Intuitively, defining poverty doesn't seem necessary: we know it when we see it. But while we can probably agree that the inhabitants of Dickensian slums were in poverty, in many cases today things aren't so clear cut.
It's partly because of this difficulty that there are several measures available. The main government figures are for absolute and relative poverty, measured before and after housing costs.
Although the figures produced by the Department for Work and Pensions (DWP) are referred to as measuring relative and absolute low income, it's common to refer to them as measures of poverty (for example, in the Child Poverty Act or at Prime Minister's Questions).
Relative poverty generally means that a person can't afford an "ordinary living pattern"—they're excluded from the activities and opportunities that the average person enjoys.
A household is in relative poverty (also called relative low income) if its income is below 60% of the median household income.
The median is the number 'in the middle' of a set—so half of all households earn more than the median income household, and half earn less.
The government wants to measure spending power, rather than earning power, so it counts incomes after taxes and benefits.
And households need different degrees of spending power to live comfortably, depending on their size and shape. A household with one person in it needs less money to live comfortably than a household consisting of a couple and two children.
So household incomes are “equivalised” in poverty statistics. They’re adjusted to take into account the number of adults and children who live there.
If you want to know where your own household fits in the income scale you need to adjust your income in a similar way. The Office for National Statistics has published an interactive that lets you do that.
Absolute poverty is slightly trickier. The definition used by a number of international organisations (such as the UN and the World Bank) is that you cannot afford the basic needs of life—food, clothing, shelter and so on.
This is absolute in the sense that it’s measured relative to a fixed standard of living, rather than the rest of the population.
This isn’t the definition used by the UK government.
As we explained above, it defines “relative poverty” in comparison to median incomes in the current year.
It defines “absolute poverty” in comparison to the median in 2010/11.
This gives a measure of poverty that’s constant over time.
Which measure to look at?
There's been a long political debate between relative and absolute measures of poverty.
Former Prime Minister David Cameron was at the forefront. In 2006 he felt that we needed to think about poverty in relative terms. By 2011 he felt that relative measures could lead to 'illogical' conclusions. He was happy to answer a question about absolute poverty with relative measures in 2015. Then it appeared that his doubts about relative poverty surfaced again a few months later.
It's reasonable to feel conflicted about this issue; neither measure is perfect and sometimes they can tell quite different stories.
Relative poverty measures can lead to some odd outcomes. It's somewhat perverse that you can be 'lifted' out of poverty simply because people around you are poorer.
Absolute measures have their own disadvantages. They won't tell you if those at the bottom are falling even further behind those in the middle, and they won't take into account changes in what we see as the minimum acceptable standard of living.
A more general problem with these figures is that a low income doesn't necessarily imply a low standard of living. The ONS uses the "60% of the median" benchmark to indicate whether someone is 'at risk of poverty'.
As the Institute for Fiscal Studies put it, absolute and relative measures tell us different things, and "there is nothing to be gained from ignoring some of this information". Considering both together paints a fuller picture.
The DWP produces figures for poverty using income before and after housing costs, like rent or mortgage interest.
There are some advantages to looking at people's income after housing costs have been paid. The price for accommodation of a given quality is highly variable across the country (the £2 million for a 'mansion' buys you rather less in London than it does in Wiltshire). The amount that people spend on housing also varies across age groups (with pensioners spending less and younger people spending more).
Additionally, measuring income after housing costs removes the influence of housing benefits from the figures. When rents increase and housing benefits rise to offset them, this is counted as an income rise. If housing costs aren't taken into account then this can be somewhat misleading.
It could be argued that applying this approach to only one part of spending is a little inconsistent. Although housing takes up a large part of most people's budgets, other important costs are not afforded the same treatment. Increased spending on food could reflect the purchase of more (or better quality) food, or it could just represent increasing prices. Similarly, pensioners and young people might differ in their spending on fuel for commuting and heating, and prices for goods and services other than housing also vary across the country.
What's happening to poverty?
Looking at the number of individuals in poverty can be misleading. As the number of households in the economy grows, all else being equal it's likely that the number of individuals below the poverty line will also tend to grow.
Looking at the proportion of the population below the poverty line gives a more accurate idea of how poverty is changing over time. So that’s what we’ve done here.
Absolute poverty has fallen slightly over the last ten years. It’s fallen by slightly more if you ignore housing costs.The proportion of individuals in relative poverty has been broadly stable over the last decade.
Unlike the rest of the population, pensioners are less likely to be in relative poverty if housing costs are taken into account.
The proportion of pensioners in various forms of poverty has fallen significantly over the long run, but the trend flattened out towards the middle of the last decade.
Children are more likely to be living in poverty than the general population, whatever measure you use.
The estimated percentage of children in relative poverty before housing costs increased by one percentage point from 2014/15 to 2015/16.
Who do the figures miss?
The household income survey used to create these figures is, as the name suggests, a survey of private households.
Homeless people sleeping rough or living in bed and breakfast accommodation are excluded from the figures. Other people are missed out too: people living in institutions like nursing homes or halls of residence are excluded.
How badly off are these people?
It's difficult to tell. As we've said, although it's common to refer to these figures as measuring poverty they measure income. And a low current income doesn’t necessarily mean a low standard of living.
For example, people can sustain their living standards after a fall in income if they run down their savings. Data on household income and spending shows that the bottom 10% of the income distribution spend in a way that suggests that they might not be in the bottom 10% of living standards.
This makes intuitive sense; you might expect people who spend time unemployed while they move between jobs (or for other reasons are not currently working) to spend their savings.
So as with many statistics, poverty figures can raise as many questions as they answer.