World Development, Vol. 20, No. 1, pp. 119-131, 1992. 0305-750X/92 $5.00 + 0.

00
Printed in Great Britain. 0 1992 Pergamon Press plc

On Measuring the Quality of Life

PARTHA DASGUPTA and MARTIN WEALE

University of Cambridge

Summary. - It is widely recognized that per capita income is not an adequate indicator of
aggregate well-being. It has thus become customary to use a range of socioeconomic indices for
assessing the quality of life in a society. In this article we extend measures of general well-being in
current use by including ordinal indices of political and civil liberties, and we provide a ranking of
the world’s poorest countries on the basis of the Borda Rule. We then compare improvements in
socioeconomic performance with the availability of political and civil liberties during the decade
of the 1970s and observe that improvements in per capita national income. life expectancy at
birth, and infant mortality are positively correlated with the extent of political and civil liberties
enjoyed by citizens, while improvements in literacy are negatively correlated with these liberties.

1. PLURAL MEASURES OF WELL-BEING international agencies pursue both avenues at

once. and it is today a commonplace to assess the
Measures of the quality of life can take one of quality of life by studying a heterodox collection
two forms: they can reflect the constituents of of socioeconomic indicators. (See, for example.
well-being, or alternatively, they can be mea- the World Bank’s annual World Development
sures of the access people have to the determi- Report.)
nants of well-being. Indices of health, welfare, In choosing socioeconomic indicators for the
freedom of choice, and more broadly, basic purpose of cross-country comparisons, a balance
liberties, are instances of the first; those indices has to be struck between the claims of complete-
which reflect the availability of food, clothing, ness and costs. But all things considered, we
shelter, potable water, legal aid, education facili- would now seem to be arriving at something of a
ties, health care, resources devoted to national consensus that, at a very aggregate level, real
security, and income in general, are examples of national income per capita, life expectancy at
the latter. birth, the child survival rate, and the adult
In principle it does not matter which route we literacy rate arc, taken together, a comprehen-
take. Changes in a suitable aggregate of either sive package of indices. To be sure, we would
the constituents, or the determinants, can be also be interested in the distribution of well-being
made to serve as a measure of changes in the along gender, caste. class, or regional lines.’
quality of life in a society. Along the former One weakness (there are others) with the
route we would measure the constitutents indices of general well-being currently in use in
directly and aggregate them in a suitable way. such institutions as the World Bank and the
Along the latter route we would need to estimate United Nations Development Program (e.g.,
the accounting (or shadow) prices of the deter- UNDP. 1990) is that they are restricted to the
minants of well-being to arrive at an all- socioeconomic sphere of life; the political and
embracing measure of real national income.’ civil spheres are for the most part kept separate.
In practice, neither route on its own captures When they are mentioned at all, they are dealt
what we wish to see included in any reasoned with perfunctorily. Ordinal indices of political
conception of the quality of life. Thus, for and civil liberties have been in use in the political
example, there would be far too many person- science literature for some time. But to the best
specific accounting prices to contend with were of our knowledge. they have not been combined
we to attempt at an overarching measure of real with socioeconomic indicators for the purposes of
national income. At the same time, disposable arriving at a pluralist measure with which to
income as customarily measured captures aspects
of welfare and the extent of certain patterns of *We have benefited greatly from the comments of Irma
liberties, matters which are hard to come to grips Adelman. James Griffin. Oliver Hart. Paul Isenman
with directly. For this reason, governments and and Paul Streeten.

11’)
120 WORLD DEVELOPMENT

compare the quality of life across nations, or with enjoyed a per capita real national income of less
which to measure changes in the quality of life than $1,500 in 1980 international dollars.6 The
over time within a nation. One of the purposes of idea is to look at a snapshot of the quality of life
this article is to do this. in each country. The year in question is 1979980.
Our aim here is to conduct a number of simple As it happens, data on all the six constituents of
exercises with data (on national income per well-being we will study here are available for
capita, life expectancy at birth, the infant survival only 48 countries out of the more than 55 which
rate, the adult literacy rate, and political and civil should be on our list.’ Table 1 summarizes the
liberties) on countries which were in the early data. The first column of figures provides esti-
1970s among the world’s poorest in terms of mates of national income per capita for the year
income per capita.3 Our purpose is to gain a 1980. The second, third and fourth columns
preliminary understanding of the way the various present life expectancy at birth, infant mortality
constituents of general well-being are related in rates, and adult literacy rates, respectively, for
today’s world. Given the context in which such 1980.
discussions have recently been undertaken, our Of the six columns of figures in Table 1, it is
restriction to the world’s poorest countries is the last pair which will be a novelty to econom-
deliberate. In Section 2 we will use the well- ists. They represent indices of political and civil
known Borda Rule as the aggregator of the set of liberties in our sample, for the year 1979. They
six constituents being considered here to rank are taken from the valuable compendium of
countries in our sample. We will then compare Taylor and Jodice (1983). Rights to political
this new ranking with rankings based on each of liberty are taken to be citizens’ right to play a
the six chosen elements of general well-being. In part in determining who governs their country,
Section 3 we will ask if poor countries are indeed and what the laws are and will be. Countries are
faced with cruel choices among these various coded with scores ranging from one (highest
elements: political and civil liberties on the one degree of liberty) to seven (lowest degree of
hand, and socioeconomic liberties on the other.4 liberty). Values for this index are given in the
Using data from the decade of the 1970s we will fifth column of figures in Table 1.
provide a tentative, and what we feel is an Civil rights are different. They are rights the
encouraging, answer to this question. individual has vis-d-vis the state. Of particular
Such exercises as those conducted for this importance in the construction of the index in
paper can only be regarded as exploratory. Taylor and Jodice (1983) are freedom of the
Cross-country data on such variables as the press and other media concerned with the dis-
literacy rate and the infant mortality rate are well semination of information, and the independence
known to be defective. But this problem has not of the judiciary. The index measures the extent
prevented policy makers, international agencies, to which people, because they are protected by
and academics from using them; nor should it an independent judiciary, are able to openly
have prevented them from doing so. We would express their opinions without fear of reprisals.
like to encourage the adoption of a more pluralist Countries are coded with scores ranging from
stance than has been customary to date in one (highest degree of liberty) to seven (lowest
assessing the state of affairs in poor countries. degree of liberty). As these indices may not be
Such pluralism as we are advocating may be familiar, we provide their key in the appendix.
commonly subscribed to, but it has to date not Even a glance at these columns tells us that for
found much expression in quantitative studies of the most part political and civil liberties are
the quality of life in nations.5 scarce goods in poor countries. Citizens of 33
Our purpose here in appealing to the data is countries in our sample of 48 suffer from systems
purely illustrative. Many of the specific steps we that score five or more for political rights, and
will take below can be questioned. But they will those of no fewer than 40 countries from systems
not be ad hoc: we will provide justifications, even that score five or more for civil rights. As the
though alternative steps will readily suggest appendix makes clear, these scores reflect severe
themselves to the reader. We emphasize the deprivation of these basic liberties. There are
exploratory nature of our inquiry only because exceptions, of course, most notably Botswana,
the matter is a delicate one. There is a great deal the Gambia, India, Mauritius and Sri Lanka. But
remaining to be done in this field. for the most part the columns make for dismal
reading. When they are combined with the
columns which reflect the socioeconomic sphere
2. INTERCOUNTRY COMPARISONS of life, the picture which emerges is chilling.
There is nothing to commend the state of affairs
We will consider countries which in 1970 in a large number of the countries in our sample.
 QUALITY OF LIFE 121

Table 1. Living standards’ indicators in 1980

Y* Et MS L§ R,ll &ll
-
Bangladesh 540.0 48.0 140.0 26.0 4.0 4.0
Benin 534.0 47.0 124.0 28.0 7.0 7.0
Bolivia 1529.0 50.0 130.0 63.0 3.0 5.0
Botswana 1477.0 55.0 78.0 35.0 3.0 2.0
Burundi 333.0 46.0 126.0 25.0 6.0 7.0
Central African Republic 487.0 47.0 143.0 33.0 7.0 7.0
Chad 353.0 42.0 147.0 15.0 6.0 6.0
China 1619.0 67.0 41.0 69.0 6.0 6.0
Ecuador 2607.0 63.0 75.0 81.0 3.0 5.0
Empt 995.0 58.0 108.0 44.0 5.0 5.0
Ethiopia 325.0 44.0 155.0 15.0 7.0 7.0
Gamdia 556.0 40.0 159.0 15.0 2.0 2.0
Haiti 696.0 52.0 132.0 23.0 6.0 7.0
Honduras 1075.0 60.0 87.0 60.0 3.0 6.0
India 614.0 54.0 107.0 36.0 3.0 2.0
Indonesia 1063.0 53.0 105.0 62.0 5.0 5.0
Jordan 1885.0 62.0 58.0 70.0 6.0 6.0
Kenya 662.0 55.0 83.0 47.0 5.0 5.0
Korea, Republic of 2369.0 67.0 32.0 93.0 5.0 5.0
Lesotho 694.0 52.0 116.0 52.0 4.0 5.0
Liberia 680.0 52.0 100.0 25.0 4.0 6.0
Madagascar 589.0 51.0 146.0 50.0 5.0 5.0
Malawi 417.0 44.0 169.0 25 .O 6.0 6.0
Mali 356.0 44.0 184.0 10.0 7.0 7.0
Mauritania 576.0 43.0 142.0 17.0 6.0 6.0
Mauritius 1484.0 65.4 45.2 85.0 2.0 4.0
Morocco 1199.0 57.0 102.0 28.0 4.0 3.0
Nepal 490.0 45.1 142.2 19.0 6.0 5.0
Niger 441.0 42.0 150.0 10.0 6.0 7.0
Nigeria 824.0 48.0 118.0 34.0 3.0 5.0
Pakistan 989.0 49.0 124.0 24.0 5.0 6.0
Paraguay 1979.0 66.0 47.0 84.0 5.0 5.0
Philippines 1551.0 61.0 52.0 75.0 5.0 5.0
Rwanda 379.0 45.0 127.0 50.0 5.0 6.0
Senegal 744.0 45.0 147.0 10.0 3.0 4.0
Sierra Leone 512.0 38.0 172.0 15.0 5.0 6.0
Somalia 415.0 44.0 145.0 60.0 7.0 7.0
Sri Lanka 1199.0 68.0 34.0 85.0 3.0 2.0
Sudan 652.0 46.0 123.0 32.0 5.0 5.0
Swaziland 1079.0 51.7 133.4 65 .O 6.0 5.0
Tanzania 353.0 50.0 119.0 79.0 6.0 6.0
Thailand 1694.0 62.0 51.0 86.0 4.0 6.0
Tunisia 1845.0 60.4 91.8 62.0 6.0 5.0
Uganda 257.0 46.0 113.0 52.0 7.0 7.0
Yemen 957.0 42.9 163.7 21.0 7.0 7.0
Zaire 224.0 49.0 111.0 55.0 6.0 7.0
Zambia 716.0 50.1 90.4 44.0 5.0 5.0
Zimbabwe 930.0 55.0 82.4 69.0 5.0 5.0

*Y - Per capita income. 1980 purchasing power parity.

tE - life expectancy at birth (years).
SM - infant mortality rate (per 1.000).
§L - adult literacy rate (%).
/lR, - index of political rights.
lRz - index of civil rights.
122 WORLD DEVELOPMENT

Our international comparison of the quality of being at the top of the list may well be news, It
life will be based on these six indices. The nature certainly was to us.
of the data being what it is for a great many of the The relative positions of China (coming in at
countries, it is unwise to rely on their cardinal 10 from the top in terms of aggregate well-being)
magnitudes. We will therefore base our compari- and India (coming in two places behind, at 12)
son on ordinal measures. This way, systematic deserve a brief comment. China and India have
biases in claims about achievement across coun- long provided commentators with a classic ten-
tries will not affect the international comparison. sion: achievements in the field of socioeconomic
But first, we need an ordinal aggregator. Of the liberties set against those in the arena of political
many we may devise, the one most well known and civil liberties. As we see from Table 1, China
and most studied is the Borda Rule. In what beats India handsomely in each of the four
follows we will construct the Borda ranking of socioeconomic indices in our list, while India
the countries in our sample. wins over China in political and civil liberties. All
The strengths and limitations of the Borda this is consistent with general knowledge. The
Rule have been investigated by Goodman and fact, however, that the two finish so close in a
Markowitz (1952), Smith (1973), and Fine and ranking of 48 countries means that the ordinul
Fine (1974), and so we will not enter into them distunce between them in political and civil
here. In any event, our investigation is explora- liberties is large relative to their distance in the
tory. If only for this reason, the fact that the socioeconomic spheres. Indeed, other things
Borda Rule is simple, and its strengths and remaining the same, had more countries man-
weaknesses therefore transparent, provides an aged to squeeze themselves between China and
immediate justification for using it.* India in political and civil liberties. the overall
The first column in Table 2 presents the Borda ranking of these two countries would have been
ranking of nations. The other six columns present reversed. (Recall that the Borda Rule violates
the six constituents of our index of aggregate the “independence of irrelevant alternatives” ax-
well-being. Rankings range from the worst (score iom in Arrow, lY63.) On the other hand, had
of one) to the best (score of 48). more countries squeezed themselves between
It is a useful exercise first to look at the best China and India in the socioeconomic spheres,
and worst-off sets of countries. From the first the Borda gap between the two countries would
column of figures, we note that in nscending have been greater. Clearly then, the relative
order of aggregate well-being, the 10 lowest placings of China and India are sensitive to the
ranked countries in 1980 were: Mali, Ethiopia, aggregator being used. To us this is instructive.”
Niger, Chad, Yemen, Malawi, Sierra Leone, How does our list of the 10 top countries
Burundi, Somalia, and the Central African Re- compare with the list of the 10 least poor
public. How does this list compare with the countries? As it happens, they are very similar.
ranking of nations based exclusively on per capita The 10 least poor countries in our sample are in
national income? To see this, we merely note descending order: Ecuador. the Republic of
from the second column of figures that, in Korea, Paraguay, Jordan, Tunisia, Thailand,
ascending order. the 10 poorest countries in our China, the Philippines, Bolivia, and Mauritius.
sample in 1980 were: Zaire, Uganda, Ethiopia, There are seven countries in common. We
Burundi, Chad and Tanzania (tied at five), Mali, conclude tentatively that, among the poorest of
Rwanda, Somalia, and Malawi. The lists are not poor nations, rankings in terms of our index of
the same, but they are strikingly similar. All aggregate well-being are not too different from
countries except one (Yemen) are in sub-Saharan their rankings based on income per capita.
Africa, and the lists contain six countries in This is a qualitative claim, however, and it will
common. be informative to get a quantitative feel for the
Turning next to the 10 highest-ranked coun- relationship between our Borda ranking and each
tries, we note first that in terms of aggregate well- of the rankings based on the six constituents of
being they are, in descending order: Mauritius, well-being. Statistically, how close then is the
Sri Lanka, Ecuador, the Republic of Korea, Borda ranking to the other six’? To examine this
Paraguay, Thailand, Botswana, the Philippines, question we look at rank correlations. We are
Morocco, and China. aware that our rankings may be disturbed by data
The presence of Sri Lanka close to the top of inaccuracy. But it is unlikely that this would tend
our list of poor countries should not come as a to do anything except depress the rank correla-
surprise: the remarkable achievements of Sri tions below those which would be found working
Lanka (at least until recently) have been much from accurate data. Thus, the table of rank
commented upon. (See Isenman, 198Oa). Mauri- correlations may be regarded as indicative of
tius, however, is rarely talked about, and so its underlying statistical relationships.
 QUALITY OF LIFE 123

Table 2. Rankings of living standards data

Borda
rank Y* Et MS Ls: R,/l &ll
-
Mali 1 7 7 1 1 1 1
Ethiopia 2 3 7 6 4 1 1
Niger 3 11 3 7 1 8 1
Chad 4 5 3 8 4 8 12
Yemen 5 30 5 4 10 1 1
Malawi 6 10 7 3 13 8 12
Sierra Leone 7 14 1 2 4 21 12
Burundi 8 4 14 20 13 8 1
Somalia 9 9 7 11 32 1 1
Central African Republic 10 12 17 12 20 1 1
Mauritania 11 18 6 14 8 8 12
Benin 12 15 17 21 17 1 1
Uganda 13 2 14 27 29 1 1
Nepal 14 13 13 13 9 8 24
Haiti 15 25 28 17 11 8 1
Zaire 15 1 21 28 31 8 1
Rwanda 17 8 11 19 27 21 12
Tanzania 18 5 23 24 42 8 12
Pakistan 19 31 21 21 12 21 12
Gambia 20 17 2 5 4 47 45
Sudan 21 21 14 23 19 21 24
Madagascar 22 19 26 10 27 21 24
Senegal 23 27 11 8 1 39 41
Bangladesh 23 16 19 15 16 34 41
Liberia 25 23 28 33 13 34 12
Swaziland 26 35 27 16 37 8 24
Zambia 27 26 25 35 24 21 24
Nigeria 28 28 19 25 21 39 24
Kenya 29 22 33 37 26 21 24
Lesotho 30 24 28 26 29 34 24
Egypt 31 32 37 29 24 21 24
Indonesia 32 33 31 31 34 21 24
Bolivia 33 40 23 18 36 39 24
Tunisia 34 44 39 34 34 8 24
Zimbabwe 34 29 33 38 38 21 24
Jordan 36 45 41 41 40 8 12
India 37 20 32 30 23 39 45
Honduras 38 34 38 36 32 39 12
China 39 42 46 46 38 8 12
Morocco 40 36 36 32 17 34 44
Philippines 41 41 40 42 41 21 24
Botswana 42 38 33 39 22 39 45
Thailand 43 43 41 43 47 34 12
Paraguay 44 46 45 44 44 21 24
Korea 45 47 46 48 48 21 24
Ecuador 46 48 43 40 43 39 24
Sri Lanka 47 36 48 47 45 39 45
Mauritius 48 39 44 45 45 47 41

*Y - Per capita income. 1980 purchasing power parity.

tE - life expectancy at birth (years).
$A4 - infant mortality rate (per 1,000).
$L - adult literacy rate (%),
lIR, - index of political rights.
llR2 - index of civil rights.
124 WORLD DEVELOPMENT

Table 3 provides the (Spearman) correlation of 0.49 and 0.51 mean that the claim that the
coefficient for each pair of rankings from the circumstances which cause poverty are also those
seven rankings of nations. The correlation coeffi- which make it necessary for governments to deny
cients between the Borda ranking and the others citizens their civil and political liberties is simply
are: 0.84 with national income per capita; 0.91 false. There are countries in the sample which are
with life expectancy at birth; 0.88 with the infant very poor and which enjoy relatively high levels
survival rate; 0.76 with the adult literacy rate; of civil and political liberties.
0.68 with political rights; and 0.69 with civil We will note presently that the adult literacy
rights. We were not expecting this. We had no rate is a rogue index: it stands somewhat apart
reason to think that life expectancy at birth from the other indices of socioeconomic free-
would be the closest to our measure of the quality doms. The correlation coefficients between
of life. literacy and political and civil liberties are 0.24
Much has been written in recent years on the and 0.27 respectively. These are relatively
inadequacy of national income as a measure of low figures, far and away the lowest figures in
aggregate well-being. We were therefore surpri- Table 3.
sed at the closeness of our measure with national
income per capita. Nevertheless, our findings
suggest that if we had to choose a single, ordinal 3. ARE TRADEOFFS MANDATORY?”
indicator of aggregate well-being, life expectancy
at birth would seem to be the best. There must be International comparisons of well-being tell us
a moral to this. something about the nature of differing societies.
It is customary to regress national income per We have seen how we can determine whether,
capita against other socioeconomic indicators to among poor countries, those which were rela-
see how closely they are related. (See. for tively prosperous in 1979-80 were also the ones
example, Kaneko and Nidaira, 1988, who have where political and civil liberties were the least
studied cross-country statistical links between per constrained. But a cross-section study at one date
capita income and the infant mortality rate.) The cannot tell us anything about whether civil and
second column of figures in Table 3 presents political liberties are conducive to growth in the
Spearman rank correlation coefficients between socioeconomic constituents of well-being. Or, to
national income per capita and each of the other put it in euphemistic terms: do “authoritarian”
five constituents of well-being. The highest corre- governments achieve better “economic” results?
lation (0.79) is with life expectancy at birth. Or, to put it bluntly: are civil and political
Again, we were not expecting this. We also had liberties a “luxury” poor countries cannot afford,
no prior notion that correlation with the adult in that they act as a drag on economic perform-
literacy rate (0.59) would be considerably less. ance?
Richer countries seem to enjoy greater political It would be impertinent to suggest that this sort
and civil liberties. But the correlation is not of question can have an air-tight answer even for
overly high (the coefficient is 0.49 with political a given period. But cross-section data on coun-
rights and 0.51 with civil rights). Neither national tries can give us some hints, and in what follows
income per head nor political and civil liberties we will pursue this trail. We will ask if, OII a
should be thought of as being exogenously given. cross-country basis among the poorest of eco-
Thus any such link between them as we observe nomies, we can detect a conflict between the
in international data should only be seen as a acknowledgement of political and civil liberties
link, nothing more. No causal relationship can be and achievements in the socioeconomic sDhere.
presumed from the data. Correlation coefficients The model we are invoking is one whe;e the

Table 3. Correlation mutrix of livng standard rankings*

Y 0.8407 - - - -
E 0.9133 0.7895 - - - -
M 0.X797 0.6943 0.9180 - - -
L 0.7597 0.5942 0.8018 0.7934 -
R, 0.6842 0.4916 0.4105 0.4065 0.2420 -
R? 0.6881 0.5135 0.4347 0.3841 0.2654 0.7871
Borda Y E M L R,

*All correlations are statistically significant at a 5% level.

 QUALITY OF LIFE 125

political and civil spheres provide the environ- somewhat different from those concerning in-
ment within which men and women shape their creases in life expectancy at birth. But we are
lives. Each of the three broad sets of liberties is a trying to record performance in this field. A
set of freedoms. A restriction in any one of these figure of 10 per 1,000 for the infant mortality rate
sets is a form of deprivation. This makes it is about as low as it is reasonable for
imperative that we ask if, as a contingent matter, countries to aspire to for a long while yet. so;;
there is a tradeoff between them when a country we take the index of improvement to be the ratio
is poor, so that if we want fast growth in income of the decline in the infant mortality rate over the
or rapid improvements in health and education period in question (197tMO) to the base-year
we have to forego substantial political and civil infant mortality rate minus 10. All countries in
liberties. our sample have shown an improvement in infant
Toward this end we will look at data for the survival rates. The sixth column of figures in
1970s. Table 4 summarizes them. Since we are Table 4 presents values for this index of improve-
involved with a greater range of questions here, ment.
we will study a slightly larger pool of countries, The construction of an index of improvement
so that we may vary the sample size as and when in literacy rates does not pose problems of the
we need to. The criterion for inclusion is the kind we faced in connection with life expectancy
same as before. We are interested in those at birth and infant mortality rates. It is not
countries which, in 1970, enjoyed a per capita immediately apparent why it should be a lot less
national income less than $1,500 in 1980 interna- or a lot more difficult to increase the literacy rate
tional dollars. We have the information we need when people are more literate (except at the top
on 51 countries. and bottom ends of the range). This suggests that
The first column of figures in Table 4 presents we should simply measure increases in adult
the average of the 1970 and 1980 figures for real literacy rates if we want to know what net
national income. We study this average, rather improvements there have been in this field.
than income at some given year, because growth Unfortunately, we have not been able to locate
rates varied across countries during the decade. adult literacy rate figures for a number of
Since we are interested mainly in performance in countries for the year 1970. We therefore present
the socioeconomic constituents of well-being, we figures for 1960 from World Bank (1983) in the
will also be interested in the percentage change in seventh column. The net increase in literacy rates
real income per capita during the period. This is over 1960-X0 is then provided in the eighth
provided in the second column of figures. Fifteen column of figures in Table 4. It will be noticed
out of the 51 countries experienced a decline in that all countries recorded an improvement. ”
real income per capita. The ninth and tenth columns of figures in
The third column of figures gives life expec- Table 4 present indices of political and civil
tancy at birth in 1970. We wish to measure liberties in our overall sample of 51 countries,
changes in this index over the decade. This is not averaged over 1973-79. (See Taylor and Jodice,
an easy matter. Equal increments are possibly of 1983, Tables 2.1 and 2.2.) We have already
less and less ethical worth as life expectancy rises commented upon the extent to which poor
to 65 or 70 years and more. But we are measuring people living in poor countries are deprived of
performance here. So it would seem that it such freedoms.
becomes more and more commendable if, with We begin with an analysis.of rank orders. Once
increasing life expectancy, the index were to rise again, the use of rank correlation has the
at the margin. The idea here is that it becomes advantage that it makes our results relatively
more and more difficult to increase life expec- insensitive to the precise transformations used to
tancy as life expectancy itself rises. A simple calculate the indices of improvements in infant
index capturing this feature is the ratio of the mortality, life expectancy and adult literacy.
increase in life expectancy to the short-fall of the Furthermore, it remains true that data errors will
base-year life expectancy from some target, say tend to reduce rather than increase the magni-
SO years. ” The fourth column of figures in Table tude of our correlations. Table 5 consists of the
4 gives this index of improvement over the period 21 (Spearman) rank correlation coefficients asso-
1970-80 for 51 countries. As it happens, all but ciated with the seven columns of figures we are
two countries (viz. Rwanda and Uganda) re- studying: namely, real national income per head
corded an improvement. and its percentage growth; improvements in life
The fifth column of figures gives infant mortal- expectancy at birth, infant survival rates and
ity rates in 1970. The construction of an index of adult literacy rates, respectively; and the extent
improvement in infant mortality poses a similar of political and civil rights enjoyed by citizens.
problem. To be sure, the ethical issues here are The correlation matrix tells us that the alleged
 Table 4. Improvemeno in living .stundards*

Y* AYt E$ AEl Ml1 AMll L** ALtt R,$$ R&I

-
Bangladesh 499.0 17.9 45.0 8.6 140.0 0.0 22.0 4.0 4.9 4.2
Benin 552.5 -6.5 40.0 17.5 155.0 21.4 5.0 23.0 7.0 6.3
Bolivia 1383.0 23.6 46.0 11.8 153.0 16.1 39.0 24.0 5.6 4.1
Botswana 1179.0 67.7 50.0 16.7 101 .o 25.3 41.0 35.0 2.1 3.1
Burundi 324.0 5.7 45.0 2.9 137.0 x.7 14.0 11.0 7.0 6.4
Cameroon 789.0 24.5 49.0 12.9 126.0 17.2 19.0 ** 6.1 4.4
Central African Republic 499.0 -4.7 42.0 13.2 153.0 7.0 7.0 26.0 7.0 7.0
Chad 409.5 -24.2 38.0 9.5 171.0 14.9 6.0 9.0 6.4 6.4
China 1315.5 60.0 59.0 38.1 69.0 47.5 43.0 26.0 6.7 6.7
Congo 986.5 -1.1 51.0 13.8 98.0 17.0 16.0 ** 5.9 6.1
Ecuador 2005.0 85.8 58.0 22.7 100.0 27.8 68.0 13.0 6.4 3.7
Egypt 833.0 48.3 51.0 24.1 158.0 33.8 26.0 18.0 5.6 4.7
Ethiopia 333.0 -4.7 43.0 2.7 158.0 2.0 4.0 11.0 6.3 6.1
Gambia 561.0 -1.8 36.0 9.1 185.0 14.9 ** ** 2.0 2.0
Ghana 494.5 -25.9 49.0 9.7 110.0 10.0 27.0 ** 6.6 5.1
Haiti 623.0 26.5 48.0 12.5 162.0 19.7 15.0 x.0 6.4 6.0
Honduras 1001.0 16.0 53.0 25.9 115.0 26.7 45.0 15.0 6.1 3.0
India 595.0 6.6 48.0 18.8 139.0 24.8 28.0 8.0 2.1 3.3
Indonesia 811.0 90.2 47.0 18.2 121.0 14.4 39.0 23.0 5.0 5.0
Jordan 1653.0 32.7 55.0 28.0 90.0 40.0 32.0 38.0 6.0 6.0
Kenya 607.0 19.9 50.0 16.7 102.0 20.7 20.0 27.0 5.0 4.6
Korea, Republic of 1779.0 99.2 60.0 35.0 51.0 46.3 71.0 22.0 4.9 5.6
Lesotho 527.0 92.8 49.0 9.7 134.0 14.5 ** ** 5.3 3.9
Liberia 694.0 -4.0 47.0 15.2 124.0 21.1 9.0 16.0 6.0 4.3
Madagascar 631.0 -12.5 45.0 17.1 183.0 21.4 ** ** 5.1 4.4
Malawi 359.0 38.5 40.0 10.0 193.0 13.1 ** ** 6.9 6.0
Mali 336.5 12.3 40.0 10.0 204.0 10.3 2.0 8.0 7.0 6.6
Mauritania 573.0 1.1 39.0 9.8 166.0 15.4 5.0 12.0 s.9 6.0
Mauritius 1254.5 44.8 62.4 17.0 61.4 31.5 ** ** 2.7 2.3
Morocco 1037.5 36.9 52.0 17.9 128.0 22.0 14.0 14.0 4.6 4.4
Nepal 498.0 -3.2 41.6 9.1 157.4 10.3 9.0 10.0 6.0 5.0
Niger 421.0 10.0 38.0 9.5 170.0 12.5 1.0 9.0 6.7 6.0
Nigeria 121.0 30.8 44.0 11.1 158.0 27.0 15.0 19.0 5.7 4.0
Pakistan 893.0 24.1 46.0 8.8 142.0 13.6 15.0 9.0 4.3 4.9
Paraguay 1584.0 66.4 65.0 6.7 59.0 24.5 75.0 9.0 4.9 5.4
Philippines 1322.5 41.8 57.0 17.4 66.0 25.0 72.0 3.0 4.9 5.1
Rwanda 323.5 41.4 48.0 -9.4 135.0 6.4 16.0 34.0 6.9 5.3
Senegal 752.0 -2.1 43.0 5.4 164.0 11.0 6.0 4.0 5.6 4.4
Sierra Leone 485.5 11.5 34.0 8.7 197.0 13.4 7.0 8.0 5.6 5.0
Somalia 394.5 11.0 40.0 10.0 158.0 8.8 2.0 58.0 7.0 6.4
Sri Lanka 1108.5 17.8 64.0 25.0 52.0 42.9 75.0 10.0 2.0 3.0
Sudan 667.5 -4.5 42.0 10.5 149.0 18.7 13.0 19.0 5.9 5.7
Swaziland 911.0 45.2 46.1 16.5 145.2 8.7 ** ** 5.7 3.9
Tanzania 318.0 24.7 45.0 14.3 132.0 10.7 10.0 69.0 6.0 6.0
Thailand 1378.5 59.4 58.0 18.2 73.0 34.9 68.0 18.0 5.4 4.1
Tunisia 1460.5 71.5 53.9 24.9 127.2 30.2 16.0 46.0 6.0 5.0
Uganda 304.5 -27.0 47.0 -3.0 117.0 3.7 25.0 27.0 7.0 7.0
Yemen 742.0 81.6 38.6 10.4 187.8 13.6 3.0 18.0 7.0 7.0
Zaire 291.0 -37.4 45.0 11.4 131.0 16.5 31.0 24.0 7.0 6.1
Zambia 752.5 -9.3 46.5 10.7 106.0 16.3 29.0 15.0 5.0 4.9
Zimbabwe 870.0 14.8 50.5 15.3 96.2 16.0 39.0 30.0 5.9 5.0

Sources: World Bank, World Development Report (1983); World Bank (1989); Taylor and Jodice (1983, Tables 2.1 and 2.2)
*Y: per capita gross national income; average of 1970 and 1980 values at 1980 international prices.
tAY: % change in Y over the decade 1970-80.
SE: life expectancy at birth in 1970.
8AE life expectancy improvement index =
(life expectancy at birth in 1980 - life expectancy in 1970) x 100
(80 - hfe expectancy at birth m 1970)
JIM: infant mortality rate in 1970.
IAM: infant mortality improvement index =
(infant mortality rate in 1970 - infant mortality rate in 1980) X 100
(Infant mortahty rate - 10)
**L: adult literacy rate in 1960.
ttAL: adult literacy rate improvement index = (adult literacy ra e in 1980 - adult literacy rate in 1960)
$-JR,: political rights index, averaged over 1973-79 (decreasing wilh increasing liberty)
§1R,: civil rights index, averaged over 1973-79 (decreasing with increasing liberty)
128 WORLD DEVELOPMENT

tradeoff between political and civil liberties and pointing to the small number of countries where
gains in the socioeconomic sphere of life is a false citizens have had their political and civil liberties
choice, that statistically speaking, societies are severely restricted, and where growth in the
not faced with this dilemma. But the matrix tells socioeconomic sphere of life has been spectacu-
us more, and the morals which emerge from lar, and to then point by almost conditioned
Table 5 appear to be these:” reflex to India as a case in contrast. There is no
(a) Political and civil liberties are positively policy prescription flowing from such examples
and significantly correlated with real as Singapore and Hong Kong.‘” It is absurd to
national income per capita and its tell citizens to establish for themselves a one-
growth, with improvements in infant party system of government, or to locate for
survival rates and with increases in life themselves reliable and efficient dictators.
expectancy at birth.‘s “Good authoritarianism” cannot be willed by
(b) Real national income per capita and its citizens, and bad authoritarian regimes are hard
growth are positively and significantly to get rid of. A central problem with authorita-
correlated, and they in turn are positively rianism is its lack of incentives for error correc-
and significantly correlated with improve- tion. A pluralist political system has the chance
ments in life expectancy at birth and of providing political competition. This is one of
infant survival rates. its chief virtues. Of course, if civil order and
(c) Improvements in life expectancy at birth general civic responsibility have broken down
and infant survival rates are, not surpris- pretty much completely, there is no prescription
ingly, highly correlated. to be obtained either, one way or the other. To
(d) Political and civil rights are not the same. be sure also, had our main finding, as stated in
But they are strongly correlated. (a) above, gone the other way, there would have
(e) Increases in the adult literacy rate are not been something really urgent to discuss and to
related systematically to per capita in- think through. As it is. the data give us no
comes, or to their growth, or to infant compelling reason to question the instrumental
survival rates. They are positively and virtues of civil and political liberties.
significantly correlated to life expectancy Correlation does not imply causation, and we
at birth. But they are negarively and should bear in mind that indices of political and
significantly correlated with political and civil liberties can change dramatically in a nation,
civil liberties. following a coup d’Ctat, a rebellion, an election,
These observations suggest that literacy stands or whatever, and as we have used a six-year
somewhat apart from the other “goods” in our average index (1973-79) for them in Table 4, we
list. It does not appear to be driven with the three must be careful in interpreting the statistical
other socioeconomic indicators. Furthermore, results.” But we cannot imagine that these
regimes which have had bad records in political difficulties provide reasons for ignoring civil and
and civil rights have been associated with good political liberties in judging the quality of life,
performances in this field. We have no compell- even at this crude level of investigation. Subject
ing explanation for this finding. to these obvious cautions, what the evidence
These are statistical results, and they should be seems to be telling us is that, statistically speak-
seen and interpreted as such. It is simply no good ing, of the 51 poor countries on observation.
arguing against the force of such findings by those whose citizens enjoyed greater political and

Table 5. Correlation matrix of indicators of improvements in living

standards*

AY 0.5x83* - - - -
AE 0.6578* 0.4113* - - -
AM 0.7546* 0.4129* 0.7917* - - -
AL -0.0308 0.0660 0.2710* 0.0631 - -
R, 0.51X7* 0.2956* 0.2383* 0.4058* -0.3769* -
R, 0.4493* 0.2776* 0.2788* 0.3730* -0.2806* 0.7290*
Y AY AE AM AL R,

*Indicates that a correlation is significant at a 5% level. The

correlations are based on 51 observations. except that those for the
changes in adult literacy, AL. are based on 42 observations.
 QUALITY OF LIFE 129

civil liberties also experienced larger improve- head and infant survival rates. This seems to us to
ments in life expectancy at birth, real income per be well worth knowing.

NOTES

1. We are assuming that there is a social well-being being per capita income, life expectancy at birth, infant
function in the background consideration. and that the survival rate, adult literacy rate, and indices of political
appropriate technical conditions (e.g., convexity of and civil rights), adding each alternative’s scores to
relevant sets) are satisfied. Otherwise, accounting obtain its aggregate score, and then ranking alterna-
prices cannot be defined. tives on the basis of their aggregate scores. To
illustrate, suppose a country has the ranks i, i. k, 1, m
2. See, for example, Morris (1979), Sen (1981) and and n, respectively, for the six criteria. Then its Borda
UNDP (1990). The infant survival rate in a country is a score is i + j + k + 1 + m + n. The rule invariably
good index of the distribution of resource availability. yields a complete ordering of alternatives. It can be
Distributional issues therefore are not totally eschewed viewed as a social welfare function here, since the
in this aggregate analysis. criteria can be thought of as “voters” and the countries
the “alternatives.” Of Arrow’s classic axioms, the
3. Since life expectancy at birth is much influenced Borda Rule violates the one concerning the indepen-
by the infant survival rate (see Table 3), one can argue dence of irrelevant alternatives. See Arrow (1963).
we will be counting health twice. It is an easy matter to
redo all our computations by deleting data on either of 9. We persisted with the rule we chose to follow
our health indices. We would have much preferred before it became clear what the final outcome would
replacing life expectancy at birth by life expectancy at be. The final ranking has also been influenced by the
age one year, to make the measure independent of the rule we have followed as regards ties. We have given
infant survival rate. Limitations of data prevented us tied ranks a score equal to that which would be taken
from doing so. by any one of the tied group had the others all been
ranked above it. We have not checked what the
4. The breakdown of citizenship into its three outcome would have been had we followed other rules
constitutive spheres (the socioeconomic, the civil and regarding ties.
the political) was the classification in Marshall (1964).
For the link between our summary measures and 10. An earlier set of calculations concerning the
notions of general well-being as articulated in modern questions raised in this section was presented in
political philosophy, see Dasgupta (1991). Dasgupta (1990). The weakness of the earlier work was
that national income figures adjusted for purchasing
5. An exception is the pioneering work of Adelman power parity were not used. As is now well known, the
and Morris (1967). ranking of poor countries is substantially different
when international dollars are used for estimating real
6. We have worked with purchasing power parity national income.
income per capita rather than conventional income.
This is open to the objection that it overstates the per 11. Thus, an increase in life expectancy at birth from
capita income in countries which choose to devote large 35 to 40 years is less difficult to achieve than an increase
amounts of resources to social services (Isenman, from 60 to 65 years. See Sen (1981). The mathematical
1980b). Indeed it may explain some of the results representation of the index is given in the notes to
shown in Table 2, but on balance we think that the Table 4.
purchasing power parity measure is the best indicator
of income per capita currently available. 12. Japan and several countries in Western Europe
have pushed the infant mortality rate below 10 per
7. Data on per capita national income have been 1,000. (See World Bank, 1988.) The historical record
taken from Summers and Heston (1988), those on life suggests that once the infant mortality rate is down to
expectancy at birth and infant mortality rates from some 20 per 1,000 any further reduction in the rate is
World Bank, World Tables (1989), and the ones on due to further improvements in pediatric care, the
literacy rates from World Bank, World Development “marginal productivity” of diet and hygiene being
Report (1983). The decision to use a 1970 figure of pretty much exhausted.
$1,500 at 1980 international prices as the cut-off point
is, of course, a bit arbitrary, but only a bit. Our 13. The coverage here is smaller. Figures for adult
motivation will be clear in the Section 3, where we will Iiteracy rate are not available for a number of coun-
study the performance of poor nations during the tries.
decade of the 1970s.
14. Kendall rank correlation coefficients are. as it
8. It will be recalled that the Borda Rule provides a happens, quite similar in values, and so we do not
method of rank-order scoring, the procedure being to report them here. It should be noted that in computing
award each alternative (here, country) a point equal to the correlation matrix, the orderings of political and
its rank in each criterion of ranking (here, the criteria civil rights have been reversed. since the Taylor-Jodice
130 WORLD DEVELOPMENT

measure award a higher number to a country where Has it been democracy that has propelled Hong Kong
citizens enjoy fewer rights. and Singapore?” What we are suggesting in the text is
that this is not the right way of asking the question.
15. The level of significance of these figures is 5%.
17. As it happens, there were only a few countries in
16. A number of economists in conversation have which political and civil rights indicators changed
reacted to our statistical findings along the lines we are dramatically during the decade in question. For most
criticizing in the text. See also Stern (1991, p. 429), who countries, there was next to no change. See Taylor and
muses: I*_ it would be hard to be confident from Jodice (1983).
comparative history that democracy is good for growth.

REFERENCES

Adelman, I., and C. T. Morris, Society, Polirics und Kaneko, Y., and K. Nidaira, “Towards basic needs in
Economic Development (Baltimore. MD: Johns relation to public health and nutrition.” in D. E. Bell
Hopkins University Press, 1967). and M. R. Reich (Eds.), Health, Nutrition and
Arrow, K. J., Social Choice and Individual Values, 2nd Economic Crises (Dover. MA: Auburn House,
edition (New York: John Wiley, 1963). 1988).
Banks, D. L., “Patterns of oppression: An exploratory Marshall, T. H.. Class, Citizenship and Social Develop-
analysis of human-rights data,” Journal of the Ameri- ment (Garden City, N.Y.: Doubleday & Co., 1964).
can Statistical Associafion, Vol. 84. No. 2 (1989). Morris, M. D., Measuring the Condition of the World’s
Dasgupta, P., “An inquiry into well-being and destitu- Poor: The Physical Quality of Life Index (Oxford:
tion,” Mimeo (Stanford, CA: Department of Econo- Pergamon, 1979).
mics, Stanford University, May 1991). Riskin, C., China’s Political Economy (Oxford: Oxford
Dasgupta, P., “Well-being and the extent of its University Press. 1987).
realization in poor countries,” Economic Journal, Sen, A. K., “Public action and the quality of life in
Vol. 100, Supplement (March 1990), pp. l-32. developing countries,” Oxford Rulletin of Economics
Fine, B., and K. Fine, ‘Social choice and individual and Statistics, Vol. 43 (1981).
rankings, I and II,” Review of Economic &dies, Smith, J. H., “Aggregation of preferences with variable
Vol. 44, Nos. 3 and 4 (July and October, 1974), pp. electorate,” Econometrica, Vol. 41, No. 6 (Decem-
303-322 and 459-476. ber, 1973). pp. 1027-1042.
Gastil, R. D., Freedom in the World: Political Rights Stern, N. H., ‘Comments” in Roundtable Discussion
and Civil Liberties 198546 (Westport, CT: Green- on Development Strategies: The Roles of the State
wood Press, 1986). and the Private Sector, Proceedings of the World
Gastil. R. D., Freedom in the World: Political Rights Bank Annual Conference on Development Econo-
and Civil Liberties 1983-84 (Westport, CT: Green- mics 1990 (Washington, DC: The World Bank,
wood Press, 1983). 1991).
Goodman, L. A., and H. Markowitz, “Social welfare Summers, R., and A. Heston, “A new set of internatio-
functions based on individual ranking,” American nal comparisons of real product and prices: Esti-
Journal of Sociology, Vol. 5X (1952). mates for 130 Countries, 195&1985,” Review of
Humana, C.. World Human Rights Guide (London: Income and Wealth, Vol. 34 (1988), pp. l-25.
Economist Publication, 1986). Taylor, C. L., and D. A. Jodice, World Handbook of
Humana, C., World Human Rights Guide (London: Political and Social Indicators, Vol. 1 (New Haven,
Hutchinson, 1983). CT: Yale University Press, 1983).
Isenman, P., “Basic needs: The case of Sri Lanka,” United Nations Development Programme (UNDP),
World Development, Vol. 8, No. 3 (1980a). pp. 237- Human Development Report 1990 (New York: Ox-
258. ford University Press, 1990).
Isenman, P., “Inter-country comparisons of ‘real’ World Bank, World Development Report (New York:
(PPP) incomes; Revised estimates and unresolved Oxford University Press, various years).
questions, ” World Development, Vol. 8, No. 1 World Bank, World Tables 1988dlY (Washington, DC:
(1980b), pp. 61-72. World Bank, Internal Version, 1989).

APPENDIX

In this appendix, we discuss the political and civil what is not dispensed with, and so forth. That econom-
liberties indices we have used in sections 2 and 3. it is ists are accustomed to dealing with economic and social
perhaps unnecessary to emphasize that any such index statistics and are as yet unfamiliar with political and
will have a subjective element, but it is necessary to civil ones should not make us blind to this. The indices
remind ourselves that there is a subjective element in themselves, however (as opposed to their estimates),
the estimation of any index, even the familiar index of are no less objective than national income. For exam-
national income, in the construction of which it is ple, freedom from police detention without charge, or
necessary to dispense with information, to arrange the freedom to practice any religion. or the freedom to
publish books and to read them, or the right to seek associated with political systems that have trappings of
information and to teach ideas, of freedom from civil liberty, and whose governments may be success-
political press censorship, or freedom of movement fully opposed in the courts, although they may be
within one’s own country, or freedom from police threatened or have unresolved political deadlocks, and
searches of homes without warrants, or the right of may have to rely often upon martial law, jailing or
women to equality, or freedom of radio and television sedition, and suppression of publications. A score of
broadcasts from state control, or freedom from torture four is awarded to political systems in which there are
or coercion by the state, are fairly concrete ideas, most broad areas of freedom, but also broad areas of
especially perhaps for people who have suffered from a illegality. States recently emerging from a revolution-
severe denial of any of them. ary situation or in transition from traditional society
Taylor and Jodice (1983, pp. 6G61) provide an may easily fall into this category. Countries scoring five
account of their scoring system for political rights. are those with political systems in which civil rights are
Those countries which score one enjoy political systems often denied, but in which there is no doctrine on which
in which the great majority of persons and families have the denial is based. The media are often weak,
both the right and the opportunity to participate in the controlled by the government, and censored. Countries
electoral process. Political parties in these countries scoring six are those in which no civil rights are thought
may be formed freely for the purpose of making the to take priority over the rights of the state, although
right to compete for public office fairly general. criticism is allowed to be expressed in limited ways.
Countries scoring two are those which have political Finally, countries scoring seven are those which suffer
systems with open access which, however, do not from political systems of which the outside world never
always work, due to extreme poverty, a feudal social hears a criticism, except when it is condemned by the
structure, violence, or other limitations on potential state. Citizens have no rights in relation to the state.
participants and results. As with countries coded one, The Taylor-Jodice rankings of countries are based on
however, a leader or party can be voted out of office. A the state of human rights published regularly by
score of three is associated with political systems in Freedom House, recent publications from which are
which people may elect their leaders or representatives, Gastil (1983, 1986). The scores themselves are awarded
but in which coup d’bat, large-scale interference with to countries on the basis of a wide range of information,
election results, and frequent nondemocratic proce- and the indices R, and R, in our text are composites of
dures occur. A score of four is associated with systems a number of indicators of political and civil rights,
in which full democratic elections are blocked constitu- respectively. Other investigators have developed in-
tionally or have little significance in determining power dices of political and civil liberties. For example,
distributions. Systems in which elections are either Humana (1983. 1986) has undertaken parallel work, in
closely controlled or limited. or in which the results which 40 specific freedoms were rated for each country
have little significance are given a score of five. in his sample and were used to assess political and civil
Countries scoring six have political systems without rights. In contrast to the seven-point scale adopted in
elections or with elections involving only a single list of the Taylor-Jodicc compilation, Humana presents a
candidates. in which voting is largely a matter of four-point scale. There is a very high correlation
demonstrating support for the system; but where between Humana’s scores on political and civil liberties
nevertheless there is some distribution of power. and the scores presented here. Banks (1989) reports
Finally, a score of seven is associated with systems that correlation coefficients between the two scores on civil
are tyrannies, without legitimacy either in tradition or rights and the two scores on political rights to be 0.895
in international party doctrine. and 0.900, respectively. For the moment it is pretty
Taylor and Jodice (1983, pp. 64-65) provide an much of a matter of indifference which set of estimates
account of their scoring system for civil rights. Those we use.
countries which score one enjoy political systems in There are a number of other exercises that can be
which the rule of law is unshaken. Freedom of conducted with quantitative indices on political. civil
expression is both possible and evident in a variety of and socioeconomic liberties. Banks (1989) presents a
news media. Countries scoring two are those with wide-ranging statistical analysis of cross-country human
political systems that aspire to the level of civil rights. rights indices, identifying clusters of nations and
but are unable to achieve it because of violence. demonstrating that with the current data a few, identi-
ignorance, or unavailability of the media, or because fiable human rights indicators capture most of the
they have restrictive laws that seem to be greater than information we need to have on these matters.
are needed for maintaining order. A score of three is