union (Humphrey/Sneath 1999;Verdery 1996), less attention has been paid to the transformations that took place in their social welfare sectors (Cerami/Vanhuysse 2009). Furthermore, while studies can be found in social policy research circles that review or compare case studies from central and eastern Europe (Jolanta 2006;Kornai 1997;Kuitto 2018;Rosenberg 1991), or the post-Soviet states of central Asia (McKee et al. 2002), Mongolia has remained largely overlooked (Terbish/Rawsthorne 2016: 90;Smith 2015: 159). 8 Mongolia's social welfare history begins similarly to those of the former soviet and socialist nations (Cook 2007). Before the democratic revolution, the socialist state was entirely responsible for the population in the sense of providing education, employment, healthcare and pensions, based on the Marxist principle of 'full employment'; in other words, Mongolia was "a model socialist welfare state" (Smith 2015: 162). The budget for social security was subsidised by the Soviet Union (Mongolia being a member of the socialist Council for Mutual Economic Assistance), but the state covered ninety percent of the costs of social security, which was then administered by the Trade Unions. 9 The end of the socialist period marked the end of social security as Mongolians had known it and the beginning of the Mongolian social welfare system.
Following these initial chaotic years, the Mongolian parliament began to pass legislation establishing a social welfare system, modelled on welfare states of Western liberal democracies (Gochoosuren 2013: 88). Having been encouraged by international financial institutions to conform to the ideology of eco-|| 8 For an overview of social work in Mongolia see Namdaldagva (2014). 9 See Atwood (2004: 185-187;514-516) on the relationship of Mongolia and the Soviet Union. 10 Healthcare expenditure in Mongolia fell from 5.7% of GDP in 1991 to 4.4% of GDP in 1996; however, this drop was less precipitous than Mongolia's central Asian neighbours (Tajikistan's healthcare expenditure dropped from 6% of GDP to 1.1% over the same period), because of "inputs from external donors, who contributed nearly one-third of the GDP at that time" (Ulikpan et al. 2014: 2). nomic shock therapy and the Washington consensus, domestic leaders now faced the challenge of the social consequences of neoliberal economic policy. With a modest return to economic growth in 1994, in terms of GDP, the newly created welfare system was thus established to counter the negative impacts of transition by providing support for the most vulnerable members of society. These included, the elderly, the disabled, mothers and children, female-headed households, the unemployed, small-scale herders, and pensioners. As such, an entirely category-based welfare system took shape.
The expansion period from 2005 coincides with a period of economic growth in Mongolia that was fuelled largely by the expansion and liberalisation of the mining sector. This period also saw successive governments compete for domestic political support through promises to expand social welfare, particularly through direct cash transfers.
In general terms, Mongolia is considered to have a relatively generous welfare system. In 2013, government expenditure on social welfare was 2.78 percent of GDP, compared to an average of 1.6 percent among other developing countries (World Bank 2015: 11). The Ministry of Population Development and Social Protection (MPDSP) and the General Office of Social Welfare Services (GOSWS) oversees 71 programmes, 70 of which are either category or merit based. The one exception is the FSP, which is proxy means tested (PMT) and targeted at only the poorest five percent of the population. While families in the ger districts may qualify for a number of welfare benefits, the CMP and FSP touch every household and are thus the focus of this study.

The Child Money Programme (CMP) and the Food Stamps Programme (FSP)
The CMP is Mongolia's largest cash transfer programme and entitles all children to a monthly payment of 20,000MNT (USD 7) per month until they reach 18. It was initiated in 2005, following the election of the Mongolian People's Revolutionary Party (MPRP), who had made the programme a key part of their campaign. Originally the programme was meant to be means-tested and relatively limited (Araujo 2006). But, following difficulties administering the proxy means testing (the first of its kind in Mongolia), issues with inclusion and exclusion rates, and public complaints regarding the difficulty the poorest families faced in proving their eligibility, it was expanded to cover all children (0-18) in the country from January 2006 (Yeung/Howes 2015: 11). It is, by far, the largest wel-fare programme in Mongolia in terms of both coverage and budget. In 2013, the programme covered one million beneficiaries, 94 percent of children actually receiving the payment, and made up 1.41 percent of the 2.78 percent of GDP Mongolia spends on social welfare (i.e. more than half of the entire expenditure). Statistical analyses of Mongolia's welfare programmes reveal the importance of CMP payments to poor households. According to the World Bank, the single and largest benefit received by many individuals in the poorest quintile is the CMP (World Bank 2015: 36). In contrast to the CMP, the Food Stamps Programme (FSP) could be called Mongolia's flagship neoliberal welfare programme. It was established on a pilot basis by a coalition government led by the Democratic Party and financed by a loan from the Asian Development bank (ADB). The FSP is a means-targeted benefit distributed based on data collected by a household survey that is carried out nation-wide. The survey covers universal topics such as age, education, disability, and employment and also those tailored specifically for the Mongolian household, including type of residential building (house or ger), number of walls in the ger (this determines the size of the ger), and possession of any animals, their type and how many. The survey ends with a list of household items that the survey-taker should account for, whether they are present in the home and how many. These include things such as a fridge, a car, a television, a computer, a rice cooker, an oven, and rural items such as a tractor and combine harvester. The household surveys are conducted every three years by low-level bureaucratic workers who are each given a sub-section of a district to visit. Surveys are generally carried out in the household. These sector workers (hesegiin ahlagchid) then report back to the care administrator (mentioned above) who has a desk in the horoo. He or she is tasked with collecting the surveys and passing them on to the district office (düüreg) where the information is entered into the national database. Each household is then issued with a 'level' (tuwshin) between 1 and 20, 1 being the poorest, 20 being the richest. The care administrator in the horoo has access to this database and uses it to determine the types of care to which each citizen is entitled.
Although households in levels 1 to 6 are classed as 'poor', only those in level 1 are entitled to food stamps. Food stamps consist of a monthly allowance for basic groceries, originally 10,000MNT (USD 3.6) per adult, 6,000MNT (USD 2) per child, rising to 13,000MNT/adult (USD 4.7) and 6,500/child (USD 2.3) in 2016. Unlike the CMP, the FSP payments are not given in cash but preloaded on a card that can only be used in one specific supermarket in the area. 11 The card can only be used to buy Mongolian products; imported and luxury products are not covered, nor are items such as cigarettes and alcohol. Recipients of the programme are listed on a paper that is pinned to the noticeboard in the administrator's office; they must come in person every month to sign next to their names and receive the food stamps. I was told that if they do not come, they do not receive the benefit.
The basic contrasts between the CMP and the FSP are clear. One, the CMP, is a universal direct cash transfer that requires little administration, no means assessment and can be spent at the receiver's discretion; the other, the FSP, is a highly targeted, proxy-means-tested benefit that requires continual assessment and reassessment by officials on various levels and limits its recipients to only certain purchases. The differences between the two programmes likewise continue in terms of ger district residents' experience of them.

Living with CMP and (without) FSP
The importance of CMP payments among families in the ger districts of Ulaanbaatar was made very clear during my fieldwork. In essence, it functioned as a reliable, if bare-bones income for families living in otherwise volatile financial circumstances. The payment amount is very low: 20,000MNT (USD 7) for each child per month, but for those with little else it provides a modicum of security. The family with whom I lived had five children, three of whom were still under 18. Their mother, Ulzii, was fortunate to have a job as a cleaner in a private university dormitory. Her husband Badrakh, however, struggled to find anything beyond occasional work. The 60,000MNT (USD 21) was a vital supplement to Ulzii's salary, which was nominally above the minimum wage but, following automatic deductions for social insurance and repayments on a salary loan taken out to cover the wedding of her eldest daughter, provided barely enough for the most basic sustenance. A few plots away, Ulzii's younger brother Bold lives with his wife Saran and their four children in a small ger. Neither Bold nor Saran can find regular employment. Once small-scale herders in the eastern countryside, Bold had lost most of the family's inherited herds and goods through a combination of unfortunate circumstances (the socio-economic situa-|| 11 The card is used for food stamps in Ulaanbaatar. In the countryside paper vouchers are used (IPC-IG/UNICEF 2019: 111). tion described above) and drinking and gambling. For Bold's family, the CMP money is indispensable. Nevertheless, with a kilo of mutton costing approximately 6,000MNT (USD 1.60), beef 7,000MNT (USD 1.90), flour 1,000MNT (USD 0.40) and rice 2,500MNT (USD 0.90), the 80,000MNT (USD 28) per month in CMP barely allows the family to purchase enough food to survive. Indeed, the minimum wage, itself extremely low, is 192,000MNT/month (USD 67). 12 Families in Ulaanbaatar, including those in the ger districts, may have some relatives who still herd animals in the countryside. Countryside relatives can supply their city-dwelling counterparts with meat in exchange for cash or city goods, or else gift it to them when they visit them in Ulaanbaatar. These exchanges and gifts supplement a family's subsistence, although they may also be unpredictable and unreliable. On the other hand, families need more than just food to live, and on such a meagre income unexpected events, such as illness, can cause great hardship. The purpose of the FSP is to support families such as Bold's who live in the most intense poverty; however, its impact is limited: The average estimated amount received by households from the [FSP] program was MNT 529,141 (approximately USD 286) per year. Considering that this program is targeted to the very poor (the bottom 5 percent of the wealth distribution), the amount is not very generous, particularly given the large family sizes (5.8 members on average) of these beneficiary households. This means that on average, the estimated benefit size was MNT 91,231 (approximately USD 49) per person per year (World Bank 2015: 72, emphasis added).
In contrast to the CMP, the FSP is supposed to be a working example of how welfare 'should be done', at least according to the ADB. Nevertheless, for those whose lives are actually touched by the FSP, its effects are mixed to say the least. During my time researching in the horoo, the FSP stood out as a flashpoint of conflict and controversy on every level. The household survey was seen by both ger district residents and welfare bureaucrats to be inadequate. It was unable to take into account complex family situations, unusual housing issues, and especially the hidden burden of debts, debts from loans taken to purchase household items or debts that were often taken to cover salaries that failed to be || 12 In 2017 the minimum was raised to 240,000MNT/month (USD 84) and in 2019 to 320,000MNT/month (USD 112). CMP have remained the same thus far. A recent UNICEF overview of non-contributory social protection programmes in Asia reports that, according to an unnamed 'personal communication' source, starting in 2018 the CMP would become a PMTtargeted benefit aimed at 60% of all children (IPC-IG/UNICEF 2019: 110).
paid. The algorithm that calculated people's 'level' based on the survey was approached with suspicion. PMT relies on the secrecy of calculation to be effective, but this secrecy has the effect of making people feel disenfranchised.
The introduction of means testing has come under critique in other contexts. For example, proxy means testing in order to target higher education subsidies in Africa according to need is known to produce numerous errors: false positives (the exclusion of eligible families) and false negatives (the inclusion of ineligible families) (Tekleselassie/Johnstone 2004: 142). Referring to the use of assets as proxy indicator's for household wealth, the study's authors point out, "[w]hether an asset is a reasonable indicator of current income or ability to pay may depend on the culture and the economy", and furthermore, when the assets included in an assessment are highly illiquid, they can be faulty indicators of a household's financial situation. In the case of land, for example, a rising market may increase the 'value' of a home without any corresponding rise in household earnings (Tekleselassie/Johnstone 2004: 144). For other assets, such a freezer, proxy means testing can lead to a situation in which a family is forced to sell a valuable asset in order to afford food, an asset which, in a Kafkaesque bind, is itself vital for efficient food use and storage.
Welfare bureaucrats were also left with no way to explain to people why they had been put into a particular level, one that denied them assistance, except for to appeal to the fact that they knew as little about the calculations as the distraught citizen making the complaint. "We don't know anything about this 'programme'" they would insist, "the levels are calculated by the computer 'up there'!" The secrecy of the process that determines whether people are considered eligible for food stamps must also be understood within a wider context, where scandal after scandal has revealed the extent of corruption in the allocation of development bank finances (Chuluunbat/Empson 2018: 422), and where people have daily interactions that make it clear that those who have connections (tanil tal) in the right places get ahead. Thus, even if I did not witness Ganbold, the local care administrator, engaging in any outright manipulation of the FSP beneficiaries list, he still suffered from the effects of being assumed to be corrupt. As a man from the local area, he bore the brunt of his neighbour's frustrations on a near daily basis. Indeed, among the horoo workers, the FSP was highly unpopular not only for the conflicts it caused in the horoo but also the conflicts it caused in people's homes. Even if the intricacies of the algorithm remained unknown, it was obvious to people that having even one able-bodied man in the family was probably enough to scupper one's chances of receiving food stamps. Thus, people were always seeking ways to perform their poverty in order to satisfy the PMT model. It must also be remembered that poor households who do not qualify for the FSP are not only left feeling that they have been victims of corruption, but also have legitimate claim to being unfairly excluded. One third of Mongolia's population live in poverty; therefore, targeting only the poorest quintile guarantees that even those in poverty and need are by definition excluded.
The horoo workers also faced pressure to reduce the number of families on the FSP list. Removing qualified households was carried out in a meeting of the senior horoo staff, during which they discussed each case and voted whether to the let the family stay or not. In some cases, the relevant subsector worker was called upon to speak on the family's situation, but, more often than not, these workers most familiar with the living situation of the household were outvoted. Thus, even qualified households could be cut from the programme for the purpose of reducing cost. Finally, the work of administering the surveys and updating the database took up so much time from the horoo workers, it left little time for them to actually deliver assistance to anyone.
While CMP payments may be meagre, it is clear they are vital for families. Indeed, studies report that CMP can even be a main source of income for poor families; a truly desperate situation. And on aggregate, despite the small amount, the CMP is demonstrated to have poverty reducing benefits. The FSP, on the other hand, is experienced by many as a source of strife. This is particularly because it excludes so many who need it. In fact, so few people actually receive the benefit that some Mongolians do not even believe the programme exists! One of the aims of this paper is to interrogate how such programmes come to exist and the variety of actors and struggles that shape them. Thus, the following sections of this paper examine the issues around such programmes that largely remain hidden from people's everyday experiences, but nevertheless have a profound impact on what they have access to.

Borrowed Money versus Revenue Redistribution
The FSP is funded by a loan from the ADB. In 2008 the programme was started with a 12 million USD loan from the ADB, which was combined with 10 billion MNT from the Mongolian government. Among the international financial institutions that fund and assess Mongolia's welfare sector, the largest role is played by the ADB. The ADB has been "particularly active", as it states, in reforming the Mongolian social welfare system (ADB 2015: 4). The bank introduced the system of proxy means testing (PMT) on which the food stamps programme relies to 'target' its efforts on only the poorest Mongolian citizens (the lowest quintile), and it developed the 2012 Social Welfare Law that was supposed to move Mongolia away from "universal social transfer to a well-targeted safety net that protects the poor and is fiscally sustainable" (IEG 2015: 54). In 2015, the ADB agreed to loan the Mongolian government 150 million USD to support the FSP, released in two tranches in 2017. In return for the loan, the Mongolian government is required to demonstrate that it is improving its budget transparency, reducing its fiscal expenditure and moving towards the model of welfare the ADB considered most suitable.
In contrast to the FSP and all other social benefits in Mongolia that target recipients according to categories of merit or vulnerability, the CMP was founded on the concept of redistribution. Since the collapse of most industries in the country that followed the end of the Soviet Union, Mongolia has turned to the land and its vast deposits of coal, minerals and precious metals as the basis for the country's economy. Known to some as "Mine-golia" (Bulag 2009), for a brief period it looked like the state would be able to harness the country's wealth of natural resources. In 2005, the country saw its first surplus budget; in 2006 the Mongolia Development Fund (MDF) was established to utilise the profits from mining, collected under the Windfall Profits Tax Law. This law mandated that state mining revenues be used to stabilise budget deficits, make investments in small and medium domestic enterprises and support families and children (Yeung/Howes 2015: 12). In 2009 the government established the Human Development Fund (HDF) that replaced the MDF. The HDF differed from the MDF only in the way that no legislative limits were placed on its expenditures; its purpose remained the same: to redistribute mining revenues through social programmes and direct cash transfers. CMP payments thus come from this fund and are regulated by the law relating to the HDF, distinct legislation from that which governs the other social welfare programmes.
Issues of redistribution plague nations with vast national resources and inadequate checks on corruption and, unfortunately, Mongolia is no exception (Ulikpan et al. 2014). In a wider context of blatantly unequal 'sharing' of mining revenues, which are clearly being co-opted by powerful individuals (partly a heritage of rapid and 'haphazard' post-socialist privatisation), the piecemeal transfers of cash or 'shares' that politicians have actually given to Mongolia's citizens have also caused problems. Given its funding source, the CMP has, not surprisingly, been caught up in the debate over the disbursement of state funds.
As mentioned, the programme was initiated when Mongolia saw its first budget surplus in 2005. As the revenue (actual and projected) from mining continued to climb, both major political parties competed over how they would redistribute this wealth to citizens, and over the following years various transfers and promises were made, concurrent with the development of the CMP into the universal, child-centred programme it is today. Over this period, successive governments thus paid out both CMP payments and a series of cash transfers from the HDF and MDF. In theory, these were supposed to be tied to mining revenues, but in reality they were tied to political promises. For example, before the parliamentary election of 2008, the Democratic Party promised to provide 1 million tugriks (about $840) of what is called "Wealth Share" (Erdeniin huv') to every citizen every year should it be elected to power, while the MPRP outbid the DP by promising 1.5 million tugriks ($1,260) of what it called "the Motherland's Blessing" (Eh Oroni Hishig) (Bulag 2009: 132).
In 2010 and 2011, annual expenditures from the HDF actually exceeded mining revenues, leaving a gap that had to be plugged with borrowing (Yeung/Howes 2015: 16). In 2012, both major parties agreed to end the practice of promising direct cash transfers during electioneering, and the HDF expenditures, such as the CMP, no longer exceeded its revenues. Nevertheless, the CMP continues to be conceptually linked, particularly in the view of international financial institutions (IFIs) such as the ADB and World Bank, to politically motivated and inadequately regulated direct cash transfers. As such, the push to 'discipline' Mongolia's financial expenditure on social welfare by such organisations is linked directly to how these institutions view the CMP. The CMP is singled out as being 'fiscally unsustainable' (World Bank 2015: 15) and contrasted with the FSP, which is praised for being an example of a "modernized" welfare programme, because it is targeted and apparently sustainable (ADB 2015: 4). The FSP is much less expensive than the CMP; annual CMP programme expenditure being 221.5 billion MNT, but only 16.9 billion MNT for the FSP (IPC-IG/UNICEF 2019: 110-111).
Examining the funding background of the CMP and FSP highlights the duration and extent to which IFIs influence social welfare systems in countries such as Mongolia (Deacon/Hulse 1997;Deacon 1995; see Shaw 1999 on Sri Lanka). 13 Such a study also reveals the first level of conflict out of which these || 13 Although, as Kilby's research demonstrates, international financial institutions are, in turn, influenced by their nation-state donors (Kilby 2006). According to Kilby, the ADB is particularly influenced by Japan and United States (Kilby 2011). policies and programmes (that have significant effects on people's lives) emerge. On this level of international finance, business and politics, actors compete to enforce their worldview and, in doing so, seek to shape both the entities in question and the terms of their interaction. While an international development bank like the ADB claims to concern itself with welfare efficiency, the bottom line must always be the repayment of the loans. As Shaw writes on the World Bank, "While the recognition that market-oriented economic policies would not by themselves solve the problem of poverty led the Bank to adopt specific poverty-focused strategies, market-led growth remains its undisputed primary objective" (Shaw 1999: 828). Fiscal responsibility and financial sustainability (to repay debts) are therefore pushed as core values that must outweigh all others. In this struggle, the government of Mongolia is painted by IFIs as 'irresponsible'. It also appears to be the weaker party, locked into debt arrangements that tie its overly beneficent hands with strings. With the passing of the 2012 Social Welfare Law, it looked as if the ADB had succeeded in reshaping Mongolia's welfare system to its liking. However, this was not to be the caseor, at least, not yet.

Democracy and Debt
The capacity of IFIs to exert their influence over Mongolia's social welfare expenditure has, I propose, vacillated according to two factors: the economy and elections. As a general rule, the power of the ADB has risen and fallen in opposition to Mongolia's own economic strength. For example, it was in the wake of the 2008 global financial crash and the subsequent fall in commodity prices that the Mongolian government was pushed to pass the Fiscal Stability Law (FSL), designed by the World Bank and International Monetary Fund. This law tied government spending to certain measures of growth and debt, capping the fiscal deficit and public debt to 2% and 4% of GDP, respectively (ADB 2015: 3). However, with the speedy recovery of the Mongolian economy by 2010, the government found various ways to circumvent the new law and instead expand social welfare programmes, such as making the CMP a universal transfer. 14 The || 14 In order to circumvent the FSL, the government created the Development Bank of Mongolia, which it designated as being subject neither to the restrictions of FSL nor the rules of government procurement (IEG 2015: 12). ADB pushed back against such moves and encouraged the government to pass the Social Welfare Law it had designed during the economic downturn. This was lauded by the ADB as the beginning of the end of Mongolia's overly generous and financially irresponsible welfare system. The law introduced as its centrepiece a 'Poverty Targeted Benefit scheme' and also abolished the CMP. It was passed in 2012, but, despite the ADB's celebrations, Mongolia never actually put the law into practice and instead continued to distribute the popular CMP.
In 2012, the government had the confidence to brush off the concerns of IFIs thanks to GDP growth figures from 2011 that were some of the highest in the world: 17.291% (World Bank Figures 2019). The optimism of the Mongolian 'wolf economy' period, however, was short lived, and, in each of the subsequent five years, GDP went into reverse, dropping to a post 2009 low of 1.168% in 2016. Deflated by the effects of falling commodity prices and sharp drops in foreign direct investment in the country (by 55% in 2013 and 80% in 2014), Mongolia's soaring national debt and budget deficits are of serious concern to international development banks. Reports from 2015 by the World Bank, the ADB and other monitoring bodies, such as the Independent Evaluation Group (IEG), thus all make a point to criticise the ongoing existence of the CMP and its broad-brush approach to poverty alleviation.
Mongolia's swift establishment of a functioning democracy following the end of state socialism played a significant role in attracting international donors (Ulikpan et al. 2014: 24986) However, the second factor that limits the capacities of IFIs to enforce the policies that are written into loan agreements is the power of democratic elections. For example, Mongolia's brief economic recovery in 2010 that facilitated the circumvention of the FSL also coincided with an election campaign.
Furthermore, I argue that the main reason that Mongolia has never replaced the CMP with the Targeted Poverty Benefit is that any attempt to do away with the payments has proven to be politically costly, demonstrating that domestic politics, particularly in states such as Mongolia with "inclusive and stronger representative structures", may function "as a counterweight to global economic pressures" (Cook 2007: 45). The funding source of the CMP, in conjunction with its universality, have contributed to the programme's popularity among Mongolian ger district residents, especially those who have seen no other return from mining. Despite being a small amount of money, it is understood as a manifestation of people's basic right to participate in the profits of their land. Indeed, the general neglect to which ger district residents are subjected is suddenly taken very seriously in the lead up to elections, when politicians are forced to reckon with these citizens' significant voting power (Fox 2016). From this per-spective, democracy and public protest play a role in limiting the power of IFIs to reshape Mongolia's welfare policies.

Conclusion
Beyond providing insight into a unique, post-socialist welfare state that is long overdue, this paper contributes to the literature on both street-level welfare bureaucracy and the globalisation of social policy. Interweaving participant observation and ethnography with materials drawn from policy and funding documents, the paper presents a historically-informed analysis of two key social welfare programmes in Mongolia, revealing the ways that social policy is formed under multiple pressures, for example the weight of an IFI, such as the ADB, pitted against the political force exerted by a large voting bloc, such as ger district residents. The approach taken here demonstrates the importance of multi-scalar analyses for welfare policy research. For ger district residents, largely unaware of the role of the ADB in shaping Mongolia's welfare system, both the CMP and FSP are considered Mongolian government programmes. Thus, their complaints land at Ganbold's desk, who duly washes his hands of personal responsibility, referring to forces outside his control, such as the algorithm 'up there' that assesses their right to care. On the other hand, policy research carried out by IFIs routinely excludes the testimonies of welfare recipients and street-level bureaucrats. As this comparative study has demonstrated, the CMP, through its universality, provides an economic lifeline to many families in the ger district who, being rural-urban migrants, would struggle to access more restricted and regulated forms of assistance (such as FSP), even if in real terms they would be eligible. The supposed 'efficiency' of the FSP is thereby challenged by this study's ethnography and posited to be a euphemistic cover for the central aim of the ADB, which is to ensure that its loans are serviced, above all. In the clear absence of an equitable exploitation of Mongolia's natural wealth -the property of the nation -the CMP stands as living claim of future generations to a share of their lands' resources and should be understood therefore in moral, not only economic terms.

Short Biography
Elizabeth Fox is a social anthropologist and visiting scholar at the Mongolia and Inner Asia Studies Unit (MIASU), Department of Social Anthropology, University of Cambridge. Her ESRC-funded PhD (UCL, 2019) was based on long term research in the outskirts of Ulaanbaatar, Mongolia and addressed topics including infrastructure, kinship, welfare bureaucracy, language, rural-urban migration and postsocialism. Her postdoctoral work, supported by an MIASU Early Career Researcher grant, focuses on the circulation of meat in contemporary Mongolia as window on both ongoing social transformations and the tension between state-led pursuits of trade and sovereignty.