Income Inequality and Poverty

We now turn to the implications of these changes in employment and welfare recipiency for household incomes and their distribution. The scale of the

Table 4.2. Percentage of working-age adults in work, by age of household head, 2004-9

Source: Analysis of EU-SILC Ireland microdata.

impact of the GR on national income, in the context of extremely rapid growth over the previous decade, must be emphasized: by 2010 GDP per head in nominal terms had fallen by close to one-fifth compared with 2007, and in real terms it was back to levels seen a decade earlier. However, as brought out in the analysis in Chapter 2, the impact on the household sector was much more muted, because much of the decline was felt in the company sector and because of the response of social transfers. The point of departure in terms of Ireland's level of income inequality (on which see Nolan and Maitre 2000; Nolan, Maitre, O'Neill, and Sweetman 2000; Nolan and Smeeding 2005) is also worth noting. Before the onset of the GR, summary measures of income inequality such as the Gini coefficient, Atkinson's inequality measure, the Theil coefficient, and the P90/P10 ratio—published for example by Eurostat and by the OECD in Growing Unequal (2008)—show that Ireland was above average within the EU-27 and the OECD. The Gini coefficient for Ireland around 2005 was 0.32, compared with a simple average of about 0.30 within the EU-27, putting Ireland on a par with countries such as Spain, Italy, the UK, and Poland, as well as among OECD countries Australia, Canada, and New Zealand. That level of inequality is consistent with relatively low social protection spending (as a proportion of national income) and low redistributive impact of income transfers together with direct taxes, arising in particular from low spending on public pensions, which in turn reflects both a low share of older persons in the population and the flat-rate nature of the social security pension system.

In terms of the institutional structures more broadly, Ireland is aligned with what is generally termed the Liberal welfare regime and has a level of income inequality similar to most of the other countries in that grouping. This remained the case over the years of the 'Celtic Tiger', when summary inequality measures were rather stable with only a modest increase towards the end of that boom (see Figure 4.1). In terms of decile shares, over the boom years there was some increase in the share going to the top 10%, but this was mostly balanced by a decline for others in the top half rather than further down the distribution. Trends right at the top of the income distribution may be better captured by data from the income tax system, and estimates of the share of total income going to the top 1% from that source do suggest a sharper increase (Nolan 2007).

Ireland's economic boom was also notable for a substantial increase in married women's labour force participation, but this did not have a disequa- lizing effect on the household income distribution because it was as common for women married to lower-earning as to high-earning men. Social welfare support rates initially lagged behind average earnings, especially net of tax as direct taxes were cut, but subsequently made up much of that ground, as larger increases were awarded in annual budgets in the latter years of the boom.

Figure 4.1. Trends in inequality in Ireland: Gini coefficient of equivalized net household income, 1987-2010

Source: Calculated by the authors from survey microdata from ESRI 1987 Survey, Living in Ireland Survey 1994-2001, and EU-SILC 2003-9;2010 estimate based on CSO (2011). Figures for 1987-2009 use the modified OECD equivalence scale. The CSO calculation for 2010 uses a slightly different equivalence scale which if applied to earlier years yields Gini coefficient estimates that are a little lower than does the OECD scale;a small adjustment has therefore been made to the CSO figure on this basis. (The series for the Gini published by the CSO shows the same upward jump in 2010.)

While substantial in real terms, those increases in social welfare were still exceeded by the increase in average net household income (before or after equivalization to take household size and composition into account), to which increasing numbers of earners in the household also contributed.

Turning to the period from the onset of the GR, the impact on average household incomes and on the distribution was striking. Here we rely on the SILC household survey, the only annual source of data on household incomes available for Ireland, carried out by the Central Statistics Office with a sample of about 6,000 households and designed to produce data to report to Eurostat in the EU-SILC framework. (In contrast with the situation for Germany, Italy, and Sweden in Chapters 3, 5, and 6, we are using the same source of household survey data that was drawn on in the comparison of household incomes in European countries in Chapter 2: section 2.6.) Whereas EU-SILC income data for most other EU countries use the previous calendar year as the income reference period, Ireland uses the 12 months prior to the interview date, with interviewing carried out throughout the year, and thus data are more up to date. At the time of writing, detailed results from the 2009 survey have been published and the microdata up to that year have been analysed for this chapter, so that is where our discussion concentrates. Preliminary headline results for 2010 have also been published (Central Statistics Office 2011) and will be described. However, since only limited information has been released as yet and unit record data are unavailable, it has not been possible to explore them in any depth.

Table 4.3 shows that average net equivalized income continued to rise in 2008 but fell by over 5% in 2009, with the median falling by 7%; similar trends were seen for gross income, before direct taxes and social insurance contributions were deducted. Changes in the composition of total income coming into households in the survey, underlying these declines in the average, are also shown: there is a decline in the share of total income coming from employment in both 2008 and 2009 (with much of that fall in self-employment income) while investment income also falls, with a sharp rise in the importance of cash social transfers and a marginal decline in direct tax and social contributions as a share of net income.

To see where in the income distribution these changes will have had their effects, Table 4.4 compares the composition of net income in 2007 and 2009 for those in the bottom 30% of the distribution, the next 50%, and the top 20%. Focusing first towards the bottom, there was relatively little change for

Table 4.3. Mean and median net and gross equivalized household income, and composition of net income, 2004-9

Source: Analysis of EU-SILC Ireland microdata. Price deflator is Consumer Price Index available from the CSO at http://www.cso.ie/statistics/consumpriceindex.htm

Table 4.4. Shares of income sources in net household incomes (%), by income group, 2004-9

Source: Analysis of EU-SILC Ireland microdata.

the bottom 30%: two-thirds of its income already came from social transfers in 2007 and this went up by a modest one percentage point, with a corresponding fall in income from employment and investment. For the net 50% of the distribution the changes were more pronounced: income from employment fell by eight percentage points and investment income by two percentage points, with a sharp rise in the share coming from cash transfers and a fall in the share of income tax and social contributions. For the top 20%, there was much more stability in income composition, with some decline in investment income and increase in the share of income tax and social contributions.

Focusing on the distributional implications, the share of income going to each decile group and summary inequality measures for net equivalized income (using the square root of household size equivalence scale) are shown in Table 4.5. Strikingly, it is the share of the top 10% that sees most change, falling from close to 25% before the crash to 23.2% by 2009, whereas the shares of each of the bottom five decile groups rose, so the share going to the bottom half of the distribution went up by 1.3% of total income. This is reflected in a decline in the summary measures of inequality, with the Gini coefficient for example down to 0.30 by 2009, a decline of 6%. (Indeed, the distribution for 2009 Lorenz-dominates that for 2007.)

Up to that point, then, the impact of the recession was equalizing rather than disequalizing, indeed quite considerably so. In addition to the most immediate effects impacting substantially on income towards the top (via profits and income from capital and self-employment), the nature and phasing of the policy response via the tax and social welfare systems is particularly important in this case. As we discuss in the next section, increases in taxation dominated in 2008 and 2009 and were rather progressive in character, whereas from 2010 reductions in social welfare were also implemented, which will have affected the shares of those towards the bottom.

Table 4.5. Decile group shares (%) and inequality indices (net equivalized household income), 2004-9

Source: Analysis of EU-SILC Ireland microdata.

The headline results for 2010 suggest that this, and other factors, contributed to a dramatic increase in inequality in that year, with the Gini coefficient rising from 0.30 to over 0.34—see Figure 4.1. This increase may also reflect the impact of lump-sum payments to those retiring, notably from the public service where early retirement was incentivized to reduce the pay bill, since these lump sums are counted as income in the year received. It will take some time before the medium-term impact of the GR on income inequality becomes clear, but it is highly significant that the initial inequality-reducing impact in the Irish case has not been sustained.

To give some indication of the impact of the recession on different types of person and household, Table 4.6 shows how changes in mean income differed by household type, and for persons classified by age and gender. Each comparison reveals a dramatic difference between older persons and the rest of the population: between 2007 and 2009 those aged 60 or over and their households saw substantial increases in average net income in real terms, of 10% or more, while other households, and both adults of working age and children, saw declines of 3% to 6%. This reflects the impact of declines in employment and in income from self-employment on those of working age and their families, together with the remarkable extent to which pensioners have been insulated from the effects of the crisis in income terms.

As well as the overall impact of the recession on the distribution of income, the impact on incomes towards the bottom, in other words on income poverty, is of particular interest from a policy perspective. With the widely-used relative approach to deriving an income poverty threshold, incorporated for

Table 4.6. Mean real net equivalized income (Euros, 2004 prices), by household and person type, 2004-9

Source: Analysis of EU-SILC Ireland microdata; price deflator is Consumer Price Index, CSO: http://www.cso.ie/statistics/ consumpriceindex.htm

example into the EU's set of commonly-agreed social inclusion indicators, a threshold of 60% of median equivalized income in the country in question is often employed. In the case of Ireland, relative income poverty measured in this fashion increased in the earlier years of the Celtic Tiger boom, despite the sharply rising levels of employment and incomes from work, largely because those remaining reliant on social transfers fell behind. Much of this ground was made up in the latter part of the boom, so that by 2005-7 Ireland's relative income poverty rate was above the EU- and OECD-averages but not an outlier, see Table 4.7.

The immediate impact of the GR was to reduce this relative income poverty measure: in 2007, 20% of persons were in households below 60% of median equivalized household income, and this fell to 19% in 2008 and 17.5% in 2009. This reflects the factors already adverted to in discussing the increasing share of total income going to the bottom deciles in the distribution—relating to the incomes most affected, the nature of the policy response, and timing—

Table 4.7. Poverty indicators, 2004-9 (% of individuals)

Source: Analysis of EU-SILC Ireland microdata.

as well as the nature of this poverty measure (on which see the discussion in Chapter 1). Those towards the bottom of the income distribution were already heavily reliant on social transfers as the main source of income, since the households involved largely comprise older persons and those of working age who, for a range of reasons, are not in sustained employment. The increase in unemployment and corresponding decline in income from employment, which was the most obvious effect of the recession, will thus have left many of them unaffected. The evolution of income support rates for the unemployed, pensioners, and others relying on social protection is also important. These cash transfers are not formally indexed to prices or wages; instead any increases are entirely at the discretion of the government of the day, generally announced along with taxation changes in the annual budget statement. Support rates provided in weekly social transfers were actually increased for 2009, as discussed in more detail in the next section. These increases in income support took place at a time when, most unusually, poverty thresholds framed in purely relative terms were going down since average (median) incomes across all households were declining. So the relative position of those relying on social transfers improved considerably, serving to offset the impact of increasing numbers relying on those transfers. These factors are reflected in the changes in composition of those falling below relative income poverty thresholds: in 2007, 9% of those below the 60% relative income threshold were unemployed, but this was up to 13% by 2009; the proportion made up of those who are retired fell by about the same amount. Headline results for 2010 show that the relative income poverty threshold fell by 10%, much more rapidly than in 2009, and despite this the percentage of persons falling below it rose by about 1.5 percentage points. This increase was concentrated among children and working-age adults, with the rate for older persons unchanged.

In such a situation it is useful to also look at what happened to poverty rates calculated using income thresholds that remain constant in purchasing power terms rather than moving in line with average incomes—often referred to as 'anchored' poverty thresholds. Taking a low-income cut-off in 2007 as 60% of median income and subsequently moving in line with the consumer price index, the percentage of persons falling below that threshold was stable to 2009, in contrast to the fall in relative income poverty over those years, and headline results suggest it rose sharply in 2010.

It is also relevant to note trends in deprivation over these years, as captured by a range of non-monetary indicators included in the same household survey. These showed increasing levels of deprivation in 2008 and a sharper rise in 2009. Taking 11 deprivation items used in monitoring poverty in Ireland, the percentage reporting deprivation on two or more items rose from 12% in 2007 to 14% in 2008 and 17% in 2009. The largest increases were seen for items such as being able to afford an afternoon or evening out or to replace worn-out furniture. Headline results for 2010 show a further sharp increase in deprivation, with 22.5% reporting deprivation on two or more of these items.