Although the UNRISD study might have been classified in this section, we believe that the two studies by Adelman and Morris that we will now describe, offer a slightly more dynamic approach to the problem.

In a book, Society3 Politics and Economic Development (1967), these two authors developed a number of development indicators that they use in two articles we will discuss.

In their first study (1965), they want to analyse the inter­

dependence of economic and non-economic factors in development. They use

74 less-developed countries and 22 social, economic and political indica­

tors. They allot an ordinal score for each country on each indicator.

Then they apply the methods of factor analysis to these indicators and

GNP and extract four factors explaining 66% of the intercountry variations

in per capita GNP. They identify the factors as socio-cultural concomm-

itant of the industrialisation and urbanisation process, westernisation

of the political institutions, character of leadership and social tension.

They also apply the same method to geographical sub-groups and find only

slight differences with the main study.

In their second study (1968), they intend to group 73 less-

developed countries according to their level of development. They use as

original criteria the rate of growth of per capita GNP to perform a

preliminary classification into 3 groups then they apply a discriminant

analysis to 29 social, economic and political indicators, in order to

select the indicators that add most to the explanation of the variance

between group means given the other variables already included. Thus

they obtain a discriminant function of four variables; improvement in

effectiveness of financial institutions; improvement in physical over­

head capital; degree of modernisation of outlook and leadership and

commitment to development, which permits them to regroup the countries

and to classify those left unclassified.

The main feature of these studies rests on the indicators

chosen. They are complex indicators often based on judgmental elements.

As a consequence, they can usually only be given ordinal score. The first

study yields four rather vague socio-political factors. Similarly,

the second study results in a discriminant function of four of the most

Rayner (1970) made a thorough criticism of the methods involved; he points out a certain irrationality in using GNP as one of the variables in the factor analysis study if GNP is to be explained. In effect, factor analysis treats all variables equally and does not preclude any causality like regression analysis. Rayner also notes that, in the discriminant analysis study, Alderman and Morris first group the indicators according to a certain success criterion, the rate of growth of GNP, so this may influence their conclusions when they regroup the countries with the discriminant function.

In a more recent development oriented study, two authors,

Takamori and Yamashita (1973) plan to construct several composite indica­ tors reflecting different dimensions of socio-economic development and to assign scores on these indicators for each country. The study is based on the assumption that development is an overall process of social changes all going in the same direction, but at different paces according to the different stages of development. They choose 45 monetary and non-monetary

indicators and break down their 79 countries into 2 groups of developed and underdeveloped countries. Then they perform a factor analysis for each group of countries in order to identify categories of indicators representing certain socio-economic concepts. Once a category is branded, a separate principal component analysis is performed on these indicators only in order to build an index for this socio-economic concept. These variables are then excluded from the total pool of indicators and the whole process is repeated until they finally identify 6 categories;

economic activity; standard of living; cultural level; industrialisation; urbanisation and agricultural proportion. They also use as a criteria the correlation between GNP and the indicators. It is not clear why they do not use methods of discriminant analysis and why they go to so much trouble to achieve a rather obvious grouping. This study demonstrates the pitfalls of the methods of multivariate analysis which will always

yield the preconceived results if it is fed and set up accordingly.

Finally, their analysis of the various countries scores on the 6 indices

based on a straightforward principal component analysis, is much more

commendable than the grouping method.

Divatia and Bhatt’s study Q.969) is the only non-monetary pro­

ject concerned with inter-temporal considerations, rather than performing

cross-country comparisons. Also, unlike the authors just quoted in the

development section, Divatia and Bhatt are perfectly aware of the short­

comings and of the limitations of the multivariate analysis method they

choose to use. Their purpose is to correct the general rate of growth

of the Indian economy as measured by the rate of growth of per capita GNP.

They affirm that this rate does not represent the real nor the potential

rate of growth of the economy because of the presence of many disturbing

factors related to development which blunt the image. Consequently, they

want to determine the rate of growth of the sectors which are the determ­

inants of development, as they argue that this is where growth is most

crucial. They carefully identify these sectors as entrepreneurial/

managerial ability, capital, skills, employment of labour and technical

change. They choose 21 monetary and non-monetary indicators representative

of these sectors. Although they plan to study only changes in stocks,

the availability of such indicators is so scarce that they have to include

changes in flows too. They apply factor analysis to a time series (10

years) of these 21 indicators (standardised) in order to obtain the weights.

As a result, the first factor, identified as an index of growth, explains a

very high proportion of the total variance. However, as they planned to

obtain only one factor they might as well have used the slightly more

straightforward technique of principal component analysis. By weighting

the standardised indicators by the factor loadings or weights obtained,

index over the 10 years. They conclude that the rate of growth on this

index is twice as high as the rate of growth of GNP. However, from a

purely mathematical point of view, such conclusions would seem to be

unwarranted, as one cannot construct an index of growth with a factor or

a principal component, as these are essentially relative measures.

Divatia and Bhatt’s study can be regarded as one of the most interesting

attempts to aggregate non-monetary indicators for a definite purpose.

The last two studies that we will now briefly describe are

interesting for their methods, but they are really only on the borderline

of economics.

A study by a geographer, Berry (I960), represents the first

application of factor analysis to non-monetary indicators, but was over­

looked by most economists as it was published in a collection of geograph­

ical essays. It is an attempt to identify and differentiate the under­

developed countries using a sample of 95 developed and underdeveloped

countries. Berry extracts out of 43 economic, social and demographic

indicators (mainly non-monetary) three main factors, a technological scale,

a demographic scale incorporating features of population pressure and a

group of poor trading nations,which permit him to find patterns and

similarities between groups of countries.1 It is a descriptive approach

with no ambition to explain purely economic facts.

Finally, a study by Harbison and Marhunic (1970) must be men­

tioned as it introduces a rather unusual technique, the taxonomic method,

which was developed by a group of Polish mathematicians. The authors'

aim is to classify and rank 112 countries on the basis of 40 economic,

social, human resource and health indicators. In some way, each country

11 With the help of a discriminant function, he also inquires into the

existence of regional underdevelopment and finally, with the help of regression analysis, he tests simple hypothesis concerning some characteristics of the underdeveloped countries.

is defined uniquely in an n-dimensional space by the difference between its score for each indicator and the scores of an ideal country defined as having the highest scores on each indicator. This method provides a single ranking for any number of indicators. It is also possible to derive the shortest distance between countries and to link them. Groups of countries with similar characteristics can then be determined.

This method can also be extended to derive targets for development and to predict missing scores for a country by using the indicator of its closest country. This is a purely descriptive method based on no economics. Consequently, the applications mentioned above are


In conclusion, the main aspect that links these studies to­ gether is their use of non-monetary indicators to measure some aggregate index with economic overtones. We will now present the non-monetary indicators and briefly discuss our main designs for them.


In document International measurements of economic and socio-economic aggregates using non-monetary indicators (Page 81-86)