At the United Nations’ big gathering in late September, world leaders signed on to an ambitious pledge: “By 2030, eradicate extreme poverty for all people everywhere, currently measured as people living on less than $1.25 a day.” But just 10 days later, the goalposts shifted. The World Bank – which is in charge of setting the global poverty line — announced it was raising the line from $1.25 to $1.90 a day.

What gives? The cynical explanation is that the World Bank wants to make it harder to eradicate poverty by widening the definition; the fewer poor people there are, after all, the less there is for the World Bank to do.

The real answer is less alarming: The bank is just trying to make sure poverty data stays consistent over time. But the sudden, jarring change is an important reminder that you can’t capture the actual condition of the world’s poor in just one simple number.

Chart by Kenny and Sandefur.

Why does the "dollar a day" line keep moving?

The World Bank’s global poverty line is supposed to measure the bare minimum amount a person needs to survive in the poorest countries of the world. This definition has some theoretical appeal, but it’s hard to measure in practice. The economist Mattias Lindgren has tried to calculate the physical minimum line, which includes “the nutrition, water and warmth needed to survive and to produce what is needed to maintain the same consumption in the future.” He suggests that the baseline might be 67 cents a day in 2005 PPP, but that under different historical circumstances — and in particular different prices for basic foodstuffs — that could vary between 11 cents and $1.42. Nailing down a single point for the whole world just isn’t doable.

So historically, the bank has measured the cost of bare subsistence simply by taking an average of the poverty lines from some of the poorest countries worldwide. But this makes the line’s placement bounce around a fair bit. Back in 1990, this process yielded the World Bank’s original “dollar a day” poverty line. Then in 2005, the bank removed some big and rapidly developing countries like China and India from the basket and switched to a basket of poverty lines from 14 African countries plus Nepal. Subsequent research showed that this switch in the basket of poverty lines led to a jump in the global poverty rate of half a billion people.

When updating the numbers this year, the World Bank stuck to the same basket of 14 African countries plus Nepal that was used for the 2005 calculation, adjusting only due to price changes and not updating for any new national poverty lines. This compromise falls in an awkward no-man’s land. It is not fixed in US dollars. Nor is it based on the latest poverty lines from a group of newly picked poor nations. Nor is it based on an attempt to measure the very minimum consumption required to live.

You down with PPP?

A market in Vitshumba, Democratic Republic of the Congo. Comparing the prices of the goods here with prices in the US is very, very difficult — but the World Bank does it.

The poverty line is also very sensitive to how you compare prices between countries. Usually, international poverty calculations rely on estimates of “purchasing power parity,” or PPP. A popular and simplified version of a purchasing power exchange rate is the Economist’s Big Mac Index — how much do you pay for two all-beef patties, special sauce, lettuce, cheese, pickles, and onions on a sesame seed bun across the world? For the International Comparisons Program, housed at the World Bank, surveyors fan out across countries in every region and record the price of a whole range of items, like 100 grams of uncooked Nile perch served up in a paper wrapper, or a 300-gram bag of generic cornflakes. And they use that data to develop purchasing power exchange rates based on local prices. They completed one round of surveying for 2005 data and another, more recently, for 2011 prices.

Calculating purchasing power parity exchange rates is hard. The Big Mac index takes a highly standardized product and compares its price worldwide as a measure of purchasing power, for example. But not everyone eats Big Macs, and no one only eats Big Macs. Try comparing the goods consumed by the average poor person in rural Tanzania with those of someone living in Manhattan. There’s simply no overlap. And even if there were, the relative quantities and qualities of those goods would vary dramatically. A thatched hut 100 miles from Dar es Salaam may provide shelter and have the same square footage as an apartment on Central Park, but that will be about where the similarity ends.

So seemingly technical changes in how you compare prices among countries can wind up making a huge difference for poverty numbers. The new PPP numbers using 2011 prices released last year pushed China past the US as the world’s largest economy as of late 2014, rather than the previous forecast of 2019. We calculated with our former colleague Sarah Dykstra that global poverty — measured against a $1.25 line in real 2005 dollars — approximately halved as a result of the new 2011 PPP estimates.

But the World Bank does not keep the poverty line fixed in dollars, so things get a bit more complicated.

The same day the World Bank announced the new $1.90-a-day line — the biggest upward revision of the poverty line in 25 years — it also announced it would revise downward its official estimate of the number of poor people living in extreme poverty in 2011. That fell from 1,010 million to 987 million extremely poor people, or 14.2 percent of the world population.

This sounds like a contradiction: If the threshold for being poor went up, there should be more poor people. But it is possible because of purchasing power revelations. Recent analysis suggests that the currencies of poor countries could buy more than we thought. That raised the dollar value of poor countries’ poverty lines, but also the dollar value of poor people’s consumption. It turns out the second effect was slightly larger than the first. People in poor countries were buying more than we thought, so fewer of them had their consumption fall below the poverty line, and the count of poor people fell.

The poverty line had moved so that eradicating poverty wouldn’t be too easy — or too hard

In reaction to the new $1.90 poverty line, Arvind Subramanian, the chief economic adviser to the government of India, surmised, for the World Bank, “Poverty is always that level whose elimination hovers tantalizingly between being attainable and unattainable over the next decade.”

The narrative for the past few years has been that there are about a billion extremely poor people in the world, and getting that number down to zero by 2030 won’t happen automatically but might be possible with concerted effort. That made some change to the calculation of the extreme poverty line a political necessity: Without adjusting the method, the World Bank (and the UN, and the US president) would never be able to declare success. The poorest countries in 2030 would never set a poverty line so low that few or none of their population was below that a line. Any system of calculating global extreme poverty based on the most recent poverty lines of the poorest countries would always end up with a lot of people worldwide in absolute poverty.

In short, the World Bank’s old method of setting the poverty line meant that “eradicating” poverty was probably impossible. The line would always move up, just out of reach. Merely adjusting for inflation, like the bank did this time around, at least gets around that problem.

There is no “right” way to make the comparison of costs of living across people who consume very different things. At best there are more or less wrong ways. The process requires innumerable seemingly minor judgment calls, all of which can have a really dramatic effect on poverty numbers. A lot of people worldwide consume somewhere in the region of $1.25, so small changes in the line can have huge impacts on the number of poor. Andy Sumner and Peter Edward estimate that changing the poverty line by only 10 cents can change the global poverty line by 100 million. That means seemingly small choices or adjustments have a massive impact on the number of poor people worldwide.

It also gives a lot of power to the people doing the calculations. Between 2005 and 2011, the best available measures of relative prices around the world underwent seismic shifts — we now think India and China are much richer than we previously thought. But the World Bank has combined its updating approach with a number of tweaks to get to the 2011 poverty line and poverty numbers — using the new PPP numbers in some countries but not all, the consumer price index to inflate some prices, and survey techniques in other countries, etc. And as it happens, when you add up all the different tweaks the World Bank has made, they almost perfectly cancel out. The 2011 World Bank estimate for global poverty under the old PPPs was a 14.5 percent extreme poverty rate. The new estimate is 14.2 percent. This is a rather convenient coincidence.

Poverty is definitely decreasing, no matter how you look at it

While the headline global poverty rate didn’t change much, some places saw big shifts. Twenty-three million people in sub-Saharan Africa who lived in extreme poverty last week are officially no longer poor, while 11 million Latin Americans found themselves reclassified in the other direction. But ask people in Rio de Janeiro or rural Tanzania, and they’ll hotly deny anything is different. They can afford to buy the same amount of food, for which they’re paying the same prices. This revision has exposed how fickle our measures of global poverty are, and how weak some of the underlying data is.

Chart by Kenny and Sandefur.

Ultimately, picking a poverty line is pretty arbitrary. The main virtue of the line the World Bank has picked is political: It keeps the total number of global poor roughly where it was last month when world leaders at the UN signed up to eradicate poverty by 2030. But it did so by moving the line quite a bit. In the future, it’d be nice to have a measure of poverty that doesn’t swing wildly — or necessitate a change in methodology — every time exchange rates move.

But for all the uncertainty around poverty lines and numbers, and despite the fact that $1.90 is still obscenely low, one thing is certain: We do know the number of people living on $1.25 in 2005 dollars has been dropping — as has the number living on less than $1.90 in 2011 dollars. Most of the very poorest worldwide are able to buy more of what they need than they could 10 or 20 years ago. Perhaps the extreme poverty line is fuzzier than we thought, but the progress against extreme poverty remains clear.

Charles Kenny and Justin Sandefur are a senior fellow and a research fellow, respectively, at the Center for Global Development. Their views are their own and do not necessarily reflect those of CGD.