An assessment of Labour s record on income inequality and poverty

Oxford Review of Economic Policy, Volume 29, Number 1, 2013, pp. 178 202 An assessment of Labour s record on income inequality and poverty Robert Joyce* and Luke Sibieta** Abstract We document the evolution of average incomes, poverty and inequality over the period of Labour government from 1997 to 2010, comparing these trends with those seen over other periods in recent history. We also relate these changes to Labour s stated distributional objectives, which we argue were clear in relation to the bottom of the income distribution (and absolutely explicit in relation to child poverty), but much less clear in relation to inequality more generally. It is thus perhaps no surprise that we observe substantial falls in child and pensioner poverty alongside increases in overall income inequality, with the latter largely driven by growth in top incomes. We also use micro-simulation techniques to demonstrate the crucial importance of direct tax and benefit reforms in driving the changes observed over this period. Key words: inequality, poverty JEL classification: D31, I30, I32 I. Introduction When Labour came to power in 1997, income inequality stood at historically high levels, and relative income poverty had risen substantially since the party was last in office in the late 1970s. In this article, we analyse the record of the Labour governments from 1997 to 2010 in addressing income inequality and poverty. We assess how the income distribution changed over the period and how these changes related to Labour s policies. This draws on new results from a tax and benefit micro-simulation exercise. It is important to try to discern what Labour s distributional objectives were, and the extent to which policy as opposed to other economic and societal factors contributed to trends in poverty and inequality. It quickly became apparent that Labour was focused on reducing poverty among families with children and pensioners. In 1999, Tony * Institute for Fiscal Studies, e-mail: robert_j@ifs.org.uk ** Institute for Fiscal Studies, e-mail: luke_s@ifs.org.uk The authors gratefully acknowledge funding from the ESRC Centre for the Microeconomic Analysis of Public Policy at the Institute for Fiscal Studies. The paper also draws partly on work supported by the Joseph Rowntree Foundation. The authors would like to thank Anthony Atkinson, Mike Brewer, Paul Johnson, John Micklewright, and participants at the Oxford Review s seminar on Labour s economic record for their comments on previous versions of this paper. The Family Resources Survey is crown copyright material and is reproduced with the permission of the Controller of HMSO and the Queen s Printer for Scotland. It was obtained from the Economic and Social Data Service at the UK Data Archive. doi:10.1093/oxrep/grt008 The Authors 2013. Published by Oxford University Press. For permissions please e-mail: journals.permissions@oup.com

An assessment of Labour s record on income inequality and poverty 179 Blair made a famous commitment to end child poverty within a generation; and Gordon Brown promised to end pensioner poverty in our country at the 2002 Labour party conference. As discussed in this paper, these commitments clearly motivated policy and, in particular, the substantial increases in welfare spending on low-income pensioners and families with children. All else being equal, reductions in poverty will tend to reduce inequality as well. However, it is much less clear that Labour took a strong view on the appropriate level of inequality within the top half of the income distribution. Its 1997 manifesto deplored the fact that There is a wider gap between rich and poor than for generations, but was thin on commitments to address income inequality. Indeed, statements made by senior government figures suggested a keenness to emphasize a lack of such intentions: Peter Mandelson, a key architect of New Labour, famously stated that he was intensely relaxed about people getting filthy rich as long as they pay their taxes. Labour also made firm commitments not to increase the basic or top rates of income taxation, which limited its ability to use the tax system to influence income inequality. If this is an accurate summary of Labour s objectives, then we show that outcomes closely reflected those objectives, although the scale of the changes achieved did not always match the considerable ambition. Poverty among families with children and pensioners both fell particularly strongly, although the extremely ambitious relative child poverty target was still missed in 2010/11. However, the incomes of low-income working-age adults without dependent children, a group not favoured by tax and benefit reforms, rose very little over the period, and hence their relative poverty rate actually increased. As a result, Labour s period in office saw a convergence in the risks of poverty across the major demographic groups. Incomes at the very top of the distribution saw the strongest growth of all during Labour s period in office. Largely as a result of that, most measures of income inequality nudged up. The main focus of this paper is on household incomes, and poverty and inequality statistics based upon the distribution of household incomes. This is consistent with most of the academic literature on the income distribution, and with the standard measures of poverty and inequality tracked by governments (for example, see Department for Work and Pensions (DWP, 2012)). But we recognize that there are other important and relevant dimensions of distribution, such as intra-household allocation and gender inequalities. Although a comprehensive treatment of all such dimensions is beyond the scope of this paper, we note these issues where we believe they are relevant and refer the reader to relevant research. Elsewhere in this issue, Lindley and Machin (2013) analyse trends in wage inequality under the 1997 2010 Labour governments. The paper proceeds as follows. Section II provides important context to the discussion of poverty and inequality by briefly summarizing trends in average living standards under Labour. Section III analyses what happened to income inequality and what drove these trends. Section IV examines trends in poverty and the extent to which they reflect the policy priorities and reforms. Section V concludes. II. Average incomes Before discussing how income inequality and poverty changed over Labour s period in office, it is important to describe how average living standards changed. Table 1 compares five measures

180 Robert Joyce and Luke Sibieta Table 1: Measures of income growth compared (%) GDP per head Household a final consumption expenditure per capita Real household disposable income per head Mean HBAI income (BHC b ) Median HBAI income (BHC) Conservatives 1979 to 1996/7 Labour 1996/7 to 2009/10 of which: Fast growth (1996/7 to 2001/2) Weak growth (2001/2 to 2007/8) Recent recession (2007/8 to 2009/10) Post-recession: 2009/10 to 2010/11 2.0 2.6 2.6 2.1 1.6 1.8 2.0 1.9 1.9 1.6 3.3 4.2 3.4 3.4 2.9 2.4 2.1 1.2 0.9 0.8 3.6 3.3 0.7 1.2 0.6 1.4 0.3 2.8 5.7 3.1 Notes: a And non-profit institutions serving households. b BHC: before housing costs have been deducted. Incomes have been measured before housing costs have been deducted. National accounts measures are for the UK as a whole, while HBAI measures relate to Great Britain only. Source: Authors calculations using ONS series IHXW, IHXX, and IHXZ, and HBAI data. of growth in average living standards. Three are derived from the National Accounts: gross domestic product (GDP) per capita, real household disposable income (RHDI) per capita, and household final consumption expenditure (HFCE) per capita. The remaining two are mean and median household income. These are based on the Family Resources Survey (FRS) and take the same definition of income as used by the government in Households Below Average Income (HBAI), i.e. net, equivalized household income from all sources. The different series measure living standards in different ways. Of the National Accounts measures, real GDP per head is a widely used measure of economic wellbeing, showing the estimated market value of all final goods and services produced in the UK economy, divided by the total number of people in the UK. One important advantage of using GDP per head to measure living standards is that it also captures expenditure on public services and thus will not fall one-for-one with any tax increases used to pay for more public services. In contrast, household income measures will fall with such tax increases, with no account taken of the extra spending on public services or its impact on living standards. Real household disposable income, as the name implies, focuses on the household sector, 1 and so excludes the incomes of companies and the government. Household final consumption expenditure (including 1 Though the household sector used for this measure also includes charities and universities.

An assessment of Labour s record on income inequality and poverty 181 the expenditure of non-profit institutions serving households) is a measure of spending, rather than income, and is also derived from the National Accounts. It captures expenditure incurred by households on consumption of goods and services. The HBAI measures focus on current household disposable income, adjusted for family size and composition. These measures allow us to look at the whole distribution of income (unlike the National Accounts measures). Despite these definitional differences, growth over Labour s period in office was broadly similar across all five measures, at around 2 per cent per year on average. The only exception is median income growth, which grew less quickly than the other four measures (all of which are means). This reflects continued increases in inequality within the top half the income distribution, as discussed in the next section. Growth in mean living standards was generally higher under the period of Conservative government between 1979 and 1997 than under Labour, particularly for household disposable income and consumption. As suggested by previous work, there were three quite distinct phases in growth in average living standards under Labour (Cribb et al., 2012). During the first phase (1996/7 to 2001/2), average income growth was relatively fast, with most measures growing by around 3 per cent per year (and household consumption growing even faster at more than 4 per cent per year). This was followed by a period of sluggish growth in average incomes between 2001/2 and 2007/8, with only about 1 per cent average annual growth. Growth in household consumption and GDP per capita also slowed, but not to the same extent. Interestingly, this meant that average household consumption grew by a total of 39 per cent over Labour s first 10 years of office, while measure of disposable income grew by 27 per cent or less. This suggests that part of the growth in household consumption was driven by household borrowing. The gap between GDP per capita growth and disposable incomes over the 2000s may partly reflect the increase in the tax burden over this period (Chote et al., 2010a). However, it is difficult to place a monetary value on the value to households of the increased spending on public services such as health and education over the same period (Chote et al., 2010b). The last years of the Labour government were dominated by the financial crisis and recession. GDP per capita fell substantially, by more than 7 per cent between 2007/8 and 2009/10. Nevertheless, average disposable incomes actually continued growing slowly over that period. Previous work (Cribb et al., 2012) suggests that this largely reflects a combination of very low and falling inflation, increased welfare payments, and a fiscal stimulus which included a temporary cut in VAT from 17.5 to 15 per cent. This delayed impact of the late-2000s recession on household incomes was an experience shared by other countries (for a detailed cross-national comparison, see Jenkins et al. (2012)). But the respite was temporary and a large fall in average disposable incomes followed in 2010/11. Interestingly, falls in household consumption clearly preceded falls in household income, and occurred more contemporaneously with falls in GDP, perhaps reflecting the earlier adjustment of consumption patterns in light of the effects of the recession on expectations about future income. For detailed analysis of household spending during and after the recent recession, see Crossley et al. (2011). To summarize, average income growth under Labour was slightly lower than under the Conservative governments between 1979 and 1997, but this masks considerable variation over time. Income growth was largely concentrated in the first 5 years of Labour s period in office, and the subsequent slowdown far predated the late 2000s

182 Robert Joyce and Luke Sibieta recession. From 2002 onwards, average income growth was sluggish, and it finally went into reverse after the recession and financial crisis. III. Income inequality Labour had no specific targets relating to overall income inequality. It was very clear that it wanted to see a reduction in relative income poverty (at least among pensioners and children), which implies a narrower gap between the bottom and the middle of the income distribution. But, as indicated by the earlier quotation from Peter Mandelson, there were no clear goals with regard to the level of income inequality in the top half of the income distribution and thus the overall level of income inequality. Things started to change at least implicitly after the onset of the global financial crisis in 2008. To raise revenue to close a widening budget deficit, those on high incomes were expected to bear the biggest burden. Alistair Darling, the then Chancellor, announced plans to introduce a new 50 per cent marginal rate of tax on incomes above 150,000 from April 2010, as well as other measures aimed at increasing the tax take from individuals with very high incomes: withdrawing their personal income tax allowance and restricting the generosity of tax relief on their pension contributions. Darling said: I considered a number of options to raise revenue in future years. And I have chosen those which are fairest and affect those who have done best out of the growth of the last decade (Hansard, 2008). There is no explicit reference to inequality, but the intended distributional consequences of the changes were clear. The rest of this section discusses trends in income inequality between 1997 and 2010. We then discuss potential explanations for these trends, including policy reforms and wider socio-economic changes. (i) Trends in income inequality under Labour Figure 1 shows the evolution of the Gini coefficient in Great Britain since 1979. 2 Inequality as measured by the Gini rose dramatically over the 1980s, from around 0.25 in 1979 to around 0.34 in the early 1990s. The scale of this rise in inequality has been shown to be unparalleled in recent British history and was larger than the growth in income inequality occurring at the same time in most other developed countries (the United States being a key exception). 3 Between the early 1990s and 2009/10, the changes in income inequality were less dramatic. After falling slightly over the early to mid-1990s, the Gini rose again during Labour s first term, reaching a new peak of 0.35 in 2000/1. During Labour s second term, however, the Gini fell. It then crept up again during the first 3 years of Labour s third term, but held steady during the financial crisis at just under 0.36 a historic high 2 Data for the UK as a whole are only available from 2002/3 onwards. We therefore focus on inequality within Great Britain only. Between 2002/3 and 2010/11 the trends for the UK as a whole were very similar to those for Great Britain only. 3 See Atkinson (1999), Goodman et al. (1997), and Gottschalk and Smeeding (1997).

An assessment of Labour s record on income inequality and poverty 183 Figure 1: The Gini coefficient (Great Britain) Note: The Gini coefficient has been calculated using incomes before housing costs have been deducted. Source: Authors calculations using Family Expenditure Survey and Family Resources Survey, various years. since our consistent data series began in 1961. Overall then, the Gini coefficient rose from 0.33 in 1996/7 to 0.36 in 2009/10. Other summary inequality measures exhibit a similar pattern (see Cribb et al., 2012). Although Labour was in power for only 1 month of the 2010/11 financial year, it would be very unusual for an incoming government to significantly change the tax and benefit system midway through a financial year, and it is unlikely that any other mechanism by which the government could have affected the income distribution would have had substantial immediate effects. It therefore makes sense to include 2010/11 when considering Labour s record. The Gini coefficient in Great Britain fell from 0.36 to 0.34 in 2010/11, returning inequality to just below its 1997/8 level. This 1-year decline in the Gini was statistically significant 4 and the largest 1-year fall for at least 50 years. Of course income changes in that year cannot all be directly attributed to the policies of either the incoming or outgoing governments, as they were obviously strongly related to the global financial crisis. But Labour s policy did have some role to play when it comes to changes in top incomes, and hence summary measures of income inequality. The introduction of the new 50 per cent marginal rate of income tax in April 2010 will clearly have reduced the after-tax incomes of high-income individuals before accounting for any changes in behaviour. After behavioural response, the effect on net incomes at the top of the distribution is likely to have been even larger as a result of financial incentives for these individuals to reduce their hours (assuming the substitution effect dominates the income effect), engage in tax avoidance or evasion, and/or emigrate. However, it will also have provided strong incentives for high-income individuals to bring forward their income into 2009/10 in order to reduce their overall tax burden, artificially inflating measured top incomes in 2009/10 and reducing them in 2010/11. Hence, some of the reduction in the Gini coefficient in 2010/11 is almost certainly due 4 Standard errors were calculated by bootstrapping. See Cribb et al. (2012) for further details.

184 Robert Joyce and Luke Sibieta to the unwinding of the temporary forestalling behaviour of very rich individuals in the previous year. It is therefore difficult to ascertain the extent to which the reduction in income inequality in 2010/11 will prove to be a permanent or temporary phenomenon (especially with more planned changes to top rates of tax in the pipeline: see Cribb et al. (2012) for more details). It is important to note, however, that these issues affect only the very richest individuals (approximately the top 1 per cent), whereas inequality in 2010/11 declined right across the distribution. This wider reduction in inequality was largely because real earnings fell by more than real state benefit levels (see Cribb et al., 2012). A large part of the trends in summary measures of income inequality such as the Gini coefficient under Labour can be explained by the top of the distribution. To illustrate this, Figure 2 shows the level of the Gini coefficient over time if we were to exclude the richest 1, 5, and 10 per cent of individuals in any given year. This provides some guidance as to how sensitive the levels and changes in the Gini coefficient have been to the top of the distribution. As one would expect, the Gini coefficient goes down in any given year if one excludes the top of the distribution. The changes under Labour between 1997 and 2010 also become much less pronounced, suggesting that trends in the Gini over this period were largely being driven by the top of the distribution. Finally, it is noteworthy that all measures of the Gini coefficient show a fall in 2010/11, reflecting the fact that income inequality fell unambiguously in that year across the entire income distribution (and, in particular, that the fall in inequality was not driven only by temporary or permanent effects of the new 50 per cent marginal income tax rate). To further understand the changes in income inequality under Labour, Figure 3 moves away from summary measures of inequality and shows how incomes changed right across the distribution between 1996/7 and 2010/11. Growth was quite evenly spread, with the highest growth in the second and third poorest decile groups and the Figure 2: The Gini coefficient (Great Britain), excluding various top income groups Note: The Gini coefficient has been calculated using incomes before housing costs have been deducted. Source: Authors calculations using Family Expenditure Survey and Family Resources Survey, various years.

An assessment of Labour s record on income inequality and poverty 185 Figure 3: Real income growth by percentile point, 1996/7 to 2010/11 (Great Britain) Notes: The changes in income at the 1st and 99th percentiles are not shown on this graph. Incomes have been measured before housing costs have been deducted. The differently-shaded bars refer to decile groups. Source: Authors calculations using Family Expenditure Survey and Family Resources Survey, various years. lowest growth towards the top of the distribution (the seventh, eighth, and ninth decile groups). By itself, this would suggest falling inequality. However, the trends in the tails of the distribution explain the overall increase in summary measures of inequality. The poorest 10 per cent experienced lower-than-average growth, and the richest 10 per cent saw strong income growth. Indeed, inequality was increasing quickly even within the top decile group, with the top 1 per cent seeing the fastest growth of all. This pattern reinforces the point that increases in overall income inequality over this period were largely driven by trends right at the top of the distribution. The figure makes two further important points. First, the grey line shows average annual income growth across the distribution from 1996/7 to 2009/10 (the last full financial year of Labour s period in office). The difference between this and the bars shows the impact of including data from 2010/11 on the average annual rate of growth in incomes at each percentile since 1996/7. The difference is striking: for instance, the fall in income in 2010/11 reduced average annual median income growth over the entire period from 1.6 to 1.2 per cent. The largest difference between growth to 2009/10 and to 2010/11 is at the very top of the income distribution, where the falls in income in the most recent year of data were largest (as might be expected, given the changes to top rates of tax and the resultant incentives). Second, the black line shows the average annual income growth from 1979 to 1996/7, when the Conservatives were in office. It shows that increases in incomes in the bottom part of the distribution were greater under Labour and increases in the top part of the distribution were greater under the Conservatives. This also explains, in part, why income growth at the mean was higher under the Conservatives. Figure 4 breaks income trends under Labour into four separate periods: fast growth in average incomes (1996/7 to 2001/2); weak growth (2001/2 to 2007/8); the recession (2007/8 to 2009/10); and 2010/11. Apart from at the very bottom of the distribution, 1996/7 to

186 Robert Joyce and Luke Sibieta Figure 4: Real income growth by percentile point in four periods since 1996 97 (Great Britain) Notes: The changes in income at the 1st, 2nd, and 99th percentiles are not shown on this graph. Incomes have been measured before housing costs have been deducted. Source: Authors calculations using Family Resources Survey, various years. 2001/2 saw the strongest income growth of all. Growth over this period was also inequality-reducing across most of the distribution, with the highest growth at around the 30th percentile. However, the tails of the distribution were a different matter, with fast growth for the top 5 per cent and much slower growth for the bottom 10 per cent. The period between 2001/2 and 2007/8 saw much slower income growth across the board. For most of the distribution, income growth was relatively flat at around 1 per cent per year. However, strong growth at the top and real falls in income at the bottom acted to increase income inequality. Income growth between 2007/8 and 2009/10 was similar to that for the preceding 6 years for the top two-thirds of the income distribution. But growth was noticeably higher in the bottom third during the recession than over the previous 6 years; and the difference was most pronounced for the poorest 20 per cent. This is likely to reflect a number of factors that led to real-terms increases in benefits and tax credits: falling inflation and the uprating rules for state benefits and tax credits (by which their rates are generally increased annually in line with a lagged measure of inflation); and discretionary changes to benefits and tax credits. This meant reduced inequality across most of the income distribution during the recession. In 2010/11, there were falls in real income right across the income distribution, much of which reflects the unwinding of some of the factors supporting real disposable incomes during the recession, such as the temporary cut to VAT. The pattern of income falls was unambiguously inequality-reducing, with real falls as large as 5.1 per cent at the 90th percentile. This clearly shows that the fall in income inequality in 2010/11 was driven not just by changes at the very top of the distribution: there was a pattern of inequality-reducing growth right across the board. In Figure 5, we show the consequences of these patterns of income growth on the 90/10, 50/10, and 99/50 ratio measures of inequality. During the 1980s, each of these

An assessment of Labour s record on income inequality and poverty 187 Figure 5: Income ratios: 90/10, 50/10, and 99/50 ratios (Great Britain) Note: The Gini coefficient has been calculated using incomes before housing costs have been deducted. Source: Authors calculations using Family Expenditure Survey and Family Resources Survey, various years. ratio measures increased, as one would expect given the unambiguously inequalityincreasing pattern of income growth documented in Figure 3. The 99/50 ratio saw the fastest increase of all: income at the 99th percentile was three times the median in 1979, but more than five times the median by the early 1990s. Trends under Labour were much less dramatic. The 90/10 and the 50/10 ratio both hovered at around 4.0 and 2.0, respectively. This suggests little change in the gap between top and bottom incomes, or between middle and bottom incomes. Note, however, that the pattern of income growth by percentile shows a slight narrowing of the gap between the 20th (close to the relative poverty line) and 50th percentiles: this will be important for considering how relative poverty changed under Labour. The rise in the 99/50 ratio continued under Labour (particularly in the late 1990s and the late 2000s). By 2009/10, the gap between the very top and the middle had reached its highest level for at least 50 years. There was then a slight dip in 2010/11. As discussed, the extent to which this just reflects the temporary effects of income forestalling in response to tax changes is highly uncertain. In summary, we saw a small increase in overall measures of income inequality during Labour s period in office. Following on from much larger increases that occurred during the 1980s, income inequality thus remained at historically high levels. This seems to have been driven largely by income trends at the very top: inequality actually declined a little within the middle chunk of the distribution. Labour left office towards the beginning of 2010/11, during which there were large real reductions in household incomes, reflecting the delayed effects of the recession. Those income reductions were also inequality-reducing, with larger falls in income higher up the distribution.

188 Robert Joyce and Luke Sibieta (ii) Drivers of income inequality under Labour Given that Labour s tax and benefit reforms tended to benefit poorer households by proportionately far more than richer ones (Browne and Phillips, 2010), it might seem surprising that overall income inequality did not fall over Labour s tenure. But, as in the period of Conservative government between 1979 and 1997, there were other changes acting to increase the level of inequality in private incomes. Labour s redistributive tax and benefit reforms were mitigating such changes, but were not able to offset them fully. What was driving increases in inequality? Brewer et al. (2009b) decompose changes in income inequality in Great Britain over recent decades and show that much of the overall rise in income inequality during the 1980s was driven by increased earnings inequality between workers, with increased wage gaps between education groups and occupations featuring particularly prominently. A variety of explanations for these kinds of trends have been proposed, both for the UK and elsewhere: skill-biased technological change (Katz and Murphy, 1992; Nickell and Bell, 1995; Katz and Autor, 1999); institutional factors such as the removal of wage/price controls and reduced trade union powers (Goodman and Shephard, 2002); and, at the household level, increased polarization between those that are work-rich and those that are workpoor (Gregg and Wadsworth, 2008). But the extent to which these factors also explain rising pre-transfer inequality under Labour is less clear. Machin (2002) argues that shifts in the relative demand for skilled workers were moderated from the 1990s onwards. There was no further major weakening of the power of trade unions or the scope of collective wage-bargaining. And there was some progress made under Labour in reducing the number of workless households, at least some of which can probably be attributed to welfare reform (see section IV). Crucially, we have already seen that the rise in income inequality under Labour was driven largely by trends in incomes right at the top of the distribution. An explanation of rising pre-tax income inequality under Labour should thus focus mostly on what happened among that relatively small group of people. It is useful to place trends in UK top incomes in an international context. Figure 6 illustrates the changes in the share of income of total income received by the top 1 per cent over time across G-7 countries (excluding Germany). This illustrates the fact that the share of income received by the top 1 per cent in the UK followed a similar pattern to that seen for Canada and the US, generally increasing over time. However, there have been much smaller increases over time in Italy and France. Atkinson et al. (2011) further analyse trends in top incomes over a very long time frame and across a wider range of countries. One of their key findings is that there is a set of countries in which top income shares have broadly followed a U-shaped pattern, with top income shares declining in the immediate post-war period before showing continuous increases from about 1979 onwards. This group includes the UK and a number of other Anglo-Saxon economies such as the USA, Canada, Australia, and New Zealand, as well as Sweden and Norway. They further show that a large part of these increases in top income shares can be accounted for by rising wage income. Other countries, largely those from continental Europe, have seen an L-shaped pattern, with little or no increase in top income shares in recent history. Atkinson et al. (2011) also discuss potential explanations for these trends, focusing on the roles of political economy, macroeconomics, global economic forces, and progressive taxation.

An assessment of Labour s record on income inequality and poverty 189 Figure 6: Top income shares across G7 countries, 1980 2010 Note: Data for Germany are not shown as data are only available for a small number of years. Source: Data downloaded from World Top Incomes Database on 20 December 2012 (http://g-mond.parisschoolofeconomics.eu/topincomes/) Kaplan and Rauh (2010) attempt to distinguish which groups make up the top income groups in the USA over time and the likely explanations for these trends. They conclude that a large and increasing proportion of those with top incomes work in financial corporations. This stands in contrast to Dew-Becker and Gordon (2005), who suggest that the top income group is mostly made up of CEOs, celebrities, and athletes. Kaplan and Rauh (2010) also conclude that the most plausible driving forces behind the growth in top income shares are skill-biased technological change (Katz and Murphy, 1992), greater scale (Gabaix and Landier, 2008), and superstar effects (Rosen, 1981). The latter two explanations are strongly connected, with globalized markets and increased information flows allowing those at the top of their markets or fields to capture even greater returns. Such explanations have also been advanced by Feenberg and Poterba (1993) and Atkinson (2003). These explanations for rising top incomes seem plausible for the UK as well, and would be a very good candidate for further research. Brewer et al. (2008) show that a large proportion of those with the highest incomes in the UK work in the financial sector. Furthermore, they note a strong correlation between the performance of the stock market and the growth in top incomes. This is consummate with analysis by Stewart (2010) of trends in earnings inequality between 1970 and 2010, which finds that a large part of the increase in earnings inequality since the mid-1990s can be attributed to growing inequality within the financial sector and within London. Finally, it is important to note that we have focused on measures of inequality using snapshots of household incomes. To understand fully the implications of these changes for the scale of differences in household living standards, it is also helpful to consider

190 Robert Joyce and Luke Sibieta other measures of inequality, including those based on consumption or long-run measures of income. Blundell and Preston (1998) show that both consumption and income inequality increased during the 1980s, first reflecting mostly increases in permanent income inequality in the early 1980s and then increases in short-term income risks in the late 1980s. Blundell and Etheridge (2010) corroborate this view and show that the increase in income inequality during Labour s first term reflected an increase in permanent income inequality as well as a mild increase in short-term income volatility. Brewer and O Dea (2012) compare the evolution of consumption inequality, income inequality, and inequality in broad income (encompassing imputed income from housing and motor vehicles). Between 1997 and 2009, they show no change in either measure of income inequality, but a statistically significant decline in consumption inequality. Jenkins and Van Kerm (2011) use longitudinal data to follow the same individuals over time, rather than comparing quantiles of repeated cross-sections (the approach taken in this paper). They show that the poorest individuals at the start of Labour s period in office experienced the fastest growth of all over the period. This pattern which was particularly stark between 1998 and 2002, when poverty fell by the most. IV. Income poverty We now analyse trends in low incomes under Labour. In contrast to income inequality across the distribution as a whole, this is an area in which Labour had very clear distributional objectives, or at least a much stronger desire to change the distribution that it inherited. This is evident both from its tax and benefit policies (as we show here) and, in the case of child poverty, its explicitly defined targets: Labour aimed to halve incomebased child poverty measures between 1998/9 and 2010/11. The adoption of numerical child poverty targets was unusual in a European context only Ireland was earlier in adopting numerical poverty targets. The European Union has since adopted a target to reduce the number of people living below national poverty lines by 25 per cent by 2020, with individual targets for member states (EC, 2010). In the rest of this article, we document the trends in income poverty under Labour, before discussing the reasons for the changes observed. (i) What happened to poverty under Labour? We first document the key trends in measures of absolute and relative income poverty under Labour, and present the series back to 1979 for historical context. These measures indicate how the incomes of the poor developed in absolute terms and relative to median income. We focus on the standard measure of relative poverty used in the European Union (and, indeed, by the previous Labour governments), with a poverty line equal to 60 per cent of contemporary median income. To track trends in absolute poverty we fix the poverty line at 60 per cent of the 1998/9 median in real terms. Conceptually, the precise choice of absolute poverty line is arbitrary. We choose this one for its policy relevance, as it was the line used by the government to define its absolute child poverty target for 2010/11.

An assessment of Labour s record on income inequality and poverty 191 Figures 7 and 8 show that poverty among the whole population fell substantially under Labour. Relative poverty fell from about 20 per cent in 1997/8 to about 16 per cent in 2010/11. This ran contrary to the recent historical trend of for the most part rising relative poverty, although it had fallen for 3 consecutive years in the early 1990s. The rapidity of the decline in absolute poverty also compares favourably to recent history, with approximately a halving of the absolute poverty rate since 1997/8 (although it is striking how little of this reduction occurred in Labour s latter years, as discussed below). Both of these facts fit naturally with the analysis presented in the previous section: this showed that income growth under Labour tended to be higher towards the bottom of the income distribution (excluding the extreme tails), contrary to the pattern observed between 1979 and 1997/8; and that real income growth towards the bottom of the distribution was generally higher in absolute terms than it was over that prior period (while lower in the top half). The figures highlight two further points. First, the reductions in income poverty under Labour were not spread evenly throughout its period in office: a large majority of this reduction had already occurred by 2004/5, with relative and absolute poverty each falling by no more than 1 percentage point over the subsequent 6 years. The 3 years between 2004/5 and 2007/8, in particular, stand out from the rest of Labour s tenure as a period when poverty actually rose. As we shall see, in the case of child poverty these 3 years now look critical in accounting for the government s failure to meet its own 2010/11 targets. Figure 7: Relative poverty rates since 1979 Note: Years refer to calendar years up to 1993 and financial years thereafter. Poverty rates are measured as the percentage of the group with income below 60 per cent of the population-wide median. Incomes have been measured before housing costs have been deducted. Source: Authors calculations based on Family Expenditure Survey and Family Resources Survey, various years.

192 Robert Joyce and Luke Sibieta Figure 8: Absolute poverty rates since 1979 Note: Years refer to calendar years up to 1993 and financial years thereafter. Poverty rates are measured as the percentage of the group with income below 60 per cent of the 1997/8 population-wide median. Incomes have been measured before housing costs have been deducted. Source: Authors calculations based on Family Expenditure Survey and Family Resources Survey, various years. Second, there was much variation in low-income trends by demographic group. In particular, the headline trends in poverty both the overall reductions and the slowdown or stalling of those reductions in the mid-2000s were driven by pensioners and families with children. The rates of relative and absolute poverty among children fell by about one-quarter and one-half, respectively, between 1997/8 and 2004/5. In the case of absolute child poverty, this meant that the government had achieved its 2010/11 target level 6 years early. In the case of relative child poverty, it meant that it could have met its 2010/11 target by maintaining approximately the same rate of progress in the period after 2004/5. Rates of pensioner poverty also fell sharply. Absolute pensioner poverty halved between 1997/8 and 2010/11 (again, mostly because of progress up to and including 2004/5). It had also been falling over the preceding 18 years: new retirees typically have higher private incomes than the cohorts that preceded them they have both higher employment rates and higher private pension incomes so there is a tendency for the composition of the pensioner population to shift towards higher-income households over time (Brewer et al., 2007). But relative pensioner poverty also fell substantially, from 25 per cent in 1997/8 to 17 per cent in 2010/11; and the relatively sustained nature of these reductions in relative pensioner poverty, which occurred during times of both boom (the late 1990s/2000s) and bust (the late 2000s), differs markedly from recent history. Previously, low-income pensioners had consistently fallen further behind median income during economic booms and had caught up only when recessions hit. In stark contrast, low-income working-age adults without children fell further behind median income under Labour and their incomes did not change much even in absolute

An assessment of Labour s record on income inequality and poverty 193 terms, continuing the recent historical trend. Absolute poverty among the group fell by just 1.6 percentage points over the 13 years between 1997/8 and 2010/11; and their relative poverty rate rose from about 12 per cent to about 15 per cent over the same period. (ii) Reasons for changes in poverty under Labour What was driving the poverty trends discussed above? Figures 9 and 10 show the results of a counterfactual simulation exercise using TAXBEN, the Institute for Fiscal Studies s tax and benefit micro-simulation model. We hold the 2010/11 population constant but replace the 2010/11 personal tax and benefit system which determined net household incomes in 2010/11, given economic behaviour and demographic characteristics with unreformed systems from each year between 1997/8 and 2009/10. An unreformed system from a particular year is defined as the system as it would have been if nominal tax thresholds and benefit rates from April of that year had simply been uprated until April 2010, using the default indexation rules that Labour inherited (mostly price indexation). This allows us to uncover the direct impacts of direct tax and benefit reforms on income poverty for each year since 1997/8. (The first of Labour s reforms to direct taxes and benefits were in April 1998.) Note that this particular exercise holds behaviour constant (as it was in 2010/11), and so does not account for indirect effects of reforms via behavioural responses (e.g. labour supply changes, or rates of take-up of Figure 9: Effects of direct tax and benefit reforms under Labour on relative poverty rates Notes: Years refer to financial years. Poverty rates are measured as the percentage of the group with income below 60 per cent of the population-wide median. Incomes have been measured before housing costs have been deducted. Source: Authors calculations based on 2010/11 Family Resources Survey and assumptions specified in the text.

194 Robert Joyce and Luke Sibieta Figure 10: Effects of direct tax and benefit reforms under Labour on absolute poverty rates Notes: Years refer to financial years. Poverty rates are measured as the percentage of the group with income below 60 per cent of the 1997/8 population-wide median. Incomes have been measured before housing costs have been deducted. Source: Authors calculations based on 2010/11 Family Resources Survey and assumptions specified in the text. means-tested support 5 ). Evidence on the effects on poverty of labour supply responses to these reforms is discussed separately below. But it is clear from the figures that the direct effects of the reforms alone are extremely powerful in explaining movements in income poverty over the period. The figures highlight that children and pensioners, the major demographic groups which experienced large reductions in poverty risk under Labour, were also the groups clearly favoured by Labour s reforms to the direct tax and benefit system. In fact, without those reforms the simulations suggest that absolute child poverty would have stayed approximately at its 1997/8 level, and that relative child poverty would have continued to increase as low-income households with children would have fallen further behind the median (this is evident by comparing actual poverty levels in 1997/8 with their simulated 2010/11 levels under an unreformed 1997/8 tax and benefit system). Low-income pensioners would have improved their position in absolute terms even without these reforms most likely due to the continuing cohort effects described above but by far 5 The simulations do allow for non-take-up. Those simulated as entitled to a benefit or tax credit in 2010/11 but who did not report receiving it are assumed not to take up that benefit under any other tax and benefit system either. For Family Credit and Working Families Tax Credit, which existed in previous years but not in 2010/11, we have to make assumptions about which eligible people would take up entitlements to them if they still existed in 2010/11. To do this, we randomize non-take-up using official estimates of the takeup rates of these payments, disaggregated by family type, for the last full financial year in which they existed (DWP, 2000; Inland Revenue, 2005).

An assessment of Labour s record on income inequality and poverty 195 less than their position actually did improve; and relative pensioner poverty would have stayed approximately flat. Which specific reforms are important in driving the patterns visible in Figures 9 and 10? Low-income households with children benefitted, in particular, from increases in Income Support and the inception and subsequent expansion of the tax credit system, with the introduction of Working Families Tax Credit (WFTC) in autumn 1999 to replace its smaller-scale forebear Family Credit, and then WFTC s own replacement with the current system of Child Tax Credit and Working Tax Credit in April 2003. By 2010/11, 25 billion per year was spent on tax credits, most of which went to low-income households with children. Low-income pensioners gained, in particular, from the introduction of the Minimum Income Guarantee in 1999 and its subsequent replacement, Pension Credit, in October 2003 (both of which were increased in real terms year-on-year in line with average earnings), and Winter Fuel Payments in 1998. Overall, reforms between 1997 and 2010 amounted to an 18 billion annual increase in spending on benefits for families with children and an 11 billion annual increase in spending on benefits for pensioners (Browne and Phillips, 2010). On an entitlements and liabilities basis, the poorest half of children were about 4,390 (28 per cent) better off in terms of annual net household income in 2010/11 than they would have been under an unreformed 1997/8 direct tax and benefit system; and the corresponding annual gain for the poorest half of pensioners was about 1,970 (19 per cent). (Note that not everyone actually claims what they are entitled to, and this non-take-up is accounted for in the modelling underlying Figures 9 and 10.) The increase in the generosity of state support for low-income families with children and pensioners stands in stark contrast to the lack of priority in this area attached to the childless working-age population. Indeed, direct tax and benefit reforms under Labour had almost no net impact on poverty rates among that group. This is arguably unsurprising given budgetary and political constraints, increases in welfare of the scale of those that benefitted mainly pensioners and families with children could perhaps not realistically have been implemented for everyone. It is striking that the low-income demographic group whose real entitlements to state benefits were not substantially increased under Labour also experienced such little real income growth over the period. And the period in which tax and benefit reforms did little or nothing to reduce income poverty more generally between 2004/5 and 2007/8 was the period in which poverty stopped falling, and, indeed, rose slightly. Both of these facts are evidence that poverty reduction under Labour was heavily dependent on the direct impacts of reforms to taxes and benefits. Tax and benefit reforms were not the only thing contributing to poverty reduction: changing work patterns also played a role. To show this, Table 2 presents a simple but revealing analysis looking in more detail at the fall in child poverty. The table decomposes the overall reduction in relative child poverty into the effects of changes in poverty risk for different groups of children ( incidence effects ), and changes in the relative sizes of those groups ( compositional effects ). Groups are defined according to family type and parental work status. The compositional effects show that changes in work patterns were a poverty-reducing factor. Reductions in the proportion of children living in workless families (a performance indicator that the government was tracking) both in lone parent and couple families acted to reduce relative child poverty by about 2 percentage points. There is good evidence that this was at least partly a result of government policy. In particular, the introduction of WFTC in October 1999 strengthened