The moral and political economy of the pandemic in Bangladesh: Weak states and strong societies during Covid-19

Abstract

As the Covid-19 pandemic spread in 2020, the government of Bangladesh ordered a lockdown and promised a program of relief. Citizens complied at first, but soon returned to economic and social life; relief proved slow and uncertain, and citizens could not rely on government assistance. The government tacitly and then officially permitted the lockdown to end, despite a rising Covid-19 caseload. This article draws on theories about state capacity to make and enforce policy to understand why Bangladesh proved unable to sustain a lockdown deemed necessary to contain the pandemic in this densely populated, low income country. Drawing on original qualitative mobile phone-based research in six selected communities, this article examines how the state exercised its capacities for coercion, control over lower factions within political society, and sought to preserve and enhance its legitimacy. It concludes that despite a) the growth in the capacity of the Bangladeshi state in the past decade and b) strong political incentives to manage the pandemic without harm to economic wellbeing, the pressures to sustain legitimacy with the masses forced the state and its frontline actors to tolerate lockdown rule-breaking, conceding that the immediate livelihood needs of the poor masses overrode national public health concerns. Chronically unable to enforce its authority over local political elites, the state failed to ensure a fair and timely distribution of relief. The weakness of the Bangladeshi state contrasts with the strength of widely shared ‘moral economy’ views within society, which provided powerful ethical and political justification for citizens’ failures to comply with the lockdown, and for officials’ forbearance in its enforcement. The Covid-19 pandemic highlights both the importance of state capacity in managing novel shocks from within the global system, and the challenges in settings where weak states are embedded in strong societies.

1. Introduction

Pandemics have historically provided a rigorous test of governments, checking their capacities to design and implement public policies that pit the material interests of economic elites against the fears of the masses, and both against technocratic and medical expertise (Slack, 1995). But which governments succeed in managing pandemics effectively, and why? An influential - but unproven - hypothesis has been that authoritarian regimes have performed better in managing and containing the Covid-19 pandemic, because they possess and are willing to use more coercive power than democracies (Frey et al., 2020). But the successes and failures of managing Covid-19 have failed to respect regime types, and policy responses have instead varied depending on: learning from past pandemics, state capacity to implement and build support for policies, the nature of national leadership, state-civil society relations, and alertness to the impacts on vulnerable groups (Capano et al., 2020). It is increasingly clear that the ‘success of governmental social control depends more on voluntary compliance than on government enforcement’ (Kleinfeld, 2020). But when do citizens comply without the need for the state to resort to coercion? This article draws on theories of state capacity to make and enforce pandemic policies to address this question, focusing on the experience of Bangladesh.

In early April 2020, Bangladesh saw an attempted nationwide lockdown to contain the spread of Coronavirus, but the effort was largely abandoned by mid-May 2020, despite a rising caseload of infection. During these six weeks, citizens first tried collectively to maintain the lockdown; by May, many were returning to normal life, rejecting the lockdown as unworkable given the devastating loss of livelihood and lack of compensatory relief from the state. Joel Migdal’s concept of state capacity (Migdal, 1988) and Michael Mann’s concept of ‘infrastructural power’ or the ‘capacity of the state to actually penetrate civil society and implement its actions across its territories’ (Mann, 2008, 355), provide a starting point for analyzing the dynamics of the lockdown in Bangladesh during this crucial period. In addition to these macro-sociological perspectives we draw on institutionally grounded rational choice theoretic approaches for a better understanding of the strategic incentives of the actors involved, and the mechanisms through which they have operated in Bangladesh. In order to think through how state action and citizen responses interacted to break the lockdown we draw on Margaret Levi’s analysis of the conditions under which citizens comply with government policies, including the ideas that legitimate such compliance (Levi, 1997). Alisha Holland’s concept of ‘forbearance’ further helps us to explain the conditions under which states behave leniently with rule-breakers (Holland, 2016). Together these theorists help us specify and elaborate on the ways in which Bangladesh’s state-society relationships shaped its Covid-19 response.

The article asks, why did the lockdown hold at first, despite concerns about the subsistence crisis it engendered? And what changed to stop citizens from complying, and – finally – to push the state to withdraw the lockdown in full? Bangladesh presents a valuable case for studying the dynamics of state capacity and state-citizen relations in the pandemic for several reasons. First, there are compelling humanitarian, developmental and public health motivations: Bangladesh has a large, densely-packed population of 170 million living with high levels of poverty, vulnerability and a weak health system: the impacts of an unchecked Coronavirus epidemic could be catastrophic (World Bank, 2020). Already by week two of the lockdown, more than two-thirds of people living with poverty and almost as many above the poverty line had lost all income, reflecting the precarious and informal nature of most employment (Rahman & Matin, 2020). State capacity is of special interest in Bangladesh because the state has demonstrably grown its power in the past decade, and has greatly more fiscal and administrative capacity and greater autonomy with respect to politics and the security services than in its first three decades (Hassan & Nazneen, 2017). As this article shows, citizens’ expectations of the state also appear to have grown over time (although they also appear not to have been met, at least with respect to the Covid relief program). And finally, while Bangladesh is no ‘developmental state’ in the East Asian model (Hassan, 2013), it has nonetheless invested relatively successfully in inclusive public policies in human development and social protection (Mahmud et al., 2013), and is noted in particular for its capacity to manage disasters and subsistence crises (Hossain, 2018).

This suggests there are powerful political incentives and growing state capacities to address the pandemic. And yet as the article uncovers, Bangladeshi societal norms and preferences have remained a powerful check on state action in the moment of the pandemic: Sarah White’s observation à la Joel Migdal, that Bangladesh is a weak state in a strong society, appears to hold some 20 years later (Migda, 1988, White, 1999). As will be discussed further, the popular legitimacy of the Bangladeshi state remains closely dependent on its policy performance, notably its ability to protect its millions of precarious and vulnerable citizens from the crises of subsistence and survival to which they are frequently exposed. Our analysis here of state capacity draws attention to the significance of popular consent in pandemic statecraft, pointing to explanations of governance that emphasize interactions between states and their citizens, rather than a notion of power rooted in whether the state can dominate its population by imposing unpopular policies upon them.

An important aspect of the state’s response to Covid-19 has been the ‘moral economy’ thinking that created both a strong justification of survival for breaking or abandoning the lockdown, and an equally strong sense that the state is responsible for protecting people during the crisis (see also Jahan & Hossain, 2017). By ‘moral economy’ we follow E. P. Thompson in recognizing a widely resonant set of political cultural beliefs in the right to subsistence, and the responsibilities of public authorities to act to protect that right; this justifies popular resistance to state policies that threaten subsistence (Thompson, 1991), in this instance, making it legitimate to break lockdown in order to earn a living. State legitimacy depends so substantially on its ability to protect against disasters that, it is argued here, once it proved unable to deliver relief on the necessary scale and time-frame, the state was forced to back down from its initial plans to lockdown and contain the pandemic. An un-institutionalized but deeply held set of moral economy beliefs about the responsibilities of public authorities to protect against subsistence shocks in effect prevented or preempted a more coercive response to locking communities down, licensing both a degree of rule-breaking by citizens and forgiveness – or forbearance - by public authorities.

The article is organized as follows. The next section provides contextual background to the pandemic in Bangladesh, and explains the key concepts guiding the analysis. It then goes on to describe the research methods used in the research. The third section presents the key findings from the six research sites, organized around the two periods of the lockdown in which the research was conducted. A fourth section discusses these findings in relation to the concepts of state capacity, infrastructural power, and the moral economy. It concludes with reflections on the public policy challenges facing what remains a ‘weak state in a strong society’ during a pandemic, and on the theoretical advances made in the paper by drawing attention to moral economic values in shaping the state’s capacity to enforce public policy.

2. Theoretical framework and research methods

2.1. Concepts of state capacity: infrastructural power, contingent compliance and forbearance

To help explain what has happened in Bangladesh, we draw on Joel Migdal (1988) for a discussion of state capacity, or its capability to ‘penetrate society, regulate social relationships, extract resources, and appropriate or use resources in determined ways’ (1988, 4; emphasis in original). Migdal’s observation that “[social and political organizations] … and any others enforcing rules of the game—singly or in tandem with one another, have offered individuals the components for survival strategies” (ibid: p. 29) provides a useful frame to capture the dynamics of state-society relations as these embed and impinge on how state and societal actors cope with the unprecedented social and economic crises that Covid-19 has ushered in. Our understanding of state capacity is further informed by Michael Mann’s distinction between two types of state power - despotic and infrastructural. Despotic power indicates the “range of actions that the state elite is empowered to make without consultation with civil society groups” (Mann, 2008, 355). For our purposes, we analyze the state’s autonomy vis-à-vis political society. Infrastructural power refers to the “…capacity of the state to actually penetrate civil society and implement its actions across its territories” (ibid: 355). Such capacities include tax assessment and collection and, more relevant to Covid, the provisioning of basic services and subsistence needs, to include employment and social assistance where necessary (Mann, 1988). Infrastructural capacity also implies that the state can coordinate the activities of civil society through its own infrastructure, including bureaucratic oversight agencies.

In this article we explore the implications of multiple dimensions of state power for the lockdown: its coercive capacity; its effective autonomy vis-à-vis political society, and the preservation and enhancement of state legitimacy. However, we do not view the infrastructural power of the state as a simple matter of domination, in part because of what we already know about how the Bangladeshi state exercises power. This is that it commands few of the resources usually associated with capable states, and to the extent that it has succeeded in pushing through its policies, it has done so by pursuing policies in line with the needs and preferences of Bangladeshi society broadly, knowledge of which is enabled by a comparatively flat and homogenous social structure, and an absence of important differences between the elites and the masses they govern (Hossain, 2017). The balance of power or political settlement within Bangladesh rests on a social contract that depends on at least minimal protections for the masses, while at the same time rewarding political and economic elites for their loyalty or consent, minimizing intra-elite conflict as well as threats from below (Hassan, 2013, Khan, 2013).

For these reasons, our analysis of the power of the Bangladeshi state in relation to the pandemic explores the conditions under which citizens complied with lockdown policies, and the related matter of the conditions under which policymakers tolerated or turned a blind eye to rule-breaking. We look at how state capacity engendered (non–)compliance with help from Margaret Levi’s model of ‘contingent consent’, which recognizes that while ‘some compliance is the result of coercion or other sanctions and incentives … at least some compliance expresses a confirmation of a belief in the rightness of the policies and of the trustworthiness of the government actors implementing them’ (Levi, 1997, 18). How citizens viewed the effectiveness and fairness of the official pandemic response is, we believe, crucial to an understanding of why they did – and then did not – comply with the lockdown regime.

We also borrow from Alisha Holland’s concept of ‘forbearance’, or ‘intentional and revocable government leniency toward violations of the law, as a distinct phenomenon from weak enforcement’ (Holland, 2016, 233) to make sense of why the Bangladeshi state appears to have chosen to drop the lockdown and related policies, even though it probably had the coercive capacity to enforce them. Holland’s theory of ‘forbearance’ helps us to move beyond an analysis of the Bangladeshi state as wholly characterized by incompetence or incapacity; forbearance, or selective tolerance of rule-breaking by some people, is an active and intentional, often distributional, strategy by some politicians. Crucially, an understanding of why some politicians sometimes tolerate rule-breaking as a political or welfare strategy helps us recognize that unenforced policies are not necessarily evidence of state incapacity, but may also encode political intention.

2.2. Bangladesh: a ‘weak state in a strong society’?

Scholars who have reflected on Bangladesh’s state-society relations in earlier decades have observed that the Bangladeshi state has been weak vis-à-vis society (White, 1999, Blair, 1985), performing at the lower end of the Migdalian scale of capability. State attempts to penetrate society through institutional reforms of local government have been weakly institutionalized and vulnerable to the vagaries and whims of changing regimes. State capacities to regulate social relationships and dominant norms are similarly weak, manifest in the inability to stop practices of dowry or child marriage. Human development gains such as around gender equality have emerged chiefly where the state followed, rather than led, changes in social attitudes (Hossain, 2017, Kabeer, 2001).

The state’s dismal performance in extracting critical resources is evident from its weak revenue effort. For decades Bangladesh has failed to increase its rate of tax collection beyond single digits (tax-GDP ratio), and is consistently ranked the worst performer in South Asia (Hassan & Prichard, 2016). The local state’s capacity to tax is particularly illustrative: local councilors are reluctant to tax rural citizens in order to maintain an image as benevolent rather than resource extracting rulers, as a bid to sustain electoral popularity (Ahmed, 2020, Yunus and Sultan Hafeez, 2015).

However, successive governments have striven to meet citizens’ expectations for protection against the kinds of subsistence crises and life-threatening shocks to which they are so vulnerable (Hossain, 2017, Hossain, 2018). A powerful ‘moral economy’ or set of expectations about the rightful behavior of ruling elites in times of crisis, has shaped public policy to a significant, if generally invisible, degree (Jahan et al., 2017, Jahan et al., 2016). The role of subsistence crises and disasters in critical turning points in its political history has meant protection against mass livelihood shocks frames the social contract in Bangladesh (Hossain, 2017, Hossain, 2018), and is therefore likely to be a key determinant of state legitimacy at a time of economic and public health crisis.

Bangladesh’s transition to electoral democracy started in 1990, but instead of democratic consolidation, by 2014 had yielded a political settlement that can be described as a dominant party state of an authoritarian variant (Hassan & Raihan, 2017). The current ruling party, Awami League (AL), came to power in 2009 and immediately removed a constitutional provision that stipulated elections were to be held by an interim neutral caretaker government; it then won two elections (in 2013 and 2018), widely perceived by both domestic and international observers as extensively rigged (Riaz, 2019, ICG., 2015). The nature of the current regime can be described as a de facto party-state, whereby the party machine, which is deeply rooted in the society, dominates the state functionaries (national and local bureaucracies, elected local government). The implications of this domination will be demonstrated in our subsequent discussions on the governance of relief operations related to the Covid-19 crisis.

A major feature of the Bangladesh polity is the near absence of class-based politics and parties: left parties are insignificant actors, no party exists to represent the peasantry, and industrial trade unions are weak and largely coopted by the ruling party. Two ‘catchall’ parties - AL and BNP (Bangladesh Nationalist Party) dominate national party politics. Although both in theory represent cross-class interests, they are in effect beholden to elite segments, a class bias that has been reinforced in recent years as business elites have progressively captured both electoral politics and state policymaking processes (policy and regulatory capture). The business class is in turn dominated by the readymade garments export sector factory owners, whose influence over politics owes to their collective power over foreign currency earnings and as mass employers, as well as to their role in financing party political competitions (Hassan & Raihan, 2017).

Recent decades have also witnessed a closing of civic space, buttressed by the ruling elites’ monopolistic control over civic and professional associations, featuring self-restraining practices by non-governmental organizations (NGOs) and increased state control over the media through draconian laws and criminalization of journalists. In fact, such draconian laws have been used, in the context of Covid-19, against journalists and healthworkers who exposed the state’s inefficient and non-transparent management of lockdown, related health services and relief distributions (Amnesty International, 2020). The upshot is that the polity, economy, and society of Bangladesh largely lack countervailing powers by non-elites and civil society actors. As we discuss in this paper, such an asymmetry of power has significantly structured the incentives and behavior of the state, and consequently its capacity to deal with the management of lockdown and related economic crisis coping strategies.

The authoritarian nature of the Bangladeshi state may suggest a ‘despotic’ state in the sense of Mann as discussed earlier, but this would suggest the state has considerable autonomy vis-à-vis societal actors. This suggestion is false. The state has been progressively colonized by economic elites and lost much of its room for maneuver or ‘embedded autonomy’ (Evans, 1995) in the relevant policymaking domains. A more realistic and nuanced portrayal of the state would characterize it as soft authoritarian with relatively low despotic power. This characterization of the state would seem counter-intuitive, given the aggressive campaign of ‘disappearances’ and extrajudicial kills through which the government suppresses opposition in the political society and individual dissent (intellectuals, journalists, rights activists) in civil society. Nevertheless, it is softly authoritarian in how it compromises its own policies and plans, tolerates rule-breaking to cater to the interests and expectations of the popular classes, and reverses policies when they become plainly and widely unpopular. As our empirical narrative of the management of lockdown will show, the state’s behavior in relation to lockdown has been characterized by confusion, incoherence, and reversal.

Recent empirical evidence shows that the current regime, despite its (soft) authoritarianism, enjoys a high degree of legitimacy among the majority of the population for its political, economic and social performance (Meisburger, 2017, Taylor et al., 2018, TAF and BIGD, 2019). Evidence also indicates that citizens’ expectations and demands from the state are informed more by its ability to advance positive liberties (economic rights, for instance), and less by negative liberties (freedom of association or speech, for instance). The state is broadly perceived by the popular classes as a benevolent patron-state and provider/protector, rather than as their political representative. Such expectations and perceptions of the popular classes are on the radar of the political elites. As the current Prime Minister of Bangladesh noted recently:

If I can provide food, jobs and health care, that is human rights … What the opposition is saying, or civil society or your N.G.O.’s — I don’t bother with that. I know my country, and I know how to develop my country (Abi-Habib & Manik, 2018).

State elites, being conscious of the expectations and perceptions of the popular classes, strive to ensure basic welfare needs of the poor, in particular during episodes of subsistence crisis. Bereft of democratic, in particular electoral, legitimacy (especially among the middle class and educated both urban and rural) political elites are keen to preserve and enhance their legitimacy by ensuring positive liberties (economic welfare, poverty alleviation, infrastructure development etc.), with particular sensitivity to the demands of the popular classes.

It is with these understandings of the nature of state power in Bangladesh that primary research was designed to understand the dynamics of citizen-state relations in the lockdown.

2.3. Research methods

The research on which the present paper draws was designed to produce case studies of community dynamics in response to the COVID-19 pandemic and lockdown, with the aim of exploring issues of trust and compliance in relation to policy adherence. The research team had previously worked in each selected location, and had built rapport with local key informants and acquired basic knowledge of the area’s history, economic and social structure, as well as key contacts. The present paper draws on findings from community case studies of six locations across six districts, selected to give an illustrative sense of the experiences of the lockdown and Covid-19 response across different types of social settings. The research sites included the following types of settlement:

• •Rural haor (low-lying wetland) area, heavily agricultural, but with a high proportion of people working in the garments industry in Dhaka and surrounding areas
• •Rural coastal area with high salinity, a single rice crop; the poorest of the six sites
• •Peri-urban area in northern Bangladesh, where the economy was dependent on sales of vegetables to the capital and, to a lesser extent, on garments work in the capital
• •Peri-urban area in western Bangladesh, where commercial mango production and rice cultivation formed the major economic activity
• •A neighborhood in a major industrial city close to Dhaka, with micro industrial enterprises and, particularly, small garments factories
• •An informal settlement in the heart of the capital city.

A roughly 60-minute semi-structured interview checklist was administered to key informants through mobile phone interviews. Snowball methods were used to identify key informants for interviews. Key informants were selected from specified occupation and social categories, with the aim of reaching influential people, and those whose life experiences and conditions were likely to give them specific insights into the crisis from the perspective of vulnerable or affected groups. Interviewees were drawn from a) community leaders or elected representatives; b) teachers, NGO workers and formal sector employees; c) small farmers and local business owners; d) frontline healthcare workers; e) mothers of young children; f) daily wage earners (agricultural laborers, transport workers); g) students; h) imams or religious leaders; i) law enforcers and j) recipients of the government’s relief program. In each site, interviews were undertaken with between 7 and 10 individuals, in some cases on more than one occasion. Interviews were undertaken by mobile telephone, prearranged through pre-existing contacts. Multiple informants from different backgrounds, occupations and socioeconomic categories from within a single site also enabled a degree of triangulation of findings with respect to events and perceptions within their locality.

Limitations of the data collected include the limitations of scale: although the community case studies offer rich insights into the dynamics of the pandemic in selected locations, they cannot tell us whether and the extent to which these dynamics are present more widely. The research also did not specifically cover ethnic or religious minority groups or locations; the communities covered are predominantly majority Muslim populations. Male perspectives are greatly over-represented; of the 92 interviews, only 20 were with women. And finally, there was a deliberate over-sampling of influential or authoritative members of the community, rather than the poorest or most marginalized. This was in order to collect views of people with access to public authorities at higher levels, and/or greater capacities to gather information about events. Among the advantages of the methodology include the longitudinal or repeated interviews with the same community members, which enabled a coherent picture of the direction of change in each setting.

Telephone interviews were recorded or notes were taken with permission of the interviewee, and the researchers analyzed the Bangla transcripts. Interviews took place in two rounds. The first was in the second week of April 2020, and the second during the third and fourth weeks of May 2020 (the last week of Ramadan and the week following Eid-ul-Fitr which was on May 24).

3. Research findings

The Bangladesh government responded with a raft of policy measures ranging from containment, testing and treatment regimes, to tax and industrial strategies, and employment and income relief measures. Efforts to track governmental pandemic policy responses have measured their coverage and intensity; judged on these terms, the Bangladesh government’s response appears in line with others in the region and indeed beyond, both in their ambition and in how many sectors and people they sought to influence (see Hale et al., 2020). But efforts to catalogue such measures tell us little about the politics of their adoption or who supported them (Capano et al., 2020), and therefore cannot help us understand whether and why they lasted as long as they needed to, or staved off a public health catastrophe. Bangladesh had some experience with epidemic management, and this had given its government and people ‘a very realistic lack of confidence in their existing system capabilities to handle a well-entrenched pandemic’ (Capano et al., 2020, 299). These conditions shaped the policies selected, and their eventual abandonment.

The government’s imposition and withdrawal of a lockdown took place in an uneven and ad hoc manner, a criticism which could, however, justly be levelled at many states with far higher capacities than that of Bangladesh (Capano et al., 2020). On March 24, the government announced a ten-day “national holiday” beginning on March 26 with the closure of businesses, places of employment, and public transportation, which the army and police were deployed to enforce. This decision led to a mass exodus of workers from the cities to their villages, although unlike in India there was no firm policy regarding the movement of migrant workers, who were left to their own devices against a backdrop of frequent policy change. In early April, without public announcement, the government passed administrative orders to empower local officials to enforce stay at home rules and stepped up the deployment of the police and the army. The army was deployed briefly, before being quietly withdrawn. The “national holiday” was extended seven times, up to May 31st, through and beyond Ramadan and Eid-ul-Fitr (23-24th May). In the meantime, there were continuous tweaks to lockdown rules to create exceptions, to permit the boro rice harvest, mosques to host congregations with social distancing rules, restaurants to sell iftar foods, garments factories and shops to reopen. Around late April and early May, the police were gradually and quietly withdrawn from the streets, without public announcement. On May 28, 2020, the government issued a new circular officially ending the “national holiday,” reopening offices, businesses, and public transport, with social distancing rules where applicable.

The first phase of interviews (April 4–14) was conducted just as the government began implementing its earlier, more intensive and coercive social distancing regime, widely known as the “lockdown.” In a peri-urban site, a local government member received orders to impose a strict lockdown the day of the interview (April 11). The rural haor site was the only site not yet under a strict lockdown. Interviews were conducted just after large numbers of garments workers returned to their villages after a harrowing journey to and from workplaces in Dhaka and Gazipur, driven by rumors that factories were reopening. The second phase of interviews (May 11–13) was after the lockdown regime had been relaxed. Respondents said the government’s decision to allow shops to open with social distancing was announced on May 9 as the definitive signal of the official end to the lockdown. Respondents also reported that the intensive phase of the lockdown had ended 15 or 20 days previously, with the withdrawal of the police, suggesting the phase lasted for about two weeks, from early to late April.

3.1. The lockdown

Respondents experienced the lockdown as the near-incessant “miking” (use of mobile amplified sound systems) featuring instructions to stay at home, wash hands, maintain social distancing, and avoid crowds, combined with periodic patrols of the police and, initially, the military, on roads and in marketplaces. The police and military were deployed to vacate the streets after 2 pm, and close all but permitted shops. The police were complemented by different government agencies. In the coastal rural site, the local government, through the village police and the “Chairman’s people” (political networks and followers) guarded the inner village roads, while the police patrolled marketplaces, highways and river crossings.

In the Dhaka slum, respected elders (murubbis) and ruling party-affiliated youth volunteers selected by the local government representative patrolled the inner slum, while the police were stationed on the main road and market. In the peri-urban sites, district-level bureaucrats and the police carried out periodic inspections of marketplaces and public spaces. Thus, the state mobilized substantive coercive capacity, both direct (police and district administration) and proxy (youth volunteers affiliated with the ruling party) to impose the lockdown.

The police in its coercive role constituted the enduring memory of the lockdown. Respondents’ recollections of the lockdown focused on police enforcement. A handicrafts worker in a peri-urban site spoke of the fear caused by the sight of a “policeman with a raised lathi (baton or truncheon) chasing down people.” In the Dhaka slum, a respondent saw “for the first time in my life, police cars enter the slum with sirens blaring.” Many respondents, particularly tea-stall operators, rickshaw drivers, and daily laborers, recounted encounters with the police, at bazaars, when the police were shuttering shops, or on the main roads, where the police chased them with raised lathis. Rickshaw drivers reported that the police seized rickshaws or punctured tyres; urban day laborers reported that police administered “a beating or two” when they were discovered outside; multiple respondents in a peri-urban site discussed the large fines levied on shops that were found open in marketplaces (either BDT 5,000 or 10,000 (USD 58–116).

Initially, the lockdown and police action enjoyed strong social support, and what Levi would term ‘quasi-voluntary compliance’ (1997); this was based in popular consent, fear of the virus and the fact that many people had savings to draw on at first. “The lockdown is necessary,” was repeated through the first phase of interviews. Social support stemmed from a prevailing sense of alarm (atonko) over the spread of the virus, produced by international, national, and local news of infections and deaths, circulating through television channels, newspapers, social media, and word of mouth. A local government representative in the coastal rural site linked international news, social media, and community alarm:

People are more scared of the virus than they are of the police and the army. Because they see on Facebook, YouTube, not only news of Bangladesh but news from other countries. They are learning that thousands of people are dying of the virus in the advanced countries of the world and they are afraid that if the virus wreaks such havoc in advanced countries, what will happen to a country like Bangladesh?

Respondents watched national and local news of the pandemic. The first phase of interviews took place just as the government began increasing testing and reporting of cases: the number of tests per day passed 1,000 for the first time on April 9 and the total number of detected cases in Bangladesh passed 1,000 on April 14. A businessperson in the urban industrial site reported that people in their community followed and trusted daily televised briefings of testing and results by the government institute charged with conducting tests. Dhaka slum residents heard about detected cases from relatives in a nearby locality deemed an early hotspot. In a peri-urban site, a schoolteacher and local government representative related community alarm to an incident in a nearby village where a woman who had returned from America, attended a wedding and tested positive two days later. In the rural haor site, multiple respondents spoke of the alarm caused when a nurse at the Upazila (sub-district) Health Complex had tested positive - and how that alarm dissipated when she subsequently tested negative.

While alarm translated into broad support for the lockdown regime, support was conditioned by the immediate subsistence crisis. As an agricultural laborer in the rural coastal site said, “if this lockdown continues, I will die of hunger before I die of the disease.” Rickshaw-drivers, tea-stall operators, construction workers, garments workers, agricultural laborers, and domestic workers across the six sites provided similar narratives of the impact of the lockdown on livelihoods. Compliance with the lockdown was seen to be contingent on subsistence. As corroborated by Rahman and Matin, their incomes had been sharply reduced or had entirely evaporated, and they were surviving by sharply reducing food consumption, drawing on savings, and accessing informal loans (Rahman & Matin, 2020). People were adapting their livelihoods: rickshaw van drivers and porters were selling vegetables, fish, and meat door-to-door; tea-stall operators were retailing vegetables; and agricultural laborers had started growing vegetables at home and catching fish in village ponds.

Respondents across sites and social classes, in both phases of interviews, expressed the opinion that the lockdown could not be maintained on hungry people. “People will leave their homes out of hunger, to go search for food or work,” was repeated across the six communities. Many respondents stated that the lockdown could only be effective if the government provided subsistence for the poor and hungry. The fairness of the lockdown regime was at stake, and with it the feasibility of the lockdown. Citizen compliance was, in other words, contingent upon the effective and just distribution of relief, necessary to offset the costs of locking down (Levi, 1997).

3.2. Expectations of relief

The pandemic and the lockdown generated the strong expectation that the government would take measures to feed its citizenry. A construction worker in the urban industrial site told us, “the government should deliver food to everyone’s house, without bothering about what they may already have.” A community leader and labor leader in the Dhaka slum informed us, “in this situation, the government has to feed every household for one month.” Multiple respondents thought the scale of the relief required meant that only government could provide relief widely and in sufficient quantities to make a lockdown effective.

Expectations were formulated in response to government announcements of relief packages. The first phase of interviews was conducted right after the Prime Minister’s April 5 announcement of an economic stimulus package of BDT 725 billion (USD 8.5 billion). While the bulk of the stimulus consisted of subsidized credit to large businesses, and chiefly benefited the owners of garments factories, the Prime Minister also announced an expansion of social safety net provisions, through greater provisions of food free of cost, access to subsidized rice at BDT 10 per kg (USD 0.12), and the expanded coverage of old age and widow allowances. The second phase of the interviews was conducted soon after the announcement that five million families would receive BDT 2,500 (USD 29) each through mobile banking services before Eid. According to news reports, the Prime Minister was to inaugurate the new program on May 14, the day the interviews concluded. While these promises of relief created expectations, national media coverage of corruption in relief distribution created uncertainty. Several respondents referred to news reports on television or social media, of local government representatives caught with stocks of relief rice, and of relief going to people with political connections. These images reinforced a prior expectation that relief distribution was less effective than it should be, because politicians used it to enrich themselves or benefit their own followers. The stimulus package, in theory to pay workers’ wages, in fact prioritized garment factory owners’ interests above workers and other industries; this further reinforced this sense of injustice (Sultan et al., 2020). Such images reinforced a sense that the heaviest costs of managing the pandemic were to be borne by those least able to bear them, undermining the perceived fairness and effectiveness of the response.

Popular expectations focused on “the Prime Minister’s relief.” Several explicitly related their expectations to what they believed the Prime Minister had promised. An electronics retailer credited the Prime Minister with the promise of discreet relief to the impoverished middle-classes, who were too proud to ask for help. A bricklayer’s apprentice in the Dhaka slum told us the Prime Minister had asked landlords to forgive one month’s rent. Many respondents contrasted the Prime Minister’s promises with the paucity of relief actually distributed. The local head of a community-based organization in the urban industrial site stated:

The Prime Minister has promised to send food to every person’s home, but no one has seen that food yet. If we ask the Commissioner, he says that government hasn’t given us anything yet. In this situation people are losing faith in the government, even the Prime Minister.

Amidst these expectations, promises, and uncertainties, many people whose livelihoods had been destroyed attempted to obtain government relief. To get on the lists for relief, people were said to ask the nearest elected local government representative, or (in urban areas) local ruling party leaders. A garments worker noted: “poor people in need turn to landlords, the area’s councilor, the woman councilor, party leaders. They are the ones who put the names on the list.” The local politics around creating lists of relief recipients was at the heart of everyday experiences of the government’s relief regime.

3.3. The relief regime

As early as the first week of April, people began approaching local government officials for government relief. In the rural coastal site, a group of ten to fifteen people visited the Union Parishad (local council) chairman asking for food. An agricultural laborer said that the chairman sent them back, on the grounds that he had not received anything from the government yet. In a peri-urban site, the chairman of the Union Parishad reported that 50 to 60 people visited him daily in search of relief, and that an equal number called him. He told them that he had already distributed what he had received, and that “the Prime Minister has promised everyone will get food. When we get more aid, I will make sure you get some.”

As the lockdown progressed, the economic crisis worsened and substantive government relief failed to arrive, but the government continued to announce new aid schemes. More people began approaching local government representatives. By mid-May, the Union Parishad chairman quoted above reported 300–400 people seeking relief at his gates each morning. Information – or rumor – travelled fast. A fisherman in the rural haor site said:

this is the computer age, when there is a decision taken in Dhaka, people know about it in the village. When people hear there will be relief distributed, they go to the chairman, member.

A tea stall owner in a peri-urban site told us:

When people see on television that the government is giving this, giving that, they crowd at the Chairman and Member’s homes. They are comforting them, that they will get something. They take their names, making copies of their voter cards (National ID Cards).

The government had chalked out a mechanism for greater transparency in the distribution of relief during the pandemic, but nobody seemed aware of the government’s plans to improve accountability. Government circulars from early May directed district administrations to establish “Ward Committees” of elected officials, civil society and party representatives, and to establish two hotline numbers for people to call to request relief or inform authorities about irregularities in aid distribution. Relief recipients were to be issued a card with a unique code and the government was to establish a software system to monitor the quantities of aid distributed to each unique card-holder. None of our respondents were aware of any of these accountability innovations.

Instead of the accountable, transparent, and uniform process of creating lists of relief recipients envisioned above, the actual process of creating the list varied. In the rural coastal site, relief recipients reported importuning the local government representative repeatedly. In both peri-urban sites, multiple relief recipients stated that they had not approached anyone for relief, yet relief had been delivered direct to their homes. An agricultural worker had received relief through “Jamal grandfather,” a respected community leader connected to the ruling party and a friend of the Ward Commissioner. In the Dhaka slum, relief recipients reported approaching community leaders affiliated with the ruling party to obtain relief. Local government representatives or ruling party members visited villages, neighborhoods, and apartment buildings to prepare lists. In the urban industrial site, multiple relief recipients stated that local government representatives had been diligently preparing lists with the assistance of landlords. In a peri-urban site, a handicrafts worker had not approached anyone for relief, but the Union Parishad Member had taken his wife’s name and subsequently delivered relief.

With the exception of the urban industrial site, where respondents felt that relief distribution had been fair, relief recipients across the other sites were critical of relief distribution. In both the peri-urban sites and in the rural haor site, wealthier people who did not require relief were said to have benefited due to the nepotism of local government representatives and ruling party leaders. A relief recipient in the rural haor site stated that it was impossible to say who was making the lists of relief recipients: “one day, the Union Parishad member comes to take your name, the next day the Awami League’s people come.” In some sites, local leaders of the ruling party and local government representatives drew up separate lists that were subsequently merged into a single list of relief recipients. In the peri-urban site, multiple local government representatives stated that 20 per cent of relief recipients included in the list were linked to the ruling party.

In addition to problems with relief distribution, local government representatives and relief recipients were critical of the amounts received and the pace at which it arrived. A peri-urban local government representative rejected the insults and accusations that he had misallocated relief, noting that he had not yet received sufficient relief to distribute to everyone. Relief arrived in unpredictable batches and he had received seven allocations of relief rice one month into the lockdown, with which he could only distribute relief to 320 people out of the 800 people on his list. Local government representatives elsewhere similarly complained about the paucity and pace of government relief. On the other hand, relief recipients complained about the quantity of relief in each package, stating that it could only feed their families for a few days, forcing them to find other sources of subsistence. The withdrawal of the lockdown must to be understood in the context of the government’s failure to provide adequate quantities of relief through an accountable and transparent mechanism. With the lockdown proving to be neither fair nor feasible, it is unsurprising that compliance weakened. That weakening was justified by a moral economy view of the responsibilities of the state that had been tested and negotiated over several historical moments of disaster and subsistence crisis in Bangladesh’s history (Hossain, 2017, Hossain, 2018). Lockdown compliance was always going to be contingent on people being able to survive; when this was very soon in doubt, the end of the lockdown was in sight.

3.4. The end of the lockdown

Though the government made no formal announcement ending the lockdown, respondents in the second phase of interviews all stated that the lockdown was officially over. A series of announcements from the end of April reopening sectors of the economy were widely interpreted as signals that the state was tacitly lifting lockdown. In late April, the government permitted garments factories to open, which was significant in research sites with large number numbers of garments workers, which included the rural haor, a peri-urban site, the Dhaka slum, and urban industrial sites. Government actions to enable the rice harvest (between mid-April to mid-May) saw work and travel resume in the rural haor site and a peri-urban site. The May 9 decision to allow all shops, including “non-essentials” to reopen was interpreted as the final signal that the lockdown was well and truly over.

Respondents described the end of the lockdown as the quiet and unannounced withdrawal of the police from streets and marketplaces before these piecemeal official reopenings. As the police withdrew, so did local government representatives, community leaders, and youth volunteers; the incessant ‘miking’ of stay at home instructions ceased. As a relief recipient in a peri-urban site phrased it: “the police too had become tired.” The quiet withdrawal of the police needs to be contextualized in the worsening economic conditions of people and the government’s failure to provide adequate quantities of relief and to distribute it effectively.

The government’s failure to distribute substantive aid resulted, as many of our respondents had predicted, in increasing numbers of people leaving their homes in search of subsistence. This "opportunistic disobedience" resulted from the government’s failure to uphold its end of the bargain (Levi, 1997). In order to find subsistence they had to evade the police. A common motif of the lockdown regime, described in almost all sites, were games of hide and seek (lukachuri khela, uki jhuki khela) between police and citizens. Across our research sites, people described how shops would quickly shutter and people disperse when the police patrol arrived, only to resume commercial and social activity after they left. In the Dhaka slum, smaller shops inside the slum operated behind a curtain, while keeping an eye out for patrols of community leaders and youth volunteers. In a peri-urban site, a small tea-stall operator described an event when he could not shut his shop in time before the police arrived. He fled and hid in the fields, leaving his shop half shuttered. These games of hide and seek also took place on major roads, as people tried to sneak past police checkposts, on rickshaw or foot, in search of work or charity. A rickshaw van driver in the rural coastal site spoke of people whose vans or motorcycles were seized by the police; some were even beaten. Respondents across social classes described these attempts to evade the police as desperate acts in search of subsistence. Some commented that a police beating or two would not deter the hungry.

As government relief failed to arrive and the scale of the livelihood crisis worsened, the police found it increasingly difficult to inflict violence on those evading the lockdown. Two episodes recounted by respondents in the Dhaka slum during the first phase of interviews indicate the limits placed on police coercion by the moral claims of the hungry. The local government representative reported the following incident: a rickshaw driver returning home with food purchased from his meager earnings for a half-day’s work encountered the police. The police seized the rickshaw seat, including the groceries stored in a compartment. The Ward Councilor appealed to the police to return the groceries – what would he eat, otherwise? The police responded that the government would provide him with food. The Councilor said, but the government relief hasn’t arrived yet, and he needs to survive in the meantime. The police subsequently returned the rickshaw’s seat and groceries to the driver. Another incident was reported by a bricklayer’s apprentice:

My house is in slum number 7, lane number 1. Yesterday, two people were standing around over there. The police hit one of them and told them to go inside. The person who was hit by the police responded, ‘Sir, I won’t come out on the street if you give me food’. Hearing this, the police said nothing and left. Actually, the police want our own good, they want us to stay safe in our homes for our own safety. If I had the ability to eat and live, would I come out into street?

This statement about the contingency of compliance established the grounds for policy forbearance: in the absence of government relief, the legitimacy of coercive powers to keep the poor and hungry off the streets was forfeit. The government faced a choice: either provide sufficient relief, or permit people to work. By quietly withdrawing the police, the government appears to have forborne lockdown, with its unequal, and potentially life-threatening, consequences. Instead of protecting people from the virus through a lockdown and providing sufficient relief to its citizenry to survive during a lockdown, the government had effectively asked its citizenry to fend for itself. In the absence of a fair and feasible policy, strong and shared moral economy norms visibly licensed the relaxation of lockdown rules, with the support of state functionaries.

4. Discussion and conclusions

The Bangladesh government introduced a lockdown regime in an effort to contain the spread of Covid-19 in April 2020 in ways that paralleled the uncertain and uneven responses in many other countries, and not only those with weak health and social security systems. Although widely recognized as a necessary precaution given the density of the population, its vulnerability to disease, and the infectiousness of the Coronavirus, the lockdown held only two or three weeks before being gradually abandoned as unworkable, given the precariousness and poverty of the majority of the population. Critically, a promised relief program arrived too little, too late, and too unaccountably to provide the kind of social protection that many people believed might have enabled the lockdown to last longer. Covid-19 infection rates continue to rise at the end of May 2020 (the time of writing).

All over the world, Covid-19 has tested state capacities, in particular their ‘infrastuctural power’, or capacities to get citizens to comply with necessary but unpopular policies. Each country has taken on the challenge with a unique mix of responses applied at different times, affecting different groups variously. The choices have in turn been shaped by relations of citizen-state trust, past learning about pandemics and health system capacity, institutional autonomy from politicization, and a capacity to take account of the vulnerable (Capano et al., 2020). On the surface, the ways in which India handled its pandemic resemble the Bangladesh response: ad hoc, punitive, and in the final analysis, weak. Yet in Bangladesh, the lockdown and its quiet withdrawal did not involve coercion on quite the mass scale witnessed across India, nor indeed repression of healthworkers and others as seen in Pakistan. The point is not that Bangladesh had a more effective response, but that its weak response has been shaped by a need to protect its legitimacy with the population, and not merely by incapacity. The key area of failure in Bangladesh has been the inability to produce a transparent or accountability relief regime widely perceived as fair; this failure justified non-compliance and the forbearance with which the policies were selectively pursued.

What is clear from the Bangladesh case is the vital importance of state capacity in enforcing unpopular policies in the public interest. Despite a substantial growth in state capacity over the past decade, and a powerful set of political incentives to demonstrate its legitimacy through good policy performance, the Bangladeshi state has to date proven incapable of enforcing lockdown policies and a fair and effective relief program. As a test of the Bangladeshi state, the pandemic has served to highlight not only its institutional weaknesses but the contingency of citizen compliance with policies seen as both unfair and infeasible; in such a context, the state could have acted more coercively, but instead went for forbearance, before quietly dropping the policy. Both non-compliance and forbearance only make sense in light of the power of moral economy constructions of the role of the state in subsistence crises: Bangladesh’s political history of disasters, subsistence crises and the political turmoil they have unleashed has continually tested and renegotiated a set of principles about the primary role of the state in protecting people against subsistence crisis. The failed lockdown response to Covid-19 brings to light several features of state capacity that have evolved through its political history, as citizens and political actors have struggled and negotiated over the responsibilities of the state when faced with disasters and economic crises (Hossain, 2017). Protection against such shocks can be seen as the terms of the social contract between citizens and the state (Hassan, 2013, Hossain, 2018). In response to moral arguments against punitive enforcement of the lockdown, by late May, the state had been forced to cede to the demands of the moral economy, too weak to either enforce its lockdown policies, or even to protect its relief program from being seen as part of the political patronage machine.

In this article, we have disaggregated the concept of state capacity to obtain a better analytical understanding of its concrete political-economic manifestations in Bangladesh during the Covid-19 crisis. The three dimensions were coercive capacity, effective autonomy of the state vis-à-vis political society, and the preservation and enhancement of legitimacy of the state. Also relevant is a fourth dimension, namely the state’s economic and fiscal capacity of the state, which helps to understand state capacity and behavior in relation to the weak relief effort.

The state exhibited a reasonably high degree of coercive capacity during the initial stage of the lockdown, forcing citizens to largely comply with its ‘diktat’. ‘Diktat’ or the imposition of order or decree without popular consent—is not entirely apposite here, since it cannot fully capture the complex interplay of different factors in determining the lockdown. As the research revealed, people’s compliance was contingent on economic affordability and a heightened fear of the unknown (a particularly fearsome media portrayal of the killer virus). But people’s incentives to comply petered out when their savings diminished and their hopes of relief receded. So when a moral economic consensus began to emerge that poor and working class people would need to break lockdown to meet basic survival needs, the state stopped enforcing its own rules and then withdrew altogether. This selective enforcement was not about lack of coercive capacity, but an intentional toleration or forbearance, of lockdown rule-breaking by the poor and the hungry.

The state’s lack of autonomy from the political society was demonstrated in how bureaucrats dealt with local representatives of political society. Government attempt to improve the accountability and transparency of its relief operations came to naught. The state soon reconciled with the political-economic reality that its capacity to discipline political society is circumscribed by the political capacity of the age old local political machine, characterized by patron-clientelism and malfeasance. This machine, in the context of dominant party politics of Bangladesh, is an informal coalition of local political elites and local government representatives, dominated by the ruling party. The supervisory functions of relief delivery slipped from bureaucratic actors to political actors. Whether or not politics was less effective or fair in delivering relief, it was perceived thus; the non-transparent and clientelistic logic was widely understood to shape the governance of relief distribution. The state did not ‘forebear’ to distribute relief, it failed: it lacked the capacity to insulate the relief program from local political interest, and this created a strong sense of unfairness.

As noted earlier, the Bangladesh state enjoys a reasonably high degree of legitimacy of the state based on its perceived performance in social, political economic domains. Lacking the legitimacy that comes with free and fair elections, the state has strong incentives to protect the livelihoods of the popular classes: we could expect the state to put a high premium on performance legitimacy with regard to pandemic governance (Murphy, 2020). How can we explain its failure to manage the pandemic? We believe the state continues cont to enjoy a contingent political privilege--lack of organized countervailing powers of the popular classes--as noted earlier.

While the popular classes are too unorganized to effectively demand a fair relief program, the high degree of sensitivity to the needs and expectations of these classes and its strategic incentives to give in to their demands meant the state was compelled to restrain its impulse to coerce or punish those who broke lockdown rules. The principal reasons behind the failure to deliver adequate provisioning for the popular classes seem to be a lack of political will (thanks to the absence of countervailing powers) to commit necessary resources for the purpose and to protect those resources against capture by the party political machine. It is notable that a majority of the allocated funds were designated for readymade garment industries, whose owners are closely involved with party politics at the center (Khan, 2013). Corruption on a mass scale may not have actually occurred in the relief distribution, but citizens’ fears of corruption were not allayed by any special effort by the state to communicate about the relief program transparently. Relief remained closely tied up in the system of political patronage, and in the uncertainty about who would get what, when, and how, the program failed to provide the social protection citizens needed so desperately.

While the case of Bangladesh illustrates the importance of state capacity in managing pandemics, it also draws attention to the critical dimensions of legitimacy and embeddedness in a response that requires an extraordinary and unprecedented degree of cooperation and trust between citizens and the state. In bowing to popular moral pressure to allow citizens to seek work or relief, the state has demonstrated the strength of a common political culture that prioritizes citizens’ urgent rights to subsistence even over rational public health policy. ‘Forbearance’ in this context was certainly constrained, but it was also a more realistic policy or political option than coercion, in a context in which the moral economic consensus was that the choices were between working and dying from hunger, and not between lockdown and death from Coronavirus.

Although the first lockdown failed, the Bangladeshi state showed signs of doing as it has in the past, focusing on rapid learning and policy development. While the pandemic may or may not ultimately lead to a stronger health system in Bangladesh, the state’s palpable failure to provide adequate social protection is already motivating adaptation and innovation in its design and delivery of social safety nets. In the pressure cooker of the pandemic, state capacities are being reworked and tested, as they have been during other disasters and crises. A vital challenge remains the ability to control political patronage in the public interest: future research should focus closely on the relief program, which now provides the single greatest test of the Bangladeshi state in its quest for popular legitimacy. Until and unless Bangladesh learns how to deliver a fair and feasible relief system in times of crisis, we can expect that citizens will not comply, and policies to protect against pandemics or other crises will be quietly dropped.

Author contributions

All three authors were equally involved in the conceptualization, research design, training of 'field' researchers, data analysis, and writing up of the research.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.