DEVELOPMENT FUTURES SERIES WORKING PAPERS UNITED NATIONS DEVELOPMENT PROGRAMME Mitigating Poverty: Global Estimates of the Impact of Income Support during the Pandemic by Johanna Fajardo-Gonzalez, George Gray Molina, María Montoya-Aguirre, Eduardo Ortiz-Juarez UNDP GLOBAL POLICY NETWORK Mitigating Poverty: Global Estimates of the Impact of Income Support during the Pandemic by Johanna Fajardo-Gonzalez, George Gray Molina, Maria Montoya-Aguirre and Eduardo Ortiz-Juarez UNDP is the leading United Nations organization fighting to end the injustice of poverty, inequality and climate change. Working with our broad network of experts and partners in 170 countries, we help nations to build integrated, lasting solutions for people and planet. Learn more at undp.org or follow at @UNDP. The views expressed in this publication are those of the author(s) and do not necessarily represent those of the United Nations, including UNDP, or the UN Member States. Copyright © UNDP July 2021 All rights reserved United Nations Development Programme 1 UN Plaza, New York, NY 10075, USA Mitigating Poverty: Global Estimates of the Impact of Income Support during the Pandemic by Johanna Fajardo-Gonzalez (johanna.fajardo-gonzalez@undp.org), George Gray Molina (george.gray.molina@undp.org), Maria Montoya-Aguirre (maria.montoya-aguirre@undp.org) and Eduardo Ortiz-Juarez (eduardo.ortiz.juarez@undp.org) 1 Abstract This paper reconstructs the full welfare distributions from household surveys of 160 countries, covering 96.5 percent of the global population, to estimate the pandemic-induced increases in global poverty and provide information on the potential short-term effects of income-support programmes on mitigating such increases. Crucially, the analysis performs a large-scale simulation by combining the welfare distributions with the database of social protection measures of Gentilini et al. (2021) and estimates such effects from 72 actual income-support programmes planned or implemented across 41 countries. The paper reports three findings: First, the projection of additional extreme poverty, in the absence of income support, ranges between 117 million people under a distributive-neutral projection and 168 million people under a distributive-regressive projection —which may better reflect how the shock impacted poor and vulnerable households. Second, a simulation of the hypothetical effects of a temporary basic income with an investment of 0.5 percent of developing countries’ GDP, spread over six months, finds that this amount would mitigate to a large extent, at least temporarily, the increase in global poverty at both the $1.90- and $3.20-a-day thresholds, although poverty would still increase significantly in the poorest regions of the world. Third, the analysis of income-support programmes in 41 countries suggests that they may have mitigated, at least temporarily, the overall increase in poverty in upper-middle income countries but may have been insufficient to mitigate the increase in poverty at any poverty line in lowincome countries. Income support likely mitigated 60 percent of the increase in poverty at the $3.20-aday threshold and 20 percent at the $5.50-a-day threshold among lower-middle-income countries. This pattern is correlated with the amount of social assistance and social insurance per capita payments made in each country. 1 Johanna Fajardo-Gonzalez is Policy Specialist, Economist at the Strategic Policy Engagement Unit (SPE) at the UNDP Bureau for Policy and Programme Support (BPPS); George Gray Molina is the Head of Strategic Engagement and Chief Economist at BPPS; Maria Montoya-Aguirre is Economic Analyst at SPE-BPPS; and Eduardo Ortiz-Juarez is Economist at SPE-BPPS and Researcher at King’s College London. The authors are grateful to Jacob Assa, Nathalie Bouche, Lars Jensen, Luis F. Lopez-Calva, Marcela Melendez, Mansour Ndiaye, Christian Oldiges and the RBLAC Chief Economist Office for their valuable feedback. Special thanks to Anna Ortubia, Dylan Lowthian, Lesley Wright and Samantha Happ for their expert work on communications. The findings, interpretations and conclusions expressed in this paper are entirely those of the authors. 1. Introduction A key question arising from the pandemic policy response is: Was it robust enough to mitigate income and jobs losses around the world? While it is still early to adequately assess the welfare effects of multiple policy measures, this paper provides estimates of the potential influence of income support in mitigating, at least temporarily, increases in poverty headcount rates vis-à-vis a pure pandemic-induced shock scenario.2 Clearly, policy responses around the world included more than income support—they included tax deferrals, service payment waivers and loans and guarantees, as well as various work furlough and employment insurance programmes, among other measures. But it is also evident that income support programmes were ubiquitous and made up a significant portion of the response. Twelve months ago, two of the co-authors of this paper analysed the costs and implementation challenges of a temporary basic income (TBI) targeting poor and vulnerable people across the developing world (Gray Molina and Ortiz-Juarez, 2020). This paper revisits that exercise and provides counterfactual information on the potential short-term effects that income support has on mitigating the increase in poverty, and the associated financial costs, had countries implemented TBI schemes in response to the shock. To estimate the pandemic-induced increase in poverty and perform the simulations, the analysis retrieves the distributions of per capita income and consumption from household surveys in 160 countries (128 developing countries and 32 advanced economies) that covered about 96.5 percent of the world’s population in 2019–2020. But the paper also dives into the actual response. Specifically, the analysis exploits these welfare distributions and the database of social protection measures of Gentilini et al. (2021) to undertake a systematic, large-scale assessment of the potential short-term effects on mitigating the increase in poverty of 72 cash-based programmes across 41 countries, which together concentrate a fourth of the global population and represent a fifth of the total number of countries that have planned or implemented incomesupport measures since March 2020. There are three main findings derived from the simulations. First, spending equivalent to 0.5 percent of developing countries’ GDP, for a monthly total of $58.1 billion (2011 PPP) spread over six months, would have sufficed to mitigate, at least temporarily, the increases in global poverty at the $1.90- and $3.20-a-day poverty lines. Despite the aggregate mitigation, the number of people pushed below these poverty lines because of the crisis could still be significant within the poorest regions of the world. It is important to emphasize that the estimates rely on a distribution-neutral economic contraction. This seems unlikely, and 2 While this paper focuses on monetary poverty, it acknowledges that other dimensions of poverty such as education, employment, food security or safety are likely sensitive to the existence and timing of the income support provision. it might well be that the incomes of some segments of the population contracted more than proportionally during the crisis; e.g., low- and middle-skill workers, women, or the informally employed (see, e.g., ILO 2020, 2021; IMF, 2021a). Although there is no consistent information available on the incidence of the income contraction across households, the analysis also simulated the mitigating effects of TBI schemes under an ad hoc regressive contraction that hits proportionally harder the bottom 60 percent of each country’s population, which concentrates, on average across-countries, most of those living in poverty and at high-risk of falling into poverty (see section 4). The results suggest that the above investment could have helped to mitigate an important share of the increase in poverty, but certainly not all of it. Second, actual income support programmes potentially mitigated the short-term increase in poverty in a sample of 41 countries. Although this result is driven by upper-middle-income countries that were able to roll out generous income support, the estimations suggest that low- and lower-middle-income countries may not have provided transfers large enough to fully mitigate the shock-induced increase in poverty and even experienced short-term increases in their headcount rates. Finally, although there has been substantial heterogeneity in the generosity and coverage of the social protection response across countries, mostly conditional on fiscal capacity and budget adaptation, the limited effectiveness observed in some poorer countries suggests that there is room for action even under significant constraints. Yet, again, the success of these moderate interventions in mitigating the increase in poverty is likely fragile under a scenario in which the income contraction is harder on those at the bottom. Although with important caveats, the results presented in this paper provide some initial benchmarks on how the pandemic shock likely impacted poor and vulnerable households around the world, but also how important policy choices were in potentially mitigating those effects. The remainder of the paper is organized as follows. Section 2 reviews the evidence on the socioeconomic impacts of the COVID-19 pandemic and introduces the income support measures implemented as part of the governments’ policy response to this crisis. Section 3 discusses the construction of the distributions of per capita income or consumption and measures the increases in poverty at different poverty lines. Section 4 estimates the potential magnitude of the mitigation of poverty increase from hypothetical and actual emergency income support around the world. Finally, Section 5 discusses some policy implications and provides a conclusion. 2. Looking back at the first pandemic year At the onset of the pandemic, most developing countries were riven by pre-existing inequalities that would eventually threaten the lives and livelihoods of their most vulnerable citizens. A large share of workers in informal3 and at-risk service sectors (construction, transportation, retail, tourism and hospitality), combined with absent safety nets, would soon reveal that any social distancing measures would prevent many people from earning their usual income or earning an income at all. Indeed, following the implementation of the first lockdowns, the earnings of informal workers were estimated to have contracted by 60 percent globally in the first month of the crisis, reaching an average contraction of 80 percent among the poorest countries, whereas estimates covering the whole of 2020 suggest that, relative to 2019, the loss of labour incomes had reached US$3.7 trillion globally as a result of working-hour losses (equivalent to more than 220 million full-time jobs), with lower-middle-income countries being the hardest hit (ILO, 2020a; ILO, 2021). The rapid progression of the pandemic across developing countries and the immediate stringent disruptions to people’s livelihoods that followed sounded the alarms of a potential immediate increase in global extreme poverty rates (see, e.g., Mahler et al., 2020a, 2020b; Sumner, Hoy and Ortiz-Juarez, 2020; Valensisi, 2020). While increased poverty is perhaps the most salient and visible negative economic consequence of the COVID-19 pandemic, and the focus of this paper, other critical, related indicators of social progress have also worsened. For starters, the pandemic-induced crisis has left more people food insecure worldwide. Some estimates suggests that it has pushed the number of acutely food insecure people to 270 million in 2020, an 82 percent increase compared to pre-pandemic projections (WFP, 2020). Studies using household survey data from developing countries suggest that the main reason for this increase in food insecurity is the loss of incomes resulting from strict lockdowns and restrictions to mobility,4 while such an effect is compounded by disruptions to global and domestic markets and food value chains (see, e.g., Aggarwal et al., 2020; Amjath-Babu et al., 2020; Khan et al., 2021; Mahajan and Tomar, 2021). Other analyses suggest that the effects of the pandemic are likely to exert important adverse effects on gender equality. More women than men lost their jobs or experienced a disproportionate decline in their incomes, resulting in a widening of gaps in labour market outcomes and opportunities (see, e.g., AdamsPrassl et al., 2020; Foucault and Galasso, 2020; Dang and Viet Nguyen, 2021; Montoya-Aguirre, OrtizJuarez and Santiago, 2021). There are at least two factors behind this disparity. First, in contrast with previous crises, the coronavirus pandemic has particularly affected sectors with high female employment shares (Alon et al., 2020; ILO, 2020b). Second, the demand for childcare has increased. In response to closures of schools and day-care centres, more mothers than fathers have reduced their working hours or shifted to unemployment or even inactivity (see, e.g., Andrew et al., 2020; Blundell et al., 2020; Collins et al., 2021; Sen, Zhengyun and Hao, 2020; Oreffice and Quintana-Domeque, 2021; Reichelt, Makovi and 3 About 60 percent of total workers in developing countries make a living in non-agricultural informal markets (70 percent when including agriculture) (ILO, 2018; p 14). 4 See, for example, evidence for China (Wang et al., 2021), Guatemala (Ceballos, Hernandez and Paz, 2021), Ethiopia (Hirvonen, de Brauw and Abate, 2021), Nigeria (Amare et al., 2020) and South Africa (Arndt et al., 2020). Sargsyan, 2021); indeed, estimates suggest that the loss of women’s jobs in 2020 could reach 64 million globally, with 86 percent moving completely into inactivity (ILO, 2021). A critical gendered outcome is that domestic violence against women was also exacerbated during the pandemic, with its rise being mainly associated to lack of employment, low social support, substance abuse, increased stress and poor mental health (see, e.g., Peterman and O’Donnell, 2020). There are also potentially harmful, long-lasting consequences on human capital accumulation. Children have experienced learning losses across a range of subjects, grade levels and geographical regions due to school closures.5 There is evidence that children have devoted less time to schoolwork, even though parents and schools are providing resources to support their learning process during the pandemic (see, e.g., BacherHicks, Goodman and Mulhern, 2021; Jæger and Blaabæk, 2020; Maldonado and De Witte, 2020). Learning losses have also been amplified due to inadequate access to technical equipment for online schooling (see, e.g., Andrew et al., 2020b; Huber and Helm, 2020). Furthermore, learning delays are much more pronounced for primary-school students and students from low-income households, implying that educational inequalities may persist in the long term (see, e.g., Engzell, Frey and Verhagen, 2020; Gore et al., 2021; Tomasik, Helbling and Moser, 2020). Finally, in terms of health-related indicators, some estimates suggest that the less advantaged groups of the population are likely to suffer high COVID-19-related infections and mortality rates in the future as they often lack access to basic services and good-quality health care, and they tend to live in contexts with persistent conditions of indoor and outdoor pollution and where malnutrition, infectious diseases and other comorbidities are more prevalent (see, e.g., Alkire et al., 2020; Brown, Ravallion and van de Walle, 2020; Walker et al., 2020). There are also indirect health effects that are yet to be fully addressed. Access to essential health services has been severely disrupted, presenting major threats to meeting general and special health-care needs. Krubiner et al. (2021) summarize the evidence and report that most providers diverted to COVID-19 activities and supply chains were seriously affected. For instance, focusing on HIV services, studies report that disruptions to treatment may increase HIV deaths by 10 percent over the next five years, with Sub-Saharan Africa being particularly affected (Hogan et al., 2020; Jewell et al., 2020). Maternal health services have been negatively affected, as well. Antenatal care visits and institutional deliveries declined markedly in Sub-Saharan Africa due to lockdowns (Shapira et al., 2021), while in some Asian countries the quality of intra- and post-partum care and immunization rates experienced major reductions following the containment measures (Headey et al., 2020; KC et al., 2020). 5 Patrinos and Donnelly (2021) provide a systematic review of the evidence available for developed countries. 2.1 How did the world respond? Since the start of the pandemic, an ever-increasing number of countries and territories embarked on an aggressive social protection response comprised by social assistance, social insurance and labour market measures. Data from the comprehensive tracker compiled by Gentilini et al. (2021) shows that by the end of March 2020, a total of 283 social protection measures were planned or implemented across 84 countries and territories, whereas by December 2020 their cumulative numbers had reached 1,414 and 215, respectively, and 3,333 measures worldwide by mid-May 2021. During 2020, about two-thirds of the total responses corresponded to social assistance, both cash-based and in-kind, with the former accounting for about a third of the total responses (Figure 1), and although social assistance still dominates in number of responses, the expansion in social protection measures after December 2020 comprised mostly social insurance and labour market programmes. Figure 1. Evolution of the number of social protection measures, March 2020 to May 2021 3333 734 1107 283 107 133 Mar ’20 685 214 198 273 Apr ’20 937 283 276 378 May ’20 1024 310 311 403 Jun ’20 1055 323 315 417 Jul ’20 1179 370 354 455 Sep ’20 1414 458 412 544 Dec ’20 1492 May ’21 Social insurance and labour market programs Other social assistance Cash based social assistance Sources: Authors’ elaboration based on Gentilini, Almenfi and Orton (2020), Gentilini, Almenfi and Dale (2020a); Gentilini, Almenfi, Dale, Blomquist et al. (2020); Gentilini, Almenfi, Dale, Lopez, Mujica et al. (2020); Gentilini, Almenfi, Dale, Lopez and Zafar (2020); Gentilini, Almenfi, Dale, Palacios et al. (2020); Gentilini, Almenfi, and Dale (2020b); and Gentilini et al. (2021). The magnitude of the emergency social protection response is unprecedented. Available data on actual investment for about 15 percent of the 3,333 measures amounts to a world total of about $2.9 trillion (current US dollars) invested since the start of the pandemic (Gentilini et al., 2021).6 However, the lion’s share of this effort has been accounted for by high-income economies that have spent about $2.6 trillion, or 87 6 All monetary figures in this paragraph are expressed in current US dollars. From Section 3 onward, unless otherwise stated, all monetary figures are expressed in international dollars adjusted by purchasing power parity at 2011 prices (2011 PPP). percent of the world’s total (Figure 2). When considering social assistance alone, low- to middle-income countries have allocated $79.6 billion in cash-based and in-kind measures, equivalent to 4.6 percent of the world’s total spending of $1.7 trillion on these measures. This staggering heterogeneity in the capacity to respond is dramatic in per capita terms: while high-income countries have allocated an average of $545 in social assistance and of $847 if social insurance and labour market programmes are added, low- and middleincome countries have spent a per capita average of just $26 in social assistance and $124 in total social protection—among low-income countries only, the amounts per capita are as low as $4. The capacity to respond to the crisis was not only smaller among poorer countries, but further, not all of them were able to provide any income support to mitigate the short-term effects on income losses. Figure 2. Spending on social protection measures by income group ($billion, US dollars), March 2020 to May 2021 0.37% $10.9 (LMIC) 12.5% $366.7 87.1% $2,563.4 0.05% $1.3 (LIC) LIC LMIC UMIC HIC Source: Authors’ elaboration based on Gentilini et al. (2021). Notes: LIC = low-income; LMIC = lower-middleincome; UMIC = upper-middle-income; HIC = high-income. The evidence summarized in this section reveals how the COVID-19 pandemic has negatively affected households worldwide and how governments have provided a sizeable response to protect the livelihoods of the most vulnerable individuals. However, important questions remain. First, had countries around the world implemented temporary basic income schemes, what could have happened in terms of mitigating the increase in poverty? Second, what has been the potential magnitude of mitigating the increase in poverty from the actual investment in emergency cash support programmes among those countries that implemented them? The next section presents the analytical approach followed in this paper to answer these questions. 3. Estimating the pandemic-induced increase in poverty 3.1 Data and the counterfactual approach to measuring poverty To address the questions presented at the end of the last section, this paper built a cross-country comparable dataset to estimate the potential magnitude of the increase in poverty headcount rates resulting from the economic shock induced by the COVID-19 pandemic. To do so, the analysis exploited the latest version of the World Bank’s online dataset of harmonized household income and consumption surveys, which is the main data source to report comparable indicators of monetary-based poverty at the regional and global levels (Arayavechkit et al., 2021).7 The user of this dataset cannot observe per capita income or consumption at the household level, but rather can retrieve the distributions of those indicators for each country and year (see, e.g., Dykstra, Dykstra and Sandefur, 2014) using an algorithm applied to the dataset’s application programming interface (Castañeda Aguilar et al., 2019; Zhao, 2019).8 Specifically, the analysis focused on the most recent household surveys for 160 countries containing about 96.5 percent of the world’s population in 2019–2020.9 To retrieve each country’s distribution, the algorithm computed the cumulative share of the population with per capita income or consumption below an array of poverty lines that change in value every $0.10 a day per person (2011 PPP),10 starting from $0.10 up to a maximum value that covers 99.9 percent of the population. From these cumulative shares, individuals within each $0.10-bin were isolated and then assigned the middle value of their bin as their daily amount of per capita income or consumption. That is, for those individuals located within the interval [$0, $0.10], each one holds $0.05; for those within the interval [$0.10, $0.20], each one holds $0.15, and so on. Since not all household surveys were collected in a year that is common to all 160 countries, a distribution-neutral extrapolation of per capita income or consumption, while adjusting for population growth, was performed between each distribution’s actual year and the year 2019, just before the start of the pandemic, in those countries where data collection occurred before this year. The extrapolation follows the approach of Prydz et al. (2019),11 in which each value of the distribution is multiplied by a factor (
关注微信
