Impacts of COVID-19 on agricultural production and food systems in late transforming Southeast Asia: The case of Myanmar

The objective of this contribution is to report the initial impacts of measures taken to contain the COVID-19 pandemic on Myanmar's agri-food system. Myanmar is one of several late-transforming low-income countries in Southeast Asia where agriculture still plays a large role in rural livelihoods, and where food prices are a key factor affecting nutrition security for poor urban and rural households. Whereas the economic impacts of COVID-19 disruptions on tourism and manufacturing were obvious to policymakers, the impacts on the agri-food system were less evident and often more indirect. This resulted in the rural sector being allocated only a very small share of the government's initial fiscal response to mitigate the economic impacts of COVID-19. To correct this information gap, a suite of phone surveys covering a wide spectrum of actors in the agri-food system were deployed, including farm input suppliers, mechanization service providers, farmers, commodity traders, millers, food retailers and consumers. The surveys were repeated at regular intervals prior to and during the main crop production season which began shortly after nationwide COVID-19 prevention measures were implemented in April. While the results indicate considerable resilience in the agri-food system in response to the initial disruptions, persistent financial stress for a high proportion of households and agri-food system businesses indicate that the road to a full recovery will take time. The experience provides important lessons for strengthening the resilience of the agri-food system, and the livelihoods of households that depend on it.


Introduction
Myanmar is a late transforming economy with a very diverse but low productivity agricultural sector. Together with ancillary services, the agri-food sector as a whole employs approximately half the population and contributes just over a third of GDP, 23% in farming and another 11% in input distribution, agro-processing, trade and retailing (Cunningham and Munoz, 2018); Diao et al., 2020). Prior to COVID-19, Myanmar's economy was growing rapidly, but nevertheless faced many challenges, including climate change, a legacy of repressive economic policies, prolonged armed conflict and population displacement, and unpredictable trade policies of neighboring countries (especially China and India). In 2016, after more than five decades of military rule, Myanmar's hybrid military/democratic government developed an agricultural development strategy designed to diversify and improve crop production, promote competitive value chains, and ensure safe food systems (MOALI, 2017;. In the urban economy, garment manufacturing took off on the back of cheap labor and investment incentives, leading to growth in the urban population, whichalong with overseas migrationled to increased rural wages and rapid agricultural mechanization (Win et al., 2018). The decade prior to COVID-19 saw rapid declines in poverty and malnutrition, though these remain widespread, and clearly even many non-poor households had precarious livelihoods (CSO et al., 2019;MoHS, 2019).
COVID-19 caused major disruptions well before the first confirmed case in March 2020, starting with border trade closures with China in late January, the collapse of international tourist arrivals, and factory closures from February onwards due to shortages of imported raw materials and capital flight (Diao et al., 2020). Until August, Myanmar only registered a few hundred cases. 1 Nevertheless, the government's rapid and stringent response to the threat of COVID-19 inevitably accelerated economic consequences (Minoletti and Hein, 2020). In April the government imposed a three-week lockdown of the entire country and closed all international borders. Macro-economy simulation analysis predicted the measures would shrink economic output by 40% and lead to around 5 million people losing employment (Diao et al., 2020). Thereafter, internal movement restrictions were partially relaxed, but schools remained closed and public health measures such as restrictions on the size of gatherings, factory inspections, mandatory wearing of face masks, and restrictions on international entry remained in place. Moreover, as a second wave of infections took hold in September, the more stringent COVID-19 measures were re-imposed, sometimes with additional restrictions, especially on transport between different states/ regions.
Recognizing the economic challenge posed by COVID-19, the government established an Economic Recovery task force to develop a Comprehensive Economic Recovery Plan. Due to the government's very limited fiscal resources, the plan was modest in financial terms, at approximately $2 billion, with a focus on supporting shuttered industries such as garment manufacturing. The vulnerability of agriculture and rural livelihoods to COVID-19 was not initially well recognized by the Economic Recovery task force, which did not even include representation by the Ministry of Agriculture, Livestock and Irrigation (MOALI).
The objective of this paper is to document and understand the initial impacts of COVID-19 on Myanmar's agri-food system, covering crop production, inputs and mechanization services, marketing, retail and household income effects. 2 To do so we analyze data from 10 different phone surveys conducted at all stages of Myanmar's agri-food system. While every country is unique, the results from Myanmar are relevant to other late transforming economies (e.g., Cambodia, Laos, Vietnam), and the kinds of economic disruptions and potential mitigation measures are likely similar to those of other low-income countries.
The remainder of the paper proceeds as follows. The next section presents the data sources used. Section 3 presents the findings of phone surveys of farm households and traders and businesses upstream and downstream from the farm. Section 4 concludes.

Methods and data
To monitor the impacts of the COVID-19 crisis we designed a suite of panel phone surveys in the second quarter of 2020. The phone surveys cover almost the full spectrum of actors in the agri-food system: upstream agribusinesses (mechanization service providers and retailers, and farm input suppliers), rural farm and non-farm households, commodity traders, and urban consumers (Table 1). Each phone survey was designed as a panel (though sometimes unbalanced where replacements were needed). There are several limitations to the surveys worth noting here. First, the sampling strategies were necessarily opportunistic given time constraints, based on the availability of phone numbers from previous surveys. Hence they are typically not strictly representative of either the geographies or the agri-food system actors covered (though many surveys have samples that are relatively large in comparison to the target population). The samples primarily focus on the two main crop production areas of Myanmar, the delta and the central dry zone, which account for three quarters of national crop output value. Hence, with the exception of the community and food vendor surveyswhich sample communities from all over Myanmarthe surveys generally do not cover the hilly areas of Myanmar, Rakhine in the west, nor the southeast. 3 Second, although all surveys are panels, several of these panels were not completed at the time of writing, such that most of our results focus on earlier rounds of data pertaining to the disruptions that took place in the second and third quarters of 2020. Finally, there are the usual technical limitations of phone surveys to be noted, including attrition and non-response (which could lead to some bias, although these were not so common as to be of serious concern) and the need for quite short survey instruments and relatively simple question structures.

Upstream agribusinesses
Three rapid assessment phone surveys were developed to identify the impacts of COVID-19 on businesses operating upstream of the farm, all of which play key roles early in the monsoon season: agricultural input retailers, mechanization service providers, and equipment retailers. These surveys included questions to identify and monitor the effects of transportation restrictions on input supply chains (including fertilizer, pesticides, and seed) to learn about supply shortages, delays or lags in procurement, and their effects of prices and to understand the demandside effects of the COVID-19 policies and transportation restrictions. In all three surveys, we asked about sales levels during the pandemic along with recall questions back to the same period in 2019 to understand the year-on-year changes in demand. These questions help us understand how income shocks to rural farm households affected their use of mechanizations services and input purchasestwo important leading indicators for monsoon crop production.

Agricultural production and rural economies
Evidence about the impact of COVID-19 on rural households is derived from an urban and rural community survey. The community survey asked respondents from 85 urban wards and 223 rural villages (spread across about half of Myanmar's 356 townships) about perceptions of what is happening in their communities, such that responses are at least partially subjective. The survey included questions on (1) the impact on different rural income sources (farming, non-farm enterprises, farm and non-farm wage employment, and remittances); (2) particular impacts of COVID-19 mitigation measures on agricultural production and sales; and (3) coping mechanisms. Thematically, this survey included COVID-19 prevention measures, poverty and food insecurity, access to social safety nets or other forms of assistance, migration, agricultural production and marketing, and exposure to different kinds of shocks.
The community survey results are contrasted with a second, more geographically targeted survey focused on household member level gender dimensions of COVID-19 on rural livelihoods (Ragasa et al. (forthcoming)). This phone survey included 1072 male and female respondents from 606 households who were interviewed several times in 2020 about their individual and household's experiences related to COVID-19. The surveyed communities lie in the catchment areas of two irrigation sites in Myanmar's Central Dry Zone, which were drawn from a baseline survey (BL) conducted in January 2020 (pre-COVID) for a research project related to gender and nutrition.

Agricultural marketing and trade
Crop traders comprise the midstream of Myanmar's food supply chain and play a central role in crop marketing throughout the country. We developed a phone survey with crop traders to understand how crop marketingspecifically for post-monsoon crops which were being harvested around the onset of COVID-19 in Myanmarwas affected by COVID-19 policy responses. The survey included questions on (i) disruptions to crop buying and selling from transportation restrictions, and (ii) more detailed information on business activities including crop trading and credit offered to farmers, with recall comparisons of business in 2020 compared to the same time in 2019. Our sample of traders was randomly selected from crop trader lists collected during two inperson value chain studies conducted in 2017/18. Like the other phone surveys, we followed a panel design and called the same sample of crop traders three times at one-month intervals to track changes in crop trading activities and understand the evolving disruptions caused by transportation restrictions.

Food retail
A survey of small-scale food vendors, mostly general store owners, was conducted in the same wards/villages as the rural and urban community survey described above. Food vendors were asked about disruptions to their business, changes in consumer behavior, changes in supply, prices for representative food items in a range of food groups, whether prices, supply and demand were lower than usual for the time of year, as well as prevention measures in local wet markets.

Household welfare
To examine impacts on household welfare, a Rural and Urban Food Security Survey (RUFSS) was conducted from June onwards, in which approximately two thousand mothers were interviewed in urban Yangon (Myanmar's largest city) and rural areas of the Dry Zone (with the sample evenly split between the two areas). The questionnaire was designed to assess the welfare impacts of COVID-19 through different kinds of household income, economic, social, and health impacts of COVID-19, household coping mechanisms in response to income shocks; and questions on food security and dietary diversity. Here we chiefly focus on the income effects and briefly report findings on food security, dietary diversity and coping mechanisms.

Impact on upstream agribusiness
The interviewed mechanization service providers (MSPs) mostly provided land preparation service with tractors, typically earning 5000 USD of revenue from the business in a year. Equipment retailers were COVID-19 transportation restrictions significantly limited geographical areas of operations, and the timeliness of operations by MSPs. In June, 90% of MSPs were restricted to operating within the village tract and township, respectively. Similarly, 40% or more of equipment retailers (including more than half of independent operators) were restricted within their respective state or region. While land preparation was also delayed due to weather, 10-20% of MSPs perceived COVID-related restrictions directly delayed land preparation activities, notably in the Delta region. Partly due to movement restrictions, one-quarter to one-third of MSPs, and one-third of more of equipment retailers, reported reduced availability of machines and spare parts or attachments in May and June.
Most MSPs and ERs also faced reduced demand for their service or goods compared to a year ago, due partly to the cash constraints of customers. About half of MSPs in the Delta Zone and two-thirds of MSPs in Dry Zone reported lower demand (compared to 2019) for primary tillage services, with similar shares for other land preparation services. Similarly, two-thirds or more of equipment retailers reported reduced sales of machines, attachments and spare parts than one year ago (half of them reporting more than 50% declines in sales), with particularly higher shares among independent equipment retailers. Two-thirds of more MSPs were facing more requests from farmers for late payments than in 2019, and this share increased between May and July. Among equipment retailers, recovering credit from buyers has been one of the primary financial challenges, reported by one-third of interviewed ERs.
The consequences of these disruptions are mixed. By mid-2020 most respondents reported reduced restrictions on travel and fewer disruptions related to availability of equipment, yet over time an increasing share of ERs reported an outlook of reduced revenue in 2020 compared to 2019. About two-thirds of MSPs and equipment retailers expected their revenues in 2020 would be less than 2019, with significant numbers (20-30%) expecting more than 10% revenue reductions. Thus far there are no indications that these disruptions have significantly affected production of key crops in the 2020 monsoon season, although reliable data on this are yet to be publicly disseminated.
Input retailers also experienced large disruptions to both demand and supply caused by transportation restrictions early in the pandemic (mid-May, Fig. 1) when 64% of the sample reported negative effects on demand and 44% reported negative effects on supply. The supply-side disruptions led to longer input procurement times in 2020 compared to 2019, particularly for pesticides and fertilizer which have longer supply chains and need to travel across states and regions. Prices of these key inputs also increased in 2020 compared to 2019 with 38% of retailers reporting higher prices of fertilizer and 18% reporting higher pesticide prices. On the demand side, fertilizer and pesticide sales were lower in mid-May 2020 compared to the same time in 2019 for 69% and 58% of retailers, respectively. As a result of lower sales, 73% of the sample expected lower revenues overall in 2020 compared to 2019.
The disruptions from transportation restrictions lessened over time, partly due to both less stringent enforcement of the restrictions in some areas and to farmers and supply chains adapting to the new environment. However, collecting payments on inputs given to farmers on credit was a more persistent challenge experienced by 38% of our sample even in late July. Low repayment rates and lower revenues may have ongoing implications as input retailers may be more hesitant to offer inputs on credit in the future and may have difficulties repaying their own debts.

Impact on farming and the rural non-farm economy
The community survey results suggest significant and widespread disruptions to agriculture, both from disrupted production (perhaps more related to poor weather) and disrupted agricultural marketing (mainly from COVID-19). The community survey respondents largely suggested that crop productivity was lower than normal due to delayed rains, low overall rainfall, and insufficient irrigation water. Around onethird of communities also mention problems with pests. A much larger share of farmers had difficulties in selling their agricultural produce. Nationwide and local travel restrictions and the closure of main markets limited the movements of traders, causing difficulties for farmers to market and sell their pre-monsoon crops. The results are similar to Ragasa et al. (forthcoming) who found that 68% percent of farmers in their irrigated area Dry Zone sample expressed difficulties in selling their harvest due to COVID-19 (Table 2). Among those reporting difficulties, 32% had trouble finding traders to whom they could sell, 28% were constrained by market closures, 27% were hindered by movement restrictions, and 25% could not find adequate transportation to markets. Challenges in selling agricultural produce were accompanied by lower prices for commodities, as reported by 39% of the interviewed farmers. Thirty-four percent of farmers also expected further challenges in selling their harvest due to COVID-19. Rural respondents in the community survey confirmed that these issues were present nationwide: 52% mentioned more difficulties than usual for farmers to sell their produce, mainly due to low output prices, COVID-19 related mobility restrictions and not enough traders/brokers making purchases (Table 2).
These surveys also suggest that poor weather and disruptions to agricultural marketing from COVID-19 adversely affected demand for unskilled labor in rural areas. Sixty-nine percent of nonfarm businesses in the AgProGrS dry zone irrigation sites were affected by COVID-19. Half of these businesses reported having no work at all, while the other half reported facing less demand between February and May 2020. A small percentage of men and women in these communities, 18% and 7%, respectively, were usually engaged as wage workers in nonfarm wage employment, either in their villages or in the nearby town. The majority (74% of male and 78% of female workers) experienced a negative impact on nonfarm work and wages during the crisis. Moreover, 31% and 66% of men and women, respectively, who normally work in nonfarm wage employment did not engage in any nonfarm wage employment during this period Ragasa et al. (forthcoming). During the first months of the pandemic, fewer rural households received remittances and the value of these remittances was lower compared to months prior to the COVID-19 crisis. In January 2020, 32% of households in the AgProGrS dry zone sites received cash remittances from non-household members in the past year, with an average monthly value of about 256,000 MMK. Between February and May 2020, only 25% of households received remittances with a much lower average monthly equivalent of 185,000 MMK Ragasa et al. (forthcoming). 4

Impact on agricultural commodity traders
The principle business of crop traders in Myanmar is spatial arbitrage of agricultural commodities, but many also have strong connections to farmers to whom they offer inputs on credit. The crop traders in our sample were hindered in both of these roles by the COVID-19 policy responses. Over half of the sample of crop traders cited transportation restrictions as one of their two largest challenges from the COVID-19 crisis. In late May, 56 and 47% of traders reported disruptions to selling and buying crops, respectively (Fig. 2). These transport frictions, along with decreased exports from land border closures, reduced market access and put downward pressure on crop prices -56% of the sample cited lower crop prices as one of their two main challenges during the 4 Note that the baseline survey only included cash remittances from nonhousehold members. The phone survey question included both cash and inkind remittances, from both household and non-household members. Hence, the true difference in remittances received prior to and after the onset of the pandemic is likely much larger than what one would infer from the numbers reported here.

crisis.
Similar to the upstream agribusinesses, the effects of transportation restrictions lessened in later survey rounds. However, there were still major disruptions as late as August when 33% and 25% of traders reported challenges to selling and buying crops, respectively. Furthermore, 17% were still facing challenges in collecting payment on credit previously lent out to farmers in early August, a relatively small improvement from 25% of traders faced these challenges in late May. The sum of these effects was a decline in crop trading and lower revenues: 64% of the sample expected lower business revenue in 2020 compared to 2019, and 42% expected a revenue decline of at least 25%.

Impact on food retailers and food prices
Disruptions to agricultural production systems can have important downstream effects on food availability, food consumption and nutrition outcomes. Over the first wave of COVID-19 disruptions from April to August, the food supply chain adjusted surprisingly quickly to meet consumer demand despite early disruptions from transportation restrictions and lockdowns. Interviewed food vendors in the major urban center of Yangon did not generally notice major disruptions to food supply, but a quarter reported higher prices. Most food vendors (86%) delivered more products to homes and customers who more frequently used phones to arrange their purchases. At the same time, consumers reduced the frequency of shopping. They purchased larger quantities and re-allocated purchases to more non-perishable foods (49% of the food vendors mention this), likely because such products can be stored longer and thus require fewer (and riskier) trips to the market and/or food vendors.
In general, there has not been much abnormal volatility in prices, with only a few exceptions. Poultry prices rose sharply in the wake of the first COVID-19 wave due to input supply disruptionsespecially day-old chicks from Chinabut when these disruptions eased, prices again fell ((Fang et al., 2020). The food vendor survey conducted across a wide swathe of geographies also showed higher than normal prices for fish and poultry in some instances, but for most food groups respondents stated that prices were about normal. In September, at the start of the second wave of lockdowns and COVID-19 cases, there were major concerns about trade and marketing disruptions due to requirements that traders be tested when crossing state/region borders, and retail prices of fruits and vegetables in urban areas were starting to rise, but as these restrictions were relaxed supply began to stabilize. Even so, the events in September were a warning sign that significant disruptions to food trade can quickly have harmful consequences for consumers and farmers alike if they are not quickly resolved. Fig. 3 reports median income trends by the primary source of household income from the RUFSS survey conducted with approximately 2000 households per month in urban Yangon and the rural Dry Zone over June-October. Across all households, incomes dropped by roughly one-third from January to June, recovered somewhat from June to August as COVID-19 cases fell and the economy recovered, but then fell sharply again in September and October as the second COVID-19 wave hit from late-August onwards. The COVID-19 crisis is also striking in how pervasively it has affected different types of livelihoods (Fig. 3).

Impacts on household incomes and food security
Source: Authors' estimates from the C19 Rural-Urban Food Security   Survey (C19-RUFSS). Respondents were asked to estimate total household income in the past month in each round, as well as household income in January 2020 as a pre-COVID baseline. Estimates were then converted into daily adult equivalent terms. The USD exchange rate for most of 2020 means that 1000 kyat corresponds to around $0.75. Relative to January, incomes of farming and both skilled and unskilled labor households fell by half between January and June 2020. July and August saw some recovery, but by the time the second save had fully hit in September incomes in labor households were roughly one third of the levels reported in January, while farm-based incomes just one-quarter of January 2020 levels (although half of farm households cited normal lean season income problems as the main explanation of this). The incomes of trade-based households were also sensitive to COVID-19 disruptions, falling by just over 30% between January and June, recovering in July, but falling rapidly again from August onwards. Moreover, while salary-based households were the least affected, they were by no means unaffected, with incomes in September and October 40% lower than normal.
In the community surveywhich covered villages/wards spread across Myanmar, respondents were asked to estimate what proportion of households in their village/ward were struggling to eat sufficient food and in urgent need of external assistance. In June/July this metric typically varied between 16 and 18% across the different states/regions and rural/urban aggregates reported in Fig. 4, it fell to 11-13% by August. However, the second wave of COVID-19 infections from late August onwards, followed by further lockdown measures, saw estimates of extreme poverty rise to 27-29% in most locations, falling only slightly 23-25% in October. Those results suggest that income losses have been pervasive across Myanmar.
Coping mechanisms reported in RUFSS are also a cause of concern for longer term economic recovery. Almost half of poor households reported taking loans or making purchases on credit, raising the risk of longer-term indebtedness. In contrast, better off households used cash savings and reduced non-food expenditures (roughly 40% for both measures). Selling off assets was relatively rare in the RUFSS data (<10%), as was reducing food expenditure, although this because more common over time. Another striking finding from this survey was that urban mothers were at least twice as likely to report food insecurity experiences, irrespective of the specific measure, and also twice as likely to consume an inadequately diverse diet . This may suggest that urban livelihoods are even more dramatically affected by COVID-19 than farm households, although plenty of evidence suggests that rural non-farm (labor) households are also experiencing considering economic suffering.

Conclusions
Results from an array of phone surveys across Myanmar's agri-food system find that the economic effects of COVID-19 disruptions on farm and agricultural labor dependent households, agribusiness enterprises, and rural and urban consumers have been severe. The agri-food system has been hit by multiple shocks including domestic and foreign demand shocks, supply disruptions due to movement restrictions, and liquidity constraints. Farm households, in response to income losses and lower crop prices, cut back on investment in monsoon season crop production, with systemic effects on firms providing agricultural inputs and  mechanization services. While the agri-food system appears to have been resilient in its ability to adapt to disruption in the short run, persistent income losses among all types of actors may result in a prolonged recovery period.
Three key lessons emerge to enhance the resilience of Myanmar's agri-food system in the face of any similar future event. First, agricultural inputs, services and products must be allowed to move freely while ensuring safety measures appropriate to COVID-19 prevention. Second, additional financial liquidity should be made available to farmers and businesses, along with flexible terms, to prevent disruption of farm activities and service provision. Third, additional social protection will likely be required to avoid severe food insecurity and malnutrition among vulnerable households. This implies a larger fiscal allocation to social protection in the government budget. Expansion of mobile payment coverage could further ensure that such expanded social protection is timely. Finally, establishment of the capacity to monitor the agrifood system in a crisis situation would enable decision makers to have access to data and analysis to target fiscal resources in a timely way to minimize economic and welfare consequences.

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.