The Human Development Index (HDI)

Avantika Chilkoti - December 2011

Speed Read:
  • When selecting the variables to include in a measure of human development, controversy is inevitable as the decision-maker is essentially defining what it is that gives people greater satisfaction and a better life

  • The Human Development Index - bringing together income, health and education - is used to compare human development in different nations

  • Key changes to this indicator reflect the difficulty of such a statistical exercise and of keeping the measure relevant. For instance, income is now accounted for using the GNI rather than GDP, as the increasing flows of international remittances skew national income figures and the index is now composed using geometric, rather than arithmetic, mean to reflect the fact that the different elements are not substitutable

  • The subjectivity of human development, as a concept, rears its problematic head once again in the decision of how to weigh education, health and income in the index

The United Nations Development Programme’s Human Development Report (HDR) introduced a new indicator in 1990. Termed “The Human Development Index” (HDI), the composite index seeks to measure human development in a nation by combining indicators for income, health and education (Figure 1). The result is a number between nought and one, with higher values indicating greater development. The idea is to allow for cross-country and inter-temporal comparison. Since 1970, 135 countries, or 92% of the global population, have been covered. In 1990, the world HDI was 0.57 and by 2010 it had climbed to 0.68.


Figure 1


This index is exemplar in exhibiting the multiplicity of challenges faced when dealing with economic data. These difficulties are especially amplified when working with an issue such as human development as the meaning of development is inherently subjective (as Rist says, our conception of human development is dependent on “the way in which one person pictures the ideal conditions of social existence”) so that the factors included in an index, weightings and definitions are innately questionable. Further, as Sen (1999) takes the view that human development consists of a range of “capabilities” or “freedoms”, a composite index may be deemed redundant as these elements must be measured directly and separately.


Thus, in 1990 the HDI was explicitly presented as only the seed for what was to be an evolutionary process of mathematical tinkering, definitional adjustment and prioritisation. The report stated that the HDI “should for the time being focus on 3 essential elements of human life” - income, education and health – stating that it “will be designed over time” and that “a participatory approach is crucial to any strategy for successful human development”. This explicit acceptance that any new index will be initially suboptimal is crucial in a world where economic data and accounting, both in the public and private sectors, are increasingly an area of contention.


Thus, this paper looks at several elements of the HDI and various opinions on its suitability to illustrate just the kind of disputes that emerge in dealing with economic data.



As Deaton (2008) states, “without health, there is very little people can do, and without income, health alone does little to enable people to lead a good life” – synergies such as this require the use of a composite index when measuring human development. The selection of income, education and health as the components of human development has, however, sparked debate.


The underlying sentiment behind the development of the HDI was Robert McNamara’s espousal in the 1960s of the ideal that growth alone is an insufficient indicator of development. Though President Truman’s inaugural address in 1949 had commended the possibility of a “trickle down” effect, whereby economic growth would improve poverty, a decade later this had not been evidenced. Questioning the supremacy of economic growth as the sole determinant of human development, McNamara and Sen attempted to bring people to the centre of this discussion, conceiving the notion of “human development”.


For some time now, many have found income to be relatively unimportant to human beings in comparison to factors such as health, employment and family circumstances. In particular, Easterlin (1974, 1995) found that, over a long time period, even though income per capita may rise, national happiness does not. This links with the fact that satisfaction is thought to adapt to new income within four years, according to a study by the German Socioeconomic Panel. This decoupling of growth and development is still supported today. Tim Jackon’s report on “Prosperity without Growth” was released in 2009 and considers that, aside from the human instinct for acquisition and the psychological importance of relative wealth, rising income does not guarantee rising “prosperity” or satisfaction.


However, some authors such as Solow and Schumacher insist that this decoupling of growth and “prosperity” occurs only once basic needs are met. A current example is the prevalence of 25% youth unemployment in Spain – though this is a nation with relatively high GNI per capita, the effect that unemployment has on morale and thus on life satisfaction is significant. It is this analysis that leads to the inclusion of non-income elements in the HDI. Veenhoven (1991) takes just this view but others such as Robbins’ (1938) argue the converse, stating that it is after basic needs are met that income and happiness become positively correlated, as the “luxuries” of culture and intellect can only then be factored into well-being. Using data from the Gallup World Poll, where questions regarding life satisfaction are presented to nationals of different countries, Deaton (2008) finds that “life satisfaction is higher in countries with higher GDP per head”. When plotting GDP per capita in 2003 on the x-axis of a graph and mean life satisfaction on the y-axis, “the slope is steepest among the poorest countries…but it remains positive and substantial even among the rich countries”, so there is no “critical level” where basic needs are met and “GDP per capita above which income has no further effect on life satisfaction”.


Conflict is rife in the choice of three components for the HDI. Mc Gillivray (1991) finds that a significant and positive correlation exists between all variables within the HDI, deeming a composite index unnecessary. Similarly, Ivanova et al’s (1999) study shows that using just one measure alone rather than a composite index, the results rank nations in much the same order. As Georgiadis states, wealthier may well be healthier as health services and improved hygiene become affordable. However, this link inevitably subsists only until incomes are struggling to cover basic needs. Furthermore, the amalgamation attempted by the HDI is valuable as a procedural statement, attempting to highlight the inadequacy of analysing economic growth, even if it has no substantive advantage in altering results relative to simple comparisons of income.


In choosing the three elements of the HDI there is a trade-off between simplicity (and thus the breadth of use) and the comprehensiveness of the index, a common difficulty in the compilation of indices. This is evident in the following disagreement. In the 1990 HDR itself, Dasgupta & Weale (1992) suggested that civil and political aspects of freedom should be included within the HDI. This suggestion is highly valid – economic and social developments are reflected in incomes, health and levels of education, but the political dimension has thus far been neglected. However, this expansion of the HDI is deemed an unnecessary complication by Streeten (1994, 1995), who suggests that freedom should be considered separately from human development. His point is that political conditions are more volatile than the measures currently in the HDI, there is no trade-off between the HDI and freedom indicators and the relationship between the two requires further research before they can be combined. However, Deaton (2008) poignantly notes that freedom may not factor into life satisfaction as “people do not necessarily perceive the constraints caused by their lack of freedom”. “The child who is potentially a great musician but never has a chance to find out” is the typical example of this.


This analysis links with Collier’s (2008) work on “The Bottom Billion”, the poorest nations that diverged in growth from the rest of the developing world in the 1970s. The countries in this cohort face an average life expectancy of 50 years, where this figure lies at 77 years in other developing nations. Infant mortality is 14%, where it is 4% in other developing nations. Thus, a problem with existing economic data is that “developing countries” are often grouped together. A large economy with a small population that is growing rapidly sees its growth rate averaged along with a small economy that is declining but has a large population. Though the populous nation contains more individuals, the large economy’s figures have more impact on the average, skewing the picture that we have of “developing countries” as a group. In fact, “for the countries that have really fallen apart, there are no usable data” and leaving out these economies would mean assuming that their economic performance was that of the average developing country, an erroneous assumption.


Collier also notes that “the countries now at the bottom are distinctive not just in being the poorest but also in having failed to grow”. In his explanation of this phenomenon Collier takes into account non-economic factors: the “conflict trap” whereby certain negative economic conditions lead to a cycle of violence; the problem of leaders that “are sometime psychopaths who have shot their way to power, sometimes crooks who have bought it”; the development trap; the trap of being landlocked; and the trap of bad governance in small countries. And all this analysis of growth comes before we even consider whether the progress occurring is the “right” sort of growth, the “sustainable, pro-poor” variety that actually benefits individuals in a nation.




Within the three sections of the HDI, indicators are chosen to reflect the levels of income, health and education in a nation. Over time, the indicators used in each of the three categories have been disputed and altered. The changes made to the HDI in the 2010 HDR will be studied to exemplify some of the typical difficulties faced when working with economic data.


Life expectancy at birth was the only indicator that was unchanged, though simply because no superior alternative could be decided upon. This measure is sluggish as only the life expectancy at birth of the very youngest affects the average in a nation. Paul (1996), for instance, suggests that child mortality would prove a better reflector of the sanitation and health facilities in a nation.


GDP per capita was substituted by GNI per capita, which takes into account the balance of cross-country income. As the World Bank reported that $325 billion were remitted to LEDCs in 2010, the need to account for these flows has become greater over time and this amendment to the HDI reflects the fact that indicators must develop as the world changes.

In measures of education, literacy was substituted by mean years of schooling, an indicator which is measured more frequently in more nations and which allows for cross-national comparisons to be made. Literacy has long been considered a suboptimal indicator as it is not only difficult to measure in many nations with poor access to rural communities, but as it is difficult to compare across nations with languages of disparate complexity. Enrolment was also substituted with expected years of schooling to create consistency in the measurement of education in years.


Even still, the measurement of education in the HDI remains imperfect. Srinavasan (1994) points out that it is the quality of education that matters and even if time spent in education is used, dropout rates and the length of the school year in different nations are yet to be adjusted for. To look at the quality of education, for example, the 2003 Arab States Human Development Report constructed a measure that captures both the quantity and quality of education, adjusting mean years of schooling with average test scores and including indicators related to media, communication and the training of scientists. This is a most useful exercise but one which requires significant amounts of detailed and reliable data that do not exist for many nations. The trade-off between detailed analysis and comprehensive, global comparisons thus emerges due to problems of data availability.


Nuebler (1995) also criticises the sensitivity of the different indicators utilised in the HDI, their ability to reflect recent changes. For instance, school enrolment rates are more sensitive to development than the average number of years of schooling as the latter is a lagging indicator, reflecting changes in development for the population aged over 25 only. Another key issue is thus the comparison of stocks and flows - of slow and fast moving variables. Hopkins (1993), identifying the importance of a sensitive index, suggests that a “Human Development Improvement Index” may be the way forward, creating a flow version of all indicators to view the impacts of policy and change in particular.


The standardisation and transformation of variables compiled in a composite index is another area of contention in the creation of global economic indices.


A key example of the problems faced in this arena was highlighted in 1991, when “years of schooling” was added to literacy to gauge levels of education in a nation. Here, overall educational attainment was calculated with raw data in a mixture of percentages and numbers. This reduced the implicit weighting given to mean years of schooling, calculated as a simple number. The suggestion was thus to transform the values of the two indicators on a scale between zero and one before adding them together.


GDP is a particularly difficult factor to measure as diminishing returns to income must be considered using logarithms but discounting above the poverty line is disputed. Desai (1991) makes the key point that income above that level will still make some difference – relative income deprivation persists. Excess discounting of income above the threshold level means it makes practically no contribution to the income value whilst incomes below the level are not subject to diminishing returns at all, leading to overvaluation there. The result is a skewed set of HDI values. Dasgupta (1990) believes that all variables face increasing returns due to achievement effort but Srinavasan (1994) held that only health indicators should be modelled with increasing returns as he views life expectancy as something intrinsic rather than an effort to be achieved.




It is crucial that the indicators used in the HDI require only comparable data that is available for many nations, to allow for useful international comparisons and for enough years, to allow for intertemporal analysis of progress.


Problems with data affect the usefulness of the HDI itself. Ivanova et al (1999), for example, find the index does not provide useful results as changes in the HDI could be attributed to progress or to changes in the estimation procedure. Murray (1991) takes issue with data on income in particular, as use of national currency and differing methodologies inhibit cross-country comparison.


Srinavasan (1994) finds that 87 of 117 countries lack reliable data on life expectancy. The result is that movements in the HDI are determined mostly by income, as that is the piece of data available annually. Another trade-off emerges here: should extrapolation from real data be used so that no factor is underrepresented in the final HDI figure or should accuracy be upheld. A problem that Murray (1991) identifies is that changes in the HDI are highly correlated with the timing of censuses rather than step changes in human development, suggesting that conclusions drawn from HDI figures have little significance.


The main issue with unreliable data is that cross-country comparisons become redundant. One solution to this problem is to accept the imperfection of data collection, eliminate league tables and to only categorise countries into groups, accepting that no practical meaning can be found in ordering high-achieving HDI countries and ranking nations with marginally different HDI scores. The development of an ordinal, rather than a cardinal index may be a move forward in this vein.


With data deficiencies, resources are the key issue. Many nations lack the resources to collect regular, reliable data. In this way, the work of the HDI and the HDR is seen as a solution aiding human development, not merely external, passive measurement as these analyses encourage countries to monitor and collect data.




As with any national figures, though measuring human development at the country scale provides some material for comparison, the distribution of income and the range of education and health levels is very important. Thus, Chowdhury (1991) suggests the HDI should be adjusted to account for access to services. Another potential improvement may be the introduction of a Gini coefficient for all components (Stanton, 2006). All such enhancements would require more data to be available.




Explicitly, the components of the HDI are weighted at a third each but implicit weightings differ. For instance, Kelley (1991) points out that the income dimension is given less weight than health and education because the endpoint value of income is developed countries’ poverty line whereas for literacy and life expectancy it is the highest country value worldwide.


Annual changes to goalposts are also problematic, meaning that the lowest income country needs GDP growth of 13% to maintain its index value and countries with income measures above US $ 5,860 need to lower their value for their index worth to remain static.


If a composite index is sensitive to weights, some justification for a given weight must be offered. Just as improvements to data collection have been touted as an example of the HDI improving development as well as measuring it, Streeten (2000) calls for national participation in the selection of weights for the index’s various components. This would prove valuable in displaying the public’s priorities and preferences for the HDI and for wider policymaking too.




A key change in 2010 was the shift to using geometric mean (which measures the typical value of a set of numbers) to reflect the imperfect substitutability across the three dimensions. This captures how well rounded a country’s performance is across the three dimensions. As a basis for comparisons of achievement, this method is also more respectful of the intrinsic differences in the dimensions than a simple average. It recognizes that health, education and income are all important, but also that it is hard to compare these different dimensions of well-being and that we should not let changes in any of them go unnoticed.




 “The more we discuss what should or should not be included in the HDI, the more the debate moves away from the single-minded focus on growth that pervaded thinking about development”


The difficulties of gathering reliable data and of composing an index to measure the subjective notion of “human development” are so severe that after the HDI has been in existence for a decade, disputes still arise. Thus, the true value of the HDI may not be its ability to measure economic factors. Rather, it may be this very “debate” that is the premier value of economic data, encouraging consideration of issues such as the diminishing returns to income and the factors which contribute to human development.


Tadashi Hirai, “Review of the most important critiques of the Human Development Index in the last 20 years”