World Bank Document€¦ · social indicators and associated systems of social accounts, and composite indices of develop- ment. A ... when no correlation between unequal income discuss

World Bank Reprint Series: Number 104

Norman Hicks and Paul Streeten

Indicators of Development:The Search for aBasic Needs Yardstick

Reprinted with permission from World Developmnient, vol, 7, 1979, pp. 567-80

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

World Development Vol. 7, pp. 567-580Pergamon Press Ltd. 1979. Printed in Great Britain

Indicators of Development: The Search for aBasic Needs Yardstick

NORMAN HICKS

and

PAUL STREETEN*The World Banik, Washington, D. C.

Summary. - The measurement of development efforts in developing countries has generallyfocused on the growth of GNP per head and related concepts. Increasingly, developmenteconomists have become aware that growth of output or income by themselves are not ad-equate indicatots of development, and that the reduction of poverty and the satisfaction ofbasic human needs are goals that should shoxv up in a measure of development. There has beengrowing interest in designing better measures of development, including modifications of GNP,social indicators and associated systems of social accounts, and composite indices of develop-ment. A review of these approaches and concepts points to the conclusion that the use of socialand human indicators is the most promising supplement to GNP, p3rticularly if work on socialindicators is done in areas central to the basic needs approach.

1. INTRODUCTION The heavy emphasis on GNP, or GNP perhead, and their growth rates, as the principal

Ever since economists have tackled the devel- performance test (not normally as the 'objec-opment problems of the less developed countries, tive') of development was based oln doubtfulthe principal yardsticks for measurements of assumptions. Either it was assumed that econo-economic development have been GNP, its mic growth has a tendency automatically tocomponents, and their growth. Despite the 'trickle down' to the poor, or it was thoughtmany problems with national accounting in that, where there was no automatic tendencydeveloping countries, the national accounts for the benefits from growth to spread to thehave continued to be the main focus of dis- poor, governments would take correctivecussions of growth, the allocations between action. Some authors insisted that concerninvestment, consumption and saving, and the with greater equality of income distribution,relative influence of various sectors in total with alleviating poverty, or other social aspectsvalue added. GNP per head is widely accepted of development is premature since it wouldas the best single indicator of development, reduce savings, investment and work ilsentives,both historically and for international com- and therefore growth.parisons, despite well-known serious problems. In the light of the expericnce of the last

The use of national accounting was inspired 25 yr, neither of these assumptions turnedby the attention of Western economists to the out to be generally valid. Highly concentratedbroad aggregates of Keynesian economics, and unequal growth was observed in somewhich was itself of major influence on econ- countries for prolonged periods, so that thereomic thought at the time (1950s) when attention was no universal tendency for growthl to spread.was being increasingly paid to the let', devel- Nor did governments always show signs ofoped countries. National accounting served to correcting gross inequalities. Doubt was castintegrate, through a weighting system based onmarket prices or factor costs, such disparate *Yithout implicating them, the authors would likeitems as agriculture and industrial production, to acknowledge thle useful comments received frominvestment, consumption and government ser- S. J. 5urki, James Grant, Mahbub ul Haq, D. V.vices, In fact, national income accounting was McGranahan, Mancur Olsen and Dudley Seers. Thea tool of analysis that other social scientists views expressed are those of the authors and not thesometimes viewed with considerable envy. World Bank.

567

568 WORLD DF\VELOPMENT

on the need for inequality to promote growth and the numbers living below a poverty line, Wewhen no correlation between unequal income discuss this briefly under the 'adjustments todistribution or perpetuated poverty and high GNP' approach. The extensive literature on thisgrowth rates was established. Inequality and subject could, however, warrant a separatepoverty were found not to be a necessary review.condition of growth and indeed were often anobstacle to it.

The disappointment with GNP per head 2. ADJUST ENTS TO THE GNP MEASUREand its growth led to greater emphasis onemployment and redistribution. But it was Despite the overwhelming attention tosoon seen, on the one iand,that unemployment growth, the deficiencies of GNP per head as anin the sense in which the term is used in the indicator of ecu.nomic development becamedeveloped countries was not the problem in apparent to many even during the early years,the developing countries and that, on the other Pigou already had pointed out that economichand, redistribution from growth only yielded welfare comprises not only national income pervery meagre results. Furthermore, it is clearthat mass poverty can coexist witlh a high head, but also its distribution and the degreedegree of equality, and reductions in absolute of steadiness or fluctuation over time. Measure-poverty are consistent with increases in in- ment problems become apparent when onepoveity. are consernts whifth o icreasdicn attempts to make inter-country comparisonsequallty. The concern has shilfted to eradcation of GNP per head. Part of the problem arises

f a e p , p l b from the fact that official exchange rates doon basic humnan needs. Meeting these needs in not measure relative domestic purchasingnutrition, education, health and shelter may be pot sure a large domestic GNPachieved by various combinat.ions of growth, does not enter into world trade. In addition,redistribution of assets and income, and re- trade policies ofteni create distortions in nom-structuring of production. It is the composition inal exchange rates, so that they fail to reflectof production and its bieneficiaries, rather thani tlle true value of even that proportion of GNPindexes of total production or of income wlhich is traded,distribution that have become the principal Coln Clark (1940, 1951) was one of theconcern. This new focus on meeting basic to ark ( 1 natione of tshuman needs requires an indicator or a set of first to attempt to convert national accohmntsindicators, therefore, by which deprivation using purchasing power parities, which means

can e ju8ed nd neasredsandpolicies measuring the output of each country at somecan be judged and measured, and plce cmmnpielv,usayitraioldirected at its alleviation and eradication can common price level, usually internationalbe initiated and monitored. The problems prices. The most recent and complete work oninherent in using GNP as a measure of social purcliasing power parities has been undertakenwelfare have been recognized almost since the by Kravis et al. (1976, 1978). The results ofinception of natioxfal income accounting. This tlhs research suggest that the GNP of Ino ia, forpaper identifies and reviews four different instance, should be adjusted upward by a factorapproaches to the measurement problem: of 3.5, while most other countries would be

adjusted by a somewhat smaller margin. Even(1) adjustments to GrP, thlrouglh which these kinds of adjustment, however, cannotmodifications of standard national income eliminate all the problems of conmparing GNPaccounting conicepts are undertaken in across countries. For instance, because oforder to capture some of the welfare aspects climatic conditions greater expenditures may beof development and to improve international required for clotlling and shelter in the morecomparability; temperate parts of th- wvorld in order to sur-(ii) social it2dicators which attempt to define vive, while dry tropical zones require morenon-monetary measures of social progress; expenidituire on irrigationi and disease control.(iii) the related social accountitig sistetns E valuations of non-tradables, p)articularly publicwhich attempt to provide an organizing and other services, are difficult and subject toframework for some of these indicators; and conceptulal problems. In addition, a great deal(iv) the development of cotinpositu indices of work is necessary, covering hundreds ofwhich combine various social indicators into goods and services, in order to estimate accu-a single index of human and social develop- rately purchasing power parities. Unless a 'shortment or the 'quality of life'. cut' or reduced information approacli is devel-

In addition to these four broad areas, con- oped, it will be difficult to make wide use ofsiderable effort has been expended in defininlg this approach.an adequate measure of income distribution, Nordhaus and Tobin (1972) attempted to

INDICATOPS 01 1V \'1I 011M I NT 569

adjust GNP so that it would be a better'Measure for disease and accidents, we would not needof Economic Welfare' (MEW). This approach to incur this expenditure. The same goes forentails subtracting from GNP an allowance for shelter against the cold, for sewerage and,defence expenditures and other 'regrettable perhaps, for literacy. Even food for under- ornecessities', such as the 'disamenities' of urban- malnourished people is a necessity to preventization (pollution, congestion, crime, etc.), hunger, disease or death. A logically consistentwhile adding an estimate of the value of leisure ap)plication of the Nordlhaus Tobin principleand the services of consumer durables. At the would lead to an incluitsioni in the nationalsame time, Nordhaus and Tobin reclassified incomiie only of tIhse itenis that we do nothealth and education expenditures as invest- really need, the ine,senitiails and frills, whichment, rather than consumption. The final woukld be a paradoxical conclusion, contraryresult produced a MEW for the USA that was to the judgement of those wlho wish to exclu1desubstantially larger than GNP (about twice), all frivolous luxuiries from our national incomelargely because of the high value imputed to accounts.leisure (the measure of which raises great If it were possible to) distingUish preciselydifficulties) and other non-market activities, between 'goods', 'hads', and 'anti-bads', weThe growth rate of MEW for the USA between would deduct fromi 1nltiol0lal incorime all 'anti-1929 and 1965 was somewhat lower than that bau's': those that comnbat the 'bads' generatedfor GNP, mainly because of the larger value by potential enemies (del'enze), those thatof leisure and non-market activities in the offset the 'bads' generated by nature (heating,base year (1929), reducing the proportionate shelter, medicines) - the narrowest definitionrate of growth, and partly because of the of basic needs - and those that offset thegrowth of defence expenditure and urban 'bads' generaLi,' by the dornestic economic'disamenities'. Denison (1971) and others systeml itself ('artificially' created wants throughhave criticized this approach on the ground advei tising, einiilationi and social pressures).that GNP was never meant to measure welfare, In f'act, it is not possible to distinguiishl b tweenand attempts to adjust it only confuse the good and bad 'artificially' created wants with-concept. out introducinig value judgemen is (the desire

It might be possible to inicorporate sonie of for books, art anid mulllSiC is also 'artificially'the items captured by social indicators by created), and it is not possible to distinlgluislhGNP adjustments. Thus, life expectancy could between 'anti-bads' (the need for deodorantsbe allowed for by using expected lifetime or anti-dandruff shamlpoo created by the fearearnings instead of annual income per head or, of social ostracism) and goods (the need formore crudely, the product of average incoime literaztuLre created by tlle desire to participateper head and life expectancy. The Co01LInsumptiOn in society).benefits of literacy could be allowed for by Adjustments to GNP for distributiontalimputing the value of services fromn education value judgements call be made by weightingas a durable consumer good, etc. (The benefits different components of the national incomeof literacy as a durable investment good already according to who receives them. Such a re-show in the form of higher productivity.) definition would, however, eliminate theDistribution could be allowed for by taking distinction between the national inconme andthe median or the mode ralther than the mean its distribution. AWluI\alia and Chenery (1974)income, or by multiplying the mean income by have suggested that the growth rate of GNP1 minus the Gini coefficient, etc. in itself is a nmisleadiing indicator of' deVLlop-

From the poin1t of view of indicating the nient, sinice it is heavily weighred by the incomesatisfaction of basic needs, the Nordhalus slhzires of the rich. A growth of I04 in incomesTobin correctionis raise certain difficulties. of the upper '0'; will have miiore impact on the'Regrettable necessities' are subtracted from aggregate growth rate than1 10'', growth inGNP, because, 'we see no direct effect of incomlles of the lower 20';. They suggest twodefense expenditures on household economic alternatives: eitlher the eq(utal wveiglhting of eaclwelfare, No reasonable country (or houselold) decile of income recipients or the introduLtiollbuyb "national defense" for its own sake, If of 'poverty weights' which would place morethere were no war or risk of war, there would weight on the growth of incomes for the lowerbe no need for defense expeniditures and no 40'.; The result is a revised aggregate growthonie would be the worse without them'. But rate which makes an allowance for differencessimilar reasoning could be applied to the and changes in income distributiion.components of basic needs. We do not want Ainotlher approach would be to use simplymedical services from nurses, doctors and the absolute income level of the lower 40''; ashospitals for their own sake. If it were not the appropriate indicator to which developmiient

570 WOIRI D D)IVlNloPMElNT

policies should be related. This has the advan- rely on less efficient and more costly alterna-tage of shifting the focus away from the distri- tives in the private sector (traditional healers,bution of income, a politically sensitive suibject private water deliveries, private schools). Thein many countries, to the level of living of the importance of the public sector in these areaspoor. Progress in reducing poverty can be derives from the view of these goods andjudged, however, only is the income level of services being 'merit goods', as well as from thethe poor can be comp: - with some standard external econoiiiies present in both constimip-minimum wvhiich reflect, a 'poverty line'. 'T'he tion and production. The mnaini basis of thegeneral approach adopted by maniy is to calcu- basic needs applroach, in fact. stermis both fronmlate that level of income at which the average the view that raisilng incomes alone is insuf-family consumes a nutritionally adequate diet, ficient in view of the inefficiencies in theusually defined in terms of calories. Those consumption patterns of the poor and the lackfamilies (or individuals) not having this incnome of availability of essential -oods and services.are therefore judged to be below the poverty Thuls, any measure of poverty income, noline, and comiipr-ise the poverty target grouip. matter how carefully derived, will be inadeqUate

The shortcomings of this approaclh are for measuring basic needs.many, and will be discussed here only briefly.First, examination of famiiy income and foodlConIumLIption ignores the important problem of 3. SOCIAL INDICATORSdistribution of food and other amenities withinthe family. It seems clear that in many countries An alternative approaclh is to develop betterwomen (who, in some societies work harder indicators of hualinai, social ani economic deevel-than men) and childrenl receive less than an opment which cover areas and aslpects that can-'adequate' amount of food despite the fact that not be ceflccted in mnost incomie-based measures.the family's total consumption is judged to be Thlese so-called 'social indicators' attempt to'adequate'. Poverty line measures dlo not con- mneasure the deT'elopuIent of hlealthl, nutrition,sider how far tainilics are below the poverty liotusinig, incomne dlistribution, as well as otherline. They do not show improvements that take aspects of cultural and social developnment. Aplace below this line and suggest a 'solution' for great deal of work lhas been undertaken bythose brouglt barely above the line. They pay various agencies to compile a set of social indi-no attention to the distribution of food between cators, including the UT (1 975), OECD ( 1976),different families below the line. They therefore AID (1976), ITNFSC'O (1977) to mention aconceal the efforts required to reduce poverty. few.Sen (1973, 1975) has proposed a weighting of In theory, social indicators should be moreindividuals on the basis of how far thiey fall Useful in cross-country comiipairisons, since theybelow the poverty line, thus combining poverty avoid the exchange and valuation problem. Inline and income distribultio(n approaches. fact, the statistical basis for comparing these

In addition, the concept of 'nutritionallI indicators between countries or over timeadequate' is difficult to define since caloric remains very frail. The figures are often un-needs vary widely with clinmate, boody weight, reliable and not comparable, particularlyactivity, and height, age and other factors, and because of different definitions used in col-even for the same person in the same conditions, lecting data. In addition, many data are basedfroir day-to-day. Hlousehold incomne surveys on limited sample surveys or other, highlygenerally show that many families below tlmc inaccurate data collection methods. Differencespoverty line could consumne an a(lequlate diet by ihservcd in social indicators between co.untriespurchasing a different, and more efficient, often reflect these statistical and defini.ionialbasket of foods which are availahle but rejected variations in the indicators ratlher than realon grounds of taste, variety, etc. Families living differences in social development. But thisbelow the 'poverty line' are often found to constitutes a cliallencge to collect hetter, miioreundertake certain nun-food Cxpen1ditUre.s wlhich comiparable data.many would judge to be non-basic, such as on Unlike the national accounts which use thedrink and entertainiment. Even with anl incolme pricing miechlanism to combine heterogeneousabove the poverty line, a fanily may not be item.,s, there is no obvious way to combineable to purchase essential goods and services (lifferenlt social indicators. Consequently, prob-which are controlled and in inadeqLuate supply lems arise in absorbing the content of a largeor supplied by the public sector (such as health, number of socio-ecornomiiic indicators and ineducation, water supply), or they may have to any attenmpt to draw general conclusions, The

INDICATORS OF1 DEVELOPMENT 571

movement to develop social indicators, further- to be a clear need to develop more specificmore, has suffered from a lack of clear percep- measures related to the poor, such as lifetion of purpose. The term 'social indicators' expectancy or calorie consLlllmption indicatorshas itself been used very loosely to encompass for those in the lower quir tile of the incomea whole range of human, economic, social, distribution, for women, for rural dwvellers,cultural and political indicators. The need to etc.supplement the GNP as an indicator of economic Thirdly, while GNP per head follows andevelopment has become confused with a search ascending order from the poorest to the richestfor indicators of other aspects of dcevelopment counitries, sonme social indlicators are capableas well as for an indicator of the 'quality of of catching something of the human, social,life'. The latter concept has generally been and cultural costs of opulence (the (liseases oftaken to cover concepts such as security, peace, affluence like heart disease, stomachl ulcers orequality of opportunity, participation, and deaths in automobile accidents), as well aspersonal satisfaction, all of whiclh present poverty, They can, in principle, register some ofdifficult incasurenment problems, It has never the shared global problems, such as pollution,been clear whether the search was for an cultural deCpende%ncC or interdependence, etc.,alternative to GNP, or a complement or a and reducoe the false hierarchical and paternal-supplement. istic impression that may be created by purely

Although we do not as yet have a unifying economic indicators. As a resuilt, a differentconceptual framework for these indicators, meaning can be attached to the 'gap' betweenand despite the problems mentioned above, the so-called developed and developing countries.social indicators do have certain advantages The GNP measure points to 'catchling up' andover GNP per head. First, they are concerned suggests a race. Social indicators can point towith ends as well as means, or at least with common and shared values and problems, tointermediate ends nearer to the ultimate end alternative styles of dev,elopnment, to theof a full and lhealthy life, than aggregate average opportunities for learning from one another.production measures. Even those social in- Reducing or closing the initerniationial 'gap'dicators that neasure inputs (e.g. hospital beds in life expectancy, literacy, infant nmortality,for 1000 population or school enrolment rates) or morbidity would appear to be a morerather than results (life expectancy, morbidity, sensible objective, and can be achieved at muclliteracy) attempt to capture inputs that are lower levels of GNP per head and thereforenearer to the desirable results than GNP per much sooner, than reducing the 'income gap',head. thouglh we are perhaps even more ignorant

Secondly, many social indicators say sonme- about lhow to achieve the former thlan thething about the distribution as well as the latter,average, because skewness at the upper end ismore limited than it is for income per head.(The mode or the median for income per head 4, INPUTS VS RESULTScan, however, eliminate skewness and ieflectsome aspects of (listribtution in the average.) Whether social indicators should reflectThere is practically no limit to how much inputs or results depends upon their purpose.income a man can receive, but the maximum F1or performance testing there is somethinglife span is limited. Any increase in literacy to Le said for the approach of clhoosinig indicesreflects also a distrihUtional inimprov,nient, that measure results or outputs, since tlhesebecause the proportion of beneficiaries has are closer to what we are trying to achieve.risen. Furthermore, measures of inputs carn introduce

Some indicators are better than others for biases toward certain patterns of nmeeting

showing also the distributioni of basic needs needs w1uich may not be universal. For inistance,(leficiencies since they are construlcted on an a cotunitry with fairly acceptable healtlh stan-either- or, I ave lhave niot basis. Thus, me;Lsures dards should niot be encorLLaged to acluire thicsuch as litiracy, access to clean water, and same number of doctors as one with seriousprimary scho'1l enrolmnenit can be used to indi- health problems. We are back with the probleincate the percentage of the population having of 'regrettable necessities', which should notbasic needs dicficieclcies in each of these import- be couLjnted as final goods or as social achieve-ant sectors. Measures such as life expectaLncy, ments, Moreover, the number of doctors doesinfant nmortality, and average calorie consump- not measure the distribution of these doctorstion are less informative since they average the and medical services, or the degree of theirstatistics of rich and poor alike. There seems specialization. Resourves may be deployed in

572 WORLD DEVELOPNIFlNr

inefficient ways, failing to benefit the poor. national prudUCt' (p. 95). These contra[dictoryMeasuTes such as infant mortality and life results appear to arise from the use of difreringexpectancy, hiowever, indicate the degree to selections of indicators, soulrces of dtata, andwhich basic needs have been fulfilled, rather country samples, as well as (littering inter-than the resources expended. Likewise, literacy pr i.? ions of results. Many sclholars includemeasures the effectiveness of the educational in 'social' indicators noni-monetary IleadsuLressystem, and is, in principle, a better indicator of economic perfornmance, suchl as newsprintthan the number of students enrolled or the or enlergy consumption or the ownership ofstudent/teacher ratio. In general, outputt automobiles and radtios. These eCoLnomnic in-measures are better indicators of the level of dicators are almost always higlhly correlatedwelfare and basic needs achievement. More- with GNP, and at times, have been suggestedover, most outputs are also inputs. Health, as a shortcut to estinmating internationallyeducation and even nutrition are valued not comparable income levels (see Beckerman,only in their own right, but also because they 1966). Some researchliers exclulde the developediraise the productivity of present and future countries, whose higlh levels of both GNP andworkers, though higher productivity is valued social development might dominate the sanmple.because it contributes to a better life. Likewise, different results can be obtained

Input measures, such as doctors or hospital based on the inclusioni or exclusion of thebeds per 1000 or enrolment rates in schools, centrally plannted economies, the OPlC('on the other hand, also have their uses. They countries, and the very small LDCs.may reflect government intention, commitment Correlations based on 1970 data from theand efforts to provide public services. For World Bank's Social Data Bank are shown inpurposes of assessing policies and mzonitorinig Table 1.2 The results for 7 social indliCatorsperformance, both sets of indicators are nec- show a modest correlation with GNP (averageessary. Input measures are useful indicators of r2

- 0.50), while a sample of 5 economicresources devoted to certain objectives (though indicators shlow somlucwaht higlher correlationithese can be misdirected). To the extent to (r2= 0.71). llowever, when the social indlicatokrwhich we know how to link inputs to results, data are disaggregated intosamiples of developingi.e. have a 'production function', we can trace and developed countries, the correlationthe connections between means and ends. Even coefficients for both grotups drop significantlywhere we donothave knowledge ofa'production (r2 =0.25 for developing countries, 0.18 forfunction' (e.g. linking expenditure on family developed). Similar declines in the correlationplanning to a decline in the fertility rate), the coefficients are also found when the economiccombination of input and ouitput measures indicators are disaggregated. Conse(qUently, itpresents the raw material for research into the would appear that studies whichi examine onlycausal links between the two, particularly social variables for developing countries aresince, in a social system of interdependent apt to discover a poor relationshlip, while thosevariables, so many outputs are also inputs. that consider economic and social variables forIn addition, where output measures cannot all countries are likely to find better relation-be readily found, it might be necessary to fall ships.back on measures of inputs as useful proxies. One reason why social indicators are not

mnore highly correlated with GNI' per head isthat the relationisliips are often distinctly non-

5. GNP VS SOCIAL INDICES linear. indicators such as life *expectancy,literacy, and school enrolmlent have asymptotic

Several studies (McGranahan et al., 1972; limits wvihich reflect biological and physicalLUnited Nations, 1975) have indicated a high maxima. It is impossible, for instarce, to havecorrelation betweeni economnic indicators, in- more than ]00'; literacy. FUrtlhermnore, thesecluding GNP and social indicators. This miglht limits are often reaclhedI by imiikddle-inlcoelesuggest that GNNP can be used as a proxy countries, so that fLirtlh increases in inioncemeasure of social developiment. Morawetz show little gains in social indIicators. For in-(1977) found that there was a weak correlation stance, life expectancy reaches 70 yr of agebetween the level of GNP and indicators of for countries with intcomie per head (1970) ofbasic needs fulfilment, and even less correlation $2000, and does not increase even as incoinesbetween the grovwthl of GNP and inmproilcnments increase to $5000. Most couintries have attainedin basic needs indicators. Sheehan and Hopkins close to 1 00%,1o literacy by the time their income(1978) concluded, however, that 'the most reaches the $2500 level. Conversely, counltriesinmportant variable explaining the average level below $500 GNP per head demnonstrate a widIeof basic needs satisfaction is per capita gross variety of social developmiient wvihich is largely

INDICATORS OF DEVELOPMENT 573

Table 1. Correlation of indicators with GNP per capita (1970)

Coefficients of Determxination (r 2

All countries Developing Developed Sample size

Social indicatorsExpectation of life at birth 0.53 0.28 0.1 3 102Calorie consumption (as % of required) 0.44 0.22 0.02 103Infant mortality 0.42 0.34 0.25 64Primary enrolment 0.28 0.24 0.05 101Literacy 0.54 0.47 0.16 70Average persons per room (urban) 0.58 0.08 0.29 34Housing units without piped water (%) 0 '1t 0.13 0.36 36

A verage* 0.66 0.25 0.18

Economic indicatorstNewsprint consumption 0.79 0.20 0.46 85Automobiles 0.85 0.59 0.46 102Radio receivers 0.43 0.14 0.07 97Electricity consumption 0.67 0.30 0.24 102Energy consumption 0.82 0.28 0.49 99

Average 0.71 0.30 0.34

Source: Based on data taken from the World Bank's Social Data Bank. Excludes centrally planned economiesand countries with a population of less than one million.

* Simple unweighted arithmetic means of the rS.t All economic indicators are on a per capita basis.

Life expectancy at birth, yr

370 410 450 490 530 570 610 650 69Z 730

35990-...-.-...-.- .....- ....... ... ....... . . ... . .. . ,........ . .

GNP per head 3

(1970 $1 Regression linetLE-.47.5 +0.00689 GNPP~II.

2 590C~R ~0.53

............. ... ................................................ ..................

Fiur 1.GPadlieepcanY.90

I .,

Figur per GNPd an lif exotn'(90

574 WORLD DEVFLOPMENT

Adult literacy rate, %

00o 90. 280 380 470 570 660 760 850 940

.. ........II I 51,1

461,60I I c

I t ,

I I,

- - - - - - - -- - - - - - - - - - - ... . . . . . . .II i

IIRegression line:GNP per head ,:LIT=42.15 +0.0186 GNPPi

(1970$) . (89.6)=R 0.54

* ~780

4 C

.. . . . . . . ..... . .. .. . . . . . . . . . .. ... .. . . .. . ... .c.. . . .. . .

Figu re 2. GNP and literacy (1970)

unrelated to the level of GNP. This can be seen indicators. Stone (1975) and Seers (1977)more clearly from the two graphs on the fol- have proposed the use uf lifetime activitylowing pages for GNP with life expectancy and sequences calculated by dividing total lifeliteracy (other social indicators show similar expectancy into segments. Suich tables wouldpatterns). The cluster of points along either show the average time a person could expectaxis indicates the lack of correlation at both to spend in various mutu~lly exclusive states.the high and low income levels. It seems clear One such matrix could divide lifetime activitythat a much better correlation couild be devel- between school, work, leisuire, retirement,oped using some sort of non-linear relationship.' etc., while another might be built on a maritalA non-linear function would obscure the fact sequence, (single, married, divorced, widowed).that the correlation exists, however, only Such tables would comibine various imiportantamong the iiiiddlle-inconle countries. GNP social statistics from different fields, and wouldper head is likely to be a miisleading indicator be used to inidicate changes over time, eitherof social developmniit and progress in m-eeting actual or planned. The system presents manybasic needs, particularly when used in some problemis, however, not the least of whichi islinear fashion. Yet rankings of countries by its inability, to incorporate fuilly all aspectssocial indicators and GNP are likely to be very of social developm-ent. Somie indicators (inconlcsimiilar, because the ranking process obscuries distribution, security, police protection, pol-these non-linearities. lution) cannot readlily be transformied into life

expcctancies. Furthermore, the system goes

far beyond the data available in most countr-ies,6. SOCIAL ACCOUNTING SYSTEMS and is thus more suited for the industrialized

countries. Nevertheless, it is a concept whichSome work has been done on developing a has some future potential for integrating a

system of sociall accounts to provide a kind of large variety of social variables, and providingnational accouinting framework for social the basis for a theory linking policies to results

INI)ICATORS 01F DEV1L.O1I'IH:NT 57s

in the area of social planning. Other ideas have difficult to obtain for many countries, such asbeen developed for ' more limited social the amount of leisure time available, the numberaccounting approack. lhe Social Accoutnting of people in possession of private saving, andMaxtrix (SAM) iof Pyatt and Round (1977) the quality of housing. This makes the appli-does not utilize social indicators, but expands cation of the index very difficult, and Drew-the traditional input-output table into a nowski and Scott were forced to use short-cutmatrix which details payments made by pro- approximations even for their limited sampleductive sectors to different income recipients. of 20 countries, Furtheriuiore, the work, onceRecipients can be disaggregated in various beguLn, was not continuiied after 1966 in theways so as to indicate the distribution of in- same form.come between various factors, urban/rural McGranahan et al. (UNRISD, 1972) examl-households, or income classes. The power of ined 73 indicators which covered economic andthe SAM approach is that it integrates pro- social characteristics, and found that there wasduction and income distribution data in a way fairly high inter-correlatioin between thesethat gives a better view of the economy, and indicators. Through a process of elimination,the flows hetween sectors, It still relies, how- he conistructed a 'Development Index' based onever, on the use of GNP as a measure of welfare 18 'core indicators' which included 9 socialand is limited in its application by the absence and 9 economic indicators. The resulting indexof good income distribution data. Terleckyj was highly correlated with GYNP per head(1975) has developed a matrix framework for (r2 = 0.89), although there were some countriesanalysing the impact of government programmes (Venezuela, Chile, Japan) whose rankinig wason various social goals, as indicated by the sutbstantially different under the index. Inappropriate social indicators. Since programmllies general, the correlation of the index and GNPaffect more than one social goal, the approuclc per headl was somewhat lower for developingdevelops a matrix of inputs and outputs, and thani developed countries. This study concludedsuggests the possibility of defining the most that social development occurred at a more rapidefficient set of programmes for achievinig a pace than economic developlmlent up to a levelparticular set of goals. While this approach of about $500 per capita (1960 prices). Someprovides a useful rationale for using different of these results are themselves, however, aindicators, it does not provide a better measure product of the method employed, whereby theof growth or development. selection of the 18 'core indicators' was based,

in part, on their having a highl inter-correlationwithl the other indicators. As a result of the

7. COMPOSITE INDICES OF high inter-correlation, the composite index wasDEVELOPMENT relatively insensitive to the choice of com-

ponent variables. The UJNRISD team found,Relatively more work has gone into develop- forinstance, that the country rankings remained

ing composite indices that could be used to virtually unchanged when the number of indi-replace or supplement GNP as an indicator of cators was reduced from 18 to 10.social, economic or general development. A A study by the United Nations-- ECOSOClarge amount of work was undertakeni by the (1975) sought to analyse development byUN Research Institute for Social Development ranking 140 countries by seven indicators(UNRISD) during the 1960s to develop better other than G',NP. These included two 'social'social indicators, incluiiding composite indicators. indicators (literacy, life expectancy) and fiveFor instance, Drewnowski and Scott (UNRISD, 'economiiic' indicators (energy, mlanui;facturing1966) developed the 'Level of Living' index, share of GDP, mn;lufacturing slhare of exports,which was defined as 'the level of satisfaction employment outside agricul]tlure, number ofof the needs of the population as measured by telephonies). An overall rank for eaclh counltrythe flow of goods and services enjoyed in a was calculated by giving e(lual weight to theunit of time' (p. 1). The Level of Living Index ranks under each separate indicator. Arrangiingitself, however, goes beyond the provision of the results by quinltiles, and conmparing withgoods and services, and considers 'basic needs', GNP, the UJN indicated that the overall indexsubdivided between physical needs (nutrition, was closely associated with the ranking by GNP.shelter, health), and cultural needs (education, It should be ioted, however, that the UN indexleisure, security). 'Higher Needs' or 'Surplus was heavily weighted by economic, rather thanover Basic Needs' is taken as the surplus income purely social, indicators and thus tends toover some minimum level. The 'basic needs' replicate the findings of Beckerman (1966)part of the index includes items whiclh are very and others that show that non-monetary

576 WORLD DEVELOPMIl NI'

indicators are highly correlated with GNP. A Scaling problems arise when raw data on1similai study by the OFCD Secretariat (1973) social indicators are converted into cominponentused regression techniques for six variables to indices ranging from 0 to 100. For instance,establish a predicted GNP per head index for reasonable values for life expectancy couldl be82 developing countries. A more recent paper either 40 75 yr or 40- 100 yr. A country witlhby the OECD/DAC (1977), however, con- a life expectanlcy of 60 yr will obviously have acluded that 'per capita GNP still appears to dlifferent Iscore' depentding on the the scalingbe the best measure' of the level of develop- chosen (57 vs 33), and this will materiallyment. clhange the composite index. Furtlhermiore,

A more recent study of the use of a com- the scaling systeml need not he linear. Drew-posite index has been undertaken by the nowski used 'expert opinion' to derive a linearOverseas Development Council (ODC), under scale system reflecting set levels of basic needsthe guidance of Morris D. Morris. Morris's satisfaction. McGranahan et al. developed anPhysical Quality of Life Index (PQLI) uses elaborate system of 'correspondience poin ts'three simple indicators with equal weights to to determine the appropriate scale range, andattempt to measure the fulfilment of 'minimum uLtiliedti non-inear (logarithmic) scaling forhuman needs'; life expectancy at age one, many indicators. Morris sinmply took the rangeinfant mortality and literacy. Morris argues of the data for eaclh indlicator with the 'worst'that the use of indicators for judging perform- country being defined as zero and the 'best'ance under basic needs criteria should concen- as 100.trate on indicators of outputs or results, rather In addition, there is the even more difficultthan inputs. Input measures, he feels, do not problem of the proper weights to be used inmeasure success in meeting the desired goals, combining the component indices into theand may lend an ethnocentric bias to the coiiiposite. Drewvrnowski tried bohth eqlual fixedmeans employed. The use of only three in- weights and a systeim of slidling weights underdicators permits the calculation of the PQLI which deviations fromii tlle normal were givenfor a wide range of countries and facilitates more weiglht than indlices close to thei normnal.the examination of changes in the index over The rankings of countries bly sliding or equaltime, The term 'quality of life' is perlhaps a weights were highly correlated svitl the rankiingsmisnomer, since what is really being measured of countries by (GNP per head or consLInuptionis effectiveness in reducing mortality and per head, and the shift in the weighting systellmraising literacy. Life expectancy measures did not materially affect the rankings.the quantity, not the quality of life. (These McGranalhan's weighting system gave greateraims also have an ethnocentric bias.) Most wveight to the component indi-ators that hadimiiportantly, the weighting systemn of the PQLI the lhigliest degree of inter-correlation with theis arbitrary and there is no rationale for giving otlher indicators, a soimiewlhat dulbious tmethod.equal weights to literacy, infant mortality and One would think that the absence of correlatiolllife expectancy at age one. It is not possible would be an equally valid criterion, thoughi oneto prove that the PQLI gives a 'correct' index miglit then wish to know why there is noof progress on hunman needs, as opposed to correlation, rather than integrate them. He alsosome alternative index having different weights found that moderate chalnges in the weightingor a different selection of comiiponient inidices. system did not affect the level of eachIt is not clear what is gained1 by comiiniing the country's indlex, or its rank-inig. The insensitivitycomponent indiLes with a weigihting system of the general index to the clhoice of weightsthat cannot be defenided. Analytical work can is a logical result of having high inter-correlationibe undertakell Lusing the componient indices among the ltCOMponelts, since the highi cor-almost as easily as with the composite index, relation implies that any one component is awithout introdLucing the biases of the PQLI. goo)(I sIb stitu1tC for ylly other, The UI NWhllile iMorris' index has received muclh attention F OSOC stud(ygives equal weight to the countryin the popular press, most serious scholars fil,] ranks of the social indicators, thus avoidillg,it difficult to accept the results of a composite in a certain sense, the scaling problem. As

idex %kithout a stronger theoretical foun(latiotn. imentioned above, the PQLI gives equal weiglhtDeslpite the potential attractiveness of having to each of the three conmponents witholut

a single index of socio-economic (levelopimient, ascertaining if this implies the correct 'trade-there is little theoretical guidance to govern the off' betweeni the variouis components. None ofchoice of indicators, the correct scaling of these stuLdies in(licates that mlutichl effort wascomponent indices, or the appropriate weights. expended in developinig a theorelically soundMforeover, an indilex that relies only on rankiing rationale for the weighting system.neglects the distance between ranks. Becausc of these problemis, it nlighit well

INDICATORS Ol: 1) VEt O)PMENT 577

be argued that a composite index is either 1:fforts to develop comnposite indices haveunnecessary, or undesirable or ipnliossible to ranged from a search for better measures ofconstruct. It is unnecessary if the comiiponenlts the physical production of goods and services,are highly correlated with one anotlher, because tu a imeasure of the 'quality of life' of 'econ-then any one of the component indicators by omic' or 'social' welfare, of satisfactions,itself will serve as an adequate index. If, on the 'haippiness' and otlher objectives. The searchother hand, the components move in ditterenit for a composite index of social welfare, ana-directions in cross-coLIltry comparisons and logous to GNP as an index of prodLuction,time series, averaging would conceal the import- has been a fruitless one so far, since it hasant issues and would be undesirable. To have proven virtually iiiipo.sible to translate everythe same index for a situation in wlhiclh mortality aspect of social progress into money values oris high and literacy low, as for one in which some other readily accepted comllmon de-literacy is high and mortality low, implies nominator. The great deal of work devoted toevaluating the 'trade-off' between lite racy conmposite indices, lhowever, sulggests the needand life expectancy. Unless the basis for such for a single number whiichi, like GNP per head,an evaluation can be establislhed, all weiglhting can be (luickly grasped and gives a roughremains arbitrary and misleading, and compo- indication of social' developlmient.sition is impossible, The case for considering The current liscuission of basic needs orientedthe two indices separately is exactly the same developmient focuses on the alleviation ofas the case for having an index independent poverty through a variety of measures otherof GNP. than mierely redistribution of inicremlneltal

If the interpretation of basic needs were output. 4 Such a focus supplements attention totaken literally, so that all basic needs, being hiow much is being produced, by attention to'basic', would have to be nmet together and Iwhliat is being produced, in what w al s, fortrade-offs between different basic needs were whomz and with what itnly.tt. Ok joiusly, theruled out, a composite index would not be rapid growtth ol outpult will still be imlportanitnecessary. As long as the 'package' of basic to the alleviation of pverty, andt GNP per headneeds has not been fully met, no amount of remains an imiiportant ligure. Wh,at is re(qired,additional satisfaction of any one coinpoent in adldition, are some indicaitors of the comripo-could compensate for the slightest deficiency sition and beneficiarieN of GNP, whiclh wouldin any other, so that a copnposite indicator supplement the GNP data, not replace thenm.would be ruled out, Or 'e all basic needs had The basic needs dpproaclh. therefore, can hebeen met, again no composite index would be the instrumient for giving the necessdrY focus torequired, for tl,e indicator for any one need work on social indicators.would show Lhat all had been satisfiel. But As a first step, it niiglht be useful to definewe are not advocating such a literal imotr- the best indicator for eahll basic need. Atpretation of 'basic needs', present, the essential basic needs are considered

to cover six areas: nutrition, basic education,8. Rl-:C OMMlNll NDATIONS FOR FUTURI'F healthi, sanitation, water supply andl housing,

WORK and related infrastructure (see Streeten andBurki, 1978). This list is not exhaustive, nor do

This brief survey has revxiewed four alternla- all needs listedl have the saLmlle status. It is in-tives to GNP per head as metlods of calculalting tended to be illustrative. A limited set of coresome of the dimensions of developmlent. The indicators covering tlhese areas wou lLd be aadjulst,nent to (GNP approachl lias focused largely im,eftil device for Conenltrating effOrts. Onceon inmproving GNil as a measure of economic lelCnedmld however, this set could then serve as awelfare. Attempts to introduce other costs call for the collection ot more adetqtuate, stan-and benefits of developnient, which wvould dardiied , cu1lm parable international statistics,nmove GNP toward a broader welfare mnasuire, and tlius help f'ocuis data gaitlering eff'orts onlack a logical basis anlLd tend instead to result only the most inipoirtalni indicatomis. It is notin a conifuision of concepts. Research oni 'social' clear that becaILIse there are six b:msic needs,indicators has failed to produce an alternaltive' there net d be only six core indlicators. It mlaywhich is as readily accepte(l and conmprehended be that more than one indicator will be necess-as GNP per head t1ouggh they are useful for ary to uMeasurl1e .aleC(Itely progress in any onejuLdging social lperfornmance. Systems of social area of basic needs. Neverthieless, the basicaccounts, which could integrate socialindicators needs concept serves aSa useuLll dIeVice forinte-through some unifying concept. liave not been gratinig efforts of data galihering anmd analysis.able to Overcomile suLcessfully all the difficult Once defilned, these core blasic needs in-problemils encountered, dicators would hlave the potential for important

578 WORLD DEVI.LOMLMNT

policy analysis relating, for instance, to inter- a satisfactory measure of housing needs. Thenational comparisons of performance and only readily available indicator is people perrelative aid levels. Such indicators would be a room. but this really dues not capture much ofmore useful guide to the relative 'gap' between the qutiality of housing, only the number ofrich and poor countries, and offer a different roons.s which in turn is a very rough index ofview on the speed with which this gap was crowding. Ideally, these indicators should bewidening or narrowing. They would be useful supplemented by data about distribiutioni ofin understanding which countries were meeting c.g. calories per head, etc.their basic needs, and how their policies are If an acceptable system of weights couldrelated to tlhe growth of output, trade, invest- be developed, it might be possible to combinement, infrastructture, etc. Not enough is being the core indicators into a composite basic needsdone internationally to improve the capacity index. The chances of an acceptable system ofof developing countries to identify, collect and weights beingdeveloped,however,areextremiiel)issue better primary data on a regular, system- small. Despite considerable research on corn-atic and comparable basis. Such data collection posite indices, no one has come close to devel-can be costly, but a substantially greater effort opillg a rational weigh-ting systein. It is difficultseems to be justified. even to suggest directions for further research.

The problem of selecting the appropriate (Some may question the desirability of such aindex in each field is best taken up by teclhnical composite index, even if it could be con-experts in each sector. To give an indication, structed.)however, of the indicators which might be Instead of attempting to develop a conmpositeincluded, the following have been identified index of basic needs, a useful alternative mayas a preliminary set: be to narrow the range of indicators from six

to one or two, which correlate highly witlHealtio: Life expectancy at birth; basic nieeds developmnent. This approach wouldEducation: -P iterac h serve the need of those who desire a single

Primary school enrolment number for making quick ju(dgements on

(as per cent of - poplation social performance, without introducing theaged sup14); problelns of weighted composite indices. The

Food: calorie supply per head or prospects for doing this are considerablycalorie supply as a per cent enhanced by the fact that many of the so-calledof requirements; 'basic needs' are, in fact, inputs rather than

Water supplv: Infant mortality (per thou- ultimate goals. Certainly nutrition, watersand biTths) supply and sanitation are valued because theyPer cent of population with _impIrove the healtli status of the population.access to potable water; To a more uimitedl extent, this is also true of

Sanitatio.z: Infant mortality (per thou- housing and education. All of these can besand births) considered to be inputs into the health 'pro-Per cent of population with duction function'. They may be valued foraccess to sanitation facili- reasons other than their influence on healthties; and status, but a hiiglh association between the

Housing: None. various core indicators can be traced to theirimpact on health. Therefore, it could be argued

The core indicators identifiedl here attempt that somie measure of health, such -as lifeto follow the philosophy of the paper in stres- exCectancy at birtlh, would be a good singlesing measures of results, rather than inputs. measure of basic needs. In a sense, life expect-Infant mortality is assumed to be a good indi- ancy is a kind of Weighted 'ComIpeosite' ofcatoi of the availability of sanitation and clean progress in mneetinLg physiological basic needs.water facilities because of the sus;Cptibility of It has the ddvd1ntage of capturing the impactinfants to water-borne diseases. Furthermore, on individuals, not only of non-nmarket factorsdata on infant mlortazlity are generallY more but also of income net of taxes, transfer pay-readily available than data on access to water. ments and social services, without raising all theWhile literacy is a good general measure of dlifficuilties of incomne per head measures, suchprogress in education, the per cent of .he rel- as the appropriate unit (individual, householdevant age group enrolled in primary school or family), the appropriate magnitude (capital,is included to measure country effort. Input consulmiiption income), the appropriate set ofmeasures have also been identified for water prices (market prices, internationial prices),supply and sanitation assuipplemeneltary mneasuires. wlhat to value as final goods anid what as costs,It has not been possible, however, to identify etc, For these purposes, it might be regar(led

INDICATORS OF DEVELOPMENT 579

as superior not only to a composite index of uni-dimensional way. Life expectancy can besocial indicators but also to GNP and to indices increased by measures that affect different ageof income distribution. It is possible for two groups differently. Improved nutrition, forcountries to register the same GNP per head example, may affect life expectancy above oneand the same ratio of income accruing to the year, whereas women's education may affectbottom 20%, and yet to have different average infant mortality. The second danger is thatlife expectancies. For some purposes, e.g. for the improvement of a health indicator likedistinguishing between meeting the basic needs life expectancy will divert attention to healthof men and women, or of rural and urban measures generally, and doctors, clinics andpopulations, or for additional information if nurses specifically, whereas the 'productionlife expectancies cluster very near one another, functioin' for life expectancy may include ait would, however. be useful to add a measure number of thrusts not obviously related toof progress in education, such as literacy. It is, health, like improved jobs, earnings, environ-of course, possible to have a long and miserable ment, etc. Just as we now know that reductionslife, and one might wish to put an upper limit in the rate of population growth are not simplyto the desired life span. But at low income functions of improved family planning, solevels, there is a high correlation between improvecl health and longer life are not simplymorbidity and mortality. functions of improved health delivery systems.

In using a single indicator, it is, however, But as long as the indicators are not identifiedimportant to guard against two dangers: the with uni-dimensional results or uni-causaldanger to interpret the result in auni-dimensional remedies, there is much to be said for a simpleway, and the danger to interpret inputs in a system of recording and monitoring.

NOTFS

1. A fifth method would be to interview a sample less than one million. While the total sample includesof individuals and to ask each to place himself on a 106 countries, missing data reduce the sample size'happiness' or 'basic needs' scale between, for example, for each correlation (see Table 1).0 and 10, and to say whether he feels his basic needshad been met more adequately than at some specified 3. I"or life expectancy, a semi-log function in-date in the past. But this kind of survey is still rudi- creases the r2 from 0.53 to 0.75.mentary and does not provide us with the kind ofinformation we should need for monitoring a basic 4. It is related to 'raising levels of living'. See UNneeds approach. (1954 and 1961).

S. This assumnes that such needs are legitimate2. This sample excludes the centrally-planned and should be met: some might argue that it would

economies and all countries wvith populations of be better to ignore suCh requests.

RIl:l [RENCEL S

Adelman, Irma and Cynthia Taft Morris, Eicononic Denison, Edward F., 'Welfare measurement and theGrowthl and Social Eqtuity in Dereloping Countrics GNP', Surne)' of CurrentBusiness, Vol. 51 (January(Stanford, Calif.: 1973). 1971).

Agency forInternational Development,'Socio-economic D rewnuwski, Jan, 'Social and economic factors inperformance criheria fordevelopment'(Washington, development', United Nations Research InstituteD.C.: February, 1977). for Social Development, Report No. 3 (Geneva:

Ahluwalia, M. and 1-1. Chenery, 'A conceptual frame- February 1966).work for economic analysis', in Hlollis Chenery Drewnowski, Jan and Wolf Scott, 'The level of livinget al. (eds.), Rc'disirilution wvith Growtlh (London: index', United Nations Research Institute for1974). Social Development, Report No. 4 ((;Gncv%a

Baster, Nancy, 'Work on development indicators by Septemiber 1966).national guvernments, research institutes and Edelman, John A, 'The ODC proposal on a quality ofindividual scholars, a selective review and dis- life index', World Bank Memoranduni, (7 Decembercussion of issues and problems' (Tokyo, Japan: 1976).United Nations University, May 1977). Gilbert, M. and 1. Kravis,An Interntational Comlparison

Beckerman, Wilfred, Iniernoatiotial ConiparisonIs of of National Produicts and the Puirchasinig Power ofRealIncomes (Paris: OECD, 1966). Currencies (Paris: OFCD, 1954).

Clark. Colin, Conditions of Econzomnic Progress, 3rd Harbison, F. H., J. Marumburic, and J. R. Resnick,Edition (London: 1951), (originally published Qu)anttitatii e Analyses of Modrnization andin 1940).

580 WORLD DEVELOPMENT

Delvelopmient (Princeton, N.J.: Princeton Uni- matrices for development planning', Review ofversity, 1970). Income and Wlealtht, Series 23, No. 4 (December

King, Mervyn A., 'Economic g-rowth and social devel- 1977).opment: a statistical investigation', unpublished Schreiber, Gotz, 'Measuring economic growtlh inpaper. developing countries: problems, results, impli-

Kravis, Irving, 'The scope of economic activity in cations', presented at the Fifth World Congressinternational income comparisons', in Problems of the International Economic Association (Tokyo:in the Internzational Comparisons of Economic September 1977).Accounts, NBER Studies in Income and Wealth, Seers, Dudley, 'The meaning of development', Anter)Vol. 20 (Princeton, N.J.: Princeton University niatiotnal Development Rev-iewv, Vol. 11 (DecemberPress, 1957). 1969).

Kravis, Irving et al,, A System of Intertiational Com- Seers, Dudley, 'What are we trying to measure', Jouirnalparisons of Gross Product and Purchlasing Power of Development Studies, Vol. 8 (April 1972).(Baltimore and London: The Johns Hopkins Seers, Dudley, 'Life expectancy as an integratingUniversity Press, 1975). concept in social and demographic analysis and

Kravis, Irving et al., Intern2ational CompnparisonIs of planning', Review of Income and WYealtlh, Vol. 23Real Produict anid Pumrchalsing Power (Baltimore and (September 1977).London: Johns Hopkins University Press, 1978). Sen, A. K., 'Poverty, inequality and unemployment',

Kundu, A., 'Social indicators: system and norms', Econiomtic i and Political WVeekly, Vol. III, Specialunpublished paper (Washington, D.C.: World Number 31-33 (August 1973), reprinted asBank, 14 January 1975). 'Poverty: an ordinal approach to measurement',

Leipziger, D. M. and Maureen A. Lewis, 'A basic Econometrica,Vol. 44 (February 1976).human needs approa h to development' Sen, A. K., Poverty and Econtomic Developmten7t(Washington, D.C.: USAID, 12 September 1977). (Almedabad: 1975).

McGranahan, D. V., 'Development indicators and Slheehani, Glen and Michael Hopkins, 'Basic needsdevelopment models', Journial of Developmtent performance: an anlysis of some internationalStuidies Vol. 8 (April 1972). data' (Geneva: ILO, 1978).

McGranahan, D. V. and C. Richaud-Proust, 'Methods Singer, lHans W., 'Social development: key growthof estimation and prediction in socio-economic sector - philosophy, plans and first results of thedevelopment' (Geneva: UNRISD, n.d.). UN Researclh Institute', Initernationtal Developmtictt

MIcGranalhan, D. V., C. Richaud-Proust, N. V. Sovani, Review, Vol. 7 (March 1965), pp. 3 -9.and M. Subramanian (UNRISD), Cnontemts and Stone, Richard, Toward a Svstemn of Social and Demiio-AMeasuremtent of Socio-Econonic Developmnent graphic Statistics, (New York: United Nations,(New York: Praeger, 1972), (also published as an 1975).UNRISD staff study, 1970). Streeten, Paul, 'The distinctive features of a basic

Morawetz, David, Twentv-Five Years of Economic needs approach to development', InterniationalDevelopmentt 1950 to 1975 (W'ashingtnn, D.C.: fnelo/p)imnent Review, Vol. 19, No.3 (1977).World Bank, 1977). Streeten, Paul, 'Basic needs: premises and promises',

Morris, M. D. and F. B. Liser, 'The PQLI: measuring Journal ol Policy, Modelling, Vol. 1, No. 1 (Jan-progress in meeting human needs', Overseas Devel- uary 1979).opment Council, Communique on Development Streeten, Paul and S. J. Burki, 'Basic needs: someIssues No. 32 (1977). issues', Wlorld Development, Vol, 6, No. 3 (March

Nordhaus, William and James Tobin, 'Is growth 1978).obsolete?', in Economic Growthl (New York: Terieckyj, Nestor, Imiprovemzetits in the Quality ofNBER/Columbia University Press, 1972). Life (NWashington, D.C.: National Planning

OECD/DAC, 'Performance compendium - consolidated Association, 1975).results of analytical work on economic and social UJnited Nations. International Definition and Measure-performance of developing countries' (Paris: 1973). rinetnt of Statndards and Levels of Living (1954),

()IC.(l)/DAC, 'Socio-economic typologies or criteria and Inztertnationial nlefi.itio atnd Afeasuremnent ofand their usefulness in measuring development Levels ofLiving.' Ai Interim Guide (1961).progress', (Paris: 7 April 1977). UN Economic (Comnmission for Furope, Econiotmic

Olsen, Mancur, 'Toward a 'ocial report: II' Puiblic Surveyr of k.urope in 1969, Part I, ('hap. 4 (NewInterest No. 15 (Spring 1969). York: 1970).

Olsen, Mancur, "rhe treattment of e\tcrnalities in United Natiotns, Economic and Social Council, Coin-national income stalistics', in Lowdon W'ingo) and mittee for Development Planning, 'DevelopingAlan Evans (eds.), Ptublic Economics anid the couintries and levels of development' (15 OctoberQutality of Life (publisihed for Resources for the 1975).Future and the Centre of Fnvironnmental Sttludies; (!NFSC0. Tlie Use of Socio-Econonmic Itndicators itnBaltimore: Johns tIopkins University Press, 1977). Development Planning (Paris: UNESCO Press,

Overseas Development Council, 'A physical quality 1976).of life index', in Thle US anxd Wotld Development World Bank, Atlas (Washington, D.C.: World Bank,,Igenda, 1977 (New York: Praeger, 1977). 1976).

Pyatt, Graham and Jeffery Rotund, 'Social accounting

THE WORLD BANKHeadquarters:1818 H Street, N.W. UWashington, D.C. 20433, U.S.A.Eturopean Office:66, avenue d'Iena75116 Paris, FranceTokyo Office:Kokusai Building,1-1 Marunctuchi 3-chomeChiyoda-ku, Tokyo 100, Japan

The full range of World Bank publications, both free and for sale, isdescribed in the World Bank Catalog of Publlicationis, and of the continuingresearch program of the World Bank, in World Bank Researcli Program:Albstracts of Current Stludies. The most recent edition of each is availablewithout cha.rge from:

PUBLICATIONS UNIT

THE WORLD BANK

1818 H STREET, NW,

WVASHINGTON, D.C. 20433

U.S.A.