Professional Documents
Culture Documents
DEVELOPMENT INDICATORS*
ABSTRACT. This paper discusses some of the methodological problems involved in the
design of indicators of development. To begin with, two kinds of social indicator are
distinguished and defined: descriptive and normative. Unlike the former, the latter
involve value judgments. Secondly, the very notion of development is briefly discussed.
The idea favored by economists, that development is identical with industrialization,
is criticized. It is proposed that genuine development is not only economic but also
biological, cultural, and political, for each of these factors conditions the other three.
The matter of independent vs. dependent development is discussed, to dispel the illusion
that increase in GNP is a faithful development indicator. An indicator of dependence is
introduced. Then a whole battery of development indicators is proposed, some of them
dominant, others weak, some relative, others absolute, some stray, and others systemic.
A case for systemic or theoretical, as opposed to stray or empirical, indicators is made.
Finally some ideas on development dynamics are proposed, as a possible basis for a set of
deeper and more faithful indicators of development.
INTRODUCTION
Until a few years ago social science proper, in contrast to social philosophy,
tried to keep clear from value judgments and normative or prescriptive
statements. The increasing application of the social sciences to the solution
of social problems, as well as the birth of decision theory, have changed that
attitude. Today there is a tendency to face up to values and norms, rendering
them explicit and keeping them under control instead of just ignoring them
and thereby being at the mercy of tacit valuations and norms. However, this
change does not seem to have penetrated the important field of social indica-
tors: practically all of the work in this field is limited to descriptive indicators.
We shall argue for the need to use both descriptive and normative indicators.
But first a quick characterization of each.
The parameters of an income distribution are descriptive. On the other
hand the parameters of an optimal income distribution are normative. Like-
wise the actual degree of participation in social or political matters, as
indicated e.g. by the percentage of the population taking part in political
decision making, is a descriptive indicator. A corresponding normative
indicator would be the optimal degree of participation ensuring both a fair
distribution of power and an efficient running of the social machine.
There is nothing wrong with a normative indicator, just as there is nothing
wrong with a prescription, whether technological, medical, social or of any
other kind, intending to correct an imbalance of some sort, as long as there
are reasons to believe that the corresponding balance is better than the im-
balance. For one thing a normative indicator can be just as objective as a
descriptive one. Indeed in principle it is possible to determine which value or
values of a variable correspond to the goal or goals agreed on beforehand.
(Think of nutrition indicators.) For another, some of the normative indica-
tors are just maximal (or minimal) values of the corresponding descriptive
indicators. For example, the optimal life expectancy may be taken to be the
one actually attained in Scandinavia.
DEVELOPMENT INDICATORS 371
2. T H E C O N S T R U C T I O N O F N O R M A T I V E I N D I C A T O R S . T H E C A S E
OF EQUITY
Eq(X)= 1 -Sk(X).
Maximal equity corresponds thus to minimal asymmetry in the distribution
and conversely.
If on the other hand the optimal distribution is asymmetrical (as is the
case with longevity and the number of schooling years) then the skewness
value itself, or some linear function of it, may be taken as a measure of
equity.
Finally, in the case of a cumulative distribution (such as a Lorenz curve
of land distribution) one would presumably adopt the Gini index as a
measure of inequality.
I submit, in sum, that it is possible to construct normative indicators.
Moreover I suggest that normative indicators of economic equity, social
balance and cultural opportunity are among the most important develop-
ment indicators.
society and the corresponding emergence of the shanty town society com-
posed of former peasants hoping to be employed in the industries or in the
services, and leading an almost marginal existence. To discard these and other
effects of industrial growth is to entertain a queer notion of development.
Ad independence. If a country, region, company or any other group
pushes the development of another area for its own benefit, then such a
development is bound to be lopsided or even harmful to the developing
region. Thus, much of what goes under the name of development, when con-
trolled by foreign interests, consists in the sheer plunder of the natural resour-
ces of the developing nation or region. Just as with the case of equity, we
have got to include independence among the development indicators, for an
industrialized colony is not more highly developed than a moderately
industrialized but independent country. But before we can include independ-
ence indicators among the development indicators we must build them. Let
us then proceed to this task.
4. I N D E P E N D E N C E I N D I C A T O R S
Finally we may wish to assign each such value a certain weight wk in the
range 0 to 1. This would allow us to compute the overall independence
indicator for each country:
5. D O M I N A N T A N D W E A K D E V E L O P M E N T I N D I C A T O R S
It is well known that one of the curses of social science is not the dearth of
empirical data but rather the overload of unimportant and highly correlated
empirical information. (Other curses are the scarcity of good mathematical
models and of good social indicators - both of which go hand in hand.)
Consequently one of the problems in every field of social science is to find
dominant, independent and reliable indicators that will enable one to
dispense with the weaker, dependent and ambiguous ones. Let us take for
example the health indicators.
Both the number of physicians and the number of beds per 1 000 inhab-
itants can be misleading indicators of public health even though they are
valuable in themselves. Indeed if the doctors and hospitals are concentrated in
few and distant towns then they may not give good service. And even if they
are in plenty and well distributed this may indicate either poor health or
376 MARIO BUNGE
6. C A T E G O R I E S OF SOCIAL INDICATORS
Most of the indicators in the above list are available. The rest can in principle
be built and evaluated with the help of existing data.
It is of course debatable whether the ones singled out as being possibly
dominant and consequently marked with a + sign are in fact dominant. For
example it might be thought that both political independence and political
stability under heading 3.2 are key indicators. However it may be rejoined
that popular participation subsumes them both, in the sense that the stronger
it is the greater both independence and stability are. But a detailed discussion
of these matters goes beyond the limits of the present study. (See Bunge and
Garcla-Sucre, 1976).
If all the indicators are quantitative and are uniformly normalized, so that
their values lie within the unit interval, their ordering becomes a trivial task.
What is not trivial is, of course, the assignment of relative weights with a view
to constructing an overall indicator of some sort, be it of economic soundness
or of cultural level or of degree of development. We turn next to this problem.
7. O V E R A L L D E V E L O P M E N T I N D I C A T O R S : R E L A T I V E
AND A B S O L U T E
the sense that it takes an arbitrary country as a base line and computes the
degree of development of any given country relative to that country. It might
be of some interest to attempt to construct, in addition, an absolute overall
development indicator, i.e. one not depending upon any particular country
taken as a base line and moreover one assigning different weights to different
indicators, according as they are dominant or not. In the following we sketch
a method for constructing a whole class of overall indicators of this kind.
Let X = {X1, X2 ..... Xn ) be n development indicators defined as so many
functions
Xk : P x T ~ [0, 1]
such that Xk (p, t), for p in P and t in T, is the degree to which the kth
respect has been developed in country p at time t. Furthermore pick the
subset X i of the total set X such that no two members of X i are strongly
correlated. (That is, X i is a set of practically mutually independent indicators.
To build it we must begin by finding the value rpq of the linear correlation
coefficient of every pair (Xp, Xq) of indicators in the original set X. We
assume this statistical job to have been performed.) Suppose the set X i of
independent indicators has m ~<n members. Then the overall degree of
development of country p at time t is defined to be
m
D(p, t) = kZl wk(p, t)Xk(p, t),with ~ w k = 1,
where w k (p, t) is the weight or importance of the kth aspect for country p
at time t.
Note that the weights wg have been assumed to be country and time
dependent, for (a) what is important to one country may not be so to
another, and Co) the same factor may acquire or lose importance as the goals
are attained or recede further away. Note also that those weights are not
related to the degree of statistical independence of the corresponding variables.
The problem of the statistical independence is supposed to have been solved
prior to the selection of the subset X i of independent indicators. The weights
wg are assigned by those who design development plans, not by the statisti-
cian.
Finally suppose we reduce further the set of development indicators by
picking only the dominant ones, i.e. those which, being statistically
independent, have also maximal weight for the country and time in question.
DEVELOPMENT INDICATORS 379
That is, form the subset X a of X i and assume that there are q indicators in
this set. Suppose further that these indicators are all equally important, i.e.
set w s = 1[q for every X, in X a. Then the formula for the overall degree o f
development simplifies to
q
O(p, t) = (l/q) s~=l Xs(p, t), with X s in X a.
It goes without saying that the rate o f overall development is defined by the
time derivative of D:
These are absolute indicators. But of course they allow one to compare levels
(or rates) of development of different countries. That is, an absolute indicator
gives all the information a relative indicator conveys without introducing any
irrelevant information, namely the one referring to the base line country.
9. D E V E L O P M E N T KINEMATICS
Ak : P x T ~ R ,
where P is the set of countries (or zones of some kind), T the time interval,
DEVELOPMENT INDICATORS 381
and R the set of real numbers. Then the value Ak (P, 0 , for p in P and t in
T, will represent the level of development of country p, at time t, in respect
k. For example, if A s represents the degree of industrial independence, then
'A s (P, t) = 0.5' means that the degree of industrial independence of country
p at time t is one half. Likewise if we represent by
Rk(P, t) = d/dAk/dt
the rate of change of the kth aspect, then a formula such as 'Rs(p, t) = 0.1'
means that country p, at time t, is obtaining its industrial independence at
the rate of 10% per unit time (e.g. per annum).
Since development is a many sided process we need all the variables and
their rates of change. Moreover we had better lump each set into a column
matrix, so as to be able to deal with all aspects at the same time. In other
words, we display the level of development of country p at time t by writing
A I (p, t)
A2(p, 0
a(p, O=
an(p, 0
and the rate of development as the time derivative of the former:
Rl (p, t)
R2(p, t)
R(p, t)= " ,with Rk(p, t)= dA(p, t)/dt
Rn (p, t)
The state of development of country p at time t may be construed as the
ordered pair <A (p, t), R (p, t)> = <A (p, t), ~ ) > . This may be regarded
as a vector whose fh'~t coordinate or component summarizes the level, and the
second the rate of development. As time goes on this vector (or rather its tip)
describes a trajectory in the state space of the developing country, which
space is a 2n dimensional cartesian space. So much for the kinematics of
development.
382 MARIO BUNGE
dA (p, t) _
dt F(p, t)
where the RHS is the prescribed 'force' matrix, each element Fk of which is
supposed to 'drive' the corresponding state variable Ak. In general each Fk
is a function of time and of Ag itself as well as of some other variables in
the set A = {,41, A2 ..... An) of state variables. In other words, the develop-
ment equation is in general of the form
d__AA= G(A, t)
dt
for each country p. Knowing the initial values of the various state variables
and prescribing the F~s we shall be able to forecast the future values of all the
components of Ak. If these values do not agree with the desired or goal
values then we shall have to change the 'force' assumptions. It goes without
saying that such readjustments cannot be made arbitrarily but are constrained
by the natural and human resources of the country concerned as well as by
the psychological, economic and social laws, not to speak of considerations
of political opportunity.
The simplest (and therefore perhaps also the worst) G function is of course
the linear function. In any case we may start by hypothesizing that the rate
DEVELOPMENT'INDICATORS 383
where the aq, for i, ] = 1,2, are real numbers. For certain combinations of
these coefficients there will be progressive development on both fronts, for
others decay, for still others stagnation, and again for others a combination of
the various elementary modes of development. Thus for a n = azz = 0 and
In this case there will be growth in the two aspects just in case at I> bt; other-
wise there will be decay.
We emphasize that the above is only a general framework for development
models. The actual construction of specific models, be they descriptive or
prescriptive (normative), is a tough task that goes far beyond methodology.
Yet it may well be the case that a closer look at methodological matters could
speed up the process of modelling development.
CONCLUDING REMARKS
nations. This warning may sound pedantic but it does look necessary in view
o f the fact that development is often equated with industrial growth - which
should be just one aspect of a many sided process.
If development is conceived as an integral affair, one involving the biolo-
gical, economic, social and cultural aspects of a community, then the very
choice of development indicators is bound to differ from the usual set of
indicators. For one thing indicators of self reliance (or independence) and of
fairness (or equity) should be included among the dominant indicators of
development. Now, an indicator of equity or balance in the level or the distri-
bution of something, be it consumer goods or educational opportunities, is
not a descriptive indicator like, say, the GNP: it is a normative or prescriptive
indicator. As such it is unacceptable to anyone committed to the ideal of a
value free social science.
We have suggested that this ideal is wrong and dead anyway: that the
proper scientific attitude is not to ignore values but to reckon with them and,
whenever possible, to regard them as optima (not necessarily maxima). By so
doing we do not escape objectivity but rather on the contrary, and we come
to share the attitude of the biologist and the engineer, both of whom are
centrally interested in Finding or even prescribing the optimal values of
certain parameters, i.e. those corresponding to the most efficient functioning
of the system, be it organism or machine.
Accordingly we have proposed methods to construct quantitative
normative indicators, in particular development indicators, that should be
included in any formula for the overall development indicator. We have also
proposed an absolute indicator of the latter kind, i.e. one that does not
depend on any particular country chosen as a base line. This absolute
indicator, giving the overall development level D(p, t) of country p at time t,
allows one to compute the distance, in development level, between any two
countriesp and q at a given time t, namely thus: d(p, q, t) = ID(p, t) - D ( q , t) I.
The converse inference is of course impossible.) Finally we have sketched a
nonspeeifie model of the dynamics of development and stressed that all the
development indicators should participate in some such models, for otherwise
they bring little insight and they cannot be validated in a reliable matter.
NOTE
* An earlier version of this paper was presented to the UNESCO meeting on Indicators
of Social and Economic Change, Paris, 2 0 - 2 2 May 1974.
BIBLIOGRAPHY