The World Distribution of Household Wealth

James B. Davies*, Susanna Sandstrom†, Anthony Shorrocks†, and Edward N. Wolff‡

5 December 2006

* Department of Economics University of Western Ontario London, Canada N6A 5C2

UNU-WIDER

Katajanokanlaituri 6 B 00160 Helsinki, Finland

Department of Economics

269 Mercer Street, Room 700 New York University New York, NY 10003 USA

We thank participants at the May 2006 WIDER project meeting on Personal Assets from a Global Perspective for their valuable comments and suggestions. Special thanks are due to Tony Atkinson and Markus Jäntti. Responsibility for all errors or omissions is our own.

1

1

Introduction

Much attention has recently been focused on estimates of the world distribution of income (Bourguignon and Morrison, 2002; Milanovic, 2002 and 2005). The research shows that the global distribution of income is very unequal and the inequality has not been falling over time. In some regions poverty and income inequality have become much worse. Interest naturally turns to the question of global inequality in other dimensions of economic status, resources or wellbeing. One of the most important of these measures is household wealth.

In everyday conversation the term ‘wealth’ often signifies little more than ‘money income’. On other occasions economists interpret the term broadly and define wealth to be the value of all household resources, both human and non-human, over which people have command. Here, the term is used in its long-established sense of net worth: the value of physical and financial assets less liabilities.1 Wealth in this sense represents the ownership of capital. While only one part of personal resources, capital is widely believed to have a disproportionate impact on household wellbeing and economic success, and more broadly on economic development and growth.

Wealth has been studied carefully at the national level since the late nineteenth or early twentieth centuries in a small number of countries, for example Sweden, the UK and the US. In some other countries, for example Canada, it has been studied systematically since the 1950s. And in recent years the number of countries with wealth data has been increasing fairly quickly. The largest and most prosperous OECD countries all have wealth data based on household surveys, tax data or national balance sheets. Repeated wealth surveys are today also available for the two largest developing countries — China and India, and a survey that inquired about wealth is available for Indonesia. Forbes magazine enumerates the world’s US$ billionaires and their holdings. More detailed lists are provided regionally by other publications, and Merrill-Lynch estimate the number and holdings of US$ millionaires around the world. National wealth has been estimated

1 In some work attempts have been made to include ‘social security wealth’, that is the present value of expected net benefits from public pension plans in household wealth. We exclude social security wealth here, in part because estimates are available for only a few countries.

2

for a large number of countries by the World Bank.2 In short, there is now an impressive amount of information on wealth holdings. We believe the time has therefore come to estimate the world distribution of household wealth.3

In this paper we show, first, that there are very large intra-country differences in the level of household wealth. The US is the richest country, with mean wealth estimated at $144,000 per person in the year 2000.4 At the opposite extreme among countries with wealth data, we have India with per capita wealth of about $6,500 in purchasing power parity (PPP) terms. Other countries show a wide range of values. Even among high income OECD countries there is a range from figures of $56,000 for New Zealand and $66,000 for Denmark to $129,000 for the UK (again in PPP terms).

We also look at international differences in the composition of wealth. There are some regularities but also country-specific differences — such as the strong preference for liquid savings in a few countries, such as Japan. Real assets, particularly land and farm assets, are more important in less developed countries. This reflects not only the greater importance of agriculture, but a lack of financial development that is being corrected in some of the most rapidly growing LDCs. Among rich countries, financial assets and share-holding tend to bulk largest in those with more reliance on private pensions and/or the most highly developed financial markets, such as the UK and US

The concentration of wealth within countries is high. Typical Gini coefficients in wealth data lie in the range of about 0.65 - 0.75, while some range above 0.8. In contrast, the mid-range for

2 See World Bank (2005). National wealth differs from household wealth in including the wealth of all other sectors, of which corporations, government and the rest-of-the-world are important examples. 3 One sign of the growing maturity of household wealth data is the launching of the Luxembourg Wealth Study (LWS) parallel to the long-running Luxembourg Income Study (LIS). See http://www.lisproject.org/lws.htm. In its first phase the LWS aims to provide comparable wealth data for nine OECD countries, with the cooperation of national statistical agencies or central banks. The LWS initiative differs from ours in that its aim is not to estimate the world distribution of wealth, but to assemble fully comparable wealth data across an important subset of the world's countries. 4 All our wealth estimates are for the year 2000. Wealth data typically become available with a significant lag, and wealth surveys are conducted at intervals of three or more years. The year 2000 provides us with a reasonably recent date and good data availability.

3

35 – 0. Conclusions are drawn in Section 6. ‘complete’ financial and non-financial data are available for 18 countries. Section 4 reviews evidence on the distribution of wealth where available. again higher than is usual for income.45. except for the Czech Republic. Sources and Methods for HBS data are detailed in Appendix I. the countries for which we have data included 56 per cent of the world’s population in the year 2000 and. and imputes household wealth totals to the ‘missing countries’. we estimate. The mid value for the share of the top 10 per cent of wealth-holders in our data is 50 per cent. and sources of survey data are shown in Appendix III. In Section 5 levels and distributions are combined to construct the global distribution of household wealth. and South Africa. Fortunately for our exercise. In order to do so we need estimates of the levels and distribution of wealth in countries where data on wealth are not available. our goal is to estimate the world distribution of wealth. and then performs imputations for other countries. In the next section we study what can be learned about household wealth levels and composition across countries using household balance sheet and survey data. more than 80 per cent of its household wealth. 2 Wealth Levels Our objective in this section is to develop data on measured wealth levels in as many countries as possible. The remainder of the paper is organized as follows. A Household Balance Sheet (HBS) Data We have assembled balance sheets for as many countries as possible. These are all high income countries. but are also used in the next section to impute per capita wealth to a large sample of countries that do not have wealth data. As indicated in Table 1. which are upper 4 . Poland. Section 3 presents our empirical results on the determinants of wealth levels. The exercise begins by taking inventories of household balance sheet (HBS) and sample survey estimates of household wealth levels and composition. These data are of independent interest. While inter-country differences are interesting.income Ginis is from about 0. Careful study of the determinants of wealth levels and distribution in the countries that have wealth data allows imputations to be made for the ‘missing countries’.

according to the World Bank classification.5 We considered the data complete if there was full or almost full coverage of financial assets. and inclusion at least of owner-occupied housing on the non-financial side. these gaps in HBS data are offset to an important extent by the availability of survey evidence for the largest developing countries. The German and Italian data are to a large extent also based on central bank data but are more complete. but there are several exceptions. two developing countries (India and Mexico). There are 15 other countries that have comparable financial balance sheets. This produces a total of 18 countries. the balance sheets are compiled in conjunction with the National Accounts or Flow of Funds data. Fortunately for this study. He was able to include 15 developed market economies. data are reported by central banks and include estimates based on Financial Accounts augmented with data on housing assets.6 In geographic terms this means that coverage is sparse in Africa. with six upper middle income countries and three lower middle income. India and Indonesia. Often. The German data are based on Financial Accounts data from Deutsche Bundesbank and non-financial assets data including 5 We used the World Bank classification throughout the paper except that Brazil. equal in number to the countries for which we have complete HBS data for the year 2000. 5 . as discussed below. This is especially true for non-financial assets. methods and sources differ across countries. Russia and South Africa were moved from the lower middle income category to higher middle income. are well covered.middle income countries. and Equatorial Guinea from low to lower middle income. Also note that while there are no HBS data for Russia. Here there is better representation outside the high income countries. China. Latin America and the Caribbean. but low income and transition countries are not. Europe and North America. Asia. we do have complete HBS data for two European transition countries and financial data for eight others. As discussed in Appendix I. For countries such as New Zealand. Portugal and Spain. We made these changes since the WB classifications seemed anomalous on the basis of the Penn World Table GDP data that we use for year 2000. Table 2 summarizes key characteristics of the household balance sheet data by country. but no information on the real side. Such data tend to be developed at a relatively late stage in the development of national economic statistics. 6 Goldsmith (1985) prepared ‘planetary’ balance sheets for 1950 and 1978 and found similar difficulties in obtaining representative coverage. Country coverage in HBS data is not representative of the world as a whole. and the OECD in general. and the Soviet Union.

the US. and the UK. To the best of our knowledge. in all other cases both land and all real estate owned by households are included in the data specification. For Denmark we combined financial balance sheet data with fixed capital stock accounts reported by Statistics Denmark.housing assets. but not other real estate). The Italian data are based on the Financial Accounts from the Bank of Italy augmented with Italian statistical office (Istat) estimates of the stock of dwellings and calculations of durable goods based on Istat data by Brandolini et. we believe that while gaps for these items do have some effect on our results. and for Italy (for which all housing is included. Investment or commercial real estate is missing for the Netherlands. the mean ratio for Canada. For European countries other than the UK we used the mean ratio for Germany and Italy. the Netherlands. it is possible to do a reasonable imputation for consumer durables. Fifteen countries report data on some other real property. and six have estimates for consumer durables. 6 . Germany. The data for the Netherlands are a mix of data from Statistics Netherlands and the central bank. New Zealand. we included these imputations. The nonfinancial data from the Singapore Department of Statistics also covers only housing assets. Finally. We used the mean ratio of durables to GDP in Canada and the US to impute durables to Australia. Poland. Italy. Portugal and Singapore. Since only three countries are completely missing these items and eight countries. 8 Durables data are available for Canada. the impact should not be exaggerated. including the US have complete data. other real assets and durables. Singapore and the Netherlands are at this minimum level. New Zealand. Germany and Italy was used to do the imputation for Japan and Singapore.7 We have therefore not imputed these items. and Singapore. We found that it would be very difficult to do a satisfactory imputation for land or investment real estate. We considered whether we could make imputations that would make the non-financial coverage in these ‘complete’ balance sheets more uniform. al (2004). including land and/or investment real estate in most cases. In contrast. and the financial balance sheets are only augmented with data on owner-occupied housing. other land is missing for Denmark. Germany. whether owner occupied or not. Even if the household balance sheets are based on data from the national statistical organizations they do not necessarily have a broad coverage of non-financial assets. Since this improves estimates for twelve countries. Italy and South Africa. each of the 18 countries we have classed as having complete balance sheets have good financial data plus some information on housing. Summarizing.8 7 While the value of land occupied by included dwellings is captured in the balance sheets. the US.

Poland and Spain. In Japan and most of the European transition countries. Housing assets constitute a considerable share of non-financial assets. we can draw on the 15 countries for which we have financial balance sheets only. In the United Kingdom and some other countries.. shares and equities. The high share of financial assets makes South Africa stand out. asset coverage and methodological differences. We aimed for a household sector which covered the assets and debts of households and unincorporated business. with higher shares in the Czech Republic. However. liquid 7 . and had data allowing us to do so for the UK and US. One needs. New Zealand. There are some striking differences across countries. We would like to exclude NPOs. which is clearly related to the strength of its markets. We disaggregate into liquid assets. 2003). but may also be partly explained by its relatively cheap housing. 2005). the share of non-financial assets is unusually low in South Africa (see Aron et al. Turning to the composition of financial assets. land. the large increase in real estate prices in the late 1990’s helps to explain a high housing share. however. Table 3 reports the asset composition of household balance sheets.Table 2 also shows that there are differences in sectoral definition across countries. but due to well developed financial markets combined with continuously negative rates of return on investment in fixed property and relatively high mortgage interest rates. One would expect real assets like housing. 2006). The United States is also an outlier in the share of financial assets. agricultural assets and durables to be important in a developing country. regulation and cultural preferences (IMF. This correction is especially important for the US where NPOs account for about 6 per cent of the financial assets of the household sector (Board of Governors of the Federal Reserve System. non-profit organizations (NPOs) are sometimes grouped in with households. The asset composition reflects different influences on household behaviour such as market structure. For most countries. in addition to the 18 with complete balance sheets. to be careful when analyzing these data since the comparison may be affected by sectoral definition. and other financial assets. non-financial assets account for between 40 and 60 per cent of total assets.

Table 4 shows there is more variation in country coverage than in the HBS data.9 In transition countries this is expected due to poorly developed financial markets. Hungary and Slovenia are intermediate between these extremes. These are typically more serious for estimating wealth distribution than e. These three countries.g. Romania. The high skewness of wealth distributions makes sampling error more severe. 2000 and 2005). Estonia and Lithuania stand out as having liquid asset shares of one-third or lower. together with Finland. Italy stands out as having a particularly low share of liabilities. which is of course one of the most interesting parts of the distribution (see Davies and Shorrocks. Croatia. Latvia.assets are a large part of the total. South Africa and the UK the share of other financial assets is particularly high. Non-sampling error is also a greater problem since differential response (wealthier households less likely to respond) and misreporting are generally more important than for income. wealth surveys suffer from sampling and non-sampling errors. something that is confirmed by survey data (see below). Both sampling and non-sampling error lead to special difficulties in obtaining an accurate picture of the upper tail. India and Indonesia. we also consulted household wealth survey data. Austria. 9 Among the transition countries. Most importantly. This may be partly attributable to the importance of pension fund claims in these countries. Like all household surveys. and Mexico in the case of non-financial assets. such as Australia. shares of liquid assets in total financial assets are 60 per cent or higher for Bulgaria. In Japan the preference for liquidity has a long history but also reflects lack of confidence in real estate and shares after their poor performance in the 1990’s (Babeau and Sbano. In some countries. and Slovakia. the Czech Republic. B Survey data In order to check on our HBS data and to expand our sample. for income distributions. wealth surveys are available for the three most populous developing (and emerging market) countries: China. the Netherlands. 8 . especially to non-OECD countries. are used in regressions in the next section that provide the basis for wealth level imputations for our ‘missing countries’. 2003).

only Finland oversamples rich households. Sampling error may therefore be of some concern in the Chinese CASS survey. The total asset values for the household sector are then given as the difference between total national wealth and the sum of these two other sectors. The surveys from the three developing countries pay relatively little attention to financial wealth since it is of less importance there.11 In contrast. As found here.10 Among the four countries whose survey data are used in the regressions reported in the next section. The asset coverage and details of the surveys reflect the relative importance of specific assets in rich and relatively poor countries. well-designed wealth surveys over-sample wealthier households. after the balance sheets for the government and corporate sector are first computed. housing and vehicles. however. which again helps to reduce the effects of sampling error.In order to offset the effects of sampling error in the upper tail. and concentrate on housing. agricultural assets. The Indonesia Family Life Survey also has some coverage limitations. the Indian AIDIS survey (part of the Indian National Sample Survey round 59) and the Indonesian Family Life Survey. This may be due to differential response and/or under-reporting by those who do respond. comparisons with HBS data generally show lower totals for most financial assets in surveys. there are additional difficulties regarding the representativeness of the wealth survey sub-sample. In terms of asset coverage the Finnish survey concentrates on financial assets. 9 . which covers only a part of the provinces included in the sample of the State Statistical Bureau Household Income Survey. This is despite very high reported response rates (in excess of 90 per cent) in both China and India. as in the US Flow of Funds. 2001). In the case of the Chinese survey. certain assets and liabilities in the balance sheets for the household sector are often computed as a residual. The SSB sample itself also suffers from some degree of geographical under-coverage (Bramall. As a result. are sometimes better covered in survey data. especially housing. for example. balance sheet totals for the household sector are also prone to error. As mentioned above. non-financial assets. 11 Also. land and consumer durables. See Kennickell (2004). This is the practice in the US Survey of Consumer Finances and the Canadian Survey of Financial Security. non-sampling errors include both differential response by wealth level and misreporting (mostly under-reporting). 10 The SCF design explicitly excludes people in the Forbes 400 list of the wealthiest Americans. Wealthy households are less likely to respond to surveys. The survey is reported to be representative of 83 per cent of the Indonesian population covering 13 of the nation’s 27 provinces.

vs. underreporting of debt appears to be more severe than underreporting of assets. Again poorly developed financial markets help to explain this phenomenon.Table 4 reports asset composition in the survey data. 50. in addition. In addition. The table also shows how different is the importance of non-financial and financial assets in developed and developing countries. according to Renwei and Sing (2005). agricultural assets and consumer durables are particularly important in many developing countries. One reason is that the value of housing is reported net of mortgage debt in China. reflecting both the relative accuracy of housing values in survey data and the importance of under-reporting and non-response among rich households. Subramanian and Jayaraj (2006) estimate that debts are. In addition. most of the financial assets owned by households are liquid. But. showing that about 64 per cent of household financial assets there are liquid. underrepresented by a factor of 2. It is clear that non-financial assets bulk larger in surveys than in HBS data. 10 . financial markets are often poorly developed. This low share echoes the finding in HBS data.8 per cent for fourteen developed market economies. and likely reflects a real difference between Italy and other high income OECD countries. The two low income countries in our sample.0 per cent of national assets in tangible form in 1978. Renwei and Sing (2005) report more detailed data for urban areas of China. on average. land. the only low or middle income country for which we have some detail on financial assets. Italy also stands out as having a very low share of liabilities. who own a disproportionate share of financial assets. In our table. China does not stand out as having high shares of non-financial assets.12 This is no surprise since assets such as housing. stand out as having particularly high shares of non-financial wealth. reflecting the deepening of market oriented reforms. Another is that there is no private ownership of urban land. The ratio of liabilities to total assets is particularly low in India and Indonesia (for China only non-housing liabilities are reported).93 in the AIDIS. In India. there has been a rapid increase of financial assets especially in rural areas. 12 This echoes the findings of Goldsmith (1985) who reported that India and Mexico had an average of 65. India and Indonesia.

C Per Capita Wealth from Household Balance Sheet and Survey Data Table 5 summarizes the distribution of per capita wealth in the year 2000 among countries for which we have complete household balance sheet and/or wealth survey data.086. The last column shows real consumption per capita. again slightly higher than that of GDP per capita.) The data are given both on PPP and official exchange rate bases. the US ranks first and South Africa last. and South Africa last.519. The next column shows GDP per capita.691 on a PPP basis. preceded by Poland at $24. The coefficient of variation is 0. the US again ranks first. the US ranks first on a PPP basis (although it is surpassed by Japan at official exchange rates). at $25. The full results we present later in the paper include inequality within countries.832. and then Singapore at $113. (Data for individual countries are given in Appendix IV. and the (unweighted) coefficient of variation of GDP per capita (again among the 18 countries) is 0. Italy at $119. Of the 18 countries for which we have complete HBS data. In column 4 we show personal disposable income per capita for the same group of countries. compared to the 0.431.631.704.266. at $35. the ratio of highest 11 . The comparison here is only between countries. Differences are even greater on an exchange rate basis. with the US/South Africa ratio rising to 24. The overall range is rather large. at $8. with per capita wealth in the US 8.4. These results are a first illustration of the fact that. In the group of 18 countries with HBS data. and the ratio of highest to lowest is 5. the Netherlands at $120. which further increases the gap between income and wealth inequality. wealth is more unequally distributed than income. slightly higher than for GDP per capita. compared to 0. the range is much smaller than for net worth per capita.594 figure for wealth.480.8 times as great as that of South Africa on the PPP basis. On the exchange rate basis the CV of GDP per capita is 0. followed by the UK at $126. Japan at $124. at $16.4.017 on a PPP basis.427 for net worth per capita.858. South Africa is last. The ratio of highest to lowest GDP per capita is only 4. The US again ranks first.311. among the 18 countries rises from 0. The coefficient of variation (CV).619. with per capita wealth of $143. Once again.0. globally.727 in 2000. South Africa is again last.594 on the exchange rate basis. However.654 and the Czech Republic at $32.326. at $4.427 on a PPP basis to 0.

and is higher again in the survey data — at 0. The correlation between net worth and GDP is only 0.856. and 17. Again. China appears to be about twice as wealthy as India.513 on a PPP basis and $1.848 on an official exchange rate basis. All in all. India is last. The CV is 0. at $6.922 from survey data.975 from the balance sheet data and 0. at $7.860 from balance sheet data and 0.908 on an exchange rate basis (see Table 6). again on an exchange rate basis. and Japan at $91. the coefficients of variation for the income and consumption variables are smaller than for wealth. at $143. preceded by Indonesia.267 on a PPP basis or $2. Of the 13 countries for which we have the pertinent data. having per capita net worth of $11. In the survey data. India and Indonesia). Wealth per capita is closely related to both income per capita and consumption per capita. disposable income per capita.958 in the survey data on an exchange rate basis.597.319. the variation of net worth per capita is much greater than GDP per capita. disposable income.to lowest on a PPP basis is 4. The difference between countries is even more pronounced in the survey data results than in the HBS data.7. On the exchange rate basis inequality between countries is again greater than on a PPP basis. but that correlation rises to 0. and consumption per capita. Correlations of wealth with consumption are a little lower: 0. about the same as GDP per capita and slightly higher than that of disposable income per capita.440 using official exchange rates. Correlations of wealth and disposable income are higher from both HBS and survey sources — rising to 0.112 on an exchange rate basis. slightly higher than that of GDP per capita and slightly lower than that of disposable income per capita.739 in the HBS data on a PPP basis. due to the inclusion of three developing countries (China. 13.613 using official exchange rates. as in the HBS data.993 from the survey data using 12 . the range in per capita wealth is much larger than that of per capita GDP.857.973 on a PPP basis and $1.3 for consumption on a PPP basis. and inequality is considerably greater using official exchange rates rather than PPP. In this group.3 for both GDP and disposable income.1 for net worth per capita. The ratio of highest to lowest is 22. The highest correlations are between the logarithm of net worth per capita and the logarithm of disposable income per capita: 0. followed on a PPP basis by Australia at $101. the US again ranks first in net worth per capita. or consumption.

For a large number of countries part or all of wealth is imputed on the basis of regressions run on the 38 countries for which we have HBS or survey data. 13 Middle income Oceania was assigned a simple average of Fiji and New Zealand.official exchange rates. Liechtenstein. It is tempting to regard the results as representing the global picture. as detailed below. covering 95. However this would implicitly assume that the 79 excluded countries and people are neither disproportionately rich or poor. Iraq. Myanmar. but nevertheless an improvement over the default of simply disregarding the excluded countries. in the end. Correlations of the logarithm of net worth per capita and the logarithm of consumption per capita are slightly lower. Perhaps most importantly. Sudan and Uzbekistan each have more than 10 million population) are all classified as low income or lower middle income. Kuwait. Nepal. The goal is not to estimate a structural model of wealth-holding. This gives us 150 countries with observed and/or imputed wealth. Bermuda). Serbia. This assumption is untenable. While the omitted countries include several smaller rich nations (for example. it limits our choice of explanatory variables to those that are available not only for the countries with wealth data but also for a large number of countries without wealth data. Cuba. 3 Imputing per Capita Wealth to other Countries We next impute per capita wealth to the remaining countries of the world. The nature of this exercise limits the range of models that can be applied.2 per cent of the world’s population in 2000. Angola. 13 . To try to compensate for this bias. the Channel Islands. to each of the omitted countries we assign the mean per capita wealth of the continental region (6 categories) and income class (4 categories).13 This assumption is admittedly crude. to assign wealth levels to 229 countries. the most populous countries (Afghanistan. North Korea. It allows us. The regressions we report below are designed to help us predict wealth in countries where wealth data are missing. but to find equations that fit well in-sample and that will also allow us to predict out-of-sample.

And for Mexico. and survey data for five countries that lack HBS data (China. Having discovered that the results improved when we ran three regressions. so these equations fit fairly well. In each regression the income variable is very important. and ran separate regressions for each. but the intercept is higher with HBS data. For our dependent variables we use the household balance sheet data discussed above for 33 countries. we have an observation of non-financial assets.14 However. For the 15 countries shown in Table 1 with financial balance sheets. Initially our dependent variable was simply per capita wealth. components were available. but in bringing in more developing and transition countries. but not all three. 14 . but not all three. The slopes of the simple regression lines in the two figures are similar. since this variable is available for about twice as many countries as disposable income.A Wealth Regressions We first experimented with OLS regressions for those countries with complete wealth data. excluding the 16 countries with incomplete data shown in Table 1. India. The regressions therefore become better at predicting wealth for the ‘missing countries’. there is a strong relationship between wealth and income. Part of the reason this approach yields better results is that some variables are helpful in predicting one or two of these components. which makes the consumption 14 Figure 1 uses wealth from the HBS data while Figure 2 uses wealth from survey data. Here we highlight runs using real consumption per capita instead. and Mexico). there are observations of both financial assets and liabilities. Finland. We show the results of those runs in Appendix II. The principle independent variable was per capita income or consumption. but no data on real assets. we realized that productive use could be made of data from countries where some. we discovered that we could predict better if we disaggregated wealth into (i) non-financial assets. The best fit is obtained using disposable income per capita. There are significant gains from the greater flexibility offered by running separate regressions. This reflects the fact that survey data generally provide lower estimates of wealth than national balance sheets. Indonesia. Also. (ii) financial assets and (iii) liabilities. Adding observations from these countries has a benefit not only in increasing sample size. the relative impacts of common variables vary across the equations. As Figures 1 and 2 make evident.

Since errors in our three equations are likely to be correlated we investigated using the seemingly unrelated regressions (SUR) technique due to Zellner (1962). These differences imply that. Variable sources are given in Appendix II. The slightly greater elasticity for financial wealth seems plausible.specification far more useful for imputations. There is an even larger difference for liabilities. this is very difficult to do in STATA or (we expect) in other standard packages. Both the dependent variables and most of the independent variables are entered in log form. Estimates of the latter generally come from different sources and are prepared using different techniques from those used in financial balance sheets. Also. The results we show here for the financial assets and liabilities regressions are therefore performed using SUR. The estimated elasticities of non-financial and financial wealth with respect to consumption are 1. 16 While it is theoretically possible to apply SUR with an unequal number of observations in the equations estimated.15 This involves stacking the three equations and estimating via generalized least squares. this is less likely in comparing either of those variables with non-financial assets. 15 See also Greene.487. While OLS estimates would be consistent. 486-499. since higher income countries tend to have better developed financial markets. The gain in efficiency is expected to be greater the more highly correlated are the errors across the equations. Note first that real consumption per capita appears significant at the 1 per cent level in all of the runs.16 Table 7 shows our results with three different versions of the consumption specification. 1993. while errors in the financial assets and liabilities equations are likely to be correlated. should be small. For equations with an equal number of observations it is straightforward to apply SUR in STATA. at least. pp.028 and 1. there will be a tendency coming from the consumption variable for their imputed financial assets and (especially) liabilities to be relatively less important than their non-financial assets. 15 . and the less correlated are the regressors used in the different equations. Thus correlations in measurement error.177 respectively in our preferred runs. Our preferred specifications are b for non-financial wealth and c for financial wealth and for liabilities. labelled a through c. SUR provides greater efficiency. for the many low income countries where we make imputations. Using consumption reduces goodness of fit only slightly. which have an estimated elasticity of 1.

16 . It was insignificant in the regression for non-financial assets. this variable is not statistically significant and was dropped. Market Capitalization Rate: The value of household financial assets should be positively correlated with this measure of the size of the stock market. it loses significance in the b specification and was dropped from the final run. since public pensions may substitute for private saving. Income Gini: Some theoretical models suggest that income inequality and per capita wealth should be positively related.17 17 R2 is not a well-defined concept in generalized least squares. which is fortunate from the imputation perspective since. the survey dummy is significant at the 1 per cent level in all three runs for financial wealth. In contrast. With a value of -1. as in the case of market capitalization. this dummy reflects the well-known fact that financial assets are under-reported and under-represented in survey data. Domestic Credits Available to the Private Sector: This variable is highly significant in the liabilities regression. This variable is statistically significant in the non-financial asset regressions. This is a useful result in terms of prediction and imputations. We also considered five other independent variables: Population Density: The value of non-financial assets. However. The R2 or ‘R2’ for each equation indicates that we get a fairly good fit for our model. the variable is available for many of our ‘missing countries’. so as is customary the fraction of the variance in the dependent variable that is ‘explained’ in each regression is referred to as ‘R2’ here. While significant at the 10 per cent level in the first liabilities specification. which is not unexpected since survey data typically cover non-financial assets quite well. It is positive and significant in all three regressions for financial wealth. However. should be positively related to the degree of population density (greater density indicating a relative scarcity of land).516 in the c run. particularly housing.We tried a dummy variable indicating the data source (HBS or survey data) in all three regressions. since the variable is available for a large number of countries that do not have full wealth data. Public Spending on Pensions as a Percentage of GDP: We expected this might be negatively related to financial assets per capita. the variable turns out to be insignificant.

the number of adults (specifically. we imputed estimates based on the average proportion of adults in the region-income category used previously.8 per cent share of world population and significantly more than its 53. Looking first at the 24 high income OECD countries. there is a case for expressing wealth levels in terms of the average wealth per family (or household) or the average wealth per adult. and is discussed in more detail in Section 5 below. and a strong indication that wealth is more unequally distributed across countries than is income. In contrast. (iv) lower middle income. Thus we have an immediate indication of the high concentration of world wealth in the richest countries.18 18 When the population over 20 is not reported. since the total population of households is not reported for many countries. Here we simply note that computing average wealth per household poses practical problems. The degree of concentration is even greater if the calculations are done on an official exchange rate basis. other options may also have attractions.9 per cent of world GDP). the latter reflecting an implicit assumption that the wealth holdings of those under 20 years of age can be neglected in global terms. In particular. (iii) upper middle income. We therefore provide a second set of figures for the average wealth per adult in each country. the CV of per capita wealth is much higher when we measure wealth on an exchange rate basis. A complete list of estimated wealth by country is provided in Appendix V.6 per cent share of world GDP. 17 . The high income OECD countries then have 83. We again grouped countries into (i) high income OECD. and as we found above for countries with HBS or survey data.7 per cent. (ii) high income non-OECD.3 per cent of world household wealth (and 76.B Imputations Table 8 shows summary results for our full sample of 229 countries. The choice between the three alternative concepts becomes more significant in the context of wealth distribution. While it is natural to compare wealth levels across countries in terms of wealth per capita. and (v) low income classes. Imputed levels of wealth per adult use regional averages weighted by the number of adults rather than the total population. on a PPP basis we find a share of world household wealth of 63. and present information here for groups only. much larger than this group’s 14. the number of persons aged 20 or above) is widely available.

The 39 countries in the upper middle-income group had an average wealth just a little below the world average. but the difference is quite small.) Each of these countries had a smaller share of world wealth than of world GDP — in the case of Russia 1. Kuwait. while Mexico and Brazil were somewhat lower (81 per cent and 80 per cent respectively on a PPP basis). and that both wealth and income inequality between countries are higher within the group when we use official exchange rates rather than PPP. where adults comprise a smaller fraction of the population.19 And for the low income countries inequality is greater on a PPP basis than when using official exchange rates (true for both wealth and GDP). for example the Bahamas. and Russia stood at 67 per cent of the world mean.Table 8 shows that for the high income OECD countries wealth per adult is about a third higher than wealth per capita ($151. while having just 0. Larger proportional differences are found for lower income countries. Singapore. Mexico. On a PPP basis the 43 countries in the high-income non-OECD group accounted for 3.93 per cent of world population. $113. (There is no systematic difference between wealth inequality on a per adult vs. This group includes countries like Brazil. (See Appendix V where full details are given for all countries. For the high-income non-OECD and upper middle income groups we again find that wealth inequality tends to be greater than income inequality. for the 59 lower middle income and 64 low income countries we find different results.308 vs. In both cases per capita wealth inequality is less than inequality in per capita GDP according to our estimates. Bahrain.675 in PPP results). Taiwan.184. Poland and Russia. Average PPP-based per capita wealth for the whole group was 3.6 per 18 . Poland’s per capita wealth was close to the world average. respectively on a PPP basis). This group too showed greater inequality in per capita wealth than in per capita GDP (CVs of 0.23 per cent of world household wealth and 2.35 per cent of world GDP.484 and 0.5 times the average for the world. and the United Arab Emirates. Wealth inequality among these high income OECD countries is higher on a per adult basis than on a per capita basis. Israel. These countries include many small but wealthy countries.) However. Qatar. per capita basis.

This group also showed greater inequality in per capita wealth than in per capita GDP (CVs of 0.2 per cent of the world average and whose GDP per capita was 43. whose average per capita net worth was only 24. Egypt. respectively). Others. Its collective household net worth on a PPP basis amounted to 8.385 and 0.9 per cent of the world’s population and 11.) The mean per capita PPP wealth of this group was 20. has fairly similar shares of world wealth and GDP. For this group.9 per cent of the world average. Turkey. like Egypt and the Ukraine. like Turkey. whose per capita wealth was 30.0 per cent share of world population and their 18.cent of the wealth vs. According to our estimates.264. its wealth share is only 2.9 per cent of world GDP. 3.4 per cent for GDP. have a smaller share of wealth than of GDP.3 per cent of world GDP. The lower middle income group includes China. respectively. and the Ukraine. especially using region/income group proxies 19 . have a larger share of world wealth than of world GDP.7 per cent of the world average but whose per capita GDP was 29. at least on a PPP basis: 8.2 per cent of the world household (PPP) wealth.240 and 0. the inequality of wealth per capita was less than its inequality of GDP per capita (CVs of 0. This group consists of countries such as India.5.413. (Note that our numbers for India and Indonesia. (However. This compares to their 33.2 per cent of the world average. according to official exchange rates.8 per cent of the world average.) The collective household wealth of the 59 lower middle income countries amounted to 15. 19 It is possible that this result is due to the high rate of imputations.8 per cent of world household wealth and 10. compared to 39. which is such an important element in the world distribution.5 per cent of world GDP. respectively). like those of China.5 per cent of the world total on a PPP basis.2 per cent of world GDP.6 per cent and its GDP share is 3. it is interesting to note that China. and Indonesia. some of these countries. compared to 24. on a PPP basis). Interestingly. However. for this group the inequality of per capita wealth fell short of its inequality of GDP per capita (CVs of 0.277 and 0. Overall.9 per cent of world GDP on a PPP basis. 11. are based on actual observations rather than imputations. the upper middle-income group accounted for 9.3 per cent of world wealth.170. The last group consists of 64 low-income countries.4 per cent of the world population but 13.

The data differ in many significant respects. together with the share of the top 5 per cent and the top 1 per cent of wealth holders. looking at the world as a whole. This reflects the fact that children form a larger percentage of the population in poor countries.Finally. The results should therefore be treated with caution. We return to this point in the next section. we find that according to the CV. the large share of wealth that is held by the wealthy. the inequality of net worth per capita among the 229 countries in the sample was considerably higher than the inequality of GDP per capita.298 respectively in the per adult data). in this group. 0. The economic unit of analysis is most often a household or family. and the international outlook and investments of large wealth-holders around the world.878 and 1. Given the relatively high level of integration of world capital markets that has now been achieved. in the case of the UK. (CVs of 1. As in most of the country groupings.1 per cent. but sometimes an individual or. we find that inequality is considerably greater when measured using official exchange rates than on a PPP basis. 2 per cent. 5 per cent. 1 per cent. We selected one set of figures for each nation. (2.e. Table 9 adopts a common distribution template consisting of the decile shares reported in the form of cumulated quantile shares (i. 4 Distribution of Wealth within Countries The raw data with which we begin refer to 20 countries for which there is some information on the distribution of wealth across households or individuals. Distribution information is usually reported for share of wealth owned by each decile. Lorenz curve ordinates) plus the shares of the top 10 per cent. Also note that world wealth inequality is lower when measured on a per adult basis (both with official exchange rates and PPP).769 using official exchange rates).5 per cent and 0. ceteris paribus. adult persons. To assist comparability across countries.141 compared to 1. 20 . with a preference for the year 2000. But this pattern is far rather than the regression approach. there is a stronger argument for paying attention to the exchange rate-based inequality results than is the case when studying income inequality or poverty.

(See Kennickell. does not over-sample in the upper tail). under-reporting may nonetheless occur. (2004) have used records of the number of contacts needed to win a response to estimate the differential response relationship. including Italy (which. as discussed earlier. and under-reporting is illegal.5 per cent or even the top 0. Household sample surveys are employed in 15 of the 20 countries. by high income/wealth groups. and there are valuation problems that produce analogous results. The French and UK data are based on estate tax returns. allowing reweighting. The impact of these errors can be reduced.) Pioneering new statistical techniques have been used to correct for non-sampling error in some other countries. where the accuracy of the Survey of Consumer Finances in measuring the shape of the distribution of wealth is considered to be very high. wealth shares are reported for the top 0. on other occasions. Norway. since its distributional detail is of interest.20 Survey results are affected by sampling and non-sampling error. 2003 and 2004.1 per cent in the case of France and Switzerland. however. Nonsampling error. This occurs because wealthy households are less likely to respond. while the data for Denmark.from universal. These data sources have the advantage that ‘response’ is involuntary. and hence added countries with data considerably earlier than 2000: Ireland (1987) and Korea (1988). including the US and Canada. We are also adding Sweden. There is also an imputation for non-reported dwellings respondents own (aside from their principal dwelling). Here we wish to exploit distributional information from as many countries as possible. Brandolini et al. however. Wealth tax regulations may assign to some 20 The list of countries differs a little from the 13 used in Sections 2 and 3. although the mean from this source was not judged sufficiently reliable to be used in our levels estimates. even if it is unsafe to use the 1980s values for wealth levels. A validation study comparing survey and independent amounts from a commercial bank for selected financial assets is used to correct for misreporting. The Netherlands was dropped due to insufficient distributional detail. 21 . The most important respect in which the data vary across countries is the manner by which the information is collected. Careful imputations of asset and liabilities are also made in cases where respondents do not answer all questions. and because under-reporting is particularly severe for the kinds of financial assets that are especially important for the wealthy — for example equities and bonds. in the form of differential response and under-reporting. and Switzerland originate from wealth tax records. However. In some instances information on quantile shares is very sparse. 21 Intensive efforts along these lines are made in the US. as has been done in several countries.21 Other wealth distribution estimates used here derive from tax records. The hope is that the shape of wealth distribution in these countries was reasonably stable from the late 1980s to the year 2000. tends to reduce estimated inequality and particularly the estimated share of top groups.

p. In order to move towards estimating the world distribution of wealth. the shares of the richest groups can be depressed very significantly (see for example Davies. Germany and Sweden were discarded. with the USA towards the top end of this range at 32. a list that excludes China. Table 9 shows that wealth concentration varies significantly across countries. 2006. If survey data do not oversample the upper tail. One way to attack this problem is to replace. for a review of this form of evidence). 2003. As is evident in Table 9. the available sources provide a sparse patchwork of quantile shares. 22 . the survey estimate of the upper tail with figures derived from lists of the very rich (and their wealth) compiled by journalists and others (see Atkinson. To apply this ‘ungrouping’ utility. Germany. adding them would raise the share of the top 1 per cent by about two percentage points. While estimates have been prepared on this basis in a few countries. missing cell values were imputed using a utility program developed at WIDER which constructs a synthetic sample of 1000 observations that conforms exactly with any given valid set of quantile shares obtained from a distribution of positive values (e. ranges from 41. more complete and comparable information is needed on the distribution in each country. See Kennickell. To achieve this. and omit other assets altogether. but is generally very high. where possible. the negative wealth shares reported for Finland.7 per cent. the approach has not been widely pursued and is beyond the scope of this paper. Estimated shares of the top 1 per cent range from 10. That statistic is only reported for 12 countries. (The sampling frame for the US survey excludes the Forbes 400 richest families. but in Sweden the bottom half of the distribution collectively has more debts than assets. 1993): in the absence of corrections for non-sampling error.g.4 per cent in Ireland to 34.assets a fraction of their market value. and Japan may well reflect this phenomenon. which is available for all 20 countries.8 per cent in the USA. The surprisingly low top shares seen here in Australia. and the Nordic countries apart from Denmark. For most countries the bottom decile of wealth holders is reported as having positive net wealth.4 per cent in China to 69. 3.8 per cent in Switzerland. There are also evident differences in the way that debts are investigated and recorded. The differences in wealth concentration across countries in Table 9 are probably attributable in part to differences in data quality. Ireland.) The share of the top 10 per cent. Comparisons of wealth inequality often focus attention on the share of the top 1 per cent. a reasonable guess is that the share of the top 1 per cent may be underestimated by 5–10 percentage points. incomes).

together with the zero shares reported elsewhere, thus treating the cell values as missing observations.

The 20 countries for which wealth distribution data are available include China and India, and hence cover a good proportion of the world’s population. They also include most of the rich, populated countries, and hence cover a good proportion of the world’s wealth. However, the fact that the list is dominated by OECD members, and is further biased towards Nordic countries, cautions against extrapolating immediately to the rest of the world.

To estimate wealth distribution shares for countries for which no direct information exists, we made use of income distribution data for 145 countries recorded in the WIID dataset, on the grounds that wealth inequality is likely to be correlated — possibly highly correlated — with income inequality across countries. The WIID dataset has many observations for most of the 145 countries. Where possible, we selected the data for household income per capita across individuals for a year as close as possible to 2000, with first priority given to figures on disposable income, then consumption or expenditure. 85 per cent of the income distributions conformed to these criteria. Figures for gross incomes added a further 7 per cent, leaving a residual 8 per cent of countries for which the choices were very limited. The WIDER ‘ungrouping’ utility was then applied to obtain quantile shares for income (reported in Lorenz curve form) according to the same template employed for wealth distribution.

The common template applied to the wealth and income distributions allows Lorenz curve comparisons to be made for each of the 20 reference countries listed in Table 9. In every instance, wealth shares are lower than income shares at each point of the Lorenz curve: in other words, wealth is unambiguously more unequally distributed than income. Furthermore, the ratios of wealth shares to income shares at various percentile points appear to be fairly stable across countries, supporting the view that income inequality provides a good proxy for wealth inequality when wealth distribution data are not available. Thus, as a first approximation, it seems reasonable to assume that the ratio of the Lorenz ordinates for wealth compared to income are constant across countries, and that these constant ratios (14 in total) correspond to the average

23

value recorded for the 20 reference countries.22 This enabled us to derive estimates of wealth distribution for 124 countries to add to the 20 original countries on which we have direct evidence of wealth inequality.

The group of 144 countries on which we have inequality evidence differs slightly from the group of 166 nations for which mean wealth was estimated from actual data or our regressions in Section 2. Distributional evidence is a more common for populous countries, so the group of 144 now include Cuba, Iraq, Myanmar, Nepal, Serbia, Sudan and Uzbekistan and cover 96.6 per cent of the global population. For the rest of the world, the default of disregarding the remaining countries was again eschewed in favour of imputing wealth distribution figures equal to the (population) weighted average for the corresponding region and income class.

5

World Distribution

The final step in the construction of the global distribution of wealth combines the national wealth levels estimated in Section 3 with the wealth distribution data discussed in the previous section. Specifically, the ungrouping utility was applied to each country to generate a sample of 1000 synthetic individual observations consistent with the (actual, estimated or imputed) wealth distribution. These were scaled up by mean wealth, weighted by the population size of the respective country, and merged into a single dataset comprising 228,000 observations.23 The complete sample was then processed to obtain the minimum wealth and the wealth share of each percentile in the global distribution of wealth. The procedure also provides estimates of the composition by country of each wealth percentile, although these are rough estimates given that the population of each country is condensed into a sample of 1000, so that a single sample observation for China or India represents more than half a million adults.

22 To circumvent aggregation problems, the adjustment ratio was applied to the cumulated income shares (ie Lorenz values) rather than separate quantile income share. We intend to check the validity of this procedure by comparing the ‘true’ and ‘estimated’ aggregate wealth distribution over the 20 reference countries. 23 Our data include 229 countries. However, we dropped Pitcairn Island at this point because of its very small population (far less than 1,000 people).

24

The interpretation of data on personal wealth distribution hinges a great deal on the underlying population deemed to be relevant. Are we concerned principally with the distribution of wealth across all individuals, across adult persons, or across households or families?24 When examining the analogous issue of global income distribution, we support the common practice of assuming (as a first approximation) that the benefits of household expenditure are shared equally among household members, and that each person is weighted equally in determining the overall distribution. However, the situation with wealth is rather different. Personal assets and debts are typically owned by named individuals, and may well be retained by those individuals if they leave the family. Furthermore, while some household assets, especially housing, provide a stream of communal benefits, it is highly unlikely that control of assets is shared equally by household members or that household members will share equally in the proceeds if the asset is sold. Membership of households can be quite fluid (for example, with respect to children living away from home) and the pattern of household structure varies markedly across countries. For these and other reasons, the total number of households is not readily available for many countries. Thus, despite the fact that most of the datasets listed in Table 9 are constructed on a family or household basis, we believe that the distribution of global wealth is best interpreted in terms of the distribution across adults, on the grounds that those under 20 years of age have little formal or actual wealth ownership, and may therefore be neglected in global terms.25

Tables 10a and 10b summarise our estimates of the distribution of wealth across the global population of 3.7 billion adults with wealth measured at official exchange rates for the year 2000. The results indicate that only $2161 was needed in order to belong to the top half of the world wealth distribution, but to be a member of the top 10 per cent required at least $61,000 and membership of the top 1 per cent required more than $500,000 per adult. This latter figure is surprisingly high, given that the top 1 per cent group contains 37 million adults and is therefore

24 Note that each of these bases was used by at least one country listed in Table 9. 25 The original country level data are generally based on households. We implicitly assume that the shape of the distribution of wealth among adults is the same as that among households, an assumption which would be true if all households contained two adults, if children had zero wealth, and if wealth was equally divided between the adult members.

25

It is also worth pointing out that the relative insensitivity of the Gini coefficient to the tails of the distribution implies that our likely slight underestimation of the top wealth shares will have little impact on the estimated Gini. This compares with the bottom half of the distribution which collectively owns barely 1 per cent of global wealth. 51 per cent (25 times the average) and 40 per cent (40 times the average).795 using official exchange rates. Furthermore. The figures for wealth shares show that the top 10 per cent of adults own 85 per cent of global household wealth.801) and Switzerland (0. while wealth inequality exceeds income inequality in global terms. p.000 times richer. respectively.846). to 2 per cent and top 1 per cent are 71 per cent (14. 108) estimates the Gini for the world distribution of income in 1998 at 0. the gap between the Gini coefficients for world wealth and income inequality — about 10 percentage points — is less than the gap at the country level. Table 10b supplements these results with details of the Gini coefficient for individual countries and for the world as a whole. As mentioned earlier. Thus the top 1 per cent own almost 40 times as much as the bottom 50 per cent.547 for Japan to the high values reported for the USA (0. The entrance fee has presumably grown higher still in the period since the year 2000. This is unavoidable given that an income Gini of 0. The corresponding figures for the top 5 per cent. By way of comparison Milanovic (2005.far from an exclusive club. Our wealth Gini estimates for individual countries range from a low of 0.795 and a Gini upper bound of 1. so that the average member of this group has 8.803). limits the possibilities for higher Gini values.5 times the global average holding. and the highest values of all in Zimbabwe (0. It is interesting to note that. which averages about 30 percentage points. concentration in the upper tail of the income distribution is also probably underestimated (although to a lesser extent than for wealth). This roughly corresponds to the Gini value that would be recorded in a 10-person population if one person had $1000 and the remaining 9 people each had $1.2 times the average). wealth distribution is unambiguously more unequal than income distribution in all countries which allow comparison.845) and Namibia (0. 26 . The average member of the top decile nearly 3000 times the mean wealth of the bottom decile. The contrast with the bottom decile of wealth holders is even starker. and the average member of the top percentile is more than 13. so that the estimated gap between wealth and income inequality is unlikely to be heavily biased. The global wealth Gini is higher still at 0.892.

each regional group accounting for around one third of the richest wealth holders. New Zealand and several middle eastern states) from the remaining (mainly low income) nations. Another notable feature is the relatively constant membership share of Asian countries other than China and India. ‘Thirds’ feature prominently in describing the overall pattern of results. Indonesia. At the top end. It is also convenient to distinguish the high income subset of countries in the Asia-Pacific region (a list which includes Japan. the US is in first place. The population of Latin America is also fairly even spread across the global distribution but Africa. India dominates the bottom third of the global wealth distribution. However. although the composition changes a little in the upper tail. The number of members of the top decile depends on three factors: the size of the population. this group is highly polarised. South Korea. is heavily concentrated at the bottom end. Pakistan and Vietnam) occupying the lower tail. The middle third of the distribution is the domain of China which supplies more than a third of those in deciles 4-8. As expected. Taiwan. average wealth. The list of countries include all those which account for more than 1 per cent of global wealth or more than 1 per cent of those in the top decile. The high wealth per adult and relatively equal distribution accounts for the 27 . Europe and highincome Asia monopolise the top decile. Table 10b provides more details for a selection of countries.We now turn to the composition of each of the wealth quantiles. North America. as the figures indicate. due to their population size. Bangladesh. and wealth inequality within the country. with the North American share rising while European membership declines. with the high-income subgroup populating the top end of the global wealth distribution and the lower income countries (especially. as expected. All three factors reinforce each other in this instance: a large population combining with very high wealth per capita and relatively unequal distribution. Australia. China and India are reported separately. Table 10a provides a regional breakdown where. plus those countries with adult populations exceeding 45 million They have been arranged in order of the number of persons in the top global wealth decile. perhaps – with 21 per cent of the global top decile and 27 per cent of the global top percentile. contributing a little under a third (27 per cent to be precise) of this group. with 25 per cent of the global top decile and 37 per cent of the global top percentile. Japan features strongly in second place – more strongly than expected.

and rich people from most countries travel a great deal and may do a considerable amount of their spending abroad. Lower down the scale.fact that the number of Japanese in the bottom half of the global wealth distribution is insignificant according to our figures. too. Tables 11a and 11b provide a second set of wealth distribution estimates based on PPP comparisons rather than official exchange rates. and hence the global average. Further down the list. Italy. In this case it may be more appropriate to evaluate wealth in terms of what it would buy if liquidated and spent on consumption locally. In today’s world capital is highly mobile internationally. Neither country has enough people in the global top 5 per cent in 2000 to be recorded. reflecting the small impact of PPP adjustments within the richest nations. To address this point. for much the same reasons as Japan. however. The wealth of one millionaire goes just as far as that of another in Monte Carlo or when shopping at Harrods. For the world’s super rich it is natural to compare the wealth of people in different countries using official exchange rates. The representation of both China and India has been rising in the annual Forbes list of billionaires. but also likely supply an increasing number of people in these categories. so these countries should not only be represented in an accurate estimate of the membership of the top 5 per cent or top 1 per cent. 28 . The price for entry to the top 10 per cent rises from $61041 to $88035. While the two countries are expected to be under-represented in the upper tail because of their relatively low mean wealth. irrespective of which country he resides in. Applying the PPP adjustment increases average wealth level in most countries. has a stronger showing than expected. The admission fee for membership of the top wealth groups also increases. but entry to the top 1 per cent increases more modestly. their absence here from the top 5 per cent seems anomalous. China and India both owe their position to the size of their population. which rises from $33893 per adult to $43628 per adult. data which could be improved by making corrections for differential response and under-reporting. from $514512 to $523264. It may well reflect unreliable wealth data drawn from surveys that do not over-sample the upper tail. the benefits (and valuations) of asset holdings may depend heavily on the local prices of goods and services.

802 (although the Gini values for individual countries are unaffected).1 per cent to 71. although not the top 1 per cent. and from 39.795 = 0. switching to PPP valuations compresses the variation in average wealth levels across countries and hence provides a more conservative assessment of the degree of world wealth inequality.892 to 0. 108) reports a world income Gini of 0. the estimated share of wealth owned by the richest individuals falls: from 85. India. principally due to a decline in the share of Japan. Thus the gap between the world wealth and income Ginis appears to be larger. from 0.892 – 0. on a PPP basis than using official exchange rates.2 per cent as a result of the decline in wealth per adult from $227600 to $157146 when measured in PPP terms. Brazil. is that of China which leapfrogs into fifth position with 4. The overall global Gini value also declines. 26 Milanovic (2005. As a consequence. in absolute terms. Membership of the top 10 per cent is a little more spread regionally. and both India and China are now deemed to have representatives in the global top 5 per cent. p. it is clear that household wealth is much more concentrated. when official exchange rates rather than PPP valuations are employed. India moves a little more into the middle deciles of the global wealth distribution. for example.642 using PPP. Thus a somewhat different perspective emerges depending on whether one is interested in the power that wealth conveys in terms of local consumption options or the power to act and have influence on the world financial stage.5 per cent to 14.1 per cent for the top 10 per cent of wealth holders. 29 .9 per cent to 31. where the gap reported earlier was 0.6 per cent for the top 1 per cent. In summary. Turkey and Argentina are now all promoted into the exclusive class of countries with more than 1 per cent of the members of the global top wealth decile. Russia.097. As regards the rankings of individual countries.Because the PPP adjustment factor tends to be greater for poorer countries.1 per cent of the members. both in size distribution and geography. whose membership of the top 10 per cent falls from 20. a relatively modest rise in average wealth in China will move it up to third position in the global top decile.26 The overall picture suggested by Tables 11a and 11b is much the same as before. The most dramatic rise. Even without an increase in wealth inequality. however. and overtaking Japan is not a remote prospect.

887 compared to 0. its sampling frame explicitly omits the ‘Forbes 400’ wealthiest US families. Other respects also lead us to believe that our estimates of the top wealth shares are conservative. Table 12 begins by summarising the figures recorded earlier for the total world adult population with wealth valued according to official exchange rates (row 1) and PPP (row 2). and a Gini value of 0. with a top 1 per cent share of 37. Although the SCF survey in the USA does an excellent job in the upper tail. The final row 7 keeps the same 20 countries but discards the ‘true’ wealth distribution figures. excluding all countries other than the 20 nations listed in Table 9 for which direct data exist on wealth distribution.A number of checks were conducted to test the sensitivity of our results to the assumptions made at various stages. 30 . This leads us to conclude that our method of estimating wealth distributions from income distributions may impart a downward bias to global inequality.887 to 0.879. for example.4 per cent to 32.9 per cent. The evidence suggests that the estimation procedure tends to slightly reduce wealth inequality at the top of the distribution.892. row 4 those without income distribution data (and hence no way of estimating wealth inequality). with the share of the top 1 per cent falling from 37. apart from a small rise in wealth shares in the top decile attributable to the imputations of wealth levels. The survey data on which most of our estimates are based under-represent the rich and do not reflect the holdings of the super-rich. so it is not surprising to discover almost no impact on global wealth distribution. The last two rows take an even more extreme position. Nevertheless. although the fact that less than 20 per cent of global wealth is affected by the procedure will limit the overall impact. Row 3 discards those with imputed wealth levels. replacing them instead with the estimate derived from income distribution data that was applied to most countries. the figures in row 6 are little different from the row 1 benchmark. and row 5 those with either form of imputation. The results show that the regional-income group imputations affect less than 6 per cent of the global adult population and less than 2 per cent of global wealth. The next three rows report the corresponding figures when we omit countries for which data has been imputed on the basis of region-income averages.4 per cent compared to 39. Restricting attention to these 20 countries loses 16 per cent of the world’s wealth and 41 per cent of the world’s adults.8 per cent and the world Gini value from 0.

but it is rare for them to be captured. one would not expect great accuracy in the extreme upper tail. with evident lumpiness appearing at wealth levels above $5 million. A rough idea of the possible size of the error can be gained by noting that the total wealth of the world’s billionaires reported by Forbes for the year 2000. was 1. the relationship in the range from $250. The number estimated by Forbes magazine for the year 2000 was 492. To explore the top tail in more detail. the Merrill-Lynch numbers are not based on lists of named individuals. the degree of similarity is astonishing and gives us confidence that our figures are not too far from the mark. plotting the logarithm of the number of adults above wealth level w against the logarithm of w. This is less than our figure of 13.6 million.16 trillion. Their methods are not 31 .3 trillion. This means that our estimated shares of the top 1 per cent and 10 per cent are likely to err on the low side.Surveys in other countries do not formally exclude the very rich. of $125. $2.2 million individuals worldwide with net worth in excess of $1million. Given the fact that each of our sample observations can represent many tens of thousands of adults.000 to $5 million appears stable and remarkably well approximated by a Pareto distribution. a Pareto-type graph was constructed. as we do.27 Further support comes from the fact 27 Merrill-Lynch in its 2001 World Wealth Report estimated 7. Given the very different ways in which the two estimates were derived. Figure 3 shows this to be the case. yielded the following predictions for the number of high wealth holders and the super rich: Wealth ($ million) 1 10 100 1000 Number above 13 568 229 451 809 15 010 499 Thus extrapolating our upper tail leads us to predict 499 dollar billionaires in the year 2000. Fitting a Pareto distribution in this range and extrapolating to the highest echelons of the wealth distribution. Unlike the Forbes data.7 per cent of the total world household wealth we find here. but largely use public sources of data. However.

We have seen that the distribution is highly concentrated — in fact much more concentrated than the world distribution of income. in terms of measuring the overall degree of inequality we have seen that that is not a major deficiency. 32 . 6 Conclusion This paper has provided a first estimate of the world distribution of household wealth.e. we have estimated that for the world as a whole the share of the top 10 per cent was 85 per cent in the year 2000 and the Gini equalled 0. rather than income distribution or poverty. Less striking. this is not a big problem. but still effective steps have been taken in some of the other wealthiest countries. a fraction that again conforms very closely with the data reported in Table 10b which gave a US share of 37.795 in 1998. 40. so it is difficult to know why the number of millionaires is less than our estimate. For the wealthiest country. or the distribution of wealth within all but a few of the world’s countries.7 per cent) as US citizens. Milanovic (2005) estimates that the world income Gini was 0. While the super rich are not represented in these data.892 using official exchange rates. But one likely explanation is that the Merrill Lynch wealth valuations exclude the family home and all consumer durables.4 per cent for the top 1 per cent and suggested that the share would rise further as wealth increased. Sophisticated techniques have been used by the Federal Reserve Board to make its triennial Survey of Consumer Finance a reliable instrument. On the other hand. which supplies 25 per cent of the world’s top 10 per cent of wealth-holders on an official exchange rate basis (20 per cent using PPP). surveys in the major developing countries appear to have a particular problem in capturing well documented. While wealth (and income) concentration is somewhat less when the estimates are done on a PPP basis. Much of the data that is used here comes from household surveys.7. the US. we have argued that the large share of wealth that is owned by people who can readily travel and invest globally means that converting at official exchange rates is preferable for many purposes when one is studying the distribution of wealth.that the Forbes data classifies 200 of the 492 dollar billionaires (i. and the median Gini around 0. While a typical share of the top 10 per cent of wealth-holders within a country would be about 50 per cent.

We have also found that differences are large even for countries at a similar stage of development. Our study of wealth levels and composition across countries is not only important in providing a key ingredient in estimating the world distribution of wealth. including all of the world’s major OECD economies. 2005) and others refer to as ‘international inequality’. there is a non-negligible error bound that must be placed on them. Thus. Our empirical work has shown how the importance of both financial assets and borrowing rises sharply with per capita income and the development of financial markets. It is also of independent interest. one may hope for a significant reduction in world wealth inequality as convergence takes place in wealth between nations. Secondly.the upper tail.000 for the United States in the year 2000. we get a snapshot of what Milanovic (2002. Even if the current level of wealth inequality within countries proved to be enduring. And for 18 of those countries we also have reasonable estimates of non-financial assets. Thus. but across countries at similar income levels due to institutional and traditional differences. It also depends on the accuracy of our estimates of mean household wealth in the countries studied. for example. Conversely. we have seen how balance sheets in poor countries are still dominated by land and other tangible assets. This means that while we have reasonable confidence in our estimates. and also less on a per adult basis. On an official exchange rate basis per capita household wealth ranged from about $1. Two caveats to conclusions about wealth inequality between countries (‘international inequality’) are that the concentration is smaller when we measure on a PPP basis. it gives us a picture of how wealth composition varies. In evaluating the results obtained here it is important to keep in mind that the quality of the estimates depends only in part on the accuracy of the distributional information that has been assembled. First. not only with the stage of development.100 for India to $144. varying importance of public pensions. differences between countriesmake an important contribution to inequality in the world distribution of household wealth. Here we are on safer ground. These estimates are often formed in conjunction with Flow of Funds numbers or the National Accounts. we have data on the household sector’s financial balance sheet. There is a solid basis of good numbers from financial institutions and government statistical agencies in these exercises. and other factors. while the coefficient of variation for international 33 . that is inequality between countries. For 33 countries.

and great strides have been taken in quite a few countries. There is huge room for improvement in the study of household wealth from a global perspective. Improvements in data quality. The mere fact that regular wealth surveys are conducted in the world’s two largest and most dynamic developing countries. and almost completely absent in Africa and most of Asia as well.298 on a per adult and PPP basis. 34 . are applied in more countries. however. and as techniques for achieving greater accuracy. must also occur.141 on a per capita and official exchange rate basis. Without the relevant data it is impossible to see what progress is being made. China and India. As surveys are conducted in more developing and transition countries. The task is far from hopeless. Financial development is a key aspect of economic growth and human development. These are completely lacking in Latin America. we will get a much better picture of the composition and distribution of household wealth around the globe. Household balance sheets and wealth surveys need to be generated in more countries. particularly in survey data. is very encouraging. it is only 1. especially in the upper tail.inequality is 2.

Research in Economic Inequality 4: 159-180. Australian Bureau of Statistics (2005). Guide to Flow of Funds Accounts. Volume 1. (2005). Rome: Bank of Italy. Cannari L. ‘Estimating the Balance Sheet of the Personal Sector in an Emerging Market Country. James B.F. ‘Concentration Among the Rich’. The China Quarterly 167: 689-705. and Saez. A. Household Wealth Distribution in Italy in the 1990s. Shorrocks (2000). Income and Wealth Concentration in Switzerland over the 20th Century. Ch. January. E. (2001). South Africa 1970-2005’. and Anthony Shorrocks (2005). John Muellbauer. and Faiella I. Davies. Handbook of Income Distribution Volume I. the United States and Japan’.Termi di discussione No 530. Harrison (1978). (2004). James B. WIDER/UNU Project on Personal Assets From a Global Perspective. Household Income and Income Distribution Available from: http://www. (2006). Working paper. T. Janine. London: CEPR. Working paper. Paper presented at a meeting of the Luxembourg Wealth Study. Anthony B. James B.. Discussion Paper Series No. Aron.. André and Teresa Sbano (2003). F. Center for Economic Policy Research.J.References Ammermüller. Economic Bulletin January 2005. ‘Household Wealth in the National Accounts of Europe. A. (2005). Methods and Preliminary Results’. and Westerheide P. Atkinson.au Babeau.abs. Davies.. C. Distribution of Personal Wealth in Britain. Amsterdam: Elsevier: 605-676.. ‘The Quality of China’s Household Income Surveys’. A. OECD Statistics Working Papers 2003/2. Davies. Available Dell. Zentum für Europäische Wirtscaftsforschung. ‘The Distribution of Wealth’. Perugia. and A.gov. Abschlussbericht zum Forschungsauftrag des Bundesministeriums für Gesundheit und Soziale Sicherung: Die Entwicklung und Verteilung des Vermögens privater Haulshalte unter besonderer Berücksichtigung de Produktivvermögens. and François Bourguignon. Anthony B. ‘Wealth Holdings in Developing and Transition Countries’. Board of Governors of the Federal Reserve System (2003). Weber. Atkinson. 35 . and A. Anthony B. (1993). and Johan Prinsloo (2006). WIDER/UNU Project on Personal Assets From a Global Perspective. 11 of Atkinson. Piketty. 5090. D’Alession G. Cambridge: Cambridge University Press. ‘The Distribution of Wealth in Canada’. Banco de Espana (2005)’Survey of Household Finances (EFF): Description. Brandolini. Bramall.

Global Financial Stability Report. Branko (2002). Market Developments and Issues. Washington: The World Bank. and Song. M. Econometric Analysis. International Capital Markets Department. Levy Economics Institute of Bard College.Y. (1993). and Drolet.2002): NSS 59th round (January-December 2003). R. (2003). Australian Economic Review 38: 159-175. Arthur B. Ottawa: Statistics Canada. D. Gary Marks and Mark Wooden (2005). The Evolution of Wealth Inequality in Canada 1984-1999. Household Assets and Liabilities in India (as on 30.A. Nathan Sussman and Yishay Yafeh (forthcoming).M. King. Nolan . The Wealth of Irish Households: What Can We Learn from Survey Data? Dublin: Combat Poverty Agency.. and J. 1688-1978 Chicago: University of Chicago Press Greene. ‘The External Wealth of Nations.go. (2004). Shorrocks. 187. Kennickell. Headey. 19892001’. Dollar. B. Leape (1998).stat. Bruce. (1992). X.06.F. William H.org/External/Pubs/FT/GFSR/2005/01/index. Leipziger. National Survey of Family Income and Expenditure 1999 [Internet]. Government of India: Ministry of Statistics and Programme Implementation. Lane. Raymond W. M. No.htm Kennickell. (1985). 1988 and 1993: First Calculation Based on Household Surveys Alone’. ‘True World Income Distribution. (2005).I. A. A Study of Twenty Countries. Paulo. Zhang. Distribution of Personal Wealth. Economic Journal 112: 51-92.imf.gov. The Distribution of Income and Wealth in Korea.uk/stats/personal_wealth/menu. Inland Revenue Statistics (2005). and Gian Maria Milesi-Ferretti (2000). Available from: http://www. Sovereign Bond Spreads in 1870-1913 and Today.htm Japan Statistics Bureau (2005). Comparative National Balance Sheets. ‘The Good Shepherd: Sample Design and Control for Wealth Measurement in the Survey of Consumer Finances’. Feb 22. presented at the Perugia meeting of the Luxembourg Wealth Study. D. working paper. . Working Paper No. 36 . Journal of Public Economics 69: 155-193. Morissette. ‘Wealth and portfolio composition: theory and evidence’. A. (1991). 393. S. Available at http://www. Measures of Foreign Assets and Liabilities For Industrial and Developing Countries’. ‘A Rolling Tide: Changes in the Distribution of Wealth in the US. Philip. Milanovic.hmrc. (2002). Emerging Markets and Financial Globalization. National Sample Survey Organization. Oxford: Oxford University Press.Goldsmith. Mauro.htm International Monetary Fund (2005). Second Edition New York: Macmillan. ‘The Structure and Distribution of Household Wealth in Australia’.jp/english/data/zensho/index. Available from http://www.. January 2005.

Spånt. Accumulation et Répartition de Patrimoines. The Assets and Debts of Canadians: An overview of the results of the Survey of Financial Security. Helsinki: Statistics Finland. Piketty. Statistics Norway (2005). T. UniCredit Group. A. Wolff (ed. Edward N. and Sing. Kotitalouksien varallisuus 1998. Edward Elgar Publishing Ltd. (2005). Ottawa: Statistics Canada. (1996). Review of Income and Wealth 42: 433-451. Kessler. Strauss-Kahn. 1807-1994. Wolff. Z. Edward N.no/english/subjects/05/01/ifhus_en/ Statistics Sweden (2004). Masson and D. Statistics New Zealand (2002). Wellington: Household Economic Statistics Division Statistics New Zealand. Chinese Academy of Social Sciences. in D. Where is the Wealth of Nations? Measuring Capital for the XXI Century. Wolff. ‘Chapter 1. 37 . World Bank (2005). Source OECD National Accounts Statistics. in Edward N. International Comparisons of the Distribution of Household Wealth. (1982) ‘The distribution of household wealth in some developed countries. United Kingdom and United States’. (forthcoming). Zellner. Edward N.Financial Balance Sheets . A comparative study of Sweden.. J-L. Statistics Canada (2001). ‘International Comparisons of Wealth Inequality’. Institute of Economics. ‘The Distribution of Wealth in China’. Paris: CNRS and Economica. November 2005. Örebro: Statistics Sweden.OECD (2005). Journal of the American Statistical Association 57: 500-509. (1962) ‘An Efficient Method of Estimating Seemingly Unrelated Regressions and Tests of Aggregation Bias’. Wealth Statistics 2004.ssb. The Wealth of New Zealanders: A Report on Their Assets and Debts. Renwei..). Income and Consumption 2000:26. (1987). International Perspectives on Household Wealth. R. Germany. Postel-Vinay G. Available from http://www.Stocks Vol 2005 release 02. Wealth Concentration in a Developing Economy: Paris and France. December 2003. processed. eds. France. Statistics Finland (2000). Introduction’. Income and Property Statistics for Households [Internet]. A.. S. National Accounts of OECD Countries: Volume IIIb . Denmark. Oxford: Clarendon Press. ‘New Europe Household Wealth Monitor’. (2003).. Rosenthal. ed. Wolff.

year 2000 Upper Middle High income Complete financial and non-financial data Household balance sheets Northern America Canada United States Europe Denmark France Germany Italy Netherlands Portugal Spain United Kingdom Survey data Incomplete data Financial balance sheets Austria Belgium Greece Slovenia Switzerland Survey data: Non-financial assets Number of countries with partly or Fully imputed wealth by Regression method Number of countries with imputed wealth by mean value of group 33 16 22 8 100 18 28 36 46 95. South Croatia Estonia Hungary Latvia Lithuania Slovakia Mexico 61. % of world pop.2 Korea.5 Finland China India Indonesia 56 Asia+Oceania Australia China. Poland South Africa 15.Table 1 Coverage of wealth levels data.1 Bulgaria Romania Turkey 59.Taiwan Japan New Zealand Singapore Czech Rep.3 Income Lower Middle Income Low Income Cum. .

Table 2: Comparison of full household balance sheets Consumer Country Australia Canada China. Taiwan Czech Republic Denmark France Germany Italy Japan Sector Definition households households + NPOs households households households + NPOs households + NPOs households households households Non-housing real assets yes yes yes yes yes yes yes yes (only durables) yes durables imputed Yes Yes imputed imputed imputed Yes yes imputed Netherlands New Zealand Poland households + NPOs households households no yes (only land) no imputed imputed Imputed Portugal Singapore South Africa Spain United Kingdom United States households + NPOs households households + NPOs households + NPOs households (correted for NPOs) households (corrected for NPOs) yes (only land) no yes yes (only real estate) yes yes Imputed Imputed Yes Imputed Imputed Yes Note: NPOs are non-profit organizations serving households. .

2000 Nonfinancial assets as % of total assets Household balance sheets Australia Canada China. .Table 3: Composition of household wealth in household balance sheets. Denmark France Germany Italy Japan Netherlands New Zealand Poland Portugal Singapore South Africa Spain UK USA Austria Belgium Bulgaria Croatia b b Housing assets as % of nonfinancial assets 35 46 48 na 52 49 70 86 na 83 88 78 77 86 47 88 74 80 Financial assets as % of total assets 41 57 59 34 55 40 40 42 50 54 32 20 49 45 65 31 53 67 Liquid assets as % of financial assets 22 25 39 60 21 33 34 23 53 19 35 59 47 44 21 40 21 13 22 55 88 85 26 44 43 61 53 33 76 74 52 21 62 Shares and equities as % of financial assets 20 32 32 24 54 32 37 55 16 24 40 25 38 21 20 43 25 51 20 26 5 6 54 51 43 18 44 40 21 12 31 38 31 Other financial assets a Liabilities Total assets % 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 as % of total assets 17 18 10 9 30 11 16 3 14 16 20 3 19 18 15 10 13 15 as % of financial assets 58 43 29 16 48 35 29 21 31 57 25 17 15 35 60 17 57 36 58 19 7 9 20 4 14 21 3 27 3 14 17 41 6 59 43 41 66 45 60 60 58 50 46 68 80 51 55 35 69 47 33 Financial balance sheets Estonia Greece Hungary Korea. Taiwan Czech Rep. South Latvia Lithuania Romania Slovakia Slovenia Switzerland Turkey a b b b Note: Other financial assets include insurance and pension reserves and other accounts receivable. b Composition from year 2004.

Table 4: Composition of household wealth in survey data. 2000 Nonfinancial assets as % of total assets Australia Canada China b Housing assets as % of nonfinancial assets 79 54 73 91 na 30 47 88 77 88 51 67 55 Financial assets as % of total assets 32 29 22 17 24 5 3 18 30 21 28 13 42 Liquid assets as % of financial assets 14 19 0 51 46 92 na 40 59 37 21 40 15 Shares and equities as % of financial assets 21 23 na 36 24 5 na 39 34 43 21 38 39 Other financial assets a Liabilities as % of Total assets % 100 100 100 100 100 100 100 100 100 100 100 100 100 total assets 14 16 1 12 15 3 2 2 10 27 16 9 12 as % of financial assets 64 58 Na 14 30 3 Na Na 29 20 58 22 46 68 71 78 83 76 95 97 78 70 80 72 87 58 Finland Germany India Indonesia Italy Japan Netherlands New Zealand Spain USA a Note: Other financial assets include insurance and pension reserves and other accounts receivable. . b Housing assets are net of associated debts. Liabilities exclude housing debt.

486 12.513 20.266 22.010 60.313 1.020 12. 2000 Personal Net worth per capita Household balance sheet data PPP Mean Median Coeff.326 25.313 1. Source: The Economist Intelligence Unit.319 24.251 45.145 0.338 0.691 14.461 23.480 327 11. per capita b 85.313 5.557 0.112 a a Real GDP per capita b disposable income per capita c Real Cons.504 24.658 24.619 8.006 21.427 143.485 12.727 16.663 35.619 2.406 18.031 20.197 0. Source: Penn World Table 6.708 0. of variation Highest net worth: USA Lowest net worth: South Africa Exchange rates Mean Median Coeff.661 12.210 76.479 73.798 0.425 0.724 12.594 143.914 59.231 15.727 5.921 23.916 13. In the third panel the original survey data are from close to year 2000.977 20.480 1. of variation Highest net worth: USA Lowest net worth: South Africa Survey data PPP Mean Median Coeff. national balance sheets and financial balance sheets augmented with estimates of housing assets.708 0.522 25.496 35.672 143. US$.654 0.176 0. Source for the first panel: flow of funds data.525 24.619 2.682 90. of variation Highest net worth: USA Lowest net worth: India Exchange rates Mean Median Coeff.857 1.684 12. .906 0.917 0.739 0. of variation Highest net worth: USA Lowest net worth: India 53.619 458 10.1.811 14.017 13.917 0.527 25.480 4.Table 5: Net worth per capita from household balance sheet and survey data.614 0.836 143.311 35.672 0.342 15.946 12.480 1.857 6.519 35. The numbers have been adjusted by the real growth b c rate per capita.700 25.313 240 Notes and sources: all figures are in year 2000.947 11.

per capita b Real GDP per capita b disposable income per capita c Ln (net worth per capita) a disposable income per capita) Ln (real cons.000 1.958 0.000 0. Ln (net worth per capita) PPP Exchange rates Survey data Correlations with: A. US$.Table 6: Correlations of net worth per capita with income per capita.908 0.975 0.841 0.979 .866 0. Net worth per capita PPP Exchange rates B.955 0. per capita) b per capita a 1.890 0.940 1. Net worth per capita PPP Exchange rates B.979 0.000 1.000 0. 2000 Personal Net worth Country Household balance sheet data Correlations with: A.000 0.000 0. Ln (personal Real Cons.938 0.922 1.993 0.739 0.000 1.848 0.920 0.825 0. See Table 5 for technical notes.860 1.715 0. Ln (net worth per capita) PPP Exchange rates Note: all figures are in year 2000.923 0.

683# (1.506** (. *-5% level.487** (.325 23 .403) -1.041) .158) (2b) -2.469) .735) 1. .720** (. rate (1b) .9501 .233) 2 2 Dependent variables Log of financial wealth (2a) -2.177** (.578** (.079) Log of population density .268** (.191** (.121** (.9330 .516** (. In the non-financial assets regression.922) 1.865** (.621 38 .153) Log of domestic credits available to the private sector Income Gini -.465** (. Standard errors in parenthesis.735) 1.467 38 .334) -. Significance levels: # -10% level.527) 1.053) .822** (. Ordinary Least Squares are used.115 (.005 (.949) Log of real cons per capita 1. **-1% level.018) -1. The sample for the financial and liabilities regressions consist of 34 countries with HBS or financial balance sheet data and 4 with survey data.151) Log of liabilities (3b) -8.038** (.916** (.832# (.659** (.548 (0.110) (2c) -2.006 (.440 34 Note: The sample for the non-financial regressions consists of 18 countries with HBS data and 5 with survey data. In the financial assets and liabilities regressions.9378 .9331 .400) R .874) 1.044) Log of market cap.110) (3a) -9.119* (.164) -.9320 .748** (.502** (.335) -1.9495 .9501 .535** (.098) .459) 1.219** (.132) Log of pensions as a % of GDP -.326** (.591 34 .137) (3c) -8.435 (.336 23 .114) .009) Survey dummy .118 (. the Seemingly Unrelated Regressions (SUR) estimation method is used.737) 1.098) .427** (.172) .028** (.625 38 "R " RMSE Sample size .280 (.467 38 .9513 .Table 7: Regressions of wealth components Independent variables Log of non-financial wealth (1a) Constant .171** (.183) .

379 83.74 13.01 12.74 15.184 17.456 8.233 3.264 1.708 0.385 1.308 0.342 0.498 20.309 1.260 115. of variation (weighted) World totals (229) PPP Coeff.37 21.485 0.626 0.73 27.233 27.87 5.692 8.328 76.186 1.747 5.35 population 14.342 123.445 0.54 4.367 0. of variation (weighted) Low income (64) PPP Coeff.298 33.93 91.769 39.285 0.436 0.29 33.893 1.24 11.278 153.81 113.534 147.565 0.08 2. of variation (weighted) Exchange rates Coeff.277 1.093 6.446 0.475 0.058 0.878 9.372 0.249 0.748 0. of variation (weighted) Exchange rates Coeff.421 1. of variation (weighted) Exchange rates Coeff.520 0.23 9. of variation (weighted) Exchange rates Coeff.017 0. of variation (weighted) Exchange rates Coeff.406 3.91 0.675 0.92 53.524 0.628 1.573 0.602 5. of variation (weighted) Exchange rates Coeff.Table 8: The distribution of net worth by World Bank country group.55 Net worth per capita Net worth per adult Percent of world net worth per capita Real GDP per capita Percent of world GDP Note: all figures are in year 2000 US$.950 0. of variation (weighted) 26.824 0.551 2. 2000 Share of world World Bank country group High income OECD (24) PPP Coeff.241 0.205 1.29 63.442 0.310 5.170 4.23 21. of variation (weighted) High income non-OECD (43) PPP Coeff.485 0.141 43.430 0.233 0.285 15.484 81.67 3.02 18. of variation (weighted) Lower middle income (59) PPP Coeff. .413 0. of variation (weighted) Upper middle income (39) PPP Coeff.590 0.626 19.553 33.28 2.328 10.240 3.372 3.968 0.33 9.90 11.277 9.97 8. Figures are weighted by country population.377 151.

6 16.6 35.7 31.7 19.4 52.0 32.5 1.0 90% 56.5 2.6 0.9 8.0 12.Table 9: Wealth shares for countries with wealth distribution data.1 51.5 41.1 1.0 44.0 30.2 -4.0 2.5% 0.7 -6.0 69.3 43.0 0.0 57.6 2.6 58.0 50% 9.7 36.6 71.0 28.2 7.7 10% 0.0 34.0 75% 80% 38.9 3.2 42.9 19.4 24.8 6.9 8.3 2.0 40.0 38.6 49.1 21.8 -6.4 15.0 25.1 0.3 0.8 10.8 58.8 23.3 39.0 44.8 25.6 14.4 13.0 53.7 28.0 25.7 50.6 57.8 4.0 44.3 56.9 48.9 -0.7 10.6 29.2 1.0 47.7 5.0 -0. South New Zealand Norway Spain Sweden Switzerland UK USA Year 2002 1999 2002 1975 1998 1994 1998 Unit household family unit person household household person? household -0.3 20.4 0.6 5.3 2.5 28.4 36.3 0.2 0.5 1.9 18.9 -6.1 40.0 20% 0.5 60.0 38.7 47.2 9.9 58.0 0.0 41.0 40% 4.8 27.3 61.8 13.5 39.2 0.0 5% 32.1 16.4 17.4 42.7 2.4 15.8 9.5 34.8 4.3 56.8 2.0 70% 25.1 12.2 21.1 41. official exchange rate basis share of lowest Country Australia Canada China Denmark Finland France Germany India Indonesia Ireland Italy Japan Korea.6 7.6 7.household 1997 1987 2000 1999 1988 2001 2000 2002 2002 1997 2000 2001 household household household household household tax unit household household household family adult family .1% 2002.5 4.0 14.0 8.1 5.4 28.9 -0.4 20.8 -0.9 65.9 29.9 34.7 18.3 48.7 51.2 34.5 0. 27.0 55.9 -0.0 31.2 10% 45.4 65.5 18.9 12.0 30.3 6.3 48.1 34.0 0.7 39.7 2.8 5.3 2.0 60% 16.4 13.7 56.1 -5.7 7.0 42.0 58.0 25% 30% 1.0 share of top 2% 1% 0.0 57.

8 0.9 27.2 100 .9 36.2 5.5 0.7 0.4 17.1 5.5 6.5 0.4 14.9 22.5 1405 0.5 12.0 29.3 890 0.2 0.2 0.6 39.9 22.4 4.5 11.9 19.2 5.3 9.4 1.6 27.3 1.8 85.1 8.9 10.5 1.3 100 2.0 0.7 7.4 29.6 6.7 7.0 36.1 0.2 16. official exchange rate basis Adult pop.8 6.5 8.8 29.5 0.6 0.8 15.4 1.4 10.8 14.4 38.03 0 0.4 0.6 100 190653 18163 67232 3885 1989 172414 5952 3558 72874 33893 34.4 100 8.5 2.8 27.4 0.3 0.7 30.5 17.7 17.4 100 14.0 0.0 0.8 8.3 4.8 2161 1.4 3.2 100 4.2 100 27.0 12.6 3.2 16.2 0.1 0. Decile1 Decile2 Decile3 Decile4 Decile5 Decile6 Decile7 Decile8 Decile9 Top10 Top5 Top1 Wealth Wealth share (000s) Pop.2 26.Table 10a: Global wealth distribution in 2000.1 26.0 0.0 24.9 6318 14169 61041 150145 514512 0.2 14.2 0.9 6.1 14.8 6.5 100 39.7 17.0 14.1 7.5 0.8 35.1 24.1 1.2 100 0.7 8.3 100 1.2 0.8 40.3 2.6 16.2 464 0.8 16.2 4.8 0.3 0.7 2.6 7.1 1.0 100 28.3 14.09 192 0.2 8.1 27.3 0.1 35.1 70.8 100 225719 302915 550579 842063 570595 166532 642421 376292 20405 3697518 6.2 100 0.0 13.4 0.2 12.2 0.5 10.1 8.0 24.9 2.9 3.4 3517 2.9 9.8 7. Share per adult World wealth shares Minimum wealth Population proportions by region North America Latin America and Caribbean Europe Asia: China Asia: India Asia: high income Asia: other Africa Oceania World 0.2 19.6 0.3 100 1.1 2.0 31.4 7.

1 0.8 0.0 0.4 1.1 46416 43871 44358 32165 22764 13690 15476 33242 12046 104213 56132 23307 5497 5.2 2.9 1.9 0.2 29.0 7.1 100 4.680 0.5 0.6 0.8 3.2 0.1 2.9 2.0 2.709 0.3 0.3 3.8 15.6 1.6 1.4 3.0 100 100 40391 842063 570595 0.5 1.9 0.9 3.8 1.1 26.5 5.7 64810 5.1 0.4 0.5 100 11.9 1.2 3.0 2.669 0.0 3.6 0.1 1.7 0.4 0.6 0.0 0.5 4.7 36. Pop.1 5.658 0.2 0.5 3.0 1.1 0.1 1.2 1.1 1.7 1.8 40.9 20.9 2.7 1.7 0.9 0.4 2.7 1.1 0.6 2.1 0.1 2.4 2.1 22.649 0.7 0.6 0.8 1.9 1.4 1.1 124446 51431 67968 66483 44025 100 100 3697518 Note: blank cells indicate values less than 0.7 0.8 1.2 1.1 0.4 0.5 3.5 2.6 14.2 0.0 1.5 0.9 2.4 0.0 0.8 0.0 1. selected country details.3 0.7 7.6 0.0 0.1 0.5 35.1 26.4 202865 25.8 1.892 0.6 0.0 14.8 27.609 0.9 6.1 2.8 100 7.8 1.8 1.2 0.4 1.5 1.7 0.801 0.3 1.8 1.697 0.0 0.1 1.9 2.7 37.1 0.717 0.3 2.9 4.2 0.9 4.7 0.3 0.671 0.8 1.5 0.622 0.1 0.1 0.1 7.9 0.1 0.5 0.579 0.05 per cent.3 0.4 5.5 0.5 2.1 0.2 1.3 1.5 1.2 1.3 4.5 1.6 5.5 0.2 0.1 2.9 0.7 1.6 0.3 0.1 3.6 4.654 0.3 1.6 100 0.8 0.6 0.3 2.5 3.748 0.1 3.7 4.6 0.9 0.3 5.0 0.8 100933 9.8 0.4 3.5 2.6 2.735 0.5 0.1 0.663 0.1 107493 40160 0.3 3.1 100 0.3 100 2.5 1.5 1.1 3.8 3.2 100 201319 227600 109735 122250 169617 114650 86958 95606 94712 105613 41777 144406 15958 23305 40225 222641 16218 3885 1989 4140 6717 2421 862 2633 2424 1986 33893 32.1 38. Wealth per Wealth Gini USA Japan Germany Italy UK France Spain Canada Australia China.3 3.3 0.2 1.0 2.5 1.0 4.2 0.9 19.8 0.7 1.5 11. .9 1.2 0.9 0.0 0.0 16.2 1.4 0.9 0.5 4.9 0.4 2.3 1.3 2.1 1.2 100 24.3 2. Taiwan Korea.4 0.2 0.547 0.9 27.740 0.4 2.1 0.3 1.0 0.5 2.0 1.4 1.7 0.730 0.550 0.5 2.1 1.8 0.1 0.0 100 2.9 2.1 1.8 1.1 2.4 1.9 0.698 0.3 1.7 0.Table 10b: Global wealth distribution in 2000.6 18.2 0.4 2.4 0.2 1.8 16.3 5.4 1. official exchange rate basis Decile1 Decile2 Decile3 Decile4 Decile5 Decile6 Decile7 Decile8 Decile9 Top10 Top5 Top1 Adult pop.6 27.0 1.9 0.4 0.697 0.1 0.4 0.4 0.3 100 1.2 0.4 0.2 1.0 0.2 0.2 1.3 2.6 2.3 3.2 1.6 0.6 6.8 0.565 0.1 0.3 0.7 0.0 3.5 1.5 1.2 8.6 0.1 0.2 1.2 6.1 0.2 0. South Netherlands Brazil Mexico Argentina Switzerland Turkey China India Russia Thailand Indonesia Nigeria Pakistan Bangladesh Viet Nam World 0.7 1.2 0.803 0.2 0.5 7.1 0.5 0.2 0.7 1.6 1.7 1.4 0.8 0.1 3.6 0.3 1.0 4.1 3.9 1.7 5.5 2.5 1.3 1.8 4.783 0.6 5.2 0.6 0.2 0.763 0.

7 4.6 1.1 9.4 14.8 25.3 1.2 100 6.6 27.4 100 225719 302915 550579 842063 570595 166532 642421 376292 20405 3697519 6.1 11.1 0.5 17.2 28.4 11.0 18.5 8.0 12.4 100 2.9 22.2 1.2 0.1 2 0.0 4.0 0.3 100 .3 22.6 9.4 10.6 1978 1.0 0.1 3693 1.1 57.4 10.5 100 3.2 18.1 6.1 10.2 2.1 1.2 0.7 6.1 0.3 0.3 826 0.3 100 4.6 6.5 7.7 1.0 34.3 6.2 6.1 17.6 5724 2.3 21.3 0.1 31.3 24.9 0.5 17.2 0.2 100 5.6 0.9 0.5 18. PPP values Adult popn Decile1 Decile2 Decile3 Decile4 Decile5 Decile6 Decile7 Decile8 Decile9 Top10 Top5 Top1 (000s) Popn Wealth Wealth share share per adult World wealth shares Minimum wealth 0.9 1.1 14.7 9.5 6.3 100 9.0 15.0 0.9 4.3 2.0 7.5 16.8 21.1 14.2 8.7 20.9 71.3 9.7 0.8 1.9 100 25.9 8.2 12.0 5.5 12.0 31.0 100 39.3 19.8 9.4 17.7 6.1 8.4 35.1 3.9 23.6 100 193147 34956 81890 16749 11655 138750 18266 11730 99634 43628 27.0 100 21.2 1.7 9.4 1.0 17.4 100 3.2 14.1 32.4 4.1 2.0 17.7 2.0 6.0 22.0 4.4 2.2 14.6 6.8 15.2 9.3 37.9 35.Table 11a: Global wealth distribution in 2000.6 8399 12749 20299 35954 88035 170467 523264 Population proportions by region North America Latin America and Caribbean Europe Asia: China Asia: India Asia: high income Asia: other Africa Oceania World 1.8 2.2 100 3.2 14.7 8.2 21.7 1.6 5.1 29.2 15.1 33.7 6.4 3.3 6.6 9.3 7.1 14.

9 0.5 36.9 25.0 8.7 1.2 18. Wealth per Wealth Decile1 Decile2 Decile3 Decile4 Decile5 Decile6 Decile7 Decile8 Decile9 Top10 Top5 Top1 1.9 0.0 1.2 0.5 0.1 1.6 0.1 0.2 0.2 1.6 8.0 1.2 1.730 35188 2.4 0.7 4.8 5.7 2.3 0.9 3.697 12389 0.4 0.9 7.3 0.6 4.1 0.1 1.0 2.5 40391 1.3 0. South China.1 0.801 157146 9.0 6.1 1.7 1.709 187998 0.7 0.2 1.3 0.2 0.4 0.4 0.5 3. Pop.2 0.2 0.1 0.4 22764 0.1 0.3 0.4 29.7 2.5 1.6 1.9 2.2 0.6 0.4 1.4 0.3 0.6 2.5 1.6 3.1 0.6 5.8 1.8 1.5 37.1 23307 0.2 0.2 11.9 0.8 0.763 22678 0.1 19.3 0.8 43871 1.5 34.2 2.1 0.4 3.7 3.803 13214 0.654 126635 1.3 1.8 2.9 3.4 0.3 2.5 1.6 0.9 0.9 3.7 0.7 0.3 6.8 570595 15.9 1.7 0.1 3.5 7.9 0.1 0.5 5.4 0.2 1.7 3.2 0.5 3.1 0.6 0.669 120326 1.6 2.2 1.4 20. .717 60917 0.9 12046 0.9 1.2 32.2 0.8 1.4 1.9 0.2 0.9 0.9 1.0 2.4 0.3 0.9 2.5 1.3 0.6 3.1 16.7 2.680 2194 0.9 6.5 3.1 0.7 1.2 51431 1.6 0.4 3.7 1.0 0.9 0.1 0.7 0.9 1.1 7.565 125254 3.7 3.1 5.550 116782 2.6 1.1 0.8 66483 1.4 0.7 1.4 0.3 0.0 0.8 0.4 1.7 0.0 0.2 46416 1.0 1.4 40160 1.0 21.4 0.3 2.9 0.2 0.1 2.735 43628 100 0.3 0.9 1.7 2.7 1.1 0.2 0.2 1.802 USA Japan Germany UK Italy China Spain France Brazil India Canada Korea.6 33242 0.9 44358 1.3 0.3 0.3 2.740 13401 1.5 0.5 0.5 2.7 0.6 3.8 0.8 2.5 1.6 1.2 0.748 40202 1.4 0.2 2.1 0.7 4.0 2.2 2.8 3.9 2.3 0.3 3.3 8.3 0.2 0.4 1.3 0.2 0. PPP values: selected country details Adult pop.609 16749 8.1 0.547 114185 4.3 0.4 3697519 100 adult share Gini 201319 25.3 1.3 15.8 44025 1.8 1.1 0.7 2.649 38324 1.6 3.1 0.1 0.5 0.6 0.3 0.3 0.4 0.8 1.7 1.9 1.9 1.9 6.7 0.8 32165 0.6 0.3 0.05 per cent.697 148843 4.8 0.6 0.3 1.5 0.1 1.4 0.8 14.1 100 100 100 100 100 100 100 100 100 100 100 100 (000s) share 202865 5.2 1.9 2.0 2.658 10066 0.2 0.5 0.0 1.9 0.7 64810 1.5 100933 2.1 0.6 0.4 0.1 0.8 0.3 0. Taiwan Australia Russia Netherlands Mexico Turkey Argentina Indonesia Thailand Switzerland Pakistan Bangladesh Viet Nam Nigeria WORLD Note: Blank cells indicate values less than 0.3 14.671 172461 4.1 0.2 0.3 0.1 1.0 1.8 5.1 3.3 0.2 104213 2.3 14.0 2.0 1.4 2.7 0.0 1.0 0.0 2.5 0.8 0.7 3.1 0.7 1.3 21.1 1.6 2.7 0.4 0.8 33.1 5497 0.2 15.1 2.2 1.6 0.2 0.5 0.9 1.2 2.4 1.4 13690 0.8 14.579 143405 1.1 3.3 842063 22.9 3.2 0.Table 11b: Global wealth distribution in 2000.9 15476 0.6 23.1 1.4 2.6 124446 3.1 0.6 2.698 158484 1.6 0.663 64521 1.4 0.1 0.9 11.1 0.5 1.8 2.1 0.0 2.1 67968 1.0 0.6 0.9 5.5 0.4 1.2 0.783 11655 4.7 1.2 2.2 0.7 1.9 1.1 1.6 0.2 1.9 0.4 2.622 24011 1.3 56132 1.9 0.4 0.4 107493 2.9 2.0 1.2 0.5 0.3 2.9 5.5 0.1 0.

7 58.7 93.879 .5 67.3 1.3 4.0 39.5 1.8 93.6 98.5 0.890 0.6 57.8 80.9 84.7 70.8 84.7 82.1 0.2 4.3 4. PPP valuations 228 228 3697518 3697519 100 100 33893 43628 100 100 0.9 39.8 0.3 1.1 0.1 0.1 71.8 97.5 70.891 0.7 0.5 0.7 83.6 0.3 1.6 39.7 85.5 0.8 98.1 1.0 63.0 84.887 0.3 94.7 48295 48295 83.1 0.Table 12: Global wealth distribution under alternative assumptions World wealth shares Number countries Adult population Share of global adult population (%) Wealth per Share of global adult wealth Quintile1 Quintile2 Quintile3 Quintile4 Quintile5 (%) Top10 Top5 Top1 Gini (1) (2) All countries.5 0.4 93. official exchange rates All countries.1 70.9 4.1 0.7 1.2 9.4 0.802 Excluding regional average imputations for: (3) (4) (5) — average wealth level — wealth distribution — average wealth level and distribution 150 144 129 3540759 3596803 3491133 95.0 4.0 3.1 70.892 0.2 93.4 34960 34356 35281 98.9 37.3 0.2 4.1 39.4 32.9 31.9 93.890 Countries with wealth distribution data (6) (7) — using reported wealth distributions — imputing from income distributions 20 20 2171089 2171089 58.9 94.0 85.1 0.5 1.

Figure 1: Net worth from household balance sheet versus disposable income. official exchange rates Net worth per capita/Fitted values 50000 100000 150000 200000 JPN USA GBR NLD ITA SGP ESP FRA GER CAN AUSDNK NZL PRT CZE POL ZAF 0 0 5000 10000 15000 20000 Personal disposable income per capita Fitted values 25000 Net worth per capita .

official exchange rates Net worth per capita/Fitted values 50000 100000 150000 USA JPN AUS ITA NZL ESP CAN GER NLD FIN CHN IDN IND 0 0 5000 10000 15000 20000 Personal disposable income per capita Fitted values 25000 Net worth per capita .Figure 2: Net worth from surveys versus disposable income.

top tail of estimated world distribution of wealth 100000 number above (000) 10000 1000 100 250000 500000 1m 2.5m wealth level ($) 5m 10m .Figure 3: Pareto graph.

House values and equity in unincorporated business are estimated using the perpetual inventory method. and France the balance sheets. assets and liability totals for the household sector are reviewed in light of that data and sometimes adjusted accordingly. The objective is to estimate the balance sheet of the household sector as of a certain date—often but not always the year-end. In the direct approach data are collected from household members through sample surveys. are compiled as part of the System of National Accounts (SNA). Czech Republic. or by updating previous stock figures by adding estimates of subsequent flows. The OECD publishes the financial balance sheet of the household sector for its member countries (OECD. and Turkey. 2006). What is important is not who delivers the HBS data. who puts them together. is used for this purpose. for example. however. 2005). In the indirect approach the values for the household sector by asset type are calculated as residuals from independent totals by deducting the estimated holdings of other sectors. In some countries. sectoral definition. and how the data are constructed. Often the independent totals are counterpart data. 1 Financial balance sheets for the household sector have also been published for eight countries in ‘New Europe’ by a financial group based in Italy. or at least their major elements. Poland. fully comparable household balance sheets including non-financial assets are available for only a subset of OECD countries. That is. censuses or administrative records.Appendix I: Household Balance Sheet (HBS) Data: Methods There is considerable variation between countries in how household balance sheets are constructed. In other cases central banks or national statistical agencies issue HBS data independent of the SNA or FOF. the UK. but who and what are covered. Both stocks and flows can be measured by direct or indirect means (Aron et al. Land values are updated based on an index of land prices. Croatia. Romania. and so it is the only one of these countries whose balance sheet numbers are reported here. Canada and Japan. which is conducted every three years by the Federal Reserve Board. This can be done by trying to measure the relevant stocks. they are assembled together with Flow of Funds (FOF) data. For consumer credit no deductions are necessary. In most countries a mix of direct and indirect approaches is used. The Survey of Consumer Finances (SCF). most categories of financial assets and liabilities are calculated using the indirect approach (Board of Governors of the Federal Reserve System 2003). amounts held or owed by the other sectors are subtracted from known totals and the remainders are assumed to be the amounts held by the household sector. When micro-data are available. Elsewhere. In the United States. .1 As discussed below. The countries covered are Bulgaria. Hungary. and asset coverage. We were only able to obtain nonfinancial data for the Czech Republic. for example Australia. See UniCredit Group (2005). for example the US. for example the liabilities of the banking sector in the form of deposits. Slovakia.

whereas unlisted company shares are estimated using the estate multiplier method. However. (Brandolini et al. credit institutions. But where unincorporated business is treated as a separate sector. Where the sectors are combined. 2 If unincorporated business is treated as a separate sector. . in many cases. unincorporated business is included in the household sector. Benchmark values of the stock of dwellings and the share owned by households are obtained using census data. Debts comprise all short and long term liabilities and are estimated using the same approach. household wealth will include equity in unincorporated business. Some of the net capital stock estimates included in non-financial assets are calculated using the perpetual inventory method (Aron et al. The series is corrected to be expressed in market prices using a housing price series based on information assembled in a semiannual survey of real estate agents. The stock of durables is computed using the perpetual inventory method. Residential housing stock estimates are compiled using property tax records of local authorities. Hence households will tend to appear wealthier where the sectors are combined. in the household balance sheet. Apparent composition of household wealth will also be affected. only durables are included in fixed assets. where these sectors are separated.In the United Kingdom. financial assets are measured using the residual approach and are based on data from banks. The value of fixed assets is derived using a number of sources. Sometimes the household sector only includes households. Ordinary shares are estimated with the help of a sample survey. insurance companies etc. The stock of dwellings is based on a series provided by the Italian Statistics Office given at constant rather than market prices. as we see below. Households are the owners of unincorporated business. Sectoral definitions unfortunately are not the same across all countries. Apart from housing assets. we will just see a single entry. unincorporated businesses are in practice treated as having some net worth of their own. 2006).2 However. In principle this means that whether unincorporated business is included with households or not should not affect estimates of the total wealth of households. it is also common to include non-profit organizations (NPOs) that serve the household sector. the Share Registers Survey. 2004). Data on life assurance and pensions funds are obtained from returns made by insurance companies and pension funds. business assets and debts are separated into the various relevant categories. Also in Italy. business equity. financial assets are measured mostly using the residual approach based on data from banks and other financial intermediaries. Fisheries and Food. and farm land and buildings are estimated from data on farm sizes and prices from the Ministry of Agriculture.

Belgium OECD (2005). National Statistical Office http://eng. (Quoted from Central Bank Flow of Funds Statistics) UniCredit Group.Stocks Vol 2005 release 02.abs. National Balance Sheet Accounts 2000. National Accounts of OECD Countries: Volume IIIb .gov.asp?ctNode=1640&CtUnit= 789&BaseDSD=7 Same as for financial data Non-financial data Same as for financial data Data combined by Link to data http://www.02004-05?OpenDocument .nsf/Detail sPage/5204.gov. National Accounts and Balance of Payments. Source OECD National Accounts Statistics. Table ST01415. National Accounts of OECD Countries: Volume IIIb . Table 7.Financial Balance Sheets . Statistical Tables. Sector Accounts Times Series.statbank. Annual National Accounts ESA95. Source OECD National Accounts Statistics. Statistics Denmark.asp?w=1 024 National Wealth Statistics. Household Balance Sheet and Financial Balance Sheet Sources Country Australia Financial data Australian Bureau of Statistics (2005). Statistics Canada.dk/statbank5a/default.czso. 5204. Annual National Authors Same as for financial data http://dw. ’New Europe Household Wealth Monitor’. Table 7.cz/pls/rocenka/rocenkavyber. National Wealth Statistics.0 Australian System of National Accounts Table 51. November 2005.casrady _sek_en http://www.stat.au/AUSSTATS/abs@. Statistical Tables. Taiwan Croatia Czech Republic Denmark UniCredit Group.Stocks Vol 2005 release 02. ’New Europe Household Wealth Monitor’. Czech Statistical Office (2006). November 2005. National Accounts and Balance of Payments. Austria OECD (2005). Bulgaria Canada China.Financial Balance Sheets .Appendix II A.tw/lp. Balance Sheets for Financial Assetes and Statistics Denmark. CanSim Matrix 0751.

Same as for financial data http://www.Base 2000.eu. Table 4.. National Accounts of OECD Countries: Volume IIIb .Liabilities Table NAT10 Accounts ESA95. Source OECD National Accounts Statistics. Estonia Eurostat Financial Balance Sheets http://epp. Adjusted Italian Statistical Office data. Fixed Capital Table Nat14.cec. A. and Faiella I.Financial Balance Sheets .18. Table 2.Special Statistical Publication. Financial Accounts for Germany 1991 to 2003. Table 4515.int/portal/page?_pageid=1 996. Comptes Nationaux Annuels .eurostat. D’Alession G. 86. Authors Greece Hungary Eurostat Financial Balance Sheets OECD (2005). Japan Economic Planning Agency. Real assets data provided by Deutsches Bundesbank. Government of Japan. Termi di discussione No 530. p.Stocks Vol 2005 release 02. Rome: Bank of Italy. Household Wealth Distribution in Italy in the 1990s. Table XI p..45323734&_dad=portal&_schema=PORTAL&s creen=welcomeref&open=/fina/fina_st&language=e n&product=EU_MASTER_financial&root=EU_MAST ER_financial&scrollto=0 France INSEE. Same as for financial data . (2004).htm Germany Deutshe Bundesbank (2004). Brandolini.fr/fr/indicateur/cnat_annu/base_20 00/tableaux/comptes_patrimoine.insee. see Estonia Italy Financial Accounts by the Bank of Italy. Annual Report on National Accounts. Cannari L.

Singapore Department of Statistics (2003): Wealth and Liabilities of Singapore Households Same as for financial data . Housing Privatization and Household Wealth in transition. Financial Stability Report 1/2003.html#P320_25575 http://www.1. Household Financial Assets and Liabilities. Authors Authors see Estonia see Estonia http://www. Paper prepared for the UNU-WIDER project on "Personal Assets from a Global Perspective". National Accounts of OECD Countries: Volume IIIb . Romania Singapore UniCredit Group. p.Financial Balance Sheets . Statistics Netherlands. Cardoso F. November 2005. Household Wealth in Portugal 19982004. Same as for financial data Bank of Norway.23. Financial Accounts of the Household Sector. van den End W. http://www.nl/NR/exeres/7EA6458D-70C346B2-83E7-93166787810C New Zealand Norway Reserve Bank of New Zealand.Stocks Vol 2005 release 02. Bank of Portugal WP 4-05. Source OECD National Accounts Statistics. Latvia Lithuania Netherlands Eurostat Financial Balance Sheets Eurostat Financial Balance Sheets van Els P. (2005). Table 1. ’New Europe Household Wealth Monitor’. BIS papers No 22.norges-bank.rbnz. (2005). Table 1 p. National Accounts of OECD Countries: Volume IIIb . and van Rooij M. Housing stock estimates calculated based on data from the Central Statistical Office and the Bank of Portugal.norgesbank. Yemtsov R: (2006). Table A1.no/front/rapport/en/fs/200301/ http://www.govt.cbs.Korea OECD (2005). Table 3. Financial Behaviour of Dutch Households: Analysis of DNB Household Survey Data 2003. Portugal Financial Accounts by the Bank of Portugal. Sector Accounts: Financial Balance Sheets Table 4b.Financial Balance Sheets .G. Source OECD National Accounts Statistics. 41. and da Cunha V.Stocks Vol 2005 release 02.no/front/statistikk/en/husholdningene/ Authors Poland OECD (2005).2. Bank of Norway (2003)..nz/statistics/183499807.

uk/STATBASE/Product.html Turkey UK UniCredit Group. Table B.Stocks Vol 2005 release 02.bde.snb. Swiss Financial Accounts. ’New Europe Household Wealth Monitor’.Slovakia Slovenia South Africa UniCredit Group. Same as for financial data http://www.gov.html?file=/e/ publikationen/monatsheft/aktuelle_publikation/html/e /inhaltsverzeichnis.statistics. Table II. National Accounts of OECD Countries: Volume IIIb .htm http://www. United Kingdom National Accounts.1N. Source OECD National Accounts Statistics. CSAE WPS/2004-24.ch/e/publikationen/publi.Financial Balance Sheets .gov.htm Sweden OECD (2005).50.5. Summary indicators: Household market indicators. Financial Accounts of the Spanish Economy. (2004). ’New Europe Household Wealth Monitor’.asp ?vlnk=1143 USA Federal Reserve Statistical Release (2005). I. Table 2. Blue Book Time Series Data Table 10. Release Z. 2005.10.asp ?vlnk=376 http://www. November 2005.statistics. November 2005. Authors http://www. Flow of Funds Accounts of the United States. and Muellbauer J.es/estadis/ccffe/cfcap2e.gov/releases/z1/ .bde.federalreserve. p.100. Financial Statistics Time Series Data Table 12. National Statistics Online. Revised Estimates of Personal Sector Wealth for South Africa. Same as for financial data Spain Banco de Espana (2005). Eurostat Financial Balance Sheets Aron J.uk/STATBASE/Product. June 9. http://www. Table T11. Switzerland Swiss National Bank.es/infoest/sindie.e. Authors http://www. Banco de Espana.

(1992). and Labour Dynamics. See Banco de Espana (2005). See Statistics Finland (2000). Household Saving Survey. Survey of Household Income and Wealth. See Brandolini et al. See Leipziger et al. See Ammermuller et al. See Spånt (1982). Poverty and Usage of State Services. Einkommens und verbrauchstichprobe. See Headey. See Statistics New Zealand (2002).B. et al (2003). Indonesia Family Life Survey (own calculations). See Piketty. See Kennickel (2003). Survey of Financial Security. National Survey of Family Income and Expenditure. Korea Development Institute Survey. Survey based on county wealth tax statistics. See Statistics Norway (2005). (2004).rand. et al (2005). Inland Revenue Statistics. See Inland Revenue Statistics (2005). Encuesta Nacional sobre Niveles de Vida de los Hogares. Survey of Household Finances. Wealth statistics based on registers of total population. See Jäntti and Sierminska (2006). . Household Wealth Survey. Estate tax returns. See Japan Statistics Bureau (2005). http://www. Wealth tax records. (2005).org/labor/FLS/IFLS/ The survey of Income Distribution. See Statistics Canada 2001). China Academy of Social Science Survey. See Nolan (1991). All-India Debt and Investment Survey (NSS 59th round). See Dell et al (2005). See National Sample Survey Organization (2005). See Renwei and Sing (2005). Survey Sources Australia Canada China Denmark Finland France Germany India Indonesia Ireland Italy Japan Korea Mexico New Zealand Norway Spain Sweden Switzerland United Kingdom United States 2002 1999 2002 1975 1998 1994 1998 2002 1997 1987 2000 1999 1988 2002 2001 2000 2002 2002 1997 2000 2001 Household Income. Survey of Consumer Finances 2001. Income and Property Distribution Survey. See Statistics Sweden (2004).

289 23 .172) 1. *-5% level.9521 .410) . the Seemingly Unrelated Regressions (SUR) estimation method is used.280 23 . The sample for the financial and liabilities regressions consist of 29 countries with HBS or financial balance sheet data and 4 with survey data.387** (.118) 1. income per capita 1.934** (.008) -.012 (. In the financial assets and liabilities regressions.407 (.158) Income Gini -.477) Log of financial wealth (2a) -2. Ordinary Least Squares are used.784** (. .298 (.197 (.017) Survey dummy .683** (.148) Log of domestic credits available to the private sector .007 (.252** (.0821 (.9545 .9598 .068) 1.132) .161) 1.688) Log of disp.481 (.71** (.540 33 .472** (.070** (.215 (.441** (. rate .103) Log of population density . In the non-financial assets regression.626* (1.085* (.Appendix III Appendix Table 1: Regressions of wealth components Independent variables Dependent variables Log of non-financial wealth (1a) Constant .397) (2b) -3.049) 1.102) Log of pensions as a % of GDP -.501** (.410 29 Note: The sample for the non-financial regressions consists of 18 countries with HBS data and 5 with survey data.495 29 0. **-1% level.875) (3b) -9.784) (1b) .246** (. Standard errors in parenthesis.036) . Significance levels: # -10% level.9458 -1.076** (.9662 "R " RMSE Sample Size .146 (.340) -.805) Log of liabilities (3a) -10.196) -.036) Log of market cap.452 33 0.407) R 2 2 .941* (.137) 1.092** (.9565 .

406 2.913 15.252 35.032 14.614 23.885 4.557 12.357 113.916 20.916 2. 2000 Personal disposable Net worth Country PPP Household balance sheet data Australia Canada China China.603 10.614 14.191 19.534 100.523 11.037 12.145 12.691 12.876 25.924 25.332 19.631 16.603 15.754 24.160 93.654 53.849 18.661 18.857 35.384 11.613 20.378 119.708 Survey data 75.856 43.380 9.544 13.008 17.973 74.759 20.266 92.009 32.486 15.872 23. Taiwan Czech Republic Denmark Finland France Germany India Indonesia Italy Japan Netherlands New Zealand Poland Portugal Singapore South Africa Spain UK USA Exchange rates Household balance sheet data Australia Canada China China.034 16.169 15.198 9.871 23.Appendix IV Appendix IV Table 1: Net worth per capita from household balance sheet and survey data.193 28.708 1.534 6.965 5.672 16.539 na 8.828 1.759 20.797 70.798 12.210 13.732 15.480 24.917 2.505 na 2.486 1.796 15.684 4.348 12.253 128.975 16.190 11.603 52.496 12.282 3.252 27.700 14.994 3.937 per capita a Personal disposable Real cons.169 15.188 67.983 17.034 7.917 14.086 55.731 5.798 12.876 25.089 28.197 90.159 13.751 14. per capita c Real cons. per capitac Real GDP per capita c income per capita d Net worth per capita a Real GDP per capita c income per capita d .416 11.983 17.313 60.338 22.714 14.032 910 14.823 24.951 38.794 6.942 30.338 22.990 70.513 2.959 143.954 14.267 27.008 15.729 89.994 Survey data 101.731 15.193 28.597 72.704 124.644 8.145 12.544 17.102 25.285 15.906 89.732 13.924 25.661 3.858 120.017 19.496 12.986 57.828 11.913 15.619 15.160 18.035 22.195 15. Taiwan Czech Republic Denmark 73.661 17.975 16.493 10.513 7.614 19.195 15.434 61.480 14.159 13.954 14.519 4.217 91.313 143.619 25.008 9.844 28.238 24.654 10.727 22.037 24.603 1.431 66.

026 10.977 1.443 11.171 85.799 52.852 35.1. Source: Penn World Table 6.857 60.619 a 37.544 23.438 33. The numbers have been adjusted by the real growth rate per capita.443 11.613 41.547 23.614 22.419 10.693 126.799 240 472 11.915 1.724 9. .048 35.137 14. national balance sheets and financial balance sheets augmented with estimates of housing assets.547 23. US$.112 1.735 14.976 2.336 22.048 23.313 Note: all figures are in year 2000.957 133.261 13.977 68.557 7.960 5. Source for the first panel: flow of funds data.313 45.946 14.758 13.447 7.367 14.038 39.421 90.937 24.982 2.758 458 729 18.369 21.425 22.336 22.261 13.557 7.576 22.440 12.544 23.090 10.604 37.976 14.480 11.735 327 470 12.267 11.220 15.749 6.727 18.619 9.592 8.794 86.711 24.982 15. In the third panel the original survey data are from close to year b c 2000.176 143.257 16.257 25.711 17.137 14.914 9.832 143.604 37.480 9.Finland France Germany India Indonesia Italy Japan Netherlands New Zealand Poland Portugal Singaporee South Africa Spain UK USA 98.929 2.824 14.271 14.744 1.317 180.999 7.418 37.038 23.837 109.271 4. Source: The Economist Intelligence Unit.820 25.592 8.

837 29.693 83.05 79.240 83.731 28. South Luxembourg Netherlands New Zealand Norway Portugal Spain Sweden Switzerland UK USA 0.614 14.00 0.825 73.06 0.729 89.54 18.823 72.45 71.614 23.419 21.09 0.794 86.407 23.261 13.77 70.17 0.55 1.704 124.03 0.15 0.12 0.30 9.923 23.135 18.22 0.12 69.186 128.271 36.727 1.27 2.44 0.34 1.818 89.87 76. 2000 Percent of Net worth Percent of World Bank country group and country High income: OECD Australia Austria Belgium Canada Denmark Finland France Germany Greece Iceland Ireland Italy Japan Korea.31 0.44 74.425 22.171 85.69 0.315 22.30 0.Appendix V Appendix V Table 1: The distribution of net worth by country.14 74.754 93.97 1.60 0.71 25.12 3.63 74.51 0.08 0.18 0.191 38.369 50.20 0.02 0.990 62.01 0.74 0.461 25.44 0.40 1.78 77.254 53.18 76.04 4.27 3.031 82.11 0.18 0.59 80.63 1.401 89.13 0.34 2.327 119.28 0.17 0.05 0.33 0.09 21.759 20.19 0.968 25.15 1.41 0.42 75.44 0.733 30.828 30.89 0.197 22.04 0.11 0.418 37.008 28.00 75.33 0.95 5.30 3.50 0.858 45.78 0.46 4.918 170.86 1.15 0.619 1.07 69.42 79.539 24.69 74.193 24.924 14.35 0.604 37.26 75.906 66.34 14.37 0.959 143.030 109.50 0.338 23.65 27.916 70.876 25.836 25.67 0.07 0.49 world population Share of adult population Net worth per capita (PPP) per capita (exchange rates) Percent of world net worth (PPP) world net worth (exchange rates) Real GDP per capita (PPP) Real GDP per capita (exchange rates) Percent of world GDP (PPP) Percent of world GDP (exchange rates) .47 0.06 1.619 20.96 4.05 1.420 76.56 1.086 55.09 0.83 0.77 0.78 71.421 68.16 0.95 32.567 120.37 1.755 126.06 1.34 0.38 3.06 75.50 0.505 23.71 0.264 22.44 3.555 98.252 35.97 1.13 0.22 2.09 0.67 71.39 0.849 182.39 90.751 37.98 5.13 0.937 48.68 0.52 0.34 0.38 1.416 23.687 160.24 0.83 78.64 4.71 78.95 2.209 24.852 35.17 0.60 0.26 0.35 31.07 5.671 42.252 66.357 92.12 77.317 180.628 28.832 143.74 4.25 4.253 80.577 70.72 2.547 9.29 0.31 0.40 0.46 76.091 144.008 32.06 0.048 25.01 0.558 26.15 4.917 15.057 17.02 0.026 81.805 33.70 0.84 1.21 4.089 19.37 0.021 10.871 72.188 33.037 24.929 27.758 10.727 67.59 0.639 85.21 0.91 0.87 7.

20 71. Hong Kong China.694 76.287 76.69 77.054 90.813 90. Macao China.00 0.11 76.00 0.526 18.353 44.915 14.44 99.825 90.20 67.36 0.849 12.00 0.01 0.417 76.31 0.364 85.785 9.054 99.09 77.00 0.00 0.787 138.00 0.526 138.034 105.870 17.364 105.88 77.01 0.481 10.00 0.813 62.92 0.107 23.00 0.01 0.775 9.287 103.20 67.22 0.93 71.256 9.916 12.137 19.813 90.277 90.072 100.217 11.02 0.833 90.364 76.03 0.825 0.00 0.00 0.03 0.00 0.02 0.526 75.20 67.23 0.10 0.148 15.04 0.27 71.654 90.01 22.912 0.03 0.714 17.73 58.01 0.01 0.01 0.624 14.02 0.825 90.794 73.01 0.01 0.277 53.00 .01 0.02 0.813 62.02 0.00 77.00 0.862 12.526 76.277 53.00 0.947 0.01 27.00 0.00 0.00 0.01 1.02 0.01 0.01 0.00 0.20 67.83 0.259 0.533 34.01 0.813 11.01 0.01 0.815 42.05 0.75 68.364 85.054 99.664 22.75 60.00 0.10 61.00 0.00 0.00 0.364 85.526 99.172 14.07 0.277 59.00 0.00 0.15 0.743 17.277 173.577 53.01 0.364 76.05 66.02 0.189 73.132 0.01 0.471 53.46 69.630 53.207 130.87 63.00 0.00 0.526 64.277 53.813 62.03 16.00 0.00 0.01 0.49 0.98 61.364 20.526 202.High income: nonOECD Andorra Anguilla Antigua and Barbuda Aruba Bahamas Bahrain Barbados Bermuda British Virgin Islands Brunei Darussalam Cayman Islands Channel Islands China.833 53.46 77.22 0.00 0.04 0.01 0.28 0.05 0.74 69. Taiwan Cyprus Faeroe Islands Falkland Islands French Polynesia Gibraltar Greenland Guadeloupe Guam Isle of Man Israel Kuwait Liechtenstein Malta Martinique Monaco Montserrat Netherlands Antilles New Caledonia 0.813 130.00 0.00 0.00 0.92 0.01 0.00 0.00 0.02 0.01 0.00 0.922 36.417 76.11 0.00 0.007 13.61 69.06 0.118 19.00 0.01 24.01 0.00 0.00 0.00 0.00 0.02 0.813 53.11 10.66 67.74 69.705 15.833 58.793 15.00 0.27 0.00 0.01 0.277 40.02 0.030 0.065 8.364 99.455 0.02 0.01 19.565 76.67 77.761 10.00 0.813 53.01 0.364 99.630 53.00 0.009 99.88 0.01 0.00 0.01 0.75 67.526 76.519 11.08 1.02 0.01 0.01 0.813 130.75 60.00 0.75 77.00 0.121 136.01 0.193 18.01 0.893 23.01 0.00 0.20 72.526 76.00 0.20 63.00 0.03 0.364 105.00 0.40 0.01 0.922 14.277 136.00 0.01 0.01 0.00 0.00 0.06 67.287 99.66 67.01 0.01 0.

19 0.713 53.473 8.00 0.33 0.631 36.03 0.833 136.25 0.558 29.09 71.59 77.906 0.47 0.09 0.05 0.745 10.287 138.493 54.06 0.170 7.02 0.51 59.17 0.75 68.582 105.21 0.00 0.456 19.00 0.01 0.00 0.287 76.277 33.712 3.03 0.01 0.76 65.242 15.22 0.00 0.046 62.03 0.00 0.35 0.06 0.960 19.11 0.938 18.512 4.784 13.20 62.929 9.851 6.04 0.02 2.932 14.02 0.178 4.00 0.552 26.547 14.86 0.973 22.569 38.05 22.873 9.17 0.089 8.02 0.630 90.760 47.942 3.722 33.39 76.951 12.00 0.484 3.02 0.813 90.03 67.00 0.00 0.00 0.05 0.06 0.362 130.00 0.57 59.01 0.843 3.397 9.162 20.02 0.88 0.806 6.04 0.349 4.01 0.569 14.008 4.489 25.065 3.539 15.00 0.00 0.02 0.285 8.828 6.02 1.526 47.15 0.67 77.00 0.21 0.844 8.983 13.410 5.04 0.00 0.17 0.06 0.00 0.09 0.03 0.00 28.06 0.00 0.00 0.439 32.02 1.07 0.305 6.396 9.644 16.063 8.61 74.338 21.19 0.361 22.00 0.424 7.06 0.615 10.03 0.27 0.29 0.998 30.277 90.27 48.07 0.28 65.36 55.164 4.70 56.00 0.01 1.05 0.02 1.481 12.09 0.422 14.00 0.00 0.00 0.711 7.221 10.406 9.00 0.833 53.039 0.19 11.84 .65 62.00 0.137 15.364 105.20 71.00 0.11 0.00 0.03 0.491 9.03 0.635 4.94 63.070 0.00 0.04 0.17 0.00 0.97 72.25 0.06 0.745 3.29 0.00 0.01 1.22 0.13 76.905 0.00 0.04 2.406 2.013 3.02 0.625 8.03 2.431 12.489 10.11 0.07 0.56 53.04 0.010 32.00 0.90 0.28 0.16 0.05 0.410 11.04 0.01 0.03 2.92 0.00 0.526 17.703 7.38 0.06 0.04 0.873 7.42 59.29 0.06 0.413 4.00 0.566 13.623 3.21 0.364 99.08 75.11 0.94 0.417 99.06 0.89 0.797 5.126 15.813 130.01 0.04 63.729 7.04 0.00 0.03 0.130 10.75 0.94 77.64 68.21 47.17 49.84 59.011 4.562 6.28 0.339 22.33 0.457 21.076 6.05 1.318 16.983 22.25 0.Puerto Rico Qatar Saint-Pierre et Miquelon San Marino Singapore Slovenia Turks and Caicos Islands United Arab Emirates US Virgin Islands Vatican Upper middle income American Samoa Argentina Belize Botswana Brazil Chile Cook Islands Costa Rica Croatia Czech Republic Dominica Estonia French Guiana Gabon Grenada Hungary Latvia Lebanon Libya Lithuania Malaysia Mauritius Mexico 0.29 74.04 58.420 12.00 0.389 5.672 76.80 56.092 30.526 113.00 0.34 0.693 5.780 6.00 0.61 0.34 0.35 67.06 0.01 0.00 0.04 0.00 0.284 10.12 76.994 3.18 0.

04 0.00 0.441 11.619 8.697 22.77 0.64 0.08 0.918 2.38 0.11 0.70 0.15 73.02 0.720 11.252 6.60 67.07 0.215 12.03 0.087 1.86 63.130 33.31 0.01 0.19 0.569 2.00 0.157 2.467 830 2.632 2.14 0.61 0.08 0.00 0.723 7.46 0.05 0.698 2.968 16.034 23.285 7.09 0.02 0.05 0.01 0.450 2.18 0.32 0.504 7.00 1.59 0.26 62.629 10.938 0.996 14.01 1.01 0.904 6.05 0.01 2.796 6.00 0.112 2.356 4.00 2.04 0.754 498 1.43 44.036 11.470 5.00 0.10 0.10 0.04 53.08 62.270 9.25 0.04 0.05 0.00 0.53 0.00 0.00 0.69 0.01 3.00 0.00 0.02 0.569 18.266 53.35 0.59 0.610 21.00 3.13 0.206 16.16 55.893 7.569 15.Northern Mariana Islands Oman Palau Panama Poland Réunion Russian Federation Saint Kitts and Nevis Saint Lucia Saudi Arabia Seychelles Slovakia South Africa Trinidad and Tobago Uruguay Venezuela Lower middle income Albania Algeria Armenia Belarus Bolivia Bosnia and Herzegovina Bulgaria Cape Verde China Colombia Cuba Djibouti Dominican Republic Ecuador Egypt El Salvador Equatorial Guinea 0.615 6.20 1.06 0.406 4.04 0.06 0.009 4.429 20.03 0.587 4.977 23.00 0.03 0.05 0.00 10.40 0.17 0.38 0.48 57.644 12.50 0.014 1.00 0.01 0.45 0.57 0.771 7.03 0.64 0.864 26.155 2.00 0.654 3.00 0.02 0.02 0.16 0.355 33.302 9.00 0.79 0.22 55.00 0.04 0.097 0.10 0.00 8.232 3.01 20.01 0.01 0.14 0.911 16.15 0.548 1.639 4.02 0.36 0.544 3.03 0.017 13.732 16.92 77.47 0.03 0.720 4.83 53.00 .04 0.090 9.05 0.14 60.994 1.22 0.943 47.04 0.04 0.06 0.03 0.45 72.622 10.02 0.18 0.00 0.42 72.05 0.00 0.266 891 1.491 7.02 0.37 0.70 51.01 0.627 6.291 12.99 55.28 0.613 4.00 0.02 0.458 4.81 0.60 56.09 46.729 6.02 0.02 0.22 0.04 0.823 12.374 11.650 18.01 52.107 3.654 17.946 6.352 1.117 1.924 5.408 11.98 47.32 0.561 3.721 10.62 0.118 13.00 0.75 0.00 0.713 1.152 10.207 14.101 3.523 8.42 0.07 0.06 0.299 3.49 0.356 10.67 72.457 3.00 0.650 9.16 0.489 11.31 0.591 7.438 6.00 0.489 8.44 73.523 2.14 50.04 0.131 15.78 67.738 2.467 1.04 0.934 5.00 0.04 0.07 0.11 0.545 14.34 58.02 0.14 0.03 0.074 1.40 0.44 53.06 0.09 0.58 0.339 17.779 5.09 0.086 14.46 53.00 0.068 8.267 14.00 0.661 12.79 0.099 1.67 73.27 57.658 6.41 0.07 0.04 0.20 0.677 24.909 17.08 0.04 58.844 5.150 3.011 6.754 8.00 0.68 64.06 0.95 45.807 10.017 996 1.996 5.730 4.285 3.01 62.00 0.00 0.02 0.

00 0.33 0.131 10.02 0.05 0.00 55.24 0.07 0.06 0.00 0.090 1.470 4.04 0.03 0.19 0.153 3.29 53.754 2.86 69.652 3.343 10.00 0.04 0.93 56.02 0.04 0.480 3.338 6.464 4.287 4.01 0.03 1.77 43.76 63.00 0.480 8.562 4.00 0.153 3.17 0.01 0.24 52.65 58.217 14.02 0.09 0.906 1.03 0.802 2.343 10.00 0.328 3.00 0.00 0.25 0.282 8.00 0.31 46.75 74.00 0.01 0.17 0.11 0.00 0.164 7.73 0.958 1.02 0.09 0.329 4.838 1.634 2.57 0.00 0.85 43.153 0.764 766 2.00 0.084 13.62 0.06 0.00 0.04 0.202 3.084 4.236 574 1.153 4.00 0.01 0.934 1.04 0.603 9.00 0.769 5.789 14.00 0.25 0.09 51.01 0.00 4.480 4.38 .587 3.647 5.09 0.299 6.00 0.04 0.11 0.00 0.824 10.00 0.00 0.14 0.101 12.146 1.01 0.343 11.59 44.065 5.13 0.29 51.533 4.00 0.02 0.281 10.00 0.01 0.00 0.30 72.01 0.41 0.54 49.00 4.084 10.00 0.086 13.41 55.00 0.43 1.37 49.606 12.675 958 923 4.00 0.00 0.084 12.989 0.084 10.01 0.67 52.02 0.20 0.363 1.00 0.01 0.043 842 1.390 4.153 4.356 11.02 0.06 0.91 44.67 0.827 14.841 5.01 0.67 0.024 7.331 5.153 4.36 0.08 0.00 0.08 0.13 0.165 1.00 0.870 4.00 4.11 0. States of) Morocco Namibia Nauru Niue Palestinian Territory Paraguay Peru Philippines Pitcairn Romania Saint Helena Saint Vincent and Grenadines Samoa Serbia and Montenegro Sri Lanka Suriname Swaziland Syria Thailand Tokelau 0.20 0.234 14.86 52.21 0.825 2.02 0.368 2.90 0.00 0.219 10.00 0.742 4.27 0.00 0.70 57.084 12.239 8.98 0.17 0.416 12.22 47.00 0.621 3.084 14.11 1.698 0.22 0.01 0.617 12.30 63.20 49.11 0.01 0.715 1.86 47.27 0.34 55.09 0.910 2.149 10.119 4.01 0.058 1.118 13.00 0.14 52.26 0.506 2.02 0.182 10.01 0.084 12.153 4.03 0.013 1.457 2.153 2.00 0.24 0.00 0.16 0.00 0.821 4.046 4.687 2.335 4.01 0.02 0.00 0.523 17.02 0.18 0.328 2.48 0.70 55.37 52.00 0.00 0.03 0.90 10.096 3.11 65.Fiji Guatemala Guyana Honduras Iran Iraq Jamaica Jordan Kazakhstan Kiribati Macedonia Maldives Marshall Islands Micronesia (Fed.05 0.00 0.02 0.848 1.377 10.00 0.072 2.163 5.00 0.039 1.31 46.02 0.847 5.047 4.42 47.09 0.25 0.01 0.138 0.691 1.06 0.801 4.153 4.28 1.09 0.00 0.03 0.27 0.00 0.950 4.084 14.00 0.799 4.05 0.02 0.569 12.

00 0.81 48.09 47.06 0.00 0.57 43.343 10. Republic Côte d'Ivoire Eritrea Ethiopia Gambia Georgia Ghana Guinea Guinea-Bissau Haiti India 0.10 42.11 0.513 4.00 0.371 775 1.00 0.19 0.00 0.01 0.717 6.798 2.02 0.10 0.01 0.01 0.130 7.01 0.00 0.00 0.376 2.084 9.69 44.05 0.05 0.00 0.02 0.04 0.823 669 2.86 74.01 0.20 53.502 2.153 1.90 42.23 0.084 6.20 59.02 0.01 0.99 43.01 0.03 0.480 4.02 0.221 6.00 0.949 1.81 0.55 0.03 0.153 6.03 0.01 0.225 986 619 1.00 0.05 0.104 1.06 0.01 0.01 0.315 1.02 0.02 0.13 2.53 52.728 5.668 5.01 0.04 0.03 0.00 0.00 0.123 0.90 46.06 0.12 0.02 0.029 2.389 3.00 0.01 0.78 50.237 23.00 0.414 3.03 0.026 180 1.21 0.795 5.10 0.123 1.01 0.95 59.00 0.01 0.01 0.07 0.24 47.04 0.20 45.06 1.47 47.00 0.08 0.952 3.14 51.062 1.00 0.148 959 1.13 0.01 0.14 0.961 738 1.561 2.06 0.01 0.05 0.52 0.07 0.02 0.02 0.41 45.890 10.00 0.06 0.17 55.14 52.250 950 1.00 0.402 2.349 6.01 0.127 1.01 0.06 0.30 0.88 45.00 0.876 4.05 0.409 643 1.00 0.01 0.00 0.03 5.00 0.262 428 329 1.51 0.13 16.022 586 203 98 323 592 258 407 180 466 458 0.885 12.301 1.00 0.82 0.153 4.062 409 1.27 0.01 1.147 3.129 7.15 1.349 1.236 4.Tonga Tunisia Turkey Turkmenistan Tuvalu Ukraine Vanuatu Wallis and Futuna Islands Low income Afghanistan Angola Azerbaijan Bangladesh Benin Bhutan Burkina Faso Burundi Cambodia Cameroon Central African Republic Chad Comoros Congo.89 70.61 45.272 1.16 1.164 419 327 753 1.112 0.01 0.00 0.66 40.13 0.06 42. Democratic Republic Congo.05 0.734 1.00 0.00 0.772 1.684 737 654 347 345 234 195 108 287 551 252 183 362 86 1.39 0.45 .00 0.00 0.02 0.01 42.01 0.10 0.88 10.63 0.01 0.00 0.421 2.028 747 720 1.05 0.00 0.12 0.07 0.01 0.00 0.991 1.85 42.288 7.95 0.04 0.153 1.01 0.14 0.555 1.04 0.00 0.329 5.01 0.14 0.370 6.237 1.07 0.01 0.03 4.170 1.589 577 193 945 1.00 0.50 0.13 0.02 0.01 0.414 3.06 0.859 2.01 0.62 58.00 101 1.00 0.13 1.533 2.96 43.06 0.01 0.673 6.24 0.00 0.00 0.91 6.611 1.01 0.33 0.042 9.46 47.00 0.315 4.03 0.85 40.084 10.798 12.164 577 1.00 0.66 43.00 0.02 0.561 2.00 0.02 0.11 0.601 3.12 0.883 5.01 0.084 21.

02 0.64 46.349 2.54 44.246 6.317 1.03 0.00 0.01 0.00 0.18 48.36 0.Indonesia Kenya Korea.05 0.561 5.164 848 577 633 546 383 1.02 0.02 0.205 1.03 0.01 0.01 0.00 0.12 0.03 0.01 0.00 0.01 0.03 0.00 0.296 3.00 0.09 0.53 45.01 0.54 0.02 0. North Kyrgyzstan Laos Lesotho Liberia Madagascar Malawi Mali Mauritania Moldova Mongolia Mozambique Myanmar Nepal Nicaragua Niger Nigeria Pakistan Papua New Guinea Rwanda Sao Tome and Principe Senegal Sierra Leone Solomon Islands Somalia Sudan Tajikistan Tanzania Timor-Leste Togo Uganda Uzbekistan 3.617 2.02 0.04 0.947 902 826 2.00 0.00 0.905 6.00 0.00 0.01 0.96 40.39 43.004 638 773 1.01 0.00 0.49 43.73 47.02 40.04 0.516 729 346 476 268 328 441 184 249 165 212 409 301 379 215 877 227 473 169 325 430 645 211 315 459 127 774 235 367 140 268 457 270 257 552 1.13 0.41 59.158 2.50 44.756 960 6.93 2.75 46.01 0.14 44.01 0.03 0.164 1.24 0.02 0.230 4.01 0.19 1.05 0.10 0.00 0.02 0.01 0.10 0.164 1.138 1.04 .02 4.04 0.04 0.07 0.04 0.00 0.79 42.00 0.16 57.04 0.47 0.00 0.57 0.01 0.02 0.01 0.561 2.78 0.01 0.561 1.02 0.00 0.00 0.31 67.03 0.942 1.044 3.01 0.83 0.73 45.164 1.04 0.01 0.231 1.561 2.07 0.19 0.27 45.02 0.035 1.67 0.07 0.951 3.83 7.19 0.38 46.00 0.321 1.01 0.03 0.21 0.681 734 1.14 42.349 2.02 0.498 6.40 0.35 38.65 48.09 0.03 0.030 1.551 1.561 6.02 0.02 0.34 0.349 2.64 46.00 0.617 2.29 0.00 0.561 2.756 1.01 0.19 0.72 43.12 0.13 0.05 0.00 0.08 0.00 0.164 645 725 1.00 0.83 43.164 433 1.155 1.00 0.02 0.01 0.00 0.729 2.207 8.03 0.00 0.380 490 926 1.28 46.76 48.608 1.004 577 577 298 681 1.27 0.01 0.174 6.02 0.440 966 1.248 329 377 1.55 51.02 0.214 2.04 0.50 0.00 0.798 5.09 0.00 0.113 1.01 0.13 0.01 0.19 0.44 0.01 0.561 5.10 0.69 65.973 3.683 1.01 0.316 3.03 0.01 0.01 0.01 0.00 0.07 0.04 0.17 0.09 0.04 0.01 0.00 0.164 545 1.02 0.212 1.01 0.00 0.00 0.17 40.01 0.172 353 1.02 0.00 0.255 1.223 2.02 44.03 0.02 0.70 53.881 1.294 2.08 0.00 0.01 0.00 0.00 0.492 877 808 996 1.00 0.817 6.56 0.09 0.00 0.326 976 1.07 0.08 0.02 0.870 2.00 0.40 0.14 0.01 0.96 47.41 54.671 6.

21 55.05 0.00 0.01 0.01 0.863 1.633 2.02 0.121 6.07 0.349 1.18 0.00 0.96 52.01 0.607 397 487 313 556 0.02 .012 1.03 0.10 0.01 2.112 577 537 789 1.07 0.00 0.29 0.34 0.01 0.29 0.Viet Nam Western Sahara Yemen Zambia Zimbabwe 1.293 841 2.465 0.02 0.05 0.20 44.37 40.19 42.425 2.71 5.28 0.