Customer Center
Product Center
Free Resources
Reference Reviews
Péter's Digital Reference Shelf
March 2008
Title: Human Development Reports
Publisher: United Nations Development Programme (UNDP)
URL: http://hdr.undp.org/en/statistics/
Cost: free
Tested: March 7-18, 2008
The Context
There are many statistical sources that provide some information about some aspects of the well-being and quality of life of the citizens of the world in countries, regions or continents. These indicators are scattered in various categories from population and education, to health and economics. They are rarely presented in a disaggregated format by the sexes. Even those that provide some statistics for both men and women (or boys and girls as in elementary education statistics) often would not offer an easy way to compare the countries.
Take as an example, the life expectancy chart presented by the very good Information Please Almanac. It informs about the average of males and females combined. But this average may hide telling signs. True, the difference in life expectancy between males and females worldwide is only 4-4.5 years in favor of women. Women live 5.8 longer then men in the U.S. and 6.6 years longer in the 27 European Union countries. The average life expectancy of the population is between 78-79 years in the U.S. and in the European Union, respectively.
However, combined averages often don't reveal warning symptoms. For example, in Namibia, the combined average of life expectancy is 43.11 years according to the CIA World Factbook. This is sadly low, but not the lowest. Where that country lags behind any other country is the difference between the average life expectancy of women (41.79), and men (44.39), i.e. men live 2.60 years longer than women. This is odd even in the least developed countries. The disaggregated values appear only in the individual country profiles, not in the country rank list. This extreme disaggregated figure of Namibia, clearly suggests a strong disparity between men and women in other aspects of human development, but it is not apparent because the CIA World Factbook provides a ranked list only for the aggregate average. A disaggregated rank list would be much more informative.
Databases with disaggregated statistics by gender
The World Bank Group deserves credit for having created GenderStats, an excellent, gender-focused statistical database, and as the name implies, it does have many disaggregated datasets that provide information about several components which determine, or allude to the quality of life of men and women. I reviewed it more than two years ago in this column. The Organisation for Economic Co-operation and Development (OECD) is planning to release a promising statistical database of international (human/social) development indicators by mid-2008. It is to present "comparative data on gender equality in 161 countries".
Most of the statistics of the U.N. provide disaggregated data by sex. The sites which offer these statistics should be top destinations to get gender-specific reference data, but it is easier said than done. The problem is that the U.N. statistics are scattered across many Web sites of the agencies. There is a Women's Indicators and Statistics database on CD-ROM which combines many gender-specific data gathered by a variety of U.N. agencies, but it is very outdated, and not nearly as convenient as a Web database.
The good news is that a long overdue single online database by the U.N. Statistics Division, called UNdata, is gradually becoming available on the Web. The bad news is that the U.N. Gender Statistics component (not to be confused with the World Bank's GenderStats) is not there yet. Neither is the Gender Statistics database of the U.N. Economic Commission for Europe (UNECE) included in Undata -yet.
By virtue of extending its scope to the U.S., it has a broader coverage than its name implies, but even so, it can offer only a limited view without data for Africa, Asia, South America and Canada, by the otherwise well-disaggregated, highly relevant indicators, such as the proportion of male vs female graduates in the 52 countries covered. (It is another question that the indicator values should be always corroborated; uplifting as it may seem that 72% of the women in Albania have graduate degrees, it is not very likely, considering that it has remained one of the poorest countries of Europe even long after the xenophobic reign of communism ended there.)
True, Undata may be searched for some related indicators of UN statistics, such as life expectancy, but only the World Population Prospects database comes up as a source, which is rather cumbersome to search for ready reference purposes. Luckily, UNICEF, one of the two most technologically avant-gard agencies of the United Nations has created not only a superb, content-rich website, featuring an outstanding database called ChildInfo, but also developed and distributes free of charge DevInfo a sophisticated software system, to create, maintain and search complex statistical databases. One of them is Gender Info which illustrates the smoothness of the software, but it is short on content yet, finding only the same data series that the above mentioned Undata system found from the huge statistical data collection of the UN. Gender Info will be a worthy resource once additional databases will be added to it from all the relevant international organizations and UN agencies, including UNDP, which is the publisher of the Human Development Reports, the subject of this review.
The Content
UNDP is an agency of the UN system in charge of —among others&mdash helping less developed countries to reduce poverty, the devastating consequences of AIDS, to develop environment-friendly energy, communications, and technology infrastructure. The essential, yearly compiled statistical tables of the Human Development Reports are available in a very long but very well presented and easy to browse PDF file, and as a series of Excel files. The Excel files are not viewable direct online because they are zipped. It would be better showing them as html files (with instant re-sorting option), and redesigning the headers as well as optimizing the column width instead of using a standard 12-position column which is unnecessary and very inconvenient, requiring constant horizontal scrolling).
The basic data for the human development indices are primarily provided by the statistical offices of the countries if there are ones, and if they are ready to provide appropriate data at the appropriate time, in the appropriate format. Obviously, this is not the case in countries where human survival, rather than human development has been at stake for many years, such as Afghanistan, Liberia and Somalia.
UNDP shows a good example for collaboration, as it collects much of its raw data from other U.N. agencies, and it makes good use of already available data for further processing. It gives credit to the cooperating agencies for the sources used, and to the Statistical Advisory Panel.
The 4 indices of human development (which are 5)
Actually, UNDP is as much concerned with many of the human facets of development, as with technology development. It is no accident that it has been the key agency to develop the first measure in the sea of statistics to offer a composite index for assessing and comparing the well-being, and quality of humans around the world. It was introduced by UNDP almost 20 years ago in its country profiles. It is known as the Human Development Index (HDI).
Later, UNDP developed 3 more indicators primarily from traditional demographic and social measures: the Gender-related Development Index (GDI), the Gender Empowerment Measure (GEM), and the Human Poverty Index (HPI), to gauge the status of human development with a distinction of gender, and to produce ranked lists of countries by these indexes The four indexes are actually five, because there are two variants of HPI, one is for developing countries (HPI-1), and another for industrial countries (HP-2). It is another question what countries get listed under developing countries as I discuss and illustrate later.
The visual explanation about the components of the indices is excellent, and so is the detailed explanation of the methods used to calculate indices. This 7-page file is worth printing out as they are highly informative, and often needed. Beyond mentioning the composition of these indexes I don't provide more details about the methods of calculating them as I can't explain them any better than the 7-page information file. Unfortunately, I cannot deep-link to the appropriate section of this file as there are no anchor points in it.
The concept, and the origin of the HDI are eloquently and practically explained by its co-developers: Mahbub ul Haq, and Amartya Sen. There is a very good webliography, including papers that disagree with the indices. Many of the Human Development Reports are also available online free of charge.
The Human Development Index
The HDI is a composite number derived from three indexes, the life expectancy, education and GDP indices. These, in turn, are generated from four indicators: the life expectancy at birth, the adult literacy rate, the gross enrollment ratio (combining primary, secondary and tertiary education), and the Gross Domestic Product per capita.
Appropriately, it is meant to complement the traditional indicator of development, the GDP, which simply measures the total dollar value of goods and services produced in a country —irrespective of the cost of achieving that level. The GDP per capita in and by itself can be quite misleading, although it certainly pleased Stalin, his beloved Alexey Stakhanov, and his fellow stakhanovites who were competing who could mine 15, then 30 times more coal then the quota —hardly conducive to human development.
The huge disparity between GDP per capita and the HDI is splendidly illustrated by the animated Flash comparison of the HDI and the GDP per capita for any country you choose. The Web site designer chose Saudi Arabia and Albania for good reason. Their HDI is almost identical, even though the GDP per capita of Albania is only $5,200 and Saudi Arabia's is 13,600 (and it keeps rising with every gallon of petrol that you put in your car). If you can't install Flash on your over-protected office PC, look at the static graph to get a feel for it. From that screen any target country can be selected—as long as you have installed the Flash software.
The 177 countries for which HDI could be calculated are clustered into three groups by their HDI: high, medium, and low levels of human development. The first includes 70, the second 86, and the third 21 countries. It is surprising to see some countries which made it to the high category cluster of HDI, such as Macedonia, Albania and Belarus, but only a close analysis of the source data could shed light to the possibly inaccuracy of the component measures. Inaccuracies and distortions certainly occur in reported data, and may be introduced during data entry.
One must remember that the numbers are often reported by the country which may have an agenda to paint a rosier picture about a country for political, ideological or economic reasons which differ from the reality. It illustrates very well, how different the picture may be when on rare occasions, the UN agency adjusts the data publicly, showing the reported and the adjusted values. Unfortunately, these are shown in separate lists, rather than side-by-side, which would be a good reminder for the users. I downloaded and combined them to show that some of the differences are negligible, some countries may over-report, and many may grossly under-report certain indicators, such as Niger as an extreme example for maternal mortality. This comes with the turf when working with statistics, especially when combined by different agencies at different times.
There are 17 UN member countries for which no HDI is available, simply because apparently they did not care to provide the data, or could not provide all the data which are needed for calculating the HDI and the related indexes.
The ones in the former category (such as Andorra, Monaco, San Marino) would rank obviously quite high in the list, but it still would be useful to see their stats. The ones in the latter category such as Afghanistan, Somalia, and several small island states in the Pacific, would be in the very bottom of the low level cluster, based on other data which are available, but not used for the HDI.
The rank positions are useful, but the absolute index values must be also looked at to get a realistic picture. For the HDI these range from 0.336 for Sierra Leone at the very bottom to 0.968 for Iceland and Norway at the top (in a tie). Often, there is just a tiny difference between countries that are adjacent on the list, or there is no difference at all. For example, Japan and the Netherlands are in a tie at 0.953, and so are Denmark and Spain at 0.949. For this reason, the real distance between a country ranked 2nd and 22nd the difference may not be as large as the rank order may suggest.
The Gender-related Development Index
This index is created from gender-disaggregated data and indices for life expectancy, education and income. It illustrates well the extent of gender equality, or more often, the inequality.
There are 157 countries for which GDI could be calculated, i.e. for 20 less than the countries for which HDI could be calculated. Most of the countries without GDI are from the Caribbean, and the Asia Pacific area. Looking at the list of the countries is well informing. The highest level of gender equity is in the Nordic and Western European countries (except for Ireland), and the lowest level in Eastern and Central Africa.
I wish there were similar visual and animated depictions of the rank difference between GDP vs GDI, HDI vs GDI, GDP vs GEM (to be discussed below), as there is for the difference between the rank of the countries by GDP versus HDI mentioned above. If you care to do the comparison yourself then you will find (not surprisingly) that the largest differences between the HDI rank and GDI rank are in Saudi Arabia and Oman, indicating a large inequality between males and females. The former ranks 42nd by GDP per capita, 61st by HDI, 69th by GDI, for the latter the rank positions are 43, 58 and 66. For comparison, the USA rank positions are 2nd, 12th, and 16th, Switzerland's are 6th, 7th and 9th. It could be useful to add some additional dimensions to the GDI. One that comes to mind could be an indicator for the extent of the property ownership rights of women; another could be an indicator of health care expenditures for cervical versus prostate cancer treatment. I know, such data are difficult to collect when there are more than 25,000 patients per doctor, and barely adequate resources even for diagnosis. Somewhat easier is the maternal mortality rate, even if it has no male counterpart rate to compare, but an informative index could be worked out. After all, more than half of the women of the world give birth to one or more child(ren), so with about 1.5 billion subjects it can be a rather universal gender development factor.
The Gender Empowerment Measure
The GEM index also indicates (in)equality between the genders—but here in terms of political power, and two aspects of economic power. It is measured by the combination of various indices: the percentages of women representatives in the parliament, women's share of leading and managerial statuses in the government, as well as in the corporate world, and professional/technical positions and the ratio of females' to males' income.
Data about women representatives in the parliament is available for most of the countries. It is a no-brainer for Brunei which does not bother to have a parliament, Fiji used to have a parliament but it was dissolved years ago, and never since reconvened. As you probably guessed Saudi Arabia's parliament has no room for women, and has no voting system, the king appoints the parliamentarians. The percent of women representatives in the parliament in some countries are so high that one may have some doubt about the data. Rwanda is ranked #2, Burundi is #14-15 (in a tie with Iceland which is the current leader by HDI measures, while Burundi is #167.)
Some cynics might say, that this ratio is so high in Rwanda and Burundi because a very large proportion of males lost their lives during the Tutsi-Hutu massacres, but in that incredible genocide, no distinction was made between males and females, boys and girls as victims. The perpetrators, however, were almost exclusively males, and many of them may have been sentenced and doing time for genocide and crimes against humanity, or at least not deemed feasible candidates for peaceful jobs.
The Nordic countries are among the top 15 ones by this measure. They are especially good by the rate of participation in top level government positions (where many countries in the high human development group show disappointingly low rates, such as the U.S.A, Switzerland, and Luxembourg (all at 14.3%), and especially Italy, at 8.3%.
The female proportion in professional and technical jobs is available only for 100 countries, led —surprisingly to me— by the Baltic countries, and other republics of the former Soviet Union. The very top ones include Kazakhstan, and Moldova, and even Kyrgyzstan is ahead of Iceland, Australia, Canada, France, and the U.S.A, although by just 1% or so.
I have doubt about the top level of this rank list, but not about the deep bottom position of many of the Middle East countries for reasons mentioned earlier. Revision and adjustments may be necessary needed here, although not for all countries that seem to be positioned too high, or —in case of Switzerland— too low.
Switzerland has the biggest drop in position by this criterion among the top ten countries (by HDI indicator), but it is a "well-deserved" plummet. For the ratio of female legislators, senior officials and managers criterion, its 8% achievement is identical to that of the United Arab Emirates, Oman, and Nepal. For overall GEM, it ranks 27th, below the United Arab Emirates and Quatar. It which must have required hard work for Switzerland to get behind these two countries. For fairness, it is still ahead of Yemen and Saudi Arabia. This is not so surprising considering the historical Swiss pattern of attitude related to related matters. It was the last country in Europe to mandate paid maternity leave, and Swiss women gained their voting right the latest in Europe, six years later than in Afghanistan, and a year later than in Yemen.
The irony should not be lost on anyone, that Geneva is the headquarter of many U.N. agencies which deal directly or indirectly with issues affecting the human development of women, and hosts many conferences, which is very good for the city, the canton, and the country in terms of publicity and income.
The ratio of female to male income, by the way, is the last component of GEM. Unfortunately, the ratio is not available directly as a rank list, and the HDI listing of the ratio does not easily reveal the differences. Downloading and making a rank list shows interesting data. This data is available for 167 countries, the range is from 0.83% for Kenya to 0.16% for Saudi Arabia.
Once again, I have doubt about the validity of the data reported by and for some countries. It is hard to believe that the traditionally very gender-empowered Sweden shares the 2nd-3rd rank position with Mozambique (at 0.81%), followed by Norway and Burundi (at 0.77%), followed by many "odd couples" in the subsequent rank positions. The bottom of the list with countries where the ratio of female to male earned income is less than 25%, complies with expectations: United Arab Emirates, Morocco, the Sudan, Quatar, Egypt, Oman, and Saudi Arabia.
While this element is available for 166 countries, GEM is still available only for 93 countries, because those are the ones that have all the required indicators for GEM. Hopefully, its scope will be extended both geographically and criteria-wise, as it can be quite an informative number about the extent of gender divide in many social and human criteria of development. I wish it were enhanced to incorporate a variety of telling indices related to discrimination and violence against women, none of which are conducive to their empowerment.
Customs and regulations limiting or denying their rights in marriage, inheritance, property ownership, and even basic human rights like driving a car, or just walking to the market without a male, should be applied to calculating the GEM, as gender "dispowerment" variables. We know from good examples, that it is not a heresy to provide rights and liberties to women within and outside of the home, let alone religious tenets to deprive them of those, and creating social and human poverty. This is a good transition to the remaining two indices directly or indirectly characterizing human developments.
Human Poverty Index/Indices
It seems to me that this poverty measurement pair was rushed out without much corroboration, and even without adequate proof-reading. There is a HPI-1 and a HPI-2 measure. The latter is titled "Human and income poverty: OECD countries, Central and Eastern Europe and the CIS." The former is titled "Human and income poverty: developing countries." And therein start the problems. HPI-2 provides information about 19 countries, which include data for some of the OECD countries, but none of the Central European countries, none of the Eastern European countries and none of the CIS countries —in spite of the mouthful title of the statistical tables. (There is a table with absolute values and another with rank values. This is true for both HPI-1 and HPI-2.)
The title of HPI-1 tables refers to developing countries, which includes 108 countries and territories from Mali and Chad to Singapore and Hong Kong —quite redefining the scope of developing countries, and leaving many OECD, all Central European, Eastern European and CIS countries (many of which are quite underdeveloped) fall through the cracks without either HPI-1 or HPI-2.
This terminology creates misleading and ill-titled categories, one has far less than its title promises, the other has far more than would fit the category. It includes countries and territories which could not be more different. For example, Singapore appears and is ranked as 7th in the developing countries table for the poverty index, behind such countries by this measure as Barbados, Uruguay, Chile, Argentina, Costa Rica, and Cuba, and only two positions ahead of the Occupied Palestinian Territories. The poverty index is meant to be a composite indicator of the —most basic dimensions of deprivation: a short life, lack of basic education, and lack of access to public and private resources."
This classification is rather absurd even if poverty is not restricted to per capita or household income, and other financial indicators. It is to be noted, however, that the tables' titles specifically mention human and income poverty. This makes it even more absurd that Singapore, Hong Kong, and several South American countries are lumped together with the really underdeveloped countries that need funds and external human support. The former definitely belong much more to the developed rather than to the developing world by almost any measures in general, and any of the component measures that make up the HP-1 and HP-2 indices in particular. I show the components of HP-1 and HP2 side by side to see the differences clearly.
By the overall HDI index, for example, Hong Kong is ranked 21st and Singapore is ranked 25th out of 177 countries, by the per capita health expenditure Singapore is 37th (for Hong Kong there is no data). In terms of probability of surviving past age 40, which of the countries in the developing category rank 2nd and 5th? Hong Kong and Singapore. You find Singapore and Hong Kong and several other countries in much better positions than the developing countries in Central Europe and Central Asia by the other criteria of the Human Poverty Indexes, i.e. in terms of access to improved water sources, in terms of adult literacy, age appropriate weight of infants and children, etc. You don't need to be a demographer, or a globetrotter to know that something is flawed in calculating the poverty index or in the raw data that yields a rank list where Singapore is practically juxtaposed to the Occupied Palestinian Territory.
Footnotes in the tables (as foot in one's mouth)
This classification is so confusing that even the UNDP specialists mix them up, and add discombobulating footnotes to the HPI indices to explain the odd practices of forming the subset of the HPI-2 group. The notes says that Israel and Malta were added to the HPI-2 table even if they are not OECD countries. Actually all the 177 countries appear in the table, but only 19 has data, and neither Israel, nor Malta is one of them, they appear as data-less as the other 158 countries in the HPI-2 poverty index. This is true also when you look up the table for HPI-2 index value.
There are oddities in other elements as well. In the list of illiteracy rates it caught my eyes that Estonia has far the lowest rate of adult illiteracy at 0.2% while the closest to it —with a 1% illiteracy rate —are the most developed countries in the world (where there has been practically no illiteracy). I recalled from a note that for HDI calculation purposes only 99% literacy rate is used for countries which report 100% literacy rate, but why Estonia would have a 0.2% illiteracy (99.8% literacy) rate even if it has 100% instead of the 99% compromise for calculation purposes. That's what footnotes are for. Well, not exactly in this case.
There are many countries with a footnote, especially with footnote (2). Instead of refining and clarifying, here the footnotes just do the opposite. In the footnotes part there is an empty line for footnote (2). But footnote (5) seems to give some clue, saying that "An adult illiteracy rate of 0.2 was used to calculate the HPI-1 for Cuba". Not related to Estonia, but whetting one's appetite to know why. But there are more enigmas than this.
Scrolling back up on the list you realize that there is no illiteracy rate for Cuba on the list at all, let alone the 0.2% rate promised in footnote (5). Apparently, it was "awarded" to Estonia. How many other countries and footnotes have been messed up, and left uncorrected? I don't know, but I do know that footnote (3) is incomprehensible for me even if am not entirely illiterate in English. Not giving up immediately, I went to the adult literacy rate table where both Cuba and Estonia appear with 99.8% literacy rate (so both of them should have an illiteracy rate of 0.2 then on the other table, if I am not math illiterate). The footnotes are more messy and enigmatic than the ones for illiteracy. Here is where you find the reference to applying the 99% rate —but both for countries that don't show any data for this variable, and the ones that show higher than 99%. Go figure.
These footnotes in the tables are like one's foot in his/her mouth, and indicate that this must have been an outsourced and/or a hurried and a harried job —bypassing essential proofreading.
THE SOFTWARE
There are more than 100 indicators in the Reports, and they can be easily browsed through the fine alphabetic index, by country, and by the excellent pre-defined tables which show and link to the component elements that make up the table.
The component lists are OK (except for the very poor footnotes), but they should offer instant sorting to see the ranking of the countries by the chosen variables. As I mentioned before, the tables should be available in HTML format to be able to see them without the need to downloading, and reformatting the Excel files, which are zipped into a large file, and need unzipping. This is tedious.
The Flash demonstrations are highly functional and informative —for those who can install the Flash player. Those who are not authorized to install this software, are deprived from these excellent interactive tools.
It is a very good idea to allow users to customize tables to their own liking, by choosing their preferred level of aggregation by developmental level, geographic regions and/or individual countries. It is even better to offer the tables by themes and tables, but the themes section should be broader to read the title of the often very long themes, which are cut off. It is well done at the step when the individual indicators are displayed, and they allow the users to choose one or more indicator(s).
One has the choice to display the table by arranging the indicators in rows versus columns. This is a good and easy way to create personalized tables for a special project very quickly. This figure shows part of a simple table made of a few indicators of Arab countries. There are many choices in every regard, and the results can be exported for further processing with a spreadsheet software.
In spite of some oddities, and the irritatingly careless footnotes, this is a useful database to learn the profiles of countries, groups of countries clustered by regions (e.g. Sub-Saharan countries), by associations (OECD Countries), by level of human development or by any kind of combinations chosen by the user. The possibly erroneous data are problematic, and should be treated carefully, and the footnotes mess must be cleaned up to make the database better.
— Péter Jacsó