How poor is too poor?

The world is on track to meet the first of the Millennium Development Goals – to halve the number of people living below the poverty line by 2015. The trouble is that this line – set at a dollar a day – is a deeply flawed and unreliable measure of poverty.

More than a billion people live on less than $1.25 a day.

Desmond Boylan / Reuters

How we define poverty is critically important. Poverty is a moral concept: ‘poor’ is something we consider that people should not be. So, by setting our poverty targets according to a particular poverty line, we are saying that it is quite acceptable for people to live at that level of income, just as long as they don’t fall below it.

Millennium Development Goal One defines poverty as having an income below the dollar-a-day line – although actually this is now $1.25 per person per day, at purchasing power parity (PPP), at 2005 prices. This means that it is an income which would buy the same as $1.25-a-day in the US in 2005.

This is extraordinarily low by any standards. Living on $1 (or indeed $1.25) per person per day in the US or any other developed country would be fairly unthinkable. A few years ago, I did some rough calculations to show what a dollar a day would mean in Britain. In 2006, it was equivalent to 35 people living on a single minimum wage, with no benefits of any kind, no gifts, borrowing, scavenging, begging or savings to draw on (since these are all included as ‘income’ in poverty calculations), and no free health services or education (since these are not generally available to the poor).

It is therefore not surprising that living conditions at this level of income are little short of appalling. Adam Wagstaff, a World Bank economist, estimated in 2003 that children living at (not below) the dollar-a-day line in a typical developing country had between a 1-in-12 and 1-in-6 chance of dying before their fifth birthday, compared with about 1-in-140 in developed countries. The great majority of these deaths are undoubtedly directly or indirectly poverty-related; and between a third and half of those who survive are stunted by chronic malnutrition.

Nothing magical

It is difficult to argue that it is morally acceptable for people to live at this level of income, so long as they do not fall below it. But this is exactly what the Millennium Development Goals (MDGs) imply.

Clearly, the dollar-a-day line is too low; and it is too low because it is essentially arbitrary. There is nothing magical about an income of one dollar a day, and it certainly doesn’t provide any assurance that people will be able to lead a decent life – or even survive beyond early childhood. It was originally picked as the average of the lowest 10 national poverty lines found in a survey of developing countries, even though these were themselves set in different – and in some cases equally arbitrary – ways. It has gone on being updated in essentially the same way. The problems with this approach go far beyond the level of the dollar-a-day line, to the whole way incomes are calculated in estimating poverty. So setting the line at $2, $3 or $4 a day instead, though perhaps more realistic, still wouldn’t give us a meaningful measure of poverty.

Clearly, the dollar-a-day line is too low; and it is too low because it is essentially arbitrary

The failure of the current system arises largely from the use of purchasing power parity (PPP) – or, to be more precise, from the way it is calculated. In theory, PPP exchange rates are based on comparing the prices of the same basket of goods and services in different countries. This is more sensible than using a market exchange rate, which would set the poverty line too high in countries with an under-valued exchange rate and too low where it is over-valued.

But the PPP exchange rates used are designed for comparing countries’ national income, not people’s poverty – so they reflect total spending in the country, public as well as private. Poor people, by definition, spend much less than rich people. Even where they are a large majority of the population, they have only a small share of total spending. So PPP exchange rates overwhelmingly reflect the prices of goods and services bought by non-poor households, most of which are unaffordable by the poor anyway.

This is particularly problematic because the spending patterns of poor households differ systematically from those of the better-off. The poor spend a large proportion – often a majority – of their incomes on basic staple foods, which account for a relatively small proportion of the spending of the better-off, and therefore of the country as a whole. And the prices of basic foods vary far less between countries than other goods, and especially services, which are consumed more by the better-off. The result is to set the poverty line at an artificially low level.

A Burmese market trader counts her takings. The current ‘poverty line’ relies too heavily on income as a means of measurement.

Soe Zeya Tun / Reuters

The seriousness of this can be demonstrated by the effects of the 2008 food crisis, when the prices of basic foods such as maize and rice – on which many poor households depend – more than doubled in a matter of months. For a poor household spending half its income on maize, the effect of doubling its price is disastrous. There is no money left for anything else. But if maize represents only, say, five per cent of total spending in the country, doubling its price will increase the poverty line by only five per cent. The poverty line should go up by 50 per cent, but is actually increased by only five per cent, concealing a very considerable rise in poverty.

By its nature, the dollar-a-day approach only takes account of incomes. But this is only one part of what makes poor people poor. To live on $1 a day with access to free or affordable health services and education, and reasonable living environments and working conditions, is bad enough. Without these advantages, it is considerably worse. But these are not reflected in poverty figures, wherever the threshold is set. The UN Human Development Index and Human Poverty Index represent a creditable and worthwhile attempt to fill this gap; but they fall short of fully resolving the problem. (See box.)

Every decade or so, PPP exchange rates are recalculated using a later base year. In principle, this should make no difference to estimates of poverty, but actually the effects can be dramatic.

Using a base year of 1985, the proportion of people below the dollar-a-day line in Nigeria in 1993 was about the same as that in Mauritania. When the base year was updated to 1993, however, the poverty rate in Nigeria in the same year was 10 times what it was in Mauritania. The same updating increased the estimated level of poverty in sub-Saharan Africa by nearly half relative to that in Latin America. That a technical change which should make no difference makes such a big difference, raises serious doubts about whether the dollar-a-day estimates are meaningful or reliable at all.

When the base year is updated, the poverty line is updated too – however, it is not updated in line with any measure of price inflation, but (somewhat arbitrarily) on the basis of national poverty lines. When the base year was updated from 1993 to 2005, the dollar-a-day line was reduced by 14 per cent in real terms, having already been reduced when the base year was updated from 1985 to 1993. But the estimated level of poverty at this lower poverty line was half as much again as had been estimated using the previous higher line.

If the poverty line had instead been adjusted for inflation, poverty in 1990 – the starting point of the MDGs – would have been more than 49 per cent, nearly double the 28 per cent estimated before the 2008 update. Until two years ago, we thought the MDG was to halve poverty from 28 per cent. Now it seems, it is to reduce poverty to 24.5 per cent – only slightly below the point we thought we had started from.

A better approach

We need a new and better approach to defining and measuring poverty. First, we need to consider why we are concerned about poverty. It is not because people have less than $1 (or indeed $2, $3 or $10) a day – it is because they do not have enough income to allow them to live what we would consider a decent (or at least minimally adequate) life. So our aim should be to set the poverty line at a level where people can actually have a standard of living which we would consider morally acceptable.

Various alternative approaches have been proposed to deal with these problems – but none resolves them entirely. Either they rely on applying a single global poverty line using purchasing power parity, raising all the problems discussed above; or they depend on poor people spending their incomes on exactly the right things (and economists knowing what they are) and still only meeting their most basic need – an adequate calorie intake.

Our aim should be to set the poverty line at a level where people can actually have a standard of living which we would consider morally acceptable

I have therefore proposed a new approach, which I have called the Rights-Based Poverty Line (RBPL). This moves away from the idea of setting a single global poverty line, to country-specific lines. But it also avoids the conventional approach of estimating country poverty lines according to the cost of a basket of goods and services considered to be essential. Instead, the aim is to set poverty lines in each country which correspond to the same level of outcome indicators, reflecting health, nutrition, education, access to water and sanitation, housing, and so on.

This enables us both to link the poverty line to actual living standards – without making artificial assumptions about what people will spend their income on –and to avoid the problem of exchange rates and purchasing power parity, while maintaining consistency between countries. The difference is that consistency will be based on what it means to live at a particular level of income, and not estimates of spending power in terms of goods poor people can’t afford anyway.

Poor people spend most of their income on food. But when food prices soar, the dollar-a-day poverty line does not adequately reflect this.

Romeo Ranoco / Reuters

The starting point is to establish an agreed set of indicators which reflect economic and social rights – such as health, nutrition, education – and to set an agreed minimum level of each indicator that we consider to be morally acceptable. Based on the statistical relationship between income and each indicator in each country, we can then find the income at which the threshold minimum level is reached (on average) – and this is set as the poverty line corresponding with that indicator.

Of course, we need to decide which indicators to use, and what the threshold level should be for each. But, in a sense, this is part of the point of the exercise. Rather than allowing us to make an implicit moral judgment that a certain income is ‘enough’, without even considering what it means to live on that income, it makes us confront the question explicitly: just how much do we consider to be‘enough’? How poor is ‘too poor’?

This approach presents a more complex picture than the single headline numbers generated by the dollar-a-day system. In each country we have a number of different poverty lines, corresponding with the indicators we use; and for each line, we have two important indicators instead of one – the poverty line itself, and the proportion of the population below it. We also have two ways of reducing poverty. Either we can increase incomes, so that people move above the poverty line; or we can improve living standards at a given level of income, so that the poverty line itself is reduced.

The greater complexity of the RBPL approach can be seen as an advantage rather than a disadvantage. The simplicity of the dollar-a-day approach – its greatest triumph – is also a major limitation. Poverty is a complex phenomenon, and its nature varies considerably between different countries and communities. Any attempt to capture it in a single number will inevitably be an over-simplification. Neither can it provide us with a useful tool for policymaking.

The Rights-Based Poverty Line

To tackle poverty effectively, we need an approach which will capture this complexity, but which we can make sense of, and relate to our reasons for wanting to reduce poverty.

The dollar-a-day approach can only tell us if income poverty is going up or down – and even that it cannot do reliably. The RBPL approach can give us a great deal of additional information on the relative importance of incomes compared with social provision and other factors, differentiating between different aspects of living standards – health, education, nutrition and so on. This, not how many people are below an arbitrary global poverty line, is the information needed for priority setting and decision-making.

Don’t get me wrong – none of this is intended as a criticism of the economists who originated and developed the dollar-a-day approach. Not only was this a very considerable intellectual and technical feat, but it has had a major political impact in putting poverty on the global agenda. When it was established 20 years ago, it was the best – indeed the only – tool we had for measuring poverty. Without it, there would never have been a Millennium Development Goal to halve poverty, and global poverty would no doubt have gone on being almost entirely ignored in international discussions.

But that was 20 years ago – and over this time, the weaknesses of the dollar-a-day approach have become increasingly apparent, without the approach having been improved to deal with them. It served the purpose of the time, by putting poverty on the agenda; but now poverty is on the agenda, we need an approach which will give us a more reliable and detailed picture of poverty, so that we can make the right decisions to try to reduce it. The Rights-Based Poverty Line would be one way of doing this.

The Human Development Index

In 1997, UNDP sought to move beyond consideration of income poverty by developing a composite indicator, the Human Development Index (HDI). This combines three components, giving equal weight to each:

  • life expectancy;
  • adult literacy and enrolment in primary, secondary and tertiary education; and
  • the logarithm of GDP per capita.

Less well known is the Human Poverty Index (HPI), an indicator of the proportion of the population facing serious shortfalls in these dimensions. This combines three components, each again given an equal weight:

  • premature death (the percentage of people expected to die before the age of 40);
  • the adult illiteracy rate; and
  • standard of living (combining the percentage of the population without access to health services and without access to safe water, and the percentage of children under five suffering from malnutrition).

The relatively weak relationship between these indicators and GDP per capita and dollar-a-day poverty respectively, demonstrates both the weakness of the latter, and the value of the HDI and HPI as complements to them.

However, these indicators have some important limitations as poverty indicators.

As the UNDP itself recognizes, the selection of indicators and their weightings are inevitably arbitrary.

Since the HPI does not include an income component, it can only be seen as a complement to an indicator of income poverty such as the dollar-a-day figures.

The HPI can only provide an aggregate picture of the extent of poverty in a particular country (or sub-national population), and does not provide a means of identifying which households are poor.

Because it does not identify households or individuals as poor, it does not allow a link to be made with incomes.

In consequence, while the HPI provides an important complement to income poverty data, it does not represent an effective alternative.

David Woodward is a fellow of the London-based New Economics Foundation. This article is based on How Poor is ‘Poor’: towards a Rights-Based Poverty Line by David Woodward and Saamah Adballah. The full report is released this month and can be seen on