THE GLOBAL PATTERN OF HOUSEHOLD WEALTH

Transcription

1 Journal of International Development J. Int. Dev. 21, (2009) Published online in Wiley InterScience ( THE GLOBAL PATTERN OF HOUSEHOLD WEALTH JAMES B. DAVIES 1, SUSANNA SANDSTRÖM 2, ANTHONY SHORROCKS 3 and EDWARD WOLFF 4 * 1 University of Western Ontario, Ontario, Canada 2 UN World Food Programme, Rome, Italy 3 UNU-WIDER, Henskinki, Finland 4 New York University, NY, USA Abstract: We provide the first estimate of the level and distribution of global household wealth. The level of assets and debts for 39 countries are measured using household balance sheet and survey data centred on the year Average wealth holdings are imputed to countries lacking direct evidence. Microdata on household wealth distribution are assembled for 20 countries. Wealth inequality is imputed to other countries from the observed relation between wealth and income distribution in these countries. We find that the top 10 per cent owned 85 per cent of world wealth, and the Gini coefficient was 0.893, greater than that for the global distribution of income. Copyright # 2009 John Wiley & Sons, Ltd. Keywords: wealth; net worth; personal assets; inequality; households; balance sheets; portfolios 1 INTRODUCTION Research on economic inequality both within countries and between countries usually examines differences in income or consumption. In recent years a number of studies have extended this work to the global stage, by attempting to estimate the world distribution of income (e.g. Bourguignon and Morrison, 2002; Milanovic, 2002, 2005). The findings document the very high disparity of living standards amongst the world s citizens. In this paper we round out the picture by reporting on the first estimates of the world distribution of household net worth or wealth (Davies, 2008). 1 *Correspondence to: Edward Wolff, Department of Economics, New York University, 19 West Fourth Street 6 th Floor, New York , USA. Edward.wolff@nyu.edu 1 Wealth equals the value of a household s assets minus its debts. Assets can be physical or financial. Human capital and state pension rights are excluded. Copyright # 2009 John Wiley & Sons, Ltd.

2 1112 J. B. Davies et al. Household wealth is important for many reasons. First, it provides a means of raising long-term consumption, either through dissaving, or via the income stream of returns on assets. Second, it can enable consumption smoothing, providing for retirement and reducing vulnerability to ill health, unemployment or natural disaster. Third, it may provide business finance, either directly or as collateral for loans. These roles are less important in countries that have good state pensions, social safety nets and strong financial institutions. By the same token, lack of wealth has more significance in countries that do not have these features, as in much of the developing world. Despite these reasons for interest in wealth, data limitations have slowed research on the topic in the past. However, the situation has improved in recent years. Many OECD countries now have wealth data from household surveys, tax records or national balance sheets. Household wealth surveys have also been conducted in the three largest developing countries China, India and Indonesia. Holdings of the super rich are reported annually by Forbes magazine and other media outlets. There is therefore sufficient data to support preliminary estimates of the distribution of household wealth across the world, which we provide for the year The remainder of the paper is organised as follows. The next section summarises the sources and methods used in our study. (These are described in more detail in Davies et al., 2007). Section 3 discusses results for the estimated world distribution of wealth. Likely future trends in wealth-holding and wealth distribution are discussed in Section 4. Conclusions are drawn in Section 5. 2 SOURCES AND METHODS 2.1 Wealth Levels The estimation of wealth levels is based on national household balance sheets and sample surveys. Household balance sheets are often compiled in conjunction with the National Accounts or Flow of Funds data. They enable us to construct complete financial and nonfinancial data for 19 countries and financial data for 15 countries, where complete is interpreted as full or almost full coverage of financial assets and inclusion of owneroccupied housing, at a minimum. The country coverage of household balance sheets is not representative. 2 While Europe and North America, and the OECD in general are well covered, low- and middle-income countries are under-represented. Coverage is sparse in Africa, Asia, Latin America and the Caribbean. Fortunately for this study, these gaps were offset to an important extent by the availability of survey evidence for the three largest developing countries China, India and Indonesia. Altogether we made use of full or partial data on wealth levels for 38 countries. 3 They accounted for 56 per cent of world population in the year 2000 and, we estimate, more than 80 per cent of global household wealth. Regressions run on these 38 countries allowed wealth levels to be estimated for other countries (see Davies et al., 2007 for details). 2 See Appendix in Davies et al. (2007). 3 In the case where both survey and balance sheet data were available, as in the US, we used the survey data. Comparisons indicate that the two sources are relatively consistent. Besides the 20 countries listed in Table 1, the others are the Netherlands, Portugal, Taiwan, Singapore, Czech Republic, Poland, South Africa, Austria, Belgium, Greece, Slovenia, Croatia, Estonia, Hungary, Latvia, Lithuania, Slovakia and Turkey.

3 The Global Pattern of Household Wealth Shape of Wealth Distributions A complete picture of wealth holdings within a country requires information on the shape of the distribution as well as the level. Twenty countries have reasonably reliable estimates of wealth distribution (see Table 1). The list includes the largest countries, both rich and poor the United States, Japan, Germany, Britain, France and Italy on one hand, and China, India and Indonesia on the other. Scandinavia and Englishspeaking Commonwealth countries (Australia, Canada, and New Zealand) are also well represented. Table 1 shows that the data are most often for households or families, but can also refer to individuals or adults. Although there are gaps, we usually have decile wealth shares plus the share of the top 5 and 1 per cent. The share of the top 10 per cent is reported for all 20 countries and ranges from 39.3 per cent in Japan to 71.3 per cent in Switzerland. High wealth concentration is even more evident in the share of the top 1 per cent. Amongst the 11 countries reporting values, the share of the top 1 per cent averages 22.3 per cent and climbs to 32.7 per cent in the USA and 34.8 per cent in Switzerland. As evident from Table 1, the available sources provide a patchwork of quantile shares. In order to move towards an estimate of the world distribution of wealth, more complete and comparable information is needed on the distribution in each country. To achieve this, missing cell values were imputed using a programme developed at UNU-WIDER which constructs a synthetic sample of 1000 observations that conforms exactly with any valid set of quantile shares derived from a distribution of positive values (e.g. incomes). The Table 1. Wealth shares for countries with wealth distribution data Share of top Country Year Unit 10% 5% 1% Australia 2002 Household Canada 1999 Family unit 53.0 China 2002 Person 41.4 Denmark 1996 Family unit Finland 1998 Household 42.3 France 1994 Person Germany 1998 Household 44.4 India Household Indonesia 1997 Household Ireland 1987 Household Italy 2000 Household Japan 1999 Household 39.3 Korea, South 1988 Household New Zealand 2001 Tax unit 51.7 Norway 2000 Household 50.5 Spain 2002 Household Sweden 2002 Household 58.6 Switzerland 1997 Family United Kingdom 2000 Adult USA 2001 Family Source: Davies et al. (2007).

4 1114 J. B. Davies et al. 20 countries for which wealth distribution data are available include China and India, and hence cover a good proportion of the world population. They also include most of the large rich countries, and hence cover much of global wealth. However, the fact that the list is dominated by OECD members cautions against extrapolating immediately to the rest of the world. For most countries lacking direct wealth distribution data, the pattern of wealth distribution was estimated using income distribution data recorded in the WIDER WIID dataset, on the grounds that wealth inequality is likely to be correlated possibly highly correlated with income inequality across countries. The WIID dataset covers 144 countries and has multiple observations for most of them. 4 Where possible, data was chosen for household income per capita across individuals for a year close to 2000, with first priority given to figures on disposable income, then consumption or expenditure. Eighty-five per cent of the income distributions conform to these criteria. Figures for gross incomes added a further seven per cent, leaving a residual eight per cent of countries for which the choices were very limited. The ungrouping programme was then used to generate quantile shares for income according to the same template employed for wealth distribution. The common template applied to the wealth and income distributions allows Lorenz curve comparisons for each of the 20 reference countries listed in Table 1. The ratios of wealth shares to income shares at various percentile points are fairly stable across countries, supporting the view that income inequality is a good proxy for wealth inequality when wealth distribution data are not available. Thus, as a first approximation, the ratio of the Lorenz ordinates for wealth compared to income is taken to be constant across countries. These constant ratios (14 in total) correspond to the average value recorded for the 20 reference countries. 5 This generates estimates of wealth distribution for 124 countries to add to the 20 original countries which have direct evidence of wealth inequality. Because the imputations are based on only 20 reference countries, the results have to be interpreted with some caution 6 We believe it is best to disregard ownership of wealth by minors (specifically, those aged below 20 years) and to interpret the wealth distribution figures in terms of the distribution across adults Another question concerns the appropriate conversion rate for currencies. Studies of income or consumption usually use PPP (purchasing power parity) exchange rates to compensate for price variations across countries. Here, we focus on results based on official exchange rates on the grounds that wealth is heavily concentrated in the hands of the rich, whose investments will often be made in world markets and at world prices. 7 4 There are, of course, well known problems of comparability in international comparisons of income distribution unit of observation, income versus consumption data, sampling frame, etc. (see, for example, Milanovic, 2005). 5 To circumvent aggregation problems, the adjustment ratio was applied to the cumulated income shares (i.e. Lorenz values) rather than separate quantile income shares. 6 As a sensitivity test, we excluded all countries except the 20 nations listed in Table 1 which have wealth distribution data. Restricting attention to these 20 countries loses 25 per cent of the world s wealth and 41 per cent of the world s adults. Nevertheless, estimated shares of the top one per cent differ by only 0.1 per cent between the full sample of countries and the restricted sample of 20 countries and the Gini coefficients by only (about one half of a per cent). 7 Alternative estimates using PPP rates are reported in Davies et al. (2007).

5 3 THE GLOBAL DISTRIBUTION OF WEALTH 3.1 Wealth Inequality Table 2 summarises our estimated distribution of wealth across the world population of 3.7 billion adults, based on official exchange rates and figures for the year According to our estimates, adults required just $2100 in order to be among the wealthiest half. But more than $ was needed to belong to the top 10 per cent and more than $ per adult was required for membership of the top 1 per cent. The richest 2 per cent of adult individuals owned more than half of global wealth (not shown in Table), with the richest 1 per cent alone having 40 per cent. The corresponding figures for the top 5 and 10 per cent were 70.7 and 85.2 per cent, respectively. In contrast, the bottom half of wealth holders together held barely 1 per cent of global wealth. Members of the top decile were almost 400 times richer, on average, than the bottom 50 per cent, and members of the top percentile were almost 2000 times richer. Additional information on wealth inequality is provided in Table 3, which reports the world Gini coefficient as well as values by country for selected countries. The final column of Table 3 records wealth Gini estimates ranging from for Japan to for the USA and for Switzerland. The global wealth Gini is estimated to be even greater, at By way of comparison, Milanovic estimates the world Gini for income to be in 1998 using official exchange rates (Milanovic, 2005, p. 108). 3.2 Geographic Distribution of Wealth The Global Pattern of Household Wealth 1115 The world map in Figure 1 shows per capita wealth level by country. Western Europe, North America (Canada and US only), and the rich Asian-Pacific nations (principally Japan, South Korea, Taiwan, Australia and New Zealand) stand out as the richest areas, with per capita wealth exceeding $ Next come some prosperous developing and transition countries for example Mexico, Chile, Argentina, Poland, the Czech Republic and the Ukraine in the $ to $ band. Russia and China fall in the $2000 to $ range along with Turkey, Brazil, Egypt, Thailand and South Africa. Finally, below $2000 are found India, Pakistan, Indonesia and most of Central and West Africa. Regional wealth shares are interesting (see the last column of Table 2). North America has about a third (34 per cent) of the world s wealth. Europe has a fraction less (30 per cent) and Rich Asia-Pacific is close behind at 24 per cent. The rest of the world shares the remaining 12 per cent. Figure 2 shows how these wealth shares compare to population shares. North America has the largest excess of wealth over its fair share according to population, which is a mere 5 per cent. Europe has more than double the population of North America, so that its large wealth share is more aligned with its population. Figure 3 compares the asset composition of wealth across a selection of countries. In the USA, 42 per cent of gross household assets are in financial form. Among other countries with data, this high ratio is approached only by the UK. Japan, Canada and Germany have a considerably lower share of financial assets averaging just 28 per cent. Interestingly, estimated financial assets are 22 per cent of the total in China, but just 5 per cent and 3 per cent in India and Indonesia, respectively. Like Japan and several other East Asian countries before it, China has experienced high saving and growth, producing a different wealth composition than found in low-income developing countries. Household assets in the latter are heavily weighted towards land, livestock, and other

6 1116 J. B. Davies et al. Table 2. Global wealth distribution in 2000: regional details based on official exchange rates Decile Top Population share (%) % 5% 1% Wealth per adult (US$) Wealth Share (%) World wealth shares (%) Minimum wealth (US$) Percentage of adults by region North America Latin America and Caribbean Europe Africa China India Rich Asia-Pacific Other Asia-Pacific World Source: Davies et al. (2007).

7 The Global Pattern of Household Wealth 1117 Table 3. Global wealth distribution in 2000: country details based on official exchange rates Percentage of adults by country Share of Top 10% 5% 1% Adult population share (%) Wealth per adult (US$) Wealth share (%) Gini USA Japan Germany Italy UK France Spain Canada Australia Netherlands South Korea Brazil Mexico Switzerland China India Russia Indonesia World Source: Davies et al. (2007). agricultural assets. Financial development lags, and non-financial assets dominate the balance sheet. Figure 3 also suggests that debt is higher in the developed world, according to official data. However, it is possible that debts are especially under-reported in LDC sample surveys. Subramanian and Jayaraj (2008), for example, estimate that the true indebtedness of Indian households is about three times greater than reported. If so, total debt in India would be about 10 per cent of gross assets, similar to the level reported in the USA and Japan. Figure 1. World wealth levels, year 2000

8 1118 J. B. Davies et al. Figure 2. Population and wealth shares by region Asset composition differs not only across countries but also within (Jantti and Sierminska, 2008). The share of financial assets rises, and debt declines, as household wealth increases. Stocks and bonds are concentrated in the upper tail. The picture for real assets is more complex. Housing is a smaller fraction of wealth in both the upper and lower tails. In developing countries, both producer and consumer durables generally decline in importance as wealth rises. Land, on the other hand, rises in importance with wealth in poor countries, and is usually very concentrated. 8 Figure 3. Asset composition in selected countries 8 Exceptions are provided in some transition countries (e.g. China, Vietnam) and a small number of countries that have had successful land reforms, for example Japan and South Korea.

9 The Global Pattern of Household Wealth 1119 Figure 4. Regional composition of global wealth distribution Figure 4 charts the regional composition of the various global deciles. India dominates the bottom third of the global wealth distribution, contributing 27 per cent of the bottom three deciles. The middle third is the domain of China, which supplies more than a third of those in deciles 4 8, reflecting not only higher average wealth in China than India but also more equal distribution. North America, Europe and Rich Asia-Pacific monopolise the top decile, each accounting for roughly one third of this group. Another notable feature is the relatively constant membership share of Asian countries excluding China and India. This group is polarised, with a high-income subgroup at the top end and the lower income countries (especially, Indonesia, Bangladesh, Pakistan and Vietnam) occupying the lower wealth tail. The population of Latin America is fairly evenly spread across the global distribution but Africa is heavily concentrated at the bottom. Table 3 provides more details for major countries. A country s representation in the top global decile depends on three factors: its population, average wealth and wealth inequality. The US appears in first position, with 25 per cent of the global top decile (see Figure 5) and 37 per cent of the top percentile. All three factors reinforce each other in this instance: a large population combining with high wealth per capita and relatively unequal distribution. Japan features strongly in second place more strongly than anticipated, perhaps with 21 per cent of the global top decile and 27 per cent of the top percentile. The high wealth per adult and relatively equal distribution accounts for the insignificant number of Japanese in the bottom half of the global wealth distribution. Italy, too, has a stronger showing than expected, for similar reasons. Further down, China and India both owe their position to their large population. However, neither country has enough people in the global top 5 per cent in 2000 to be recorded in Table 3. While the two countries are expected to be under-represented in the upper tail because of their relatively low mean wealth, their absence from the top 5 per cent seems anomalous. It may reflect relative unreliability of their survey data in the upper tail.

10 1120 J. B. Davies et al. Figure 5. Percentage membership of wealthiest 10 per cent According to Forbes magazine there is a rising number of billionaires in both China and India, suggesting that better data might reveal more penetration of the top global wealth percentiles. 9 4 TRENDS OVER TIME We are reporting on the first study of the world distribution of wealth. Since our estimates are a snapshot for a single year, no time series exists. However, estimates of wealth inequality over time are available for several countries, and comments can be made concerning changes over time on international differences in wealth levels. It is interesting to look at these pieces of the puzzle, although somewhat hazardous to draw conclusions about the trend in global wealth inequality on the basis of the limited evidence. Long time series on wealth inequality are available for Denmark, France, Norway, Sweden, Switzerland, the US and the UK (Ohlsson et al., 2008). From the early years of the 20th century up to the mid 1970s wealth inequality declined except in Switzerland. This parallels the decline of income inequality over the same period. In contrast, wealth and income inequality behaved differently during the last three decades. Increases in income inequality have been observed in most OECD countries. While the wealth share of the top 1 per cent also increased in most countries during this period (Ohlsson et al., 2008), the rise in inequality appears to have been weaker for wealth than for income. For example, in the US while there was a mild increase in wealth concentration after the mid-1970s, 9 Ten years ago the Forbes list contained no billionaires from China. In 2007 there were 16. As late as 2004, only 9 billionaires were reported in India. This number had risen to 36 by 2007.

11 The Global Pattern of Household Wealth 1121 measured concentration fell after the late 1990s and the share of the top 1 per cent in 2001, at 33.4 per cent according to the SCF, did not differ much from the share of 33.8 per cent in Part of the explanation for the weaker increase in inequality of wealth than of income is suggested by the findings of Piketty and Saez who show that the rise in top income shares in the US has been due mostly to increased earnings dispersion. In other words, increased executive compensation and the like, rather than investment returns, is driving higher income inequality (Piketty and Saez, 2003). This is consistent with flat or slowly rising wealth inequality during a time of strongly increasing income inequality. Large increases in house prices may provide a further explanation. Since housing is relatively more important for the middle class than for the poor or the rich, increases in house prices tend to reduce top wealth shares, but have an ambiguous effect on comprehensive measures of inequality like the Gini coefficient. There is also some evidence regarding between-country trends for the former G7 countries. In 1994 the ratio of wealth to disposable income ranged from 4.72 for Canada to 7.47 for Japan. From 1994 to 1997 unweighted dispersion fell for these countries, as the wealth-income ratio declined for Japan but rose for the other countries. After 1997, dispersion rose due to strong increases in wealth in France, Italy and the UK, mostly associated with rising property prices. As a result, this group of countries showed about the same dispersion in the wealth income ratio in 2004 as in Among developing countries, only China and India offer comparisons over time. Official data indicate no trend in wealth inequality in India, where results from a large asset and debt survey are available at decennial intervals since (Subramanian and Jayaraj, 2008). On the other hand, both income and wealth inequality have been rising at a strong pace in China. Between 1995 and 2002 the wealth Gini rose from 0.40 to As noted earlier, the number of Chinese billionaires on the Forbes list has also been rising in the last few years. The disequalising effect on world wealth distribution is offset; however, by the rise in mean wealth in China, which reduces between-country inequality. Hence the net impact of trends in China on global wealth inequality is unclear. Russia and the European transition countries illustrate the link between rising wealth inequality and the shift from limited personal property under socialism to a market system (Guriev and Rachinsky, 2008). However, the increase in wealth inequality in Central and Eastern Europe has been less extreme than in Russia. In these countries, mean wealth, which started from a low level, may have been rising fast enough to offset the impact of their increasing wealth dispersion on global inequality. This cancellation has not taken place for Russia; however, since its increase in wealth inequality has been extreme and its growth performance has been poor. While it is difficult to predict future trends, a few observations may be offered. First, as in the past, growth in GDP will affect both overall global wealth and the distribution across countries. However, growth in wealth may not exactly match GDP growth. Aggregate wealth depends on asset prices, especially real estate and equity values, and therefore declined in the world financial/economic crisis of In the medium and long run one can expect a rebound of wealth relative to GDP. This trend may be more pronounced in those countries where the emphasis on private provision for retirement, education and health care widely seen in recent years remains strong. There will also be sensitivity to any further changes in property rights, such as privatisation and property registration schemes. 10 The unweighted coefficient of variation in 2004 was 0.207, compared with.203 in 1994.

12 1122 J. B. Davies et al. Changes in exchange rates are another factor to consider. Exchange rate movements have little impact on global income inequality measured in PPP dollars, since the PPP currency conversions sterilise most of the change. If estimates of global wealth distribution employ official exchange rates, the impact could in principle be large. In practice, while country rankings have been affected, we do not believe the impact on global inequality has been large since the year The impact is reduced, first, by some significant exchange rate rigidities. For example, the US dollar value of the Chinese yuan only changes slowly, and there are no changes among countries using the euro. Second, while inequality between Europe, the UK, Japan and the US changes with exchange rates, this inequality is a small component of global inequality. How global wealth inequality will change also depends on wealth inequality trends within countries, on the level of wealth inequality in the faster growing countries, and on population. These aspects suggest a crucial role for China. Strong economic growth coupled with an expansion in private property provide the foundation for a significant rise in average wealth whose impact is reinforced by population, but constrained by the managed currency peg to the US dollar. As Figure 4 makes clear, China is poised to make big inroads into the echelons of top wealth holders. The relative equality of wealth holdings in China means that even a modest rise in average wealth relative to the rest of the world will promote many into the top global wealth decile, and, given time, into the top global percentile. Although India has a similar sized population, it is unlikely that Indian nationals will rapidly occupy many of the global top wealth slots for two reasons. First, the recent growth experience has not matched that of China. Secondly, wealth inequality is much greater, so there are significantly fewer wealth holders who can expect to be promoted into the global top wealth decile. The contrast is captured by the thin right tail of India in Figure 4 compared with the fat pattern of China in the middle deciles. Russia is another country whose super rich have made headlines. However, it is unlikely that many Russians will be in evidence among the wealth elite of the world in, say, 20 years time, at least compared to Chinese. The much smaller (and shrinking) population and the higher concentration of wealth are the two principal factors limiting the expansion of Russian membership in the global top wealth decile. 5 CONCLUSIONS This paper has reported a first estimate of the world distribution of household wealth. The distribution is highly concentrated much more concentrated than the world distribution of income, or the distribution of wealth within all but a few countries. While the share of the top 10 per cent of wealth-holders within a country is typically about 50 per cent, and the median Gini value around 0.7, our figures for the year 2000 using official exchange rates suggest that for the world as a whole the share of the top 10 per cent was 85 per cent and the Gini By comparison, Milanovic estimates that the world income Gini was in 1998, and the share of the top 10 per cent was 67 per cent (Milanovic, 2005). Much of the data used in this study come from household surveys. This is not a big problem for the USA, which supplies 25 per cent of the world s top 10 per cent of wealthholders: sophisticated techniques have been used by the Federal Reserve Board to ensure the reliability of its triennial Survey of Consumer Finance. Attempts along similar lines have been made in some other wealthy countries. In contrast, surveys in the major developing countries appear to have difficulties capturing the upper tail. Thus, while we

13 The Global Pattern of Household Wealth 1123 have reasonable confidence in our estimates, a non-negligible error bound is attributable to the limitations of household surveys. The quality of our results also depends on other sources of data and on the procedures employed to estimate wealth levels and inequality within countries. Household sector financial balance sheet data exist for 38 countries, covering 56 per cent of the world s population. Estimates of non-financial assets also exist for 22 of these countries, and are often constructed in conjunction with Flow of Funds data or the National Accounts. This generates some confidence in the basic sources. One of the most fascinating aspects of our results is the light they throw on the geographic distribution of world wealth. In the year 2000 about 34 per cent of the world s wealth was held in the US and Canada, 30 per cent in Europe and 24 per cent in the rich Asia-Pacific countries. Africa, Latin America and other Asia-Pacific countries shared the remaining 12 per cent. The location of top wealth-holders is even more concentrated, with North America hosting 39 per cent of the top global 1 per cent, and Europe and rich Asia- Pacific having 26 per cent and 32 per cent, respectively. Information on the geographic distribution of wealth holders produces some conjectures about possible future global trends. For example, if the rapid growth observed in China and India continues it will likely have different consequences for the two countries representation in different parts of the global distribution. With its large current representation in the middle wealth deciles, China is poised to contribute a greatly increased number of people to the top deciles. On the other hand, India has a relatively small number of people in the middle deciles compared with China, so the consequence of continued growth may be that Indians supplant the Chinese as the largest group in the middle wealth range. If current trends continue, the bottom deciles in the world wealth distribution may be increasingly dominated by Africa, Latin America, and low-income Asia-Pacific countries. While European transition countries are currently found among the bottom deciles, their increasing integration into Europe and growth suggest the likelihood of an upward movement. The success of so many people in rapidly growing Asian countries is very positive, but continued low wealth in Africa, Latin America, and low-income Asian-Pacific countries is a concern. From a global perspective, the level of wealth in these areas is relatively lower than income. This points to a serious problem, since these are the countries where having sufficient household wealth seems the most crucial in reducing vulnerability and improving opportunity, both at household and enterprise levels. REFERENCES Bourguignon F, Morrison C Inequality among world citizens: American Economic Review 92: Davies JB. (ed.) Personal Assets from a Global Perspective. Oxford University Press: Oxford. Davies JB, Sandstrom S, Shorrocks A, Wolff E Estimating the Level and Distribution of Global Household Wealth. UNU-WIDER Research Paper 2007/77. Guriev S, Rachinsky A The Evolution of Personal Wealth in the Former Soviet Union and Central and Eastern Europe. In Personal Assets from a Global Perspective, Davies JB (ed). Oxford University Press: Oxford; Jäntti M, Sierminska E Survey Estimates of Wealth Holdings in OECD Countries. In Personal Assets from a Global Perspective, Davies JB (ed). Oxford University Press: Oxford;

14 1124 J. B. Davies et al. Milanovic B True world income distribution, 1988 and 1993: first calculation based on household surveys alone. Economic Journal 112: Milanovic B Worlds Apart, Measuring International and Global Inequality. Princeton University Press: Princeton. Ohlsson H, Roine J, Waldenstrom D Long-Run Changes in the Concentration of Wealth. In Personal Assets from a Global Perspective, Davies JB (ed). Oxford University Press: Oxford; Piketty T, Saez E Income inequality in the United States, Quarterly Journal of Economics 118: Subramanian S, Jayaraj D The Distribution of Household Wealth in India. In Personal Assets from a Global Perspective, Davies JB (ed). Oxford University Press: Oxford;