(Human) Development Indicators in Brazil: on the importance of the theoretical background
Mattos, Ely Jose (2016). '(Human) Development Indicators in Brazil: on the importance of the theoretical background' Paper presented at the annual conference of the HDCA, Tokyo 2016.
Since the very first version of the Human Development Index (HDI) in 1990, researchers, policy makers and practitioners have become more and more engaged in the difficult task of measuring development. Although that there had been a number of experiences on constructing development indicators before, the HDI proved to be groundbreaking because of two complementary elements: the theoretical conception and its subsequent implementation. On one hand, the human development paradigm has challenged the traditional monetary approach of well-being evaluation, promoting a new perspective on the concept of development itself. On the other hand, the operational methodology has been designed, by Amartya Sen and Sudhir Anand, just as adherent as possible to that theoretical foundation.
As pointed out by the quite elucidative OECD Handbook on Constructing Composite Indicators (OECD, 2008), a solid and reliable index strongly depends on a well-established theoretical understanding of the phenomenon under study and also on an anchored translation of this knowledge into the operational methodology – the HDI has been successful for more than two decades because it has managed to handle both. In this context, the main objective of this proposal is to debate the importance of dealing properly with this double challenge for the construction of good development indicators.
This study proposes a comparison of three different Brazilian experiences on constructing development indicators for municipalities – which are very important tools considering that Brazil counts on almost 6000 municipalities spread over 27 different states. The first one is the Human Development Index for Municipalities (HDIM), which is an initiative of the United Nations Development Programme in Brazil (UNDP). The HDIM is an explicit methodological adaptation of the original HDI for the municipalities. So its theoretical background is the human development approach and the methodology follows most of the principles that can be found in the traditional HDI. Considering such a straightforward index it will be the benchmark for the comparative exercise.
The other two, although sometimes claimed to be inspired by the HDI in some sense, are not directly based on it at all. The first one is the Indicador Social de Desenvolvimento dos Municipios (ISDM) by the Center for Applied Microeconomics of the Getulio Vargas Foundation. The other one is the Indice Firjan de Desenvolvimento Municipal (IFDM) by the Industry Federation of the State of Rio de Janeiro.
Those three indices are the only ones estimated for all the municipalities in Brazil. Because of discontinuities in the calculation of the indices through the years, due mainly to the different frequencies of data-sources used, this study applies a cross-sectional approach for the year of 2010. For this year it is possible to count on results for the three indices.
Regarding the theory – assumed here as the first challenging element for and index – the ISDM and the IFDM are both very unspecific. It is not clear in the technical documents of both indices what theoretical approach has been adopted. At most one can find a broad and foggy definition of development. For those two indices, development is understood like an extensive conceptualization of “quality of life” based on some specific variables. There is some clarity, though, that multidimensionality is mandatory for development evaluation, but it is not clear how to conceive it. As a result, one of the most important symptoms of theoretical misunderstanding is the absence of differentiation between means and ends (or between endowments and achievements). The HDIM, on contraire, is clearly based on the human development approach and takes advantage of all the experience of the traditional HDI.
Regarding the second challenge for a good index – the translation of theory into practice – two things can be considered for the sake of comparison: the dimensions and variables derived from the theoretical background, and the statistical techniques applied. The ISDM is composed by 28 variables categorized into five dimensions (housing, income, occupation, health, and education). The IFDM counts on 21 variables categorized into three dimensions (occupation and income, health, and education). Considering that no theoretical reference is objectively assumed, it is hard to analyze how suitable the variables are. Anyhow, it is possible to say that the set of variables points towards a very broad conception of multidimensionality where the range of different themes seems trying to fulfill an objective of completeness. The HDIM, in turn, is composed by just 4 variables and three dimensions: life expectation, education, and income. All the variables are quite similar to those in original HDI.
The three indices considered are similar in terms of statistical techniques. The variables are normalized in order to obtain a comparable scale. None of the three indices presents any method for weighting variables or dimensions which is worth noting as a novelty – they are basically arbitrary. The aggregation method is an arithmetical mean for the ISDM and IFDM and a geometrical mean for the HDIM.
In summary, three different indices are currently available for policy makers in the Brazilian municipalities. All of them claim to measure and evaluate “development”. But only the HDIM defines clearly what development is being assessed. Furthermore, distinct sets of variables are used, making even more difficult for the stakeholders to select the most suitable tool according to some specific canon.
This scenario becomes seriously complicated when the numerical results are compared. The 10 or 20 best municipalities, as well as the 10 or 20 worst, are not surprisingly almost the same for all the three indices. The problem relies in the middle of the distribution, where the most of the municipalities are placed. For that particular (and big) part of the distribution, the results are systematically contradictory for the different indices. A brief description of the distributions: HDIM, bimodal; ISDM, unimodal negatively skewed; IFDM, normal symmetrical. It means that stakeholders are at a crossroad when selecting a development index. Different results for different indices are expected; the actual problem appears, however, when the respective theoretical background does not properly explain those differences - which is the case presented here.