Tags

, , , , , ,

Hat tip: Global Collateral

You can read the study in its entirety here: https://advances.sciencemag.org/content/7/6/eabe0997.full

Source: Science Advances

Falling living standards during the COVID-19 crisis: Quantitative evidence from nine developing countries

Dennis Egger1 2, Edward Miguel1, Shana S. Warren3, Ashish Shenoy4, Elliott Collins3, Dean Karlan3 5, [there are 20 additional contributors: see all authors and affiliations]

Abstract

Despite numerous journalistic accounts, systematic quantitative evidence on economic conditions during the ongoing COVID-19 pandemic remains scarce for most low- and middle-income countries, partly due to limitations of official economic statistics in environments with large informal sectors and subsistence agriculture. We assemble evidence from over 30,000 respondents in 16 original household surveys from nine countries in Africa (Burkina Faso, Ghana, Kenya, Rwanda, Sierra Leone), Asia (Bangladesh, Nepal, Philippines), and Latin America (Colombia). We document declines in employment and income in all settings beginning March 2020. The share of households experiencing an income drop ranges from 8 to 87% (median, 68%). Household coping strategies and government assistance were insufficient to sustain precrisis living standards, resulting in widespread food insecurity and dire economic conditions even 3 months into the crisis. We discuss promising policy responses and speculate about the risk of persistent adverse effects, especially among children and other vulnerable groups.

[…]

RESULTS

Livelihoods during the COVID-19 crisis

Results in Table 2 document the widespread nature of economic hardships and the decline in living standards across the nine LMICs in the study. Across the 16 samples, between 8 and 87% of respondents report a drop in income during the crisis period, with a staggering median of 70% (column 1). The proportions reporting declines in employment are similarly high, ranging from 5 to 49% with a median share of 30% (column 2). The estimated magnitude of the economic shock remains stable whether comparing to preexisting baseline data or to respondent recall about their pre-COVID status as reported to us in a phone interview conducted after COVID hit. These measures capture the share of individuals or households that experienced a drop in well-being during the pandemic period rather than the net changes in income or employment. However, the proportion of respondents reporting declines in income (median 70%) exceeds those reporting rising income during the period by an order of magnitude (median across samples 7%). Appendix B in the Supplementary Materials discusses robustness of the estimates in detail.

The adverse economic shock experienced by individuals surveyed in these countries has been compounded by impediments to livelihood. In most countries, a large share of respondents report reduced access to markets, with the median share being 31% (range, 3 to 77%; column 3), likely related to the ubiquitous lockdowns and other mobility restriction policies adopted during March through June 2020. Where data are available, meaningful shares of respondents also report delays or other difficulties accessing health care (median, 13%; column 4).

Together, these drops in employment, income, and access to markets and services appear to contribute to higher levels of food insecurity. During the survey period, between 9 and 87% of respondents were forced to miss or reduce meals (median share, 45%; column 5), an issue we examine further in the next subsection. Even in Colombia (sample COL1), the country in our sample with the highest per capita GDP and thus potentially the greatest financial resources to cope with the crisis, the majority of respondents report drops in income (87%) and employment (49%), and an increase in food insecurity (59%).

Social support in response to the economic shock has been mixed in our populations of study. Across samples, the proportion of respondents who report benefiting from government or NGO crisis support runs the gamut from 0 to 49%, with a median of 11%. However, the high rate of missed meals and reduced portion sizes suggests that even when these efforts are present, they have been insufficient. For instance, Rohingya refugees in Bangladesh (BGD2) report the highest rates of assistance, given the preexisting international aid infrastructure serving those communities. Even in this sample, 27% of respondents report food insecurity. More detailed data in one sample (KEN1) indicate that households also engage in extensive dissaving, such as selling assets and spending stored cash, to stabilize consumption.

These adverse effects on employment, income, market access, and food security vary substantially both across countries and across different subsamples within countries. For example, in the subset of national surveys, the share of households experiencing a drop in income varies across countries from 25% in Kenya to 87% in Colombia. Within the Kenya samples, the share of households experiencing drops in income ranges from 8 to 69%. Thus, the median impacts shroud significant variation across settings. Especially within countries, it is likely that this heterogeneity results, at least in part, from differences across the subgroups surveyed.

At the same time, however, we find little evidence that this variation is systematic, e.g., by socioeconomic or refugee status. In most of cases, we cannot reject equality in the share of high and low SES households affected. However, the impact of an equivalent income drop may be greater among low SES households, as evidenced by the generally higher rates of food insecurity reported in these subsamples. There is similarly no clear pattern across refugee and nonrefugee populations. Levels of reported food insecurity are actually slightly lower among refugees than the host communities living near Rohingya camps in Bangladesh (BGD2 to 3). On the other hand, food insecurity is somewhat higher among refugees in Kenya compared with a national sample (KEN2–3). More detailed data collected in BGD2–3 surveys suggest that the presence of international humanitarian organizations in the Rohingya camp areas may have helped buffer the economic shock for refugees.

Impact timing, magnitude, and seasonality

We next describe the magnitude and timing of the effects on economic outcomes drawing on a subset of samples that feature more detailed panel or repeated cross-sectional data with richer measures of several key outcomes. Firm operations, a natural measure of overall local economic activity, appear to have been very adversely affected during the COVID-19 crisis where we have these data. In rural Kenya (KEN1), average firm profits and revenues dried up, falling by 51 and 44%, respectively (both with P < 0.05 relative to precrisis levels; Fig. 1A1). The analogous decline in Sierra Leone rural towns (SLE1) is a massive 50% (P < 0.05 relative to precrisis levels; A2). This evidence complements numbers on the share of the population experiencing any decline in employment or income in Table 2 by quantifying the depth of the economic decline.

Fig. 1 Evolution of key indicators over time.This figure shows the percentage difference from baseline for several indicators in rural Kenya and Sierra Leone during the COVID-19 global pandemic relative to the pre–COVID-19 or early COVID-19 levels. The Kenya sample is representative of all households and enterprises across 653 rural villages in three subcounties taking part in an unconditional cash transfer program. The Sierra Leone sample is representative of households in 195 rural towns across all 12 districts of Sierra Leone. Surveys in Kenya were conducted in two rounds. During the first round (weeks 1 through 8), 8594 households were interviewed. During the second round (week 11), 1394 households were surveyed, of which 1123 were interviewed for a second time. Surveys in Sierra Leone were conducted across 2439 households. The pre–COVID-19 levels are from questions that recall data from February (A1) and March (A2 to C2) or from a previous survey conducted in November 2019 (D2). The post–COVID-19 levels are from questions that recall data from the prior 7 days (A to D2 and C to D1), prior 2 weeks (A1 and E1), and a combination (prior 7 days for food and prior 2 weeks for nonfood expenditures in B1). The weeks on the horizontal axis refer to the start of the recall period for each observation rather than the period during which the data were collected. The dotted lines in A1 and A2 show the linear trend from the pre-COVID baseline to the first observation for each respective time series. Baseline level for D1 is 1.3 days out of seven for adults and 0.72 for children. Baseline level for D2 is 35% of adults missing any meals in prior 7 days and 25% of children. Baseline level for E1 is 8% of adults experiencing violence in the prior 7 days and 20% of children. *P < 0.05.

In the rural Kenya sample, there is also a pronounced decline in per capita consumption expenditures during the crisis (B1), with declines in nonfood expenditures of 29% (P < 0.05 relative to the first observation period) persisting through all of April and May 2020. During the same period, food expenditures in Kenya and Sierra Leone actually rose slightly, by 11% (B1) and 6% (B2), respectively, although in Sierra Leone, this appears to have been driven by higher food prices facing these households (19%, P < 0.05 relative to the preperiod; C2) rather than greater quantities consumed. In contrast, Kenyan prices were largely stable or even fell slightly during the same period (C1). These data indicate that households appear to be cutting back nonfood consumption in an effort to maintain essential food intake.

Examining food insecurity in greater detail, we observe rising rates of missed meals and reduced portions during the crisis in both Kenya (D1) and Sierra Leone (D2), respectively. In Kenya, we record a 38% proportional increase in the rate of adults missing meals (0.5 meals per week) and 69% for children (0.5 meals per week). The proportional increase in the share of adults reducing portions in Sierra Leone is 86% (30 percentage points) and for children is 68% (17 percentage points, P < 0.05 for all of these effects). The sharp rise in food insecurity among children is particularly alarming given the potentially large negative long-run effects of undernutrition on later life outcomes (26, 27).

The crisis period has been damaging for other dimensions of child development beyond nutrition. Schools in all sample countries have been closed during most or all of the study period. Nontrivial shares of respondents report reduced access to health facilities, including prenatal clinics and vaccinations (Table 2, column 4). The combination of a lengthy period of undernutrition, closed schools, and limited health care may be particularly damaging in the long run for children from poorer households who do not have alternative resources to make these critical human capital investments.

The rate of dissaving indicates there may be a range of other foregone household investments, from improved agricultural inputs to new small business opportunities. Lack of investment in both human and physical capital during a time of crisis can transmit the economic fallout of the pandemic far into the future.

The COVID-19 crisis could also have contributed to rising rates of domestic violence in the rural Kenya sample for which we have detailed survey reports (E1). Both violence against women and children—groups that are already marginalized in rural Kenyan society—rise by 4 and 13% (0.3 and 2.6 percentage points), respectively, during the crisis period, although these increases are not statistically significant. This increase in violence could generate additional negative and persistent effects on physical and mental health.

A central methodological concern in interpreting the patterns described in Table 2 and Fig. 1 is that factors other than the COVID-19 crisis could drive the evolution of outcomes over time. A leading possibility is that month-to-month seasonality, related for instance to the agricultural crop cycle, can also produce large changes over a span of a few months. It is challenging to fully address these concerns given distinct growing cycles for different crops in different countries, and sometimes even divergent harvest timings for different crops across regions within the same country. However, the consistency in outcomes across 16 different samples in nine countries on multiple continents, with a wide range of seasonal harvest and weather (and other) patterns, strongly suggests we are documenting the effects of a crisis that go beyond natural seasonal variation.

In two specific cases, we can directly contrast the excessive food insecurity experienced during the 2020 COVID-19 crisis to the natural seasonal patterns observed during those same months in previous years. In the BGD5 and NPL1 agrarian samples, there is monthly information on food insecurity during the 2016–2019 period that provides an ideal benchmark. Figure 2 clearly shows the pronounced seasonal variation in food insecurity in both Bangladesh (Fig. 2A) and Nepal (Fig. 2B) that spikes during preharvest lean or “hungry” seasons even during “regular” years. It is also apparent that levels of food insecurity are far higher during the 2020 crisis than they were during the same season in previous years: The rate of food insecurity in Bangladesh in April 2020 is roughly twice as high as in previous years, and this season-adjusted difference is statistically significant (P < 0.05). In these cases, the leading explanation for the effects we document in 2020 is the COVID-19 crisis rather than seasonal fluctuations. It is notable that in both countries, the COVID crisis occurred during the favorable postharvest period with its relatively low level of food insecurity during normal years. Baseline levels of deprivation typically rise sharply in the final 4 months of the calendar year.

Fig. 2 Food insecurity in Bangladesh and Nepal.Food insecurity in Bangladesh and Nepal with 95% confidence intervals. (A) Monthly rates of food insecurity among landless agricultural households in northern Bangladesh from sample BGD5. Food insecurity is defined as missing a meal or reducing portions for at least 15 days in a month. Note that this is a more stringent criterion than that reported in Table 2; in this figure, we restrict to cases of frequently missed meals. The 2020 rates come from an April phone survey, and “Previous year” reflects retrospective survey data spanning January 2018 through May 2019 collected in two survey rounds in February and June 2019. (B) Data from agricultural households in western Terai, Nepal, from sample NPL1. The index of food insecurity is constructed using two questions on how often households had to worry about not having enough food or had to reduce portion sizes. The data points in late 2019 and early 2020 come from six rounds of contemporaneous phone survey, and “Previous year” reflects respondents’ recollection about a prior “typical year” reported during the April–May 2020 phone survey round.

CONCLUSIONS AND IMPLICATIONS FOR POLICY

We document pronounced declines in employment, income, and food security since April 2020 across 16 survey samples in nine LMICs, with surveys covering over 30,000 households. This study provides some of the most systematic data to date on how the outbreak of COVID-19 has affected households across multiple LMICs in several major world regions. While the realities of rapidly deploying a survey over the phone during a pandemic make it difficult to reach a truly nationally representative sample, we study heterogeneous samples spanning three continents. We find that the economic shock in these countries—where most people depend on casual labor to earn enough to feed their families—leads to deprivations that seem likely to generate excess future morbidity, mortality, and other adverse longer-term consequences.

The findings highlight the importance of generating on-the-ground survey data to track well-being during the crisis to gather detail necessary to craft evidence-based policy responses. We demonstrate a path forward for gathering this information using large-scale phone surveys that rely on random sampling and standardized questions for comparability across settings. The methodology and harmonized measurement tools can readily be rolled out in new contexts to cover additional populations during this and future crises.

Following on decades of steadily increasing incomes across major world regions, the sharp rise in global poverty in 2020 that we document is unprecedented. The median proportion of respondents across our sample countries experiencing reduced income is a staggering 67%, and negative effects are experienced by households across the socioeconomic spectrum. The economic distress caused by the COVID-19 pandemic has had an immediate cost in terms of nutrition in LMICs. In addition to direct health consequences, hunger places long-run productivity and growth at risk as households compensate by reducing other investments in productive inputs such as fertilizer, selling productive assets, and lowering investment into long-run child development and education. Evidence abounds that these severe shocks to food security of children can threaten long-term health and well-being (19–21).

Humanitarian relief efforts that aim to address these problems face two added complications during the pandemic, relative to standard relief programs during regular years. The first is that further viral spread is fundamentally linked to the extent of economic deprivation, and successful disease containment requires the provision of immediate economic relief. Second, the worldwide financing of large-scale relief is constrained by the aggregate nature of the COVID crisis that simultaneously affected donor countries, as well as the large magnitude of the global economic recession.

The findings in our data highlight the first challenge. Households facing acute food shortages may be less willing to adhere to social distancing rules than others and could instead seek out income-generating opportunities even in crowded and epidemiologically risky markets. For social distancing to succeed, people must feel sufficiently secure from deprivation and hunger.

Relief programs should be carefully designed to avoid unintended adverse public health consequences—such as increased face-to-face market transactions in areas with high likelihood of viral spread. Cash or food transfers that allay this direct need could even double as tools to address disease spread by discouraging such market interactions. For example, transfers could be explicitly labeled with a “soft” form of conditionality, such as “this is money for food to reduce your need to work in crowded markets,” to further promote social distancing. Furthermore, new innovations to quickly and safely identify the poor using mobile phones or satellite data [e.g., (28)] and deliver funds remotely through mobile money transfers (29) hold promise in this context because of the minimal contact required to implement.

Our data also highlight the widespread nature of the global economic shock. Social protection programs in LMICs are underfunded even in good times. During an economic downturn, reduced tax revenue will make financing such programs even harder, and debt markets are not readily available for LMICs. Because the severity of the current crisis makes it important to expand safety net programs, international support—for instance, in the form of grants or concessionary loans—will be needed. Rich countries that are themselves under pressure from this same health and economic crisis may be tempted to focus on addressing problems at home. Yet, since disease transmission does not respect national borders, it is in the self-interest of wealthy countries to help reduce the spread of COVID-19 in LMICs, over and above any humanitarian motivations.

Policymakers in LMICs will also need to craft creative solutions to develop income-generating activities with longer gestation periods in case the risky COVID-19 disease environment or the associated economic slowdown persists for a prolonged period. For instance, “graduation programs” that combine assets and training can promote a source of livelihood that requires limited external contact and have been shown to reduce poverty in the past (30, 31). Combining these programs with immediate cash support has even been shown to help build sustainable sources of income during periods of civil unrest [e.g., (32)].

On an optimistic note, the innovation and technological adoption that takes place during emergencies can spur long-run economic development. Dealing with the economic fallout from COVID-19 will require the technological infrastructure to reach poor populations in remote areas with minimal face-to-face contact. These workarounds have accelerated the expansion of new financial technology during past political and economic crises [e.g., (33, 34)]. Solutions that arise in the current climate thus have the potential to both improve resilience immediately and durably advance the financial ecosystem.

Countries around the world face difficult policy choices along the path to economic recovery from COVID-19. While much public discussion focuses on “lives” and “livelihoods,” our data suggest this is a false dichotomy. We provide systematic evidence on how the outbreak has adversely affected households across multiple LMICs in several major world regions. A more appropriate framing of the situation in these countries could be in terms of “lives damaged or lost due to disease” and “lives damaged or lost due to economic deprivation.” We emphasize that our data do not speak to the economic consequences of imposing or relaxing specific lockdown policies. However, the evidence does have specific policy implications for how to cope with the economic hardships, to protect both lives now and in the future: fund and implement immediate humanitarian relief and long-term safety net programs to ameliorate the damage that we document.

SUPPLEMENTARY MATERIALS

Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/7/6/eabe0997/DC1https://creativecommons.org/licenses/by/4.0/

This is an open-access article distributed under the terms of the Creative Commons Attribution license, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

REFERENCES AND NOTES

  1. ↵
    1. Z. Barnett-Howell, 
    2. O. J. Watson, 
    3. A. M. Mobarak
    , The benefits and costs of social distancing in high- and low-income countries. Trans. R. Soc. Trop. Med. Hyg., traa140 (2021).Google Scholar
  2. ↵
    1. P. G. T. Walker, 
    2. C. Whittaker, 
    3. O. J. Watson, 
    4. M. Baguelin, 
    5. P. Winskill, 
    6. A. Hamlet, 
    7. B. A. Djafaara, 
    8. Z. Cucunubá, 
    9. D. O. Mesa, 
    10. W. Green, 
    11. H. Thompson, 
    12. S. Nayagam, 
    13. K. E. C. Ainslie, 
    14. S. Bhatia, 
    15. S. Bhatt, 
    16. A. Boonyasiri, 
    17. O. Boyd, 
    18. N. F. Brazeau, 
    19. L. Cattarino, 
    20. G. Cuomo-Dannenburg, 
    21. A. Dighe, 
    22. C. A. Donnelly, 
    23. I. Dorigatti, 
    24. S. L. van Elsland, 
    25. R. F. John, 
    26. H. Fu, 
    27. K. A. M. Gaythorpe, 
    28. L. Geidelberg, 
    29. N. Grassly, 
    30. D. Haw, 
    31. S. Hayes, 
    32. W. Hinsley, 
    33. N. Imai, 
    34. D. Jorgensen, 
    35. E. Knock, 
    36. D. Laydon, 
    37. S. Mishra, 
    38. G. Nedjati-Gilani, 
    39. L. C. Okell, 
    40. H. J. Unwin, 
    41. R. Verity, 
    42. M. Vollmer, 
    43. C. E. Walters, 
    44. H. Wang, 
    45. Y. Wang, 
    46. X. Xi, 
    47. D. G. Lalloo, 
    48. N. M. Ferguson, 
    49. A. C. Ghani
    , The impact of COVID-19 and strategies for mitigation and suppression in low- and middle-income countries. Science 369, 413–422 (2020).Abstract/FREE Full TextGoogle Scholar
  3. ↵P. K. Goldberg, T. Reed, The effects of the coronavirus pandemic in emerging markets and developing economies: An optimistic preliminary account, Brookings papers on economic activity (2020); www.brookings.edu/wp-content/uploads/2020/06/Goldberg-Reed-conference-draft.pdf.
  4. ↵International Monetary Fund, “World economic outlook update, June 2020” (Technical Report, International Monetary Fund, 2020); www.imf.org/en/Publications/WEO/Issues/2020/06/24/WEOUpdateJune2020.
  5. ↵World Bank, COVID-19 crisis through a migration lens. Migration and Development Brief 32 (2020); http://hdl.handle.net/10986/33634.
  6. ↵World Bank, “Projected poverty impacts of COVID-19 (coronavirus)” (Technical Report, World Bank, 2020); http://pubdocs.worldbank.org/en/461601591649316722/Projected-poverty-impacts-of-COVID-19.pdf.
  7. ↵David Beasly, Statement to U.N. Security Council, www.wfp.org/publications/covid-19-potential-impact-worlds-poorest-people, April 2020. Remarks by David Beasley, UN World Food Programme (WFP) Executive Director to a virtual session of the UN Security Council on the Maintenance of International Peace and Security: Protecting Civilians Affected by Conflict-Induced Hunger.
  8. ↵
    1. A. Deaton
    , How to monitor poverty for the millennium development goals. J. Human Dev. 4, 353–378 (2003).Google Scholar
  9. ↵
    1. A. Deaton
    , Measuring poverty in a growing world (or measuring growth in a poor world). Rev. Econ. Stat. 87, 1–19 (2005).CrossRefWeb of ScienceGoogle Scholar
  10. ↵
    1. L. Medina, 
    2. F. Schneider
    , Shedding light on the shadow economy. World Economics 21, 25–82 (2020).Google Scholar
  11. ↵International Labour Organization, Women and men in the informal economy: A statistical picture, (International Labor Organization: International Labour Office, Geneva, Switzerland, ed. 3, 2018).
  12. ↵R. Chetty, J. N. Friedman, N. Hendren, M. Stepner, The Opportunity Insights Team, “How did COVID-19 and stabilization policies affect spending and employment? A new real-time economic tracker based on private sector data” (Technical Report, Opportunity Insights, 2020); https://opportunityinsights.org/wp-content/uploads/2020/05/tracker_paper.pdf.
  13. ↵U.S. Census Bureau, Household pulse survey interactive tool (2020); www.census.gov/data-tools/demo/hhp/ [accessed 4 July 2020].
  14. ↵D. Schanzenbach, A. Pitts, Estimates of food insecurity during the COVID-19 crisis: Results from the COVID impact survey, week 1 (April 20–26, 2020). Rapid research report, Institute for Food Policy Research (2020); www.ipr.northwestern.edu/news/2020/food-insecurity-triples-for-families-during-covid.html.
  15. ↵Ahmed Mushfiq Mobarak, For the rural poor, the coronavirus crash isn’t here yet, Foreign Policy (2020); https://foreignpolicy.com/2020/06/08/rural-poor-coronavirus-pandemic-crash-prolonged-lockdown-food-insecurity-seasonal-labor/.
  16. ↵United Nations Department of Economic and Social Affairs, The great recession and the jobs crisis, in The Global Social Crisis: Report on the World Social Situation 2011 (United Nations Organization, 2011), chapter 2, pp. 15–48.
  17. ↵C. J. Kim, Labor markets in East Asia during crisis: Developments and policy implications? Eye on East Asia and Pacific 67879, World Bank (2012); http://documents1.worldbank.org/curated/en/515711468245426038/pdf/678790BRI00PUB067903B0EYE0on0EA0no2.pdf.
  18. ↵
    1. D. Almond
    , Is the 1918 influenza pandemic over? Long-term effects of in utero influenza exposure in the post-1940 U.S. population. J. Polit. Econ. 114, 672–712 (2006).CrossRefWeb of ScienceGoogle Scholar
  19. ↵
    1. S. Maccini, 
    2. D. Yang
    , Under the weather: Health, schooling, and economic consequences of early-life rainfall. Am. Econ. Rev. 99, 1006–1026 (2009).CrossRefWeb of ScienceGoogle Scholar
  20. ↵A. Nandi, S. Bhalotra, A. B. Deolalikar, R. Laxminarayan, The human capital and productivity benefits of early childhood nutritional interventions, in Child and Adolescent Health and Development, D. A. P. Bundy, N. de Silva, S. Horton, D. T. Jamison, G. C. Patton, Eds (International Bank for Reconstruction and Development/World Bank, Washington, DC, ed. 3, 2017), chapter 27, pp. 385–402.
  21. ↵
    1. M. E. McGovern, 
    2. A. Krishna, 
    3. V. M. Aguayo, 
    4. S. V. Subramanian
    , A review of the evidence linking child stunting to economic outcomes. Int. J. Epidemiol. 46, 1171–1191 (2017).PubMedGoogle Scholar
  22. ↵Pew Research Center, Mobile divides in emerging economies, Technical report, November 2019.
  23. ↵FAO, Kenya beats back desert locust upsurge for now, but East Africa remains at risk, www.fao.org/news/story/en/item/1300659/icode/.
  24. ↵
    1. J. Kaminski, 
    2. L. Christiaensen, 
    3. C. L. Gilbert
    , Seasonality in local food markets and consumption: Evidence from Tanzania. Oxf. Econ. Pap. 68, 736–757 (2016).CrossRefGoogle Scholar
  25. ↵M. Smale, V. Theriault, R. Vroegindewey, Dietary Patterns in Mali: Implications for Nutrition. Feed the Future Innovation Lab for Food Security Policy Research Papers 303052, Michigan State University, Department of Agricultural, Food, and Resource Economics, Feed the Future Innovation Lab for Food Security (FSP) (2019); https://ideas.repec.org/p/ags/miffrp/303052.html.
  26. ↵
    1. C. G. Victora, 
    2. L. Adair, 
    3. C. Fall, 
    4. P. C. Hallal, 
    5. R. Martorell, 
    6. L. Richter, 
    7. H. S. Sachdev; Maternal and Child Undernutrition Study Group
    , Maternal and child undernutrition: Consequences for adult health and human capital. Lancet 371, 340–357 (2008).CrossRefPubMedWeb of ScienceGoogle Scholar
  27. ↵
    1. S. Baird, 
    2. J. H. Hicks, 
    3. M. Kremer, 
    4. E. Miguel
    , Worms at work: Long-run impacts of a child health investment. Q. J. Econ. 131, 1637–1680 (2016).CrossRefPubMedGoogle Scholar
  28. ↵
    1. J. Blumenstock, 
    2. G. Cadamuro, 
    3. R. On
    , Predicting poverty and wealth from mobile phone metadata. Science 350, 1073–1076 (2015).Abstract/FREE Full TextGoogle Scholar
  29. ↵J. Aron, J. Muellbauer, The economics of mobile money: Harnessing the transformative power of technology to benefit the global poor. Oxford martin school policy paper, Oxford Martin School, 2019.
  30. ↵
    1. A. Banerjee, 
    2. E. Duflo, 
    3. N. Goldberg, 
    4. D. Karlan, 
    5. R. Osei, 
    6. W. Parienté, 
    7. J. Shapiro, 
    8. B. Thuysbaert, 
    9. C. Udry
    , A multifaceted program causes lasting progress for the very poor: Evidence from six countries. Science 3480, 1260799 (2015).Google Scholar
  31. ↵
    1. O. Bandiera, 
    2. R. Burgess, 
    3. N. Das, 
    4. S. Gulesci, 
    5. I. Rasul, 
    6. M. Sulaiman
    , Labor markets and poverty in village economies*. Q. J. Econ. 132, 811–870 (2017).Google Scholar
  32. ↵Guadalupe Bedoya Arguelles, Aidan Coville, Johannes Haushofer, Mohammad Razaq Isaqzadeh, Jeremy Shapiro, No household left behind : Afghanistan targeting the ultra poor impact evaluation. Impact evaluation series, World Bank (2019); http://documents.worldbank.org/curated/en/855831560172245349/No-Household-Left-Behind-Afghanistan-Targeting-the-Ultra-Poor-Impact-Evaluation.
  33. ↵
    1. W. Jack, 
    2. T. Suri
    , Risk sharing and transactions costs: Evidence from Kenya’s mobile money revolution. Am. Econ. Rev. 104, 183–223 (2014).CrossRefGoogle Scholar
  34. ↵
    1. T. Suri, 
    2. W. Jack
    , The long-run poverty and gender impacts of mobile money. Science 354, 1288–1292 (2016).Abstract/FREE Full TextGoogle Scholar
  35. P. Lopez-Pena, C. A. Davis, A. M. Mobarak, S. Raihan, Prevalence of COVID-19 symptoms, risk factors, and health behaviors in host and refugee communities in Cox’s Bazar: A representative panel study. Bulletin of the World Health Organization (Preprint) (2020); http://dx.doi.org/10.2471/BLT.20.265173.
    1. G. Bryan, 
    2. S. Chowdhury, 
    3. A. M. Mobarak
    , Underinvestment in a profitable technology: The case of seasonal migration in Bangladesh. Econometrica 82, 1671–1748 (2014).CrossRefGoogle Scholar
  37. G. Bryan, A. M. Mobarak, K. Naguib, M. Reimão, A. Shenoy, No lean season 2017–2019 evaluation. AEA RCT Registry ID AEARCTR-0002685, 2019.
  38. N. Meriggi, M. Humphreys, A. B. Kamara, M. Krupoff, M. Levine, H. Mcleod, A. M. Mobarak, W. Prichard, A. Shridhar, P. van der Windt, M. Voors, Tracking the economic consequences and responses to COVID-19 in Sierra Leone. Technical report, International Growth Centre, 2020.

Acknowledgments: We are indebted to study participants for generously giving their time. We are grateful to the staff of Vyxer Remit Kenya; Center for Effective Global Action (CEGA); Yale Research Initiative on Innovation and Scale (Y-RISE); Northwestern University Global Poverty Research Lab; National Planning Department of Colombia; National Nutrition Council, and Departments of Education, Social Welfare and Development, and Labor and Employment in the Philippines; Rwanda Education Board; Centre for the Study of Labour and Mobility (CESLAM) and Backward Society Education (BASE) in Nepal; a2i Bangladesh, Gender and Adolescence: Global Evidence (GAGE/ODI); World Bank Poverty and Equity Global Practice (GPVDR); and IPA Policy, Global Research and Data Support, IPA Poverty Measurement teams, IPA staff in Bangladesh, Burkina Faso, Colombia, Ghana, Philippines, Rwanda, and Sierra Leone. See the Supplementary Materials for a full list of additional contributors. Funding: This research was supported by grants from the Applied Research Programme on Energy for Economic Growth (EEG) led by Oxford Policy Management (funded by the UK Government through UK Aid), UNOPS Sierra Leone, Bill & Melinda Gates Foundation, GLM/LIC and STEG research programs of DFID and IZA, Evidence Action, GAGE/ODI (funded by the UK Government through UK Aid), Givewell, Global Innovation Fund, Golub Capital Social Impact Lab at Stanford, 3ie, International Growth Centre, IPA’s Peace & Recovery Program (funded through the UK DFID), MasterCard Center for Inclusive Growth, Mulago Foundation, Private Enterprise Development in Low-Income Countries, U.S. National Science Foundation, Yale MacMillan Center, Yale Program on Refugees, Forced Displacement, and Humanitarian Responses (PRFDHR), Weiss Family Fund, World Bank, and anonymous donors to IPA and to Y-RISE. The studies received IRB approval from University of California Berkeley, George Washington University, Innovations for Poverty Action, Maseno University, the Office of the Sierra Leone Ethics and Scientific Review Committee, the Burkina Faso Institutional Ethics Committee for Health Science Research, Rwanda National Institute for Scientific Research, Rwanda National Ethics Committee, Yale University, and Wageningen University. D.P. acknowledges support from the Wellspring Philanthropic Fund. Author contributions: The order of author names was randomized. All authors contributed to methodology and investigation, while the first 17 authors (listed in random order) were involved in formal analysis and writing. This joint research effort was coordinated by multiple institutions: Center for Effective Global Action at UC-Berkeley (E.M., director), Global Poverty Research Lab (D.K. and C.U., codirectors), Innovations for Poverty Action, and Yale Research Initiative on Innovation and Scale (A.M.M., director). Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Data and code for replication are available at https://doi.org/10.7910/DVN/SBUFNN. Additional data related to this paper may be requested from the authors.

  • Copyright © 2021 The Authors, some rights reserved; exclusive licensee American Association for the Advancement of Science. No claim to original U.S. Government Works. Distributed under a Creative Commons Attribution License 4.0 (CC BY).