Another Perspective on the Human Development Index

A few days back I wrote a post claiming that “for all the work that goes into the Human Development Index, it just doesn’t tell you much that you wouldn’t learn from simple comparisons of G.D.P. per capita.” Subsequently, Francisco Rodriguez, who heads research at the UN Human Development Report Office, touched base to tell me that he thought I hadn’t told the whole story. Francisco is a terrific macroeconomist (in fact, he was the TA when I took my graduate macro classes at Harvard), and so he kindly agreed to write a guest post filling in the missing pieces.

What Does the Human Development Index Really Measure?
By Francisco Rodriguez
A Guest Post

Last week, Justin Wolfers presented a simple yet apparently powerful critique of the Human Development Index (H.D.I.) — a summary index of per-capita income, health, and education indicators published by the United Nations Development Programme. The correlation between the rank of H.D.I. and that given by G.D.P. is 0.95, and a scatterplot of one against the other looks like a 45 degree line plus measurement error. Justin’s critique is not new, and neither is my response. Indeed, the literature on the links between human development and per-capita income is quite vast, including the entire 1996 Human Development Report.

The criticism that Justin levies at the H.D.I. is based on the high correlation between the rank of the H.D.I. and the rank of G.D.P. If you think that development economics is purely about ranking countries, then Justin is right; both indicators do give similar comparisons.

But we care about human development not just because we want to know how to rank countries. Rankings may be a good way to compare NCAA teams or American Idol contestants, but are not necessarily the best way to think about differences in living standards across countries. For example, a cross-sectional rank correlation cannot tell you much about two key issues that have been the focus of much of the cross-country macro development literature: understanding what drives improvements over time in well-being and understanding how inequality across countries has evolved.

On both of these questions, the H.D.I. and G.D.P. give considerably different answers.

Consider first changes over time. Improvements over time in human development differ significantly from growth rates of per-capita income. The figure below shows the relationship between the growth rate of per-capita income and the change in H.D.I.; the correlation is 0.43. The correlation between changes in the non-income components of H.D.I. and the growth rate of G.D.P. is 0.03. That is, the information contained in changes in health and education appears to be very different from that contained in changes in income.

INSERT DESCRIPTION

These differences between the H.D.I. and G.D.P. suggest some very different priorities.

For example, suppose we run two regressions: one to try to explain changes over time in per-capita income, and another to explain changes in H.D.I. Let’s use the same exact set of independent variables, which come from the list of usual suspects from the growth literature: initial values of income, schooling and life expectancy, and current values of openness, inflation, and the rule of law.

Here’s what we find: inflation is negatively and significantly related to changes in H.D.I. but not to growth. Trade openness and the rule of law are positively and significantly related to G.D.P. growth but not with H.D.I.; in fact, openness gets a negative though insignificant coefficient in the H.D.I. regression. In other words, the implications of cross-country regressions for development policy depend crucially on whether you are interested in raising G.D.P. growth or in increasing the H.D.I.

Alternatively, let’s think about international disparities in living standards. Researchers have tried to sort out whether inequality across countries is increasing or decreasing. As the figure below illustrates, while cross-country dispersion in (log) income per capita has been increasing, dispersion in H.D.I.’s has been declining. Again, looking at H.D.I. and looking at per-capita income gives substantively different answers.

INSERT DESCRIPTION

Research carried out at the Human Development Report Office over the past 20 years has been devoted to understanding how different national and sub-national policies can make a difference to the expansion of individuals’ substantive freedoms to lead the lives they value. Covering topics as diverse as gender inequality, cultural liberty, and climate change, our reports have often found that the best policies for enlarging people’s choices are not necessarily the best ones for raising per-capita incomes. These results are of course contestable and should be debated. But this is a discussion that we could not even begin to have were it not for the concept of human development.

Leave A Comment

Comments are moderated and generally will be posted if they are on-topic and not abusive.

 

COMMENTS: 11

View All Comments »
  1. frankenduf says:

    this post quantifies the critique of ‘free’ trade- that it negatively impacts an affected community’s HDI via lowering wages and export-driven economies

    Thumb up 2 Thumb down 0
  2. Jonathan says:

    I’m not sure I understood all this, but if GDP and HDI both result in the same ranking of countries then wouldn’t a change in one have to eventually be matched by a change in the other? that is, an increase in GDP would eventually lead to an increase in HDI in order for the two to consistently give matching rankings.

    Thumb up 4 Thumb down 0
  3. jonathan says:

    Different jonathan here, but I was also confused by the post. It doesn’t address the criticism, that the correlation is .95 to GDP, but instead says there’s value in the various decompositions. OK. But there a lot of things going on inside what appears to be an inert rock that are interesting to those who look at the insides of inert rocks, but they don’t then present an index for inert rocks because they’d have a correlation of .95 (or more). In other words, why have the index? What does it say? That a lot of stuff goes on inside countries but they end up dispersed according to GDP? I found this post unsatisfying.

    Thumb up 0 Thumb down 0
  4. GVS says:

    I, like Jonathan, am a little lost. How does the ranking of countries by (income excluded) HDI compare to GDP and HDI? If the purpose of the HDI becomes apparent only after taking out income factors, why include it in the first place?

    Sounds like the income coefficient has the greatest impact on HDI and once that’s removed, subsequent factors play a greater role…

    Thumb up 2 Thumb down 0
  5. Mike, London says:

    I thought this was an excellent response to Justin’s initial post and raises some very interesting points (particularly that mentioned above by Frankenduf RE the implication of free trades impact on HDI).

    I think the Jonathans and GVS however, have totally missed the point. Firstly, the correlation between GDP and HDI over time, which is a far more important measure than a simple cross section snapshot between the two, demonstrates clearly that the HDI does not rely entirely on income factors (the correlation coefficient is less than .5).

    The second point which Fransisco makes is also significant. The disparity shown between living standards across countries is improving according to the HDI even if income inequality is not (according to GDP) This obviously poses some very interesting policy decisions, and moreover, makes the HDI a very important and worthwhile measure.

    Thumb up 2 Thumb down 1
  6. jonathan says:

    Mike – #5 – as the small j jonathan – I got the points but they don’t explain why they have an index. Again, there may be a lot of movements within any set of data, meaning the decompositions are valuable, then you have a bunch of questions to answer such as:

    - why aren’t the decompositions showing at top?
    - why are you using an index which correlates differently than the decompositions you then point to as important?

    It sounds to me like the Index is a political operation which is adequately described exactly as the original criticism stated, but that the data within the Index has some utility. It thus seems to me they should be compiling a different Index or coming up with an actual defense of why this Index is what it is other than saying that it hides certain useful stuff.

    Thumb up 0 Thumb down 0
  7. Adam Berkan says:

    I think the important criticism of the HDI is that too much of it is decided by GDP. It’s fine to say that excluding GDP from HDI produces useful data, but then I have to look at both numbers to see what’s happening.

    The HDI needs to have it’s income component scaled down so that non-income elements become significant. Then the 45 degree line starts to look a little more like a scatterplot.

    Thumb up 1 Thumb down 0
  8. johnleemk says:

    The regressions don’t prove anything about the impact of trade on HDI. That’s the whole meaning of “statistically insignificant.” All we can say about trade is that we have no evidence it impacts HDI in a meaningful way, whether positive or negative. The regressions certainly do not say that trade is clearly negatively correlated with HDI.

    Thumb up 0 Thumb down 1