As COVID-19 (coronavirus) has spread across the world, the World Bank has projected extreme poverty to increase for the first time since the Asian crisis in 1998, putting at risk the global goal of reducing extreme poverty to 3% of the world’s population by 2030. The duration and scale of impacts are highly uncertain and expected to vary widely within and across countries and over time, which makes it really important to closely monitor the impacts of the crisis on households and firms for designing policy responses. The World Bank’s high-frequency monitoring phone surveys, which are going on in nearly 100 countries, have sought to fill the information gap that traditional in-person surveys are ill-suited to fill during a pandemic. They are a window into how the pandemic is affecting every aspect of the lives of almost every household in the developing world. A similar initiative of Business Pulse Surveys adds to this picture by tracking the pandemic’s impacts on firms in 51 countries.
The extensive country coverage of the phone surveys opens up opportunities to view the impacts of COVID-19 on people and firms with a wider lens , which has not been available before at this scale. With some adjustments, the data from different countries can be compared, combined, grouped, and analyzed together to make comparisons between countries, find patterns across different types of countries, and examine policy impacts. The COVID-19 High Frequency Survey Global Dashboard released last week is an important step in making the harmonized indicators available to all, even as we continue to add more countries, surveys, and indicators, including those from Business Pulse Surveys.
Widespread impacts on several dimensions of well-being
The early insights from the harmonized household data from 40 countries, which account for nearly one-fifth of the world’s population, yield a picture that is alarmingly consistent with the severe welfare impacts we had projected using past data. While we cannot directly estimate poverty (since detailed data on consumption cannot be collected from phone respondents), the simple questions we ask about job and income losses provide an overall sense of changes in well-being. On average, 36% of those working in a country prior to COVID19 stopped working during April-July, and 62% of households reported reduction in total income. Large-scale job and income losses are broadly consistent with the impacts on firms reported by Business Pulse Surveys.
Households’ losses were not limited to labor earnings. More than 60% of households receiving remittances in an average country before the pandemic reported a drop in remittances they received. This seems to confirm the fears that international and domestic remittances, which act as a key source of income for many in the developing world, particularly in rural areas, are now frayed by job losses in the (urban) services sector, where migrant workers tend to be employed in large numbers.
Households achieve some degree of resilience by adopting a variety of coping strategies. Reducing consumption is the most common, followed by drawing down emergency savings; selling assets like property is less frequent but not negligible, reported by an average of 8% of households. For the vulnerable, these strategies can have serious consequences for future incomes, resilience to other shocks, and nutrition. On average, 16% of households report at least one adult going without food for a full day in the last week before the survey. We do not know how much of the food insecurity is linked to pre-existing conditions or non-COVID shocks. But COVID is at least partly to blame – in almost all countries, food insecurity is reported more frequently among households where job losses after COVID were reported (figure 1).
The burden of impacts is higher in poor countries, and on certain groups within countries
As one would expect, there is wide variation across countries in these household-level impacts. The lower the per capita GDP of a country, the more likely are households to report partial or no payment of wages and an adult going without food for a day in the previous week. Estimates from a recent Bank paper combining phone survey data with data from pre-COVID surveys in four countries in Sub-Saharan Africa illustrate the scale of income impacts in low-income countries. In these four countries alone, 256 million individuals (77% of the population) are estimated to live in households that have lost income due to the pandemic.
Disruptions to education are also much more severe in poorer countries (figure 2), as school closures and the lack of access to digital services make it much less likely for children to complete a teacher-provided assignment. In the long-term, lost months of learning, nutritional deficiencies, and difficulty in accessing health care (reported in some countries) can affect inequality of opportunity, social mobility and productivity. That these risks are much higher in low-income countries, which are home to most of the world’s extreme poor, should be a cause for worry.
Some of the patterns emerging from the phone surveys are broadly consistent with our earlier projections about the “new poor.” Those with lower levels of education are far more likely to lose their jobs than college-educated workers in almost all countries. Jobs were lost at higher rates in the industry and services sectors than in agriculture. But even so, job losses and food insecurity in rural areas are on a similar scale as in urban areas, which suggests that the impacts have permeated throughout the economy including rural areas, even though urban areas are likely to have taken the most direct and immediate hit. Early numbers hint at wide gender disparities. Among the respondents to our surveys (who may not necessarily represent women in a country), women were more likely to lose their jobs than men in all but two of the countries.
Are direct assistance programs reaching those who need help?
The harmonized data allows us to get a snapshot of the coverage of social assistance programs, which have been expanded significantly in many countries. Our general conclusion: Given the magnitude of impacts, public social assistance interventions in the early months of the pandemic were highly inadequate in most countries. On average across countries, 20% of households reported receiving assistance, with the number ranging from nearly 70% in Indonesia and Mongolia to less than 10% in countries in Sub-Saharan Africa. In most countries, social assistance does not appear to be well-targeted to those who are affected by the crisis (food-insecure or job losses) or vulnerable before it struck (e.g. those with lower education). Average coverage is much lower for IDA countries than non-IDA countries, which could reflect, at least to some extent, pre-existing differences in the coverage of safety net programs.
These findings are not surprising, given the challenges of expanding programs rapidly and in ways that reach both the existing and the new poor in countries with limited capacity and resources. The (eventual) coverage of these programs may also tick up much higher, since the first wave of surveys may have been conducted too soon after the onset of the crisis to be able to detect the effects of expansions in social assistance in many countries.
Are things improving?
On this critical question, we don’t yet have the coverage of data to form a definitive view. However, in the 15 countries for which we have harmonized data for two waves of surveys, we do see fledgling signs of improvement, particularly in food security. There has also been a small but significant increase in the share of households receiving social assistance in these countries.
In many respects, the early findings from the harmonized phone survey data conform to expectations: widespread impacts that amplify pre-existing inequalities between rich and poor countries and between haves and have-nots within countries. Even as economies rebuild gradually, the focus of policymakers and the development community must remain on the vulnerable countries and households , who will find recovery to be harder and slower. In future blogs, we will have more to say on what the uneven distribution of impacts might mean for the longer-term outlook on poverty reduction and inequality.
Join the Conversation