Disaggregation of HDI of relatively large countries at medium level or at the edge of high human development indicate to regional disparities. The results on China, Mexico, Columbia and Brazil show similar disparities. The policy implication are regional equity at medium level of development. Another common feature is that the regional disparities usually coincide with ethnical problems.

The variation of HDIs in Mexico is almost as wide as in China (TABLE 7). The state with the highest HDI has 6.8 times higher HD compared to the lowest state (0.874/0.128). These disparities have caused in Mexico very recently serious unrests.

Source: Population Reference Bureau (1993)

The disaggregated HDI's are calculated for Colombian "departamentos" by FRESNEDA, 1993. Colombia exhibits a wide range of HD, too. The indexes range from low (0.255) to high HD (0.870) in Bogota. The improvement of average HD is in Colombia very fast. From 1978 to 1992 the average increased from 0.614 to 0.794. Yet one tends to say, that the recent (1985) distribution of HD among regions seems to be as uneven as in 1973. The disparities persist.

Source: O. FRESNEDA (1993)

The persistence of regional disparities along improvement in HD can be demonstrated best for Brazil. The study on Brazil (A.G.SPINDOLA,1993) allows a graphical analysis (Graph 6). Except the shift between south and south-east region from 1970 to 1980 the ranking of the regions remained. The disparities among regions have improved from 1970 to 1980 but it increased in 1990 compared to 1980 in spite of an increase in HD. Further disaggregation within regions could have demonstrated this problem much clearer. The disparity index for 1970, 1980 and 1990 are 22%, 15% and 16% respectively.

Source: A.G.SPINDOLA (1993)

Urban - Rural disparities

Disaggregated HDIs indicate at medium level of development to another important disparity; disparity between urban and rural areas. The studies on Egypt and Turkey, two countries at the two opposite edges of medium development make this problem apparent.

The study on Egypt (El-laithy 1993) deepens the disaggregation of HDIs one step further. Here, the disaggregation starts with regions, it is then extended to subregions "governorates" and the study differentiates further between rural and urban areas in each region and governorate. Two steps of this disaggregation is presented in Graph 7. The five step, ladder type disaggregation is at regional level. There are five regions. The detailed disaggregation is by governorates and rural-urban differentiation. Deepening of the disaggregation increases usually the area of inequality, here, only slightly.

Source: H. EL-LAITHY (1993)

Source: H. EL-LAITHY (1993)

The disparity index is 16% and 18% if measured by regions and then by governorates, respectively. The difference between five region disaggregation and thirty eight rural and urban governorate disaggregation is more or less the same. The difference is only two percent. Regional disparities in Egypt amount to 16% of its HDI. This means if regions are ranked according to their HDI's, the amount of the HDI (area) to the right of the average is eight percent and eight percent is needed for the rest to reach the average level. The disaggregation into urban and rural areas does not change these disparities. The regions have most probably proportionate urban and rural characters. They capture urban and regional differences also without disaggregation. A closer look at the results in the study show that the first disaggregation (at national average level) into urban an rural areas gives a range of HD between 0.38 and 0.54. The first subdivision into regions widens the range only from 0.34 to 0.57 (EL-laithy 1993, 50).

Is further disaggregation possible? Yes, the disaggregation could have continued by disaggregating HDI for males and females. The study on Egypt calculates these numbers as 0.27 for females and 0.57 for males as national averages. The HDI for woman is lower than the index of any rural area. If the male and female indexes could have been calculated for the rural and urban areas separately, for each governorate, the widest range would have been observed, most probably, between man in urban parts of Lower Egypt and woman in rural areas of Upper Egypt.

A similar study calculates disaggregated HDIs for 67 provinces in Turkey.

The cumulative distribution function which has the steeper slope represents province aggregates, that one with the flatter slope, which starts by a lower HDI at the bottom end ends by a higher HDI at the top is achieved by disaggregation of the provinces into rural and urban areas. Only this data is presented in Table 10. The deepening of the analysis here to urban and rural areas increases the disparity index from 18% to 23%.

Both countries exhibit similar urban rural disparities. Turkey at a higher level of medium HD exhibits both wider variation and higher disparity. The urban and rural disparities point in both countries mainly to the deficit in education, especially to the education problem of rural female.
 

Large versus small countries

The First HDR had produced separate HDI ranks for small countries and for those above one million population. This distinction has disappeared after the Second Report. However, in some cases it is still important. The regional, urban-rural disparities are for small countries, islands usually nonexistent. In Trinidad and Tobago, which is a relatively small country at a relatively high level of HD, the largest distances anywhere in the country do not exceed two hours. Under such conditions, the concept of a region, urban-rural difference becomes irrelevant; organization of education and health services become quite easy. The only disparity which may come to the fore at high human development, in a small country is ethnical disparity. The ethnicity becomes a source of problem not necessarily because of the qualities of ethnicity, but rather because of the initial conditions, where ethnical differences correspond to a certain, usually colonial, division of labor. Economic policy changes affect this division of labor in time not equally.

Source: R.M. HENRY (1994)

Papua New Guinea gives the opposite example. It is larger than Trinidad and Tobago but it consists of several islands, yet the average HD level in this country is quite low and the disparities are tremendously wide (Table 12), mainly because of initial conditions which are closely related to ethnicity and high income disparities. The study on Papua New Guinea (N.A. FERNANDO, 1992) allows the presentation of disparities by the help of a cumulative distribution function (Graph 9). It should be noted that the profile has made a full leftward shift. This pattern is probably true for Trinidad and Tobago, too. However the HDR's don't indicate to such developments in these countries. Papua New Guinea gives in this study the unique example of a full leftward shift, which as might be expected, causes increasing disparities. The disparity index for Papua New Guinea has increased by this shift between 1972 and 1980 from 38 to 44 percent.

Source: N.A. FERNANDO (1992)

High human development and disaggregation:

High HD suggests potentially limited disparities. Development is achieved by reducing urban-rural and regional differences to acceptable levels. What remains are ethnical, economic and gender disparities. They are, for example, all observed in USA, however, in a much moderate range than in South Africa at the medium level of development.

Other disparities or problems which are existent in many of the top ranking countries, are not reflected to their HDI's yet. Well known problems are crime and drug abuse, besides the so called migrant, guest workers who are potential minorities of these countries are still statistically not members of respective countries.
 

Disaggregation from the perspective of gender
 

As HD progresses the disaggregation stresses at each level a different problem with respect to gender. At low levels it indicates mainly to health problems. In many regions where HD is low, the life expectation of the women is also quite low compared to males. For example, in many Indian states life expectation of females are quite close or lower than male life expectancies. At medium level of development it becomes more a problem of education, which is caused by the large distances and traditions of rural life. The most disadvantaged women there lives probably in the rural part of an ethnically disadvantaged region. At higher levels of development gender problems are linked more to ethnical and, or economic disparities. At quite high economic performance they still persist as economic disparities.
 
 

Disaggregated HDI of the world and average world HDI
 

The relationship between HDI and disaggregated HDI may be evaluated best by a reflection. One may consider the World itself as a country and the countries as regions of the world. Then the list of HDIs of countries are nothing but disaggregated HDIs of the World. Using the data in HDR 1993 (HDI's and population estimates) the world HDI profile has been presented in Graph 10. According to these results HDI for the World is calculated as 0.56.

The relatively long vertical range just above the world HDI-line represents China in Graph 10, but recalling Graph 1, the indexes there, were calculated according to the same maximum and minimum values. Is it now possible to claim that Shanghai is the most developed part, and Tibet is the least developed part of the World? To put the question more precise, are we allowed to use the HDI scale in HDR 1993 in order to evaluate the disaggregated HDI's in China or can we say, that understanding of China's Human Development is the same as understanding the World? Technically, there should be no objection, as long as the same maximum and minimum values have been used for all components of HDI. However, if we consider our starting point we shouldn't. Disaggregated HDI's have been calculated because national averages have been found as insufficient for national policy formulation. The evaluation of "improved" indexes by "inadequate" ones may pose several inconsistencies. The comparison would have been an appropriate one if each country would have produced disaggregated HDI's by the same technical conventions and if the World's profile would have been scaled by these new findings.

As a last attempt Graph 8 shows the effect of incorporating China's disaggregated data into the World data. The vertical part in Graph 10 disappears, which means an increase in disparities. The disparity index for the world was without disaggregation of China 36%.

Suggestions for future work in disaggregation
 

Aggregation and disaggregation of development indexes for the purpose of policy formation are becoming quite relevant as recent developments indicate. It is Germany which needs new policies after integration with East Germany. All new republics of former Soviet Union search also for new policies after disintegration. Many countries are trying to become a member of a new regional integration. All these developments require specific policies within a whole, which is nothing else but the philosophy of disaggregated HDI.

Some experience derived out of the studies reviewed are worth to share. The studies cited here are selected out of a larger number. Those not mentioned have still insufficient data for the completion of the disaggregation. Data requirement is a serious constraint. Those who had or collected the data, had to proceed by making restrictive assumptions. This seems to be an inescapable feature of disaggregation. However, almost all studies differ where they have done assumptions. They are specific to their country and data. More common guidelines could ease comparability. As all HD efforts indicate that relevant policies are derived mainly by international or domestic comparison. The comparability of disaggregation studies are essential.

The disaggregations have measured, probably, already well-known disparities, gaps in those countries. Their measurement by the HDI helps, first, to rank them and in future it may help to follow them up, in what direction they are changing? The authors themselves evaluate their work as useful but feel that the measurement and causes of the disparities should be linked by additional studies. There is a marked attempt to link HDI to other known (disparity) measurements. Many studies try to relate HDI and Poverty Indexes.

The graphical tool introduced here may also be refined. It may be worth to fit the observed cumulative distribution functions into known mathematical functions. If the cumulative distribution functions fit only a specific functional form, the mathematical properties of that function could be useful for the interpretation of the human development profile. The cumulative distribution function here is in this respect the first step.
 
 

Conclusion
 

HDI measures the average HD and disaggregation measures the distribution of HD, ie., disparities. There is a fine difference between improvement of HD and improvement of HD without causing inequality. Those who are seriously interested in closing gaps will need the disaggregated measurement.

S. ADAMU (1993) Disaggregated Human Development Index within Nigeria, (unpublished study for UNDP).

A.H.AKDER (1992) UNDP's Human Development Report and Turkey:Country Profile", in Report on the First National Human Development Conference, 7-8 September 1992, Ankara- TURKEY, p. 13-32.

S. ANAND, A. SEN (1993) Human Development Index: Methodology and Measurement, (unpublished study for UNDP).

B. DUTTA, M. PANDA, W. WADHWA (1994) Human Development in India, An Inter-State Analysis, (unpublished study for UNDP).

H. EL-LAITHY (1993) The Disaggregation of Human Development Index for Egypt, (unpublished study for UNDP).

N.A. FERNANDO (1992) Provincial Human Development in Papua New Guinea: Its Level and Inequality 1972 and 1980.

O. FRESNEDA (1993) Informe de Avance Sobre Recopilacion de Indicadores de Desarrollo Humano Para Columbia, (unpublished study for UNDP).

R.M. HENRY (1994) Trinidad and Tobago: Human Development Index 1993, (unpublished study for UNDP).

A.K.S. KUMAR (1990) The UNDP's Human Development Index: A Computation for 17 Indian States,
(unpublished study for UNDP).

Y.K. LENG, T. AZIZ (1993) Disaggregated Human Development Index (HDI) of Malaysia, (unpublished study for UNDP).

J. MIJAKOWSKA, L. BOLEWLAWSKI, H. JAKUBOWSKA (1993) National Human Development Report, Poland. Study on Operationalizing Human Development, (unpublished study for UNDP, prepared under the guidance of Prof. A. Ochocki).

J.C. OBAME (1993) Etude sur les Indices Sectoriels du Development Humain au Gabon, (unpublished
study for UNDP).

S.P. PAL, D.K. PANT (1994) An Alternative Human Development Index, (unpublished study for UNDP).

A.G.SPINDOLA (1993) The HDI and other Development Indicators of Brazil, (unpublished study for UNDP).

UNDP (1993) Human Development Report 1993, New York, Oxford University Press.

C. ZHIZHOU (1994) Disaggregation and Analysis of Human Development Indicators in China, (unpublished study for UNDP).