Between 1990 and 2010, the share of the population of the developing world living in extreme poverty (under $1.25 a day) was cut in half. This would appear to bring the end of extreme poverty within reach.

In a new paper, we assess the prospects for ending extreme poverty by 2030. We begin by developing a range of plausible poverty outcomes based on alternative scenarios for how mean consumption and the distribution of consumption in each country might evolve.

Our range of poverty outcomes for 2030 is large, implying that the future trajectory of global poverty is highly uncertain. Getting to the “zero zone”, defined here as a poverty rate of under 3 percent, by 2030 is unlikely to occur through stronger than expected consumption growth or an improving distribution alone. Both factors are needed simultaneously.

To better understand the challenge of ending poverty, we look at the pattern of poverty reduction in terms of the movement of poor people across the poverty line. In this figure, the area under the curve represents the population of the developing world arranged according to individuals’ level of daily consumption. The height of the curve at any given consumption level reflects the number of people who meet that exact standard of living. Thus the area under the curve to the left of the vertical poverty line (at $1.25) captures the number of people living in poverty. Press the play button to see changes over time.

Over the past twenty years global poverty reduction was made possible by a consistently large mass of people lining up behind the poverty line each year, and sufficient consumption growth to carry many of these individuals across the threshold.

Today there are more people living around the $1.25 mark than at any other consumption level in the world, as illustrated by the peak of the curve. Equitable growth will result in more movement of people across the poverty line than across any other level of consumption.

The future could follow various paths as illustrated by the three divergent scenarios. However, in both the baseline and best case scenarios, the height of the curve at the poverty line diminishes precipitously, meaning that the number of people just below the $1.25 threshold falls. Sustaining the trend rate of poverty reduction requires that each year a new set of individuals is primed to cross the poverty line. Our analysis suggests that this will become increasingly difficult as some of the poorest of the poor struggle to make enough progress to be in a position to escape poverty over the next twenty years.

A focus of poverty research over the past twenty years has been to understand the extent to which the poor benefit from economic growth.

This figure compares the scale of global poverty reduction recorded each year since 1990 with the reduction which would have occurred had the consumption growth recorded in each economy been truly equitable. In most years, actual poverty reduction fell short of this benchmark, keeping millions in destitution.

Sustained progress against poverty depends on future consumption growth being fully shared by those people living below the poverty line.

To explore this topic from a different angle, we look to the performance of China, India and sub-Saharan Africa, combining official estimates with our baseline scenario. The 40-year period resembles a relay race in which responsibility for leading the charge on global poverty reduction passes between these three giants.

China’s relay leg is the most striking. It undergoes a dramatic transformation in which its population, which is initially predominantly poor, disperses over a range of consumption levels beyond the poverty line, driven by rapid, though often inequitable, consumption growth.

With China’s poverty rate now down in the single digits, the baton has been passed to India. Large numbers of its population are on the precipice of escaping poverty, and others are not far behind, meaning that India has the capacity to deliver sustained progress on global poverty reduction over the next decade based on modest assumptions of equitable growth.

Once India’s poverty is largely exhausted, it will be up to sub-Saharan Africa to run the final relay leg and bring the baton home. Herein lies the challenge. Sub-Saharan Africa has historically struggled in its fight with poverty. In recent years, progress has begun yet the region still contains hundreds of millions of poor people, most of whom are located far behind the poverty line. Much of the movement in its curve occurs to the right hand side of the poverty line, implying that some portion of Africa’s population could remain impervious to a further 20 years of global economic progress and integration. Crucially, at no point in any of our scenarios are there as many poor people immediately below the $1.25 threshold in sub-Saharan Africa as has been recorded by China and India, which would facilitate more abrupt reductions in poverty.

Underlying the slowdown in the rate of poverty reduction in our scenarios is the property of diminishing returns. Diminishing returns are observed at two levels.

First, the rate of poverty reduction within each country is expected to slow as it approaches zero. We observe this slowdown in our scenarios for both China and India, which explains why they cede responsibility for leading global poverty reduction efforts before their poverty is eliminated.

Second, as global poverty approaches zero, it becomes increasingly concentrated in countries where the record of and prospects for poverty reduction are weakest. Today, a third of the world’s poor live in fragile states but this share is set to rise to half in 2018 and nearly two-thirds in 2030 according to our baseline scenario.

Eliminating poverty requires, among other things, the incorporation of isolated or excluded sub-national populations into the orbit of their economies; more deliberate and efficient targeting of the poor, including the poorest of the poor, at a country and sub-national level; and country-by-country progress in transitioning fragile and conflict-affected states onto a stable path.