Five things we learnt about poverty dynamics in world’s fourth most populous country

I recently had the pleasure of working with the United Nations Children’s Fund (UNICEF) and the Indonesian Central Bureau of Statistics (BPS) on a year-long project on poverty and child well-being. It was a productive collaboration: we ran a series of technical workshops with statisticians; crunched data from a range of new household surveys; investigated the incidence, causes and effects of child poverty and poverty dynamics in Indonesia; and explored alternative policy options.

In this post, I present personal reflections on our latest report in the form of five points that are critical for better understanding poverty and designing effective social protection systems. They are: (1) poverty lines are somewhat arbitrary; (2) poverty statistics usually underestimate the scale of poverty and income insecurity; (3) moving out of poverty is not a smooth, upward journey; (4) ‘the poor’ are not a homogenous group; and (5) monetary and multidimensional measures do not necessarily classify the same group of people as poor. You can download the full report here

1. Poverty lines are somewhat arbitrary

Poverty is usually defined in relation to an income (or expenditure) threshold: those whose income falls below this threshold are classified as poor; those whose income exceeds it are not. In Indonesia, the government updates the poverty line annually, based on an estimate of the minimum expenditure required to satisfy basic food and non-food needs. In our study, we found that 11.3 million children – or 13.3 percent of the total child population – were living below this official poverty line in 2016.

However, the average value of the national poverty line is very low, at around 11,650 Indonesian Rupiah per person per day. Consider, for example, that one large coffee at a Starbucks in the capital city of Jakarta costs 45,000 Rupiah (or US$3.3 at current exchange rates). This means that this one beverage costs nearly four times more than the threshold used to classify people as poor.

We therefore introduced a range of alternative thresholds and found that indicators of poverty are highly sensitive to the choice of poverty line. For instance, based on the international poverty line for lower-middle income countries, one third of Indonesian children can be regarded as living in extreme or moderate poverty. And, doubling the value of the official poverty line (to about half the cost of a Starbucks coffee) would lead to a more than four-fold increase in the child poverty rate (to 57 percent).

The ambiguity of poverty lines should be better recognised in national and international policy discussions. Rather than fixating on a single poverty line, it is important to look at the entire distributions of income or expenditure and make comparisons across a wide range of poverty thresholds to get a more complete and robust picture.

Bjorn Blog_image1

2.Typical poverty statistics underestimate the scale of poverty

Poverty statistics are usually calculated from information collected in cross-sectional household surveys – that is, interviews with a random sample of households across the country. Such surveys give us a snapshot of the population at a particular moment, but they do not tell us much about poverty dynamics and mobility over time. It’s like looking at a photo of a football match; all the players are motionless in the image though in reality, of course, they were moving up and down the field.

In our study, we analysed data from a longitudinal survey conducted by BPS that interviewed the same 10,000 Indonesian households every year between 2011 and 2015. The results show that static headcount poverty rates can be misleading and underestimate the scale of poverty and insecurity. For instance, while the official child poverty rate was about 13 percent in 2016, twice as many children (26 percent) spent at least one year below the national poverty line in the preceding five-year period. Measured against the international poverty line for lower middle-income countries, 64 percent of children lived through at least one spell of moderate or severe poverty.

So, many people experience periods of low income and poverty at some point in their lives. And, changes in national poverty rates hide two opposing effects: movements out of poverty and movements into it.

Bjorn Blog_image23. Moving out of poverty is not a smooth, upward journey

In the physical world, there are precise parameters that signal a transition between two different states of matter. Ice will melt when the temperature rises above the freezing point, and a liquid will become a gas when it’s heated to the boiling point. Many people employ a similar logic when thinking about poverty transitions. For example, ‘graduation’ programmes often portray moving out of poverty as a one-directional upward journey, kick-started by better access to financial assistance and livelihoods opportunities.

Our analysis indicates, however, that improvements in living standards are rarely a linear upward trajectory. Many families who experience a temporary improvement in their financial situation are not able to sustain it. Between 2011 and 2015, over half of Indonesian children below the official poverty line in one year were not classified as poor a year later. Yet about a quarter of those who moved above the poverty line fell back a year later. The results are even more pronounced when using higher cut-offs to define poverty: a third of children with family incomes rising above the $3.1-a-day line fell back below that threshold within a year.

This illustrates that nothing ‘special’ happens when households cross poverty lines; they are merely analytical devices to help describe and monitor changes over time.

Bjorn Blog_image3

4. ‘The poor’ are not a homogenous group

The prevailing dichotomous approach to poverty measurement leads us to neatly split society into two groups: the poor and the non-poor. But the term ‘the poor’ is itself misleading, because it groups together individuals who may have nothing more in common than an experience of low income.

Indeed, ‘poor’ children in Indonesia are not a homogenous group but have very diverse characteristics. For instance, while child poverty rates are higher in rural areas, between 31 to 41 percent of Indonesian children below the official poverty line actually reside in urban areas. Children of single parents have a higher relative risk of poverty, but more than eight in 10 children in poverty live with both parents. For about half of them, their families’ main source of income is agriculture, but the other half depend on income from work in the service or industrial sector. And poverty is not just a problem associated with a lack of education: a quarter of children below the official poverty line have a household head who has completed secondary education or higher.

Moreover, the composition of the group of ‘poor’ people is constantly changing, as many families move above the poverty threshold and others fall below it every single year. In fact, only a small minority of Indonesian children were persistently below the official poverty line during 2011 and 2015. This implies that poverty should be conceptualised as an experience or a situation that people experience at certain points in their lives, and not a permanent characteristic.

5. Monetary and multidimensional indicators of poverty do not overlap neatly

The Sustainable Development Goals (SDGs) adopted by the United Nations General Assembly in 2015 aim to ‘end poverty in all its forms everywhere’ by 2030. The SDG targets include an explicit focus on multidimensional measures, to complement the traditional monetary indicators of poverty. In our report, we therefore analysed a composite measure of child poverty that combines 15 indicators on critical dimensions of child well-being related to health, nutrition, education, shelter, basic utilities, and child protection.

One striking finding is that monetary and multidimensional measures are correlated, but they do not necessarily classify the same group of people as poor. Multidimensional measures tend to yield much higher poverty rates: nearly nine in 10 Indonesian children experienced some form of deprivation in at least one of the six dimensions of well-being, while 65 percent was deprived in two or more dimensions. And family income is an important driver of child well-being, because the average number of deprivations declines monotonically when levels of household expenditure increase.

But that’s not the whole story. Many children who live far above the poverty line still experience significant forms of deprivation. Conversely, there is a small group of children below the poverty line who are not classified as deprived according to the multidimensional measure (between 1 and 12 percent, depending on the thresholds used). And, two provinces with similar levels of monetary child poverty can have very different levels of child deprivation.

Bjorn Blog_image4

Bjorn Gelders is blah and has xx years experience of blah

Bjorn Gelders is a Senior Social Policy Specialist and Economist at Development Pathways, specialising in poverty, inequality, vulnerability and social protection programming. His areas of expertise include mixed-methods research design, statistical analysis, microsimulation and policy analysis.


The five points outlined in this blog have important implications for the design of social protection systems. The composition of the group of people below the poverty line is constantly changing, with many families moving above and others falling below the threshold every single year. And, many people higher up in the income distribution still experience high levels of insecurity or deprivations in other dimensions of well-being. As a result of these poverty dynamics, there is no fixed, static group of ‘poor’ people that is easily identifiable and that can be targeted accurately for programme interventions. Indeed, reality is more volatile and complex than often assumed.

2 Responses to “Five things we learnt about poverty dynamics in world’s fourth most populous country”

  1. Dominic Oyaya Reply

    Very interesting findings which are very true and practical. They should be used by policy developers in social protection sector in coming up with practical solutions to poverty reduction. It has always been asummed that all poor people are homogeneous but it is forgotten that the causes of poverty are different for different people and even their needs and wants not the same. Once again thanks for sharing this very wonderful finding

  2. Richard Morgan Reply

    Thanks Bjorn, a brilliant post. It seems to imply that a ‘proxy (or any) means test’, to be accurate and to avoid moral risks of excluding people in need, would have to be carried out on a daily basis. I’m sure there are better and more practical ways, however, to take these insights around the volatility and complexity of poverty on board in the design of targeted social protection programmes …

Leave a Reply

Your email address will not be published. Required fields are marked *