Economic and Political Weekly

Poverty and Inequality in India: A Re-Examination
Author(s): Angus Deaton and Jean Dreze

Source: Economic and Political Weekly, Vol. 37, No. 36 (Sep. 7-13, 2002), pp. 3729-3748
Published by: Economic and Political Weekly
Stable URL: http://www.jstor.org/stable/4412578 .
Accessed: 09/08/2013 15:53
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
Economic and Political Weekly is collaborating with JSTOR to digitize, preserve and extend access to
Economic and Political Weekly.
http://www.jstor.org
This content downloaded from 147.26.11.80 on Fri, 9 Aug 2013 15:53:20 PM

All use subject to JSTOR Terms and Conditions
Special articles
Poverty and Inequality in India

A Re-Examination
This paper presents a new set of integratedpoverty and inequality estimatesfor India
and Indian states for 1987-88, 1993-94 and 1999-2000. The poverty estimates are broadly
consistent with independentevidence on per capita expenditure,state domestic product
and real agricultural wages. They show that poverty decline in the 1990s proceeded more or
less in line with earlier trends. Regional disparities increased in the 1990s, with the
southern and western regions doing much better than the northern and eastern
regions. Economic inequality also increased within states, especially within urban areas, and
between urban and rural areas. We briefly examine other development indicators,
relatingfor instance to health and education. Most indicators have continued to improve in
the nineties, but social progress has followed very diverse patterns, ranging from
acceleratedprogress in some fields to slow down and even regression in others. Wefind
no supportfor sweeping claims that the nineties have been a period of
'unprecedentedimprovement'or 'widespreadimpoverishment'.
ANGUS DEATON, JEAN DREZE
P overty trends in India in the nineties sustained poverty decline in most states The evidenceon inequalityis discussed
have been a matter of intense con- (and also in India as a whole) during the in SectionIII, wherewe focus mainlyon
troversy.' The debate has often reference period. It is important to note, the period between 1993-94 and 1999-
generated more heat than light, and con- however, that the increase in per capita 2000. Based on furtheranalysis of Na-
fusion still remains about the extent to expenditure associated with this decline in tional Sample Survey data and related
which poverty has declined during the poverty is quite modest, e g, 10 per cent sources, we argue that there has been a
period. In the absence of conclusive evi- or so between 1993-94 and 1999-2000 at markedincreasein inequalityin the nine-
dence, widely divergent claims have flour- the all-India level. ties, in severalforms.First,therehasbeen
ished. Some have argued that the nineties In Section II, we consider related evi- strong 'divergence'of per capitaexpen-
have been a period of unprecedented im- dence from three additional sources: the ditureacrossstates,withthealreadybetter-
provementin living standards.Othershave CentralStatisticalOrganisation's 'national off states(particularly in the southernand
claimed that it has been a time of wide- accounts statistics', the 'employment-un- western regions) growing more rapidly
spread impoverishment.2 Against this employment surveys' of the National thanthepoorerstates.Second,rural-urban
background,this paperpresents a reassess- Sample Survey, and data on agricultural disparitiesof per capitaexpenditurehave
ment of the evidence on poverty and in- wages. We find that these independent risen.Third,inequalityhasincreasedwithin
equality in the nineties. sources are broadly consistent with the urbanareasin most states.The combined
So far, the debate on poverty in the revised poverty estimates presented in effects of these differentforms of rising
nineties has focused overwhelmingly on Section I. In particular, real agricultural inequalityarequitelarge.Intheruralareas
changes in the 'headcount ratio' - the wages in different states (which are highly of some of the poorest states, there has
proportion of the population below the correlated with headcount ratios of rural been virtuallyno increase in per capita
poverty line. Accordingly, we begin (in poverty) have grown at much the same rate expenditurebetween 1993-94 and 1999-
Section I) with a reassessment of the as the corresponding NSS-based estimates 2000. Meanwhile,the urbanpopulations
evidence on headcount ratios and related of per capita expenditure in rural areas. of most of the better-offstates have en-
poverty indexes, based on National Sample While each of these sources of informa- joyed increasesof per capitaexpenditure
Survey(NSS) data.Inparticular,we present tion, including the National Sample Sur- of 20 to 30 per cent, with even larger
a new series of internally consistent vey, has important limitations, they tend increasesfor high-incomegroupswithin
poverty indexes for the last three 'quin- to corroborateeach other as far as poverty these populations.
quennial rounds' (1987-88, 1993-94 and decline is concerned, and the combined SectionIV takesup somequalifications
1999-2000). The broad picture emerging evidence on this from different sources is andconcerns.We pay specialattentionto
from these revised estimates is one of quite strong. theapparent declineof cerealconsumption
Economic and Political Weekly September 7, 2002 3729

in the nineties, which is not obviously indeed,in thelastof these,the54thRound, Briefly,theproblemis as follows. After
consistentwiththenotionthatpovertyhas the survey was only in the field for six the 50th Round, the National Sample
steadilydeclinedduringthat period.We monthsratherthanthecustomaryyearand Surveyintroducedan experimentalques-
also considerthe possibilityof impover- is thereforemost unlikely to be compa- tionnairewith differentrecallperiodsfor
ishmentamongspecific regionsor social rablewithanyprevioussurvey.Thesethin differentclasses of goods, in additionto
groups,in spite of the generalimprove- rounds suggested not only that poverty the traditional'30-day recall' question-
ment in living conditions. Finally, we remainedmoreorless unchangedbetween naire.Theexperimental questionnaire used
commenton the unresolvedpuzzle of the 1993-94 andthe first six monthsof 1998 a seven-dayrecallperiodforfood,pan,and
'thin rounds'. (thereferenceperiodfor the 54thRound), tobacco,as well as a 365-dayrecallperiod
In SectionV, we arguefor supplement- but also thataverageper capitaexpendi- for less frequentlypurchasedgoods such
ing expenditure-based datawith otherin- turestagnatedduringthis periodof rapid asdurables,clothing,footwear,educational
dicatorsof living standards,focusing for economicgrowth.This is verydifficultto and institutionalmedical expenditures.
instanceon literacyrates,healthachieve- square with independentevidence, e g, Priorto 1999-2000,thetraditional'30-day
ments,nutritionallevels, crimerates,and fromnationalaccountsstatistics.As things recall'questionnaireandtheexperimental
thequalityof theenvironment. Thisbroader stand,we do not have a good understand- questionnairewereadministeredto differ-
approachshedsa differentlighton poverty ingof whythethinroundsgive whatappear ent (and independent)samplesof house-
trends in the nineties. In particular,it to be anomalousresults, and until that holds. These alternativequestionnaires
promptsus to acknowledgethat social puzzle is resolved,our confidencein our producedtwo independentseries of ex-
progresshas been unevenacrossthe dif- other results must remainqualified.We penditureestimates,with a fairly stable
ferent fields. For instance, the nineties shall returnto this issue in Section IV.3, 'ratio'of the lowerestimatesbasedon the
havebeena periodof fairlyrapidincrease and ignore the thin roundsin the mean- traditionalquestionnaireto the higher
in literacyandschoolparticipation.Onthe time. estimatesbasedon theexperimentalques-
otherhand,therehasbeena markedslow- Incontrastto thethinrounds,theofficial tionnaire.In 1999-2000,the 30-dayrecall
downin the rateat whichinfantmortality countsfromthe latestquinquennialround and seven-day recall periods for food,
has been declining,and a significantin- (the55thRound,pertainingto 1999-2000) panand tobacco were used for the same
crease in economic inequality.An inte- suggest considerablepovertydecline be- households,in two adjacentcolumns on
grated assessmentof changes in living tween1993-94and1999-2000.According the same pages of a single questionnaire.
conditionshas to be alive to these diver- to official estimates,widely relayed,the This effectively 'new' questionnairede-
sities. We also discussotherimplications all-Indiaheadcountratiodeclinedfrom36 sign led to a sudden'reconciliation'of the
of this broaderapproachto the evaluation to 26 percentoverthis shortperiod.As is results obtained from the two different
of living standards,going beyond the well known,however,the 55th Roundis recall periods,perhapsreflectingefforts
standardpovertyindexes. notdirectlycomparableto the50thRound, to achieve 'consistency' on the part of
The concluding section sums up the due to changes in questionnairedesign. investigators and/or respondents. This
insightsof this enquiry.
Table la: All-India Headcount Ratios
(Per cent)
PovertyIndexesinthe 1987-88 1993-94 1999-00

Nineties Rural
Officialestimates 39.4 37.1 26.8
1.1,Official Estimates Adjustedestimates:
Step 1: Adjustingforchanges in questionnairedesign 39.4 37.1 30.0
Step 2: Revisingthe povertylines 39.4 33.0- 26.3
Webeginwithanexaminationof house- Urban
hold per capitaconsumptionand the as- Officialestimates 39.1 32.9 24.1
sociatedpovertyestimates.Consumption Adjustedestimates:
is only one elementof well-being,but it Step 1: Adjustingforchanges in questionnairedesign 39.1 32.9 24.7
Step 2: Revisingthe povertylines 22.5 17.8 12.0
is animportant element,andmuchinterest
is rightly attached to the Planning Source: Planning Commission, Press Releases (March 11, 1997, and February22, 2001), Deaton
(2001a, b), and Table 2a below.
Commission's periodical estimates of
poverty-based on NationalSampleSurvey
data.The mostwidely-usedpovertyindi- Table lb: All-India Poverty-Gap Indexes
cator is the 'headcountratio' (hereafter 1987-88 1993-94 1999-00
HCR),i e, theproportionof thepopulation Rural
below the povertyline. Estimatesfromunadjusteddata and officialpovertylines 9.4 8.4 5.2
The latest year for which relatively Adjustedestimates:
uncontroversial HCRestimatesare avail- Step 1: Adjustingforchanges in questionnairedesign 9.4 8.4 6.4
ableis 1993-94,corresponding to the50th Step 2: Revisingthe povertylines 9.4 7.0 5.2
Urban
Roundof the NationalSampleSurvey,a Estimatesfromunadjusteddata and officialpovertylines 10.4 8.3 5.2
'quinquennial'round. This round was Adjustedestimates:
followed by a series of so-called 'thin Step 1: Adjustingforchanges in questionnairedesign 10.4 8.3 5.9
rounds',involving smaller samples and Step 2: Revisingthe povertylines 4.8 3.7 2.3
somewhat different sampling designs; Source:Authors'calculationsfromunitrecorddata fromthe 43rd, 50th, and 55th Roundsof the NSS.
3730 Economicand Political Weekly September7, 2002

reconciliationis likelyto boosttheexpen- indicator, the 'poverty-gap index'. Each questionnaire.The second is that the re-
ditureestimatesbasedon 30-daydata,and of these departures calls for further lationbetweenintermediate-goods expen-
thereforeto pulldowntheofficialpoverty discussion. ditureand total expenditureis much the
counts,which are basedon these 30-day The possibility of 'adjusting' the 1999- same in 1999-2000 as in 1993-94.5The
expenditures.In addition,only the 365- 2000 poverty estimates arises from the fact secondassumptionwouldbe undermined
day questionnairewas used for the less that the 55th Round questionnaireretained by a majorchangein relativepricesof the
frequentlypurchaseditems,andthisaban- the '30-day recall' (and 30-day recall only) intermediate goodsrelativeto othergoods
donmentof the traditional30-day recall approach for a number of items such as in thelate 1990s.Itcanbe checkedto some
for durablesand other items also brings fuel and light, non-institutional medical extent by applyingthe proposedmethod
down the poverty count. Indeed, most care, and large categories of miscellaneous to the 'thin rounds' instead of the 55th
peoplereportno such purchasesover 30 goods and services. Further, it turns out Round,and comparingthe predicteddis-
days,butreportsomethingover365 days. thatexpenditureon this intermediategroup tribution withtheactual
of totalexpenditure
Thebottomtailof theconsumption distribu- of commodities is highly correlated with distribution.Thesecheckssuggestthatthe
tion is therebypulled up, reducingboth total expenditure.4 Expenditures on these correction procedure worksreasonably well
povertyandinequalitycomparedwiththe comparably surveyed goods can therefore [Deaton2001a, Tarozzi2001]. However,
previousdesign. For this reason,as well be used to get an idea of trends in total this shouldnot be regardedas a definitive
as becauseof possiblereconciliationbe- expenditures, and hence, of trends in validationof the proposedmethod,given
tween seven-dayand 30-day reports,the poverty. the ambiguitiesassociatedwith the thin
latest headcountratios are biased down This procedure is valid if two assump- rounds (Section IV.3). There are other
comparedwith what would have been tions hold. The first is that reported ex- possibleapproaches toadjustment thathave
obtainedon the basis of the traditional penditures on the intermediate goods, for not yet been explored,and furtherwork
questionnaire. which the recall period is unchanged, are may lead to differentconclusions.Mean-
Thereis another,quitedifferentproblem unaffected by the changes elsewhere in the while, we regardour adjustedfigures as
withtheofficialestimates,whichdoes not
concernthe55thRoundspecifically.This Table 2a: State-Specific Headcount Ratios
relates to the state and sector specific (Per cent)
povertylinesthatareusedby thePlanning OfficialMethodology AdjustedEstimates
Commissionto computethe povertyes- 1987-88 1993-94 1999-00 1987-88 1993-94 1999-00
timates.In severalcases the povertylines
Rural
areimplausible,particularly theverymuch AndhraPradesh 21.0 15.9 10.5 35.0 29.2 26.2
higher urban than rural lines in several Assam 39.4 45.2 40.3 36.1 35.4 35.5
states.The sourceof the problemlies in Bihar 53.9 58.0 44.0 54.6 48.6 41.1
the use of defectiveprice indexes in ad- Gujarat 28.6 22.2 12.4 39.4 32.5 20.0
Haryana 15.4 28.3 7.4 13.6 17.0 5.7
justmentsof thepovertylineovertimeand HimachalPradesh 16.7 30.4 7.5 13.3 17.1 9.8
between states. In the next section, we Jammuand Kashmir 25.9 30.4 4.7 15.3 10.1 6.1
discussways of overcomingthis problem Karnataka 32.6 30.1 16.8 40.8 37.9 30.7
andotherlimitationsof theofficialpoverty Kerala 29.5 25.4 9.4 23.8 19.5 10.0
MadhyaPradesh 42.0 40.7 37.2 43.7 36.6 31.3
estimates. Maharashtra 41.0 37.9 23.2 44.3 42.9 31.9
Orissa 58.7 49.8 47.8 50.4 43.5 43.0
1.2 Proposed Adjustments Punjab 12.8 11.7 6.0 6.6 6.2 2.4
Rajasthan 33.3 26.4 13.5 35.3 23.0 17.3
TamilNadu 46.3 35.9 20.0 49.0 38.5 24.3
In this paper,we presenta new series UttarPradesh 41.9 42.3 31.1 34.9 28.6 21.5
of consistentpovertyestimatesforthemost West Bengal 48.8 41.2 31.7 36.3 25.1 21.9
recent quinquennialrounds (1987-88, All-IndiaRural 39.4 37.1 26.8 39.0 33.0 26.3
Urban
1993-94 and 1999-2000).3 Essentially, AndhraPradesh 41.1 38.8 27.2 23.4 17.8 10.8
these involvefourmajordeparturesfrom Assam 11.3 7.9 7.5 13.6 13.0 11.8
the official estimates.First,an attemptis Bihar 51.9 34.8 33.5 38.1 26.7 24.7
madeto 'adjust'the55th-Roundestimates Gujarat 38.5 28.3 14.8 16.4 14.7 6.4
Haryana - 18.4 16.5 10.0 11.8 10.5 4.6
to achievecomparabilitywith the earlier HimachalPradesh 7.2 9.3 4.6 1.7 3.6 1.2
rounds.Second, we use improvedprice Jammuand Kashmir 15.0 9.3 2.0 3.8 3.1 1.3
indexesto updatethe 'povertyline' over Karnataka 49.2 39.9 24.6 26.0 21.4 10.8
Kerala 39.8 24.3 19.8 21.0 13.9 9.6
time, andto derivestate-specificpoverty
MadhyaPradesh 47.3 48.1 38.5 20.7 18.5 13.9
linesfromtheall-Indiapovertyline.Third, Maharashtra 40.3 35.0 26.7 21.2 18.2 12.0
a similarprocedureis used to derive an Orissa 42.6 40.6 43.5 20.8 15.2 15.6
explicit estimateof the appropriategap Punjab 13.7 10.9 5.5 6.6 7.8 3.4
Rajasthan 37.9 31.0 19.4 19.8 18.3 10.8
betweenruralandurbanpovertylines (in TamilNadu 40.2 39.9 22.5 26.2 20.8 11.3
contrastwith the often implausiblerural- UttarPradesh 44.9 35.1 30.8 29.3 21.7 17.3
urbangapsthatareimplicitin the official West Bengal 33.7 22.9 14.7 22.3 15.5 11.3
Delhi 15.1 16.1 9.2 4.7 8.8 2.4
estimates).Fourth,in additionto corrected All-IndiaUrban 39.1 32.9 24.1 22.5 17.8 12.0
'headcountratios', we presentestimates
of a potentiallymoreinformativepoverty Source:Authors'calculationsbased on NSS unitrecorddata from43rd, 50th and 55th Rounds.

the best currentlyavailable in terms of deflatorof the urbanpoverty line. The prices used in the official deflatorsare
dealingwith the changein questionnaire price indexes for each state show rather outdated.
design, withoutpretendingthatthey rep- modest differences from one state to The third departureconcerns the gap
resentthe final word on the topic. another.They also differ from those im- between rural and urbanpoverty lines.
Turningto the price adjustments,one plicitintheofficialpovertylines,although Fromthemid-1970suntiltheearly 1990s,
limitationof the price indexes that have the two sets of deflatorsare correlated. therewereonlytwopovertylinesforIndia,
been traditionallyused to updatepoverty This patternis consistentwith the fact one for ruralandone forurban.Theurban
lines over time (e g, the ConsumerPrice thatrelativepricesacrossstatesvarysome- line was around15 per cent higherthan
Indexfor AgriculturalLabourers)is that what over time, and that the interstate the ruralline, andbothwereheldfixed in
they are based on fixed and frequently
outdatedcommodity'weights'.It is pos- Table 2b: State-Specific Poverty-Gap Indexes
sible to calculatealternativepriceindexes
OfficialMethodology AdjustedEstimates
using the informationin the consumer 1987-88 1993-94 1999-00 1987-88 1993-94 1999-00
'expenditure surveysthemselves.Formore
than170 commodities,householdsreport Rural
AndhraPradesh 4.4 2.9 1.8 8.0 5.8 4.8
bothquantitiesandexpenditures,and the Assam 7.4 8.3 8.5 6.5 5.7 6.1
ratioof'the latterto the formerprovides Bihar 12.9 14.7 8.7 13.2 10.7 8.5
anestimateof thepricepaid.Theseprices Gujarat 5.5 4.1 2.2 8.4 6.8 3.8
canthenbe combinedintoconsumerprice Haryana 3.6 5.6 1.3 2.8 3.0 0.7
HimachalPradesh 2.6 5.6 1.0 2.1 3.0 1.5
index numbersthat allow comparisons Jammuand Kashmir 4.5 5.6 0.6 2.4 1.6 0.7
across states, and if we use data from Karnataka 7.9 6.3 2.7 10.5 8.6 6.1
differentrounds,for statesandthe whole Kerala 6.4 5.6 1.5 4.8 3.9 1.7
MadhyaPradesh 10.6 9.5 7.7 11.2 8.2 6.6
countryat differentpoints in time. One Maharashtra 9.6 9.3 4.4 10.8 11.2 7.6
limitationof these price indexes is that Orissa 16.3 12.0 11.7 13.0 9.7 10.5
their coverage of commodities is only Punjab 2.0 1.9 0.8 1.0 1.0 0.3
Rajasthan 8.6 5.2 2.1 9.2 4.4 3.0
partial(a little morethanhalf the budget TamilNadu 12.6 7.3 3.8 13.7 9.1 4.6
in the 55thRound,thoughmorein earlier UttarPradesh 9.9 10.4 5.8 7.5 5.8 3.9
rounds),so thatthey cannotcaptureprice West Bengal 11.6 8.3 6.5 7.7 4.2 3.5
changesin importantitems such as trans- All-IndiaRural 9.4 8.4 5.2 9.2 7.0 5.2
Urban
portation,housing,most non-foodgoods, AndhraPradesh 10.6 9.3 5.6 4.9 3.4 1.9
and services.However,CPIALdatasug- Assam 1.5 0.9 1.5 2.0 2.0 1.9
gest thatthe inflationratefor the uncov- Bihar 13.0 7.9 6.7 8.2 5.6 5.0
ered items is not very differentfromthat Gujarat 8.2 6.2 2.4 2.8 2.6 1.0
Haryana 3.6 3.0 2.0 2.3 1.9 0.7
applyingto the covereditems.6The price HimachalPradesh 0.7 1.2 0.6 0.2 0.5 0.2
indexesfromthe surveyshavethe advan- Jammuand Kashmir 2.4 1.2 0.2 0.5 0.5 0.2
tage of being based on several million Karnataka 14.1 11.4 5.6 5.7 4.5 2.1
actualpurchasesin eachround.Theyalso Kerala 10.4 5.5 3.9 4.5 2.7 1.7
make it possible to use formulasfor su- MadhyaPradesh 13.6 13.4 9.5 4.1 3.5 2.6
Maharashtra 12.3 10.1 6.7 5.3 4.6 2.8
perlativeindexes,suchas the Fisherideal Orissa 11.1 11.4 11.1 4.2 3.0 3.0
index or the Tomqvist index, that allow Punjab 2.3 1.7 0.6 1.0 1.1 0.4
for substitutionbehaviouras households Rajasthan 9.6 7.0 3.4 4.0 3.2 1.7
TamilNadu 11.5 10.2 4.8 6.2 4.5 2.0
adaptto relativepricechangesover time. UttarPradesh 12.2 9.0 6.6 6.3 4.6 3.3
ThecalculatedTomqvistindexesforthe West Bengal 7.4 4.5 2.5 4.2 2.9 1.9
43rd and 50th Rounds are reportedin Delhi 2.8 3.9 1.5 0.7 1.7 0.4
Deaton and Tarozzi (2000), and were All-IndiaUrban 10.4 8.3 5.2 4.8 3.7 2.3
updatedto the 55th Round by Deaton Notes to Table 2:

(2001b).7Thesepriceindexesdifferfrom Table2a. The headcountratioslabelled"officialmethodology"are computedfromthe unitrecord
data using the officialpovertylines, as well as the officialproceduresforassigning povertyrates (or
the official indexesin a numberof ways.
povertylines) to small states. We have also followedthe officialtreatmentof Jammuand Kashmir.
In particular,they rise somewhat more The all-Indiapovertyrates are computedby adding-upthe numberof poorin each state and dividing
slowlyovertimethando theofficial price by the totalpopulation.Because the PlanningCommissionuses interpolationratherthancomputations
fromthe unitrecorddata, there are minordifferencesbetween these numbersand those publishedin
indexes,especiallyin the ruralsector.For the officialreleases. The adjustedestimates are computedas described in the text (and morefullyin
example,theall-IndiaruralTormqvist index Deatonand Tarozzi,2001, and Deaton,2001b); they use priceindexes computedfromthe unitrecord
rises by 69.8 per cent from 1987-88 to data, and correctforthe changes in questionnairedesign in the 55th Round.The finalcolumnis a
1993-94 and by a further54.5 per cent somewhat refinedversionof the correspondingcolumnin Deaton (2001b). The estimates forJammu
from 1993-94 to 1999-2000, compared and Kashmirare calculateddirectly,and not by assuming the povertyline or povertyrateforany
otherstate (as in the officialmethodology).
with 78.7 per cent and 59.1 per cent for Table2b. The poverty-gapindexes labelled"officialmethodology"are computedfromthe unitrecord
the deflatorsimplicit in the official all- data using the officialpovertylines, and using rules forassigning poverty-gapindexes to small states
India ruralpoverty line. For the urban (and to J and K)that mirrorthe rules used by the PlanningCommissionforcomputingthe official
headcountratios.The adjustedindexes use the recomputedprice indexes to updatethe povertylines,
sectoroverthe two periods,the Tomqvist and correctforthe changes in questionnairedesign in the 55th Round.Allnumbersare directly
priceindexesriseby 73.8 and57.7 percent computedfrompovertylines and unitrecorddata foreach state, and the all-Indiaestimates are
versus73.5and61.4percenrfortheimplicit calculatedas weightedaverages of the state estimates.
3732 Economic and Political Weekly September 7, 2002

real terms, with updating on the basis of compare privateconsumptionlevels across properties. For instance, an income trans-
approximate price indexes such as the sectors, bearing in mind that this is at best fer from a very poor person to someone
Wholesale Price Index or the CSO's pri- a partialpicture of the relevant differences who is closer to the poverty line may lead
vate consumptiondeflator.The initialrural- in living standards.9 These comparisons to a decline in the headcount ratio, if
urban gap of 15 per cent is anchored in can be made by anchoring poverty estima- it 'lifts' the recipient above the poverty
1973-74 calorie consumption data, but it tes in a single poverty line, adjusted where line. Similarly, if some poor households
is essentially arbitrarysince the urbanand appropriateto take into accountrural-urban get poorer, this has no effect on the
rural 'calorie norms' themselves (2,100 price differences, using the same method headcount ratio.
and 2,400 calories per person per day, as that described earlier for adjusting A related issue is that changes in HCRs
respectively) have a fragile basis.8 More poverty lines over time andbetween states. can be highly sensitive to the number of
recently, the Planning Commission has Based on this procedure,the urbanpoverty poor households nearthe povertyline (since
adopted a modified version of the poverty line tends be about 15 per cent higher than changes in the HCR are entirely driven by
lines recommendedby a 1993 ExpertGroup the rural poverty line, though there are 'crossings' of the poverty line). If poor
[Government of India 1993]. The Expert variations across states. As it turnsout, this households are heavily 'bunched' near the
Groupretainedthe original ruraland urban rural-urbandifference in poverty lines is poverty line, a small increase in average
lines, but adjusted them for statewise broadly consistent with the original per capita income could lead to a mislead-
differences in price levels, separately for methodology used before the adoption of ingly large decline in the headcount ratio.
urbanand ruralsectors, using estimates of the Expert Group recommendations. This 'density effect' has to be kept firmly
statewise price differences calculated from To recapitulate,the revised poverty lines in view in the context of comparisons of
NSS data on expenditures and quantities used in this paper, which are presented in poverty change, involving questions such
using similar methods to those adopted in full in Table 4 of Deaton (2001b), are as "has there been more poverty decline
this paper. The Expert Group lines used derived as follows. Our starting point is in Bihar than in Punjab during the nine-
the thenbest-availableinformationon price the official rural all-India poverty line for ties", or "has poverty declined faster in the
differences across states, both urban and the 43rd Round (1987-88): 115.70 rupees nineties than in the eighties?" Often such
rural, but the information was outdated, per person per month.10 Rural poverty questions are answered by looking at, say,
especially for the rural sector. lines for each state for the 43rd Round are the respective changes (absolute or pro-
Because the statewise adjustments were obtained by multiplying this base poverty portionate) in headcount ratios. These
done separatelyfor urbanand ruralhouse- line by the ruralprice indexes for each state changes, however, are difficult to interpret
holds, the price differences between the relative to all-India. The urban poverty in the absence of furtherinformationabout
urbanand rural sectors of each state were lines for the 43rd Round, for each state as the initial density of poor households near
derivedonly implicitly, andsome arerather well as for all-India, are calculated from the poverty line in each case.
implausible, particularly the very much the ruralpoverty lines by scaling up by the One way forward is to use more sophis-
higher urban than rural lines in several respective urban relative to rural price ticated poverty indexes such as the Foster-
states. For example, the most recent urban indexes. In all cases, we use the relevant Greer-Thorbecke(FGT) indexes or the Sen
poverty lines for Andhra Pradesh and T6rnqvist price indexes. l1 To move to the index. In this paper, we focus on the sim-
Karnatakaare around 70 per cent higher 50th Round, the original all-India rural plest member of the FGT class (other than
thanthe correspondingrurallines, with the line, 115.70 rupees, is scaled up by the the headcount ratio itself), the 'poverty-
uncomfortableresult that urbanpoverty is Tornqvist index for all-India rural for the gap index'. Essentially, the poverty-gap
much higher than rural poverty in these 50th Round relative to the 43rd Round, index (hereafter PGI) is the aggregate
two states (see Table 2 in the next section). 1.698, to give an all-India rural poverty shortfall of poor people's consumption
In Assam, by contrast, the rural poverty line for the 50th Round. This number is from the poverty line, suitably normal-
line is actually higher than the urban line, then used to generate ruraland then urban ised.13 The PGI can also be interpretedas
andbasedon these odd povertylines, Assam poverty lines for each state, following the headcount ratio multiplied by the mean
turns out to be one of India's highest- exactly the same procedureas for the 43rd percentage shortfall of consumption from
poverty states for rural areas but lowest- Round. Finally, poverty lines for the 55th the poverty line (among the poor). This
poverty states for urban areas. It is hard Round are calculated in the same way from index avoids the main shortcomings of the
to accept these and other implications of an all-India rural line, which is the 50th headcount ratio, is relatively simple to
the Expert Group poverty lines. Round all-India ruralline scaled up by the calculate, and has a straightforwardinter-
There are grounds, of course, for ques- value of the T6rnqvist index between the pretation.14
tioning whetherit is even possible to derive two surveys.12
comparable ruraland urbanpoverty lines. The motivation for the fourth departure 1.3 Adjusted Estimates
Comparisons of living standards in rural (or ratherextension) arises from the limi-
and urban areas are inherently difficult, tations of the headcount ratio as an indi- Table la presents official and adjusted
since there are large intersectoral differ- cator of poverty. The headcount ratio has estimates of the all-India headcount ratio.
ences not only in the patternsof consump- a straightforwardinterpretationand is easy In each panel, the first row gives the official
tion but also in lifestyles, public amenities, to understand. In that sense it has much estimates; the second row retains the of-
epidemiological environments, and so on. 'communication value'. Yet, the HCR has ficial poverty lines but adjusts the 1999-
One way forward is to avoid such com- serious limitations as a poverty index. For 2000 estimates for the change in question-
parisons altogether,and to focus on sector- one thing, it ignores the extent to which naire design in the way described earlier;
specific (ruralor urban)poverty estimates. different households fall short of the the third row gives fully-adjusted poverty
Yet there is a case for attempting to poverty line. This leads to some perverse estimates, which combine the adjustments
Economicand PoliticalWeekly September7, 2002 3733

for questionnairedesign and for pricein- Figure 1: Official and Adjusted Headcounts Ratios
dexes. Table lb gives the corresponding
poverty-gapindexes. 60-
As the first two rows of each panel
indicate,the official estimatesare quite
misleadingin theirown terms:the 1999-
2000 povertyestimatesare biaseddown- ' 50-
ward by the changes in questionnaire o \ Officialcounts
design.Forheadcountratios,theestimates o
Thinrounds
adjusted for changes in questionnaire 2 40-
design 'confirm' about two-thirds of the
official decline in ruralpovertybetween
1993-94and 1999-2000,andabout90 per
cent of the decline in urbanpoverty.For 30
poverty-gapindexes, the corresponding Adjustedcounts
proportions(62 per cent and 77 per cent)
arelower,especiallyfor the urbansector.
The fully-adjustedestimatesin the last 20
row of each panelshow somewhatlower 1970
I I
1980 1990 2000
ruralpovertyestimatesand much lower Year
urbanpoverty estimates for 1999-2000 Source: PlanningCommission,Press Releases ( March11, 1997, and February22,2001), Deaton(2001a,
than even the official estimates. Note, b), and Table 2a of this paper.
however,thatbecausewe arerecalculating
the povertylines backto the 43rdRound, becausewe treatthe ruralpovertyline in hasthehighestlevelof ruralpovertyamong
a good deal of the decreasetook place in the 43rd Roundas our baseline so that, all Indianstates,accordingto the adjusted
the six yearspriorto 1993-94,not only in withlargerrural-urban gapsin thepoverty 1999-2000estimates.17Reassuringly,the
the six yearssubsequentto 1993-94.The estimates, we estimate lower poverty 'anomalies'noted earlierwith respectto
fully-adjusted estimatesfor theheadcount overall.If instead,we hadtakentheurban rural-urban gaps in specific statestendto
ratiosandpovertygapindexessuggestthat povertyline as base, the adjustedfigures disappearas one moves from official to
poverty decline has been fairly evenly would have been higherthanthe official adjustedestimates.
spreadbetweenthe two sub-periods(be- figures. From 1987-88 to 1993-94, the Table2b shows the correspondingpov-
fore and after 1993-94), in contrastwith adjustedheadcountratiofallsmorerapidly erty-gapindexes.The generalpatternsare
the patternof 'acceleration'in the second thantheofficialheadcount;thisis because very much the same as with headcount
sub-periodassociated with the official our price deflatorsare rising less rapidly ratios; indeed the PGI series are highly
estimates. thanthe official ones. From1993-94,the correlatedwith the correspondingHCR
The rural-urbangaps in the poverty adjustedfiguresfall moreslowly because series,withcorrelationcoefficientsof 0.98
estimatesarealsoof interest.Lookingfirst theeffectsof thepriceadjustment aremore forruraland0.95 forurban.EventheHCR
at thebaseyear(1987-88),therural-urban thanoffset by the correctionfor question- and PGI changes between the 50th and
gap basedon adjustedestimatesis much naire design. The estimatesfor the thin 55th Rounds are highly correlated;the
largerthanthatbasedon officialestimates. rounds - which look very different - are correlationcoefficientbetweenchangesin
Indeed,the latter suggest no difference included to remind us of the residual HCR and changes in PGI is 0.95 for the
betweenruraland urbanpovertyin that uncertaintyabout our conclusions, and ruralsector,and0.96 for the urbansector.
year.This is hardto reconcilewith inde- will be discussedfurtherin Section IV.3 Thus,in spiteof its theoreticalsuperiority
pendentevidenceon living conditionsin below. over the headcountratio,the poverty-gap
rural and urban areas, such as a life- indexgives us verylittleadditionalinsight
expectancygap of about seven years in 1.4 Regional Contrasts in this case.
favouror urbanareasaroundthattime.15 In interpretingand comparingpoverty
Ourlow estimateof the urbanheadcount State-specificheadcountratiosarepre- declines over time, it is useful to supple-
ratiorelativeto the officialestimate(22.5 sented in Table 2a.16The table has the mentthepovertyindexeswithinformation
percentversus39.1 percent),andsimilar same basic structureas Table la, except oh the growthrate of averageper capita
differencesin 1993-94 and 1999-2000, thatwejumpstraightfromofficialto fully- consumption expenditure (hereafter
come fromthe fact thatwe takethe rural adjustedestimates.The lattersuggestthat APCE).State-specificestimatesof APCE
poverty line in 1987-88 as our starting the basic pattern of sustained poverty growthbetween 1993-94 and 1999-2000
point, and peg the urbanpoverty lines declinebetween1987-88and 1999-2000, are shown in Table 3, where states are
about 15 per cent higher than the rural discussedearlierattheall-Indialevel, also rankedin ascendingorderof APCEgrowth
povertylines,incontrasttothemuchlarger appliesat the level of individualstatesin for rural and urban areas combined.18
differentialsembodiedin theofficiallines. mostcases.Themainexceptionis Assam, Here,a strikingregionalpatternemerges:
Figure1 showsthe new estimatesof the wherepovertyhas stagnatedin bothrural except for Jammuand Kashmir,the low-
headcountratiostogetherwiththeofficial andurbanareas.In Orissa,therehas been growthstatesformone contiguousregion
estimatesgoingbackto 1973-74.Thefully very little povertydecline in the second made up of the eastern states (Assam,
adjusted figures are lower throughout sub-period,withtheresultthatOrissanow Orissa and West Bengal), the so-called

Figure 2: 'Divergence' of Per Capita Expenditure Across States, 1993-94 to 1999-00
relatively large absolute declines in the
headcount ratio.
6- An alternative approach is to look at
proportionate changes in HCRs or PGIs.
These turnout to be highly correlated with
A the corresponding growth rates of APCE.
A DE
c HA The point is illustrated in Figure 3, where
a)
we plot the proportionate decline in the
C.) rural headcount ratio in each state against
4-
the growth rateof APCE in ruralareas. The
TN
correlation coefficient between the two
CD A
cA A PU
PU
series is as high as 0.91. This reflects the
GU KE
c HP
fact that poverty reduction is overwhelm-
A
a, KA MA ingly driven by the growth rate of APCE,
MA ratherthan by changes in distribution- we
a,
>
<
2- shall return to this point in Section III.
UP AAI
UP
A
From these observations, it follows that if
MP
BI A RA we accept 'proportionatechange in HCR'
A AP
A (or PGI) as an index of poverty reduction,
A A
JK then the broad regional patterns identified
OR A WB earlier for the growth rate of APCE also
0- AS tend to apply to poverty reduction. In
I I I I 1 particular: (1) most of the western and
200 300 400 500 600 southern states (with the importantexcep-
Geometricmean PCE in 1993-94, rupees per head tion of Andhra Pradesh) have done com-
paratively well; (2) the eastern region has
Source: Authors'calculationsusingunitrecorddatafromthe 50thand 55thRoundsof the NationalSample achieved very little poverty reduction
Survey. between 1993-94 and 1999-2000; and
Al = All India; AP = Andhra Pradesh; AS = Assam; BI = Bihar; DE = Delhi; GU = Gujarat; HA = Haryana;
HP = HimachalPradesh;JK = Jammuand Kashmir;KA= Karnataka;KE= Kerala;MA= Maharashtra; (3) there is a strong overall pattern of 'di-
MP= MadhyaPradesh;OR= Orissa;PU = Punjab;RA= Rajasthan;TN= TamilNadu;UP = UttarPradesh; vergence' (states that were poorer to start
WB= West Bengal. with had lower rates of poverty reduction).
This reading of the evidence, however,
BIMARU states (Bihar, Madhya Pradesh, across Indian states in the nineties.20 The remains somewhat tentative, since there is
Rajasthanand UttarPradesh), and Andhra point is illustrated in Figure 2, which plots no compelling reason to accept the pro-
Pradesh. The high-growth states, for their the average growth in APCE for each state portionate decline in HCR (or PGI) as a
part, consist of the southern states (except between 1993-94 and 1999-2000 against definitive measure of poverty decline.
Andhra Pradesh), the western states the geometric mean of APCE in 1993-94. We end this section with a caveat. From
(Gujaratand Maharashtra)and the north- It is worth asking to what extent these Table 2 and Figure 1, it may appear that
western region (Punjab, Haryana and regional patterns,based on APCE data, are the 'pace' of poverty decline in the nineties
Himachal Pradesh). Further,it is interest- corroborated by regional patterns of pov- has been fairly rapid. It is important to
ing to note that this pattern is reasonably erty decline. One difficulty here is that note, however, thatthe associated increases
consistent with independentdataon growth there is no obvious way of 'comparing' the in per capita expenditure have been rather
rates of per capita 'state domestic product' extent of poverty decline across states. For modest in most cases. For instance, the
(SDP); these are shown in the last column instance, looking at absolute changes in decline of 6.6 percentage points in the all-
of Table 3. With a couple of exceptions (say) HCRs would seem to give an unfair India HCR (from 29.2 per cent to 22.7 per
on each side, all the states in the 'low 'advantage' to states that start off with cent) between 1993-94 and 1999-2000 is
APCE growth' set had comparatively low high levels of poverty, and where there driven by an increase of only 10.9 per cent
rates of per capita SDP between 1993-94 tends be a large number of households in average per capita expenditure - not
and 1999-2000 (say below 4 per cent per close to the poverty line. To illustrate, the exactly a spectacular improvement in
year), and conversely, all the states in the absolute decline of the ruralHCR between living standards. Similarly, Table 2a sug-
'high APCE growth' set had compara- 1993-94 and 1999-2000 was about twice gests that Bihar achieved a large step in
tively high annual growth rates of per as large in Bihar (7.4 percentage points) poverty reduction in the nineties, with the
capita SDP (the correlation coefficient as in Punjab(3.8 points), yet over the same ruralHCR coming down from 49 per cent
between the two series is 0.45).19 period APCE grew by only 6.9 per cent to 41 per cent. Yet, as Table 3 indicates,
This broad regional pattern is a matter in Bihar compared with 20.2 per cent average APCE in rural Bihar increased
of concern, because the low-growth states in Punjab, with virtually no change in by only 7 per cent between 1993-94 and
also tend to be states that started off with distribution in either case.21 The reason 1999-2000.
comparatively low levels of APCE or per- for this contrast is that Bihar starts off in Why are small increases in APCE asso-
capita SDP. In other words, there has been 1993-94 with a very high proportion of ciated with substantial declines in poverty
a growing 'divergence' of per capita ex- households close to the poverty line, so indexes? It is tempting to answer that the
penditure (and also of per capita SDP) that small increases in APCE can produce distribution of consumer expenditure

must have improved in the nineties. As Figure 3: HCR Declines and APCE Growth, 1993-94 to 1999-2000
discussed in Section III, however, this is (Rural)
not the case: indeed economic inequality
has increased ratherthan decreased in the 70-
nineties. The correct answer relates to the
HA
'density effect' mentioned earlier (see 60 - PU
Section 1.2): when many poor households l~~~~~~~~o~~PU
0
are close to the poverty line, modest in-
50-
creases in APCE can produce substantial
declines in standardpoverty indexes. One o
I~~~~~~~~C HP
^~KE
reason for drawing attention to this is that u 40- GU

XC JK A
the official poverty estimates have some- ~~~TN
C^~~~u~
times been used to claim that the nineties 2 30
have been a period of spectacular achieve- C
RA UP
. 'A A MA
ments in poverty reduction. In fact, when
20 -
the relevant adjustments are made, and Bl
a) WB KA
the poverty indexes are read together with WB A'
Cu A MP
the informationon APCE growth, poverty 5 10- A
*? AP
reduction in the nineties appears to be 0.
o OR
more or less in line with previous rates of 0-
progress. AS
II -10 -
FurtherEvidence 0 5 10 15 20 25 30
Six-year growth of rural APCE

11.1National Accounts Statistics Source: Tables 2a and 3.
There has been much discussion of the

consistency between National Sample consumptionexpenditure'.Overtime,the exactly the same, and also thatthereare
Survey data and the 'national accounts' CSOestimateshavetendedto growfaster majormethodologicaldifferencesbetween
published by the Central Statistical thantheNSSestimates,leadingsomecom- thetwosources.TheNSS figuresaredirect
Organisation (CSO).22 The latter include mentatorsto question the reliabilityof estimatesof householdconsumptionex-
estimates of 'private final consumer ex- NationalSampleSurvey data. penditure.The CSO figures includesev-
penditure', which is frequently compared It is importantto note that these two eral items of expenditurethat are not
with NSS estimates of 'household notionsof 'consumerexpenditure'arenot collectedintheNSS surveys;examplesare
expendituresby non-profitentreprises,as
Table 3: Growth Rates of APCE and Per Capita SDP, 1993-94 to 1999-2000 well as imputedrentby owneroccupiers
and 'financial intermediationservices
Six-YearGrowthof APCE('Adjusted'),1993-94 to 1999-2000 AnnualGrowthRate of
Rural Urban Combined Per CapitaSDP, indirectlymeasured'(the last item is es-
1993-94 to 1999-2000 sentiallythe net interestearnedby finan-
cial intermediaries,which is counted as
Assam 0.9 8.8 1.7 0.58
Orissa 1.4 -0.0 3.3 2.34 expenditureson intermediationservices
West Bengal 2.1 11.5 3.3 5.48 by households).Accordingto Sundaram
Jammuand Kashmir 5.4 8.0 5.3 2.49 andTendulkar(2002), who quotea recent
Bihar 6.9 4.8 7.1 2.10 cross-validationstudy by the National
MadhyaPradesh 6.6 14.1 7.8 2.78
AndhraPradesh 2.8 18.5 8.3 3.57 AccountsDepartment,the last two items
Rajasthan 7.0 15.4 8.6 4.60 accountfor 22 per cent of the difference
UttarPradesh 8.3 10.1 9.0 2.99 in levels betweenCSOandNSS estimates
India 8.7 16.6 10.9 4.36
Karnataka 9.5 26.5 14.0 5.82
of consumerexpenditure. Further,theCSO
Maharashtra 14.1 16.7 15.9 3.53 estimatesare 'residual'figures,obtained
Gujarat 15.1 20.9 16.8 4.88 after subtractingother items from the
HimachalPradesh 16.2 28.5 17.6 5.06 national product. Leaving aside these
TamilNadu 15.7 25.1 18.9 5.39
Kerala 19.6 18.2 19.6 4.01 comparabilityissues,thereis indeeda gap
Punjab 20.2 17.9 19.9 2.74 between the CSO-basedand NSS-based
Haryana 31.0 23.0 29.2 3.05 growth rates of consumer expenditure.
Delhi - 30.7 30.7 5.69
Accordingto CSO data, per capitacon-
Note: The states are arrangedin ascending orderof the growthrate of APCEfor ruraland urbanarea sumer expenditurehas grown at much
combined. the same rateas per capitaGDP between
Sources:ForAPCE:Authors'calculationsfromunitrecorddataforthe50thand55thRoundsof the National
1993-94 and 1999-2000- about3.5 per
SampleSurvey.ForSDP:Authors'calculationsbased on unpublisheddata kindlysuppliedbythe
PlanningCommission.The figures in the last columnshould be taken as indicative,given the cent per year in real terms.23The cor-
significantmarginof errorinvolvedin SDP estimates. respondingNSS-basedestimateassociated

withour'adjusted'APCEfiguresis around (CPIAL). The quality of this information headcount ratios in different states was
2 per cent.24 is not entirely clear, but available evidence 0.79 in absolute value, rising to 0.91 if
Thisis quitedifferentfromthe situation suggests that it is adequate for the purpose Kerala is excluded. In 1993-94, the cor-
that prevailedprior to the 55th Round, of broad comparisons.27 relation coefficient was 0.87 in absolute
when consumerexpenditurewas hardly As Figure 4 illustrates, real agricultural value, with or without Kerala. Interest-
growingat all accordingto the NSS 'thin wages in different states are highly cor- ingly, if 'official' HCRs are used instead
rounds'but gallopingforwardaccording related with expenditure-based poverty of our adjusted HCRs, the correlation
to theCSOdata.Today,in thelightof more indexes (here and elsewhere in this sec- coefficients come down quite sharply (e g,
recent estimates, the discrepancylooks tion, the focus is on rural poverty). The from 0.91 to 0.73 in 1999-2000 and from
muchsmaller.Thatdiscrepancycalls for main 'outlier' is Kerala, where real wages 0.87 to 0.54 in 1993-94, without Kerala
furtherscrutinyandresolution,butmean- are far above the 'regression line'; it seems in both cases). This can be tentatively
while, it can hardly be regardedas an that the power of labour unions in Kerala regarded as a further indication of the
indictment of NationalSampleSurveydata. has raised agricultural wages well above plausibility of the proposed adjustments.
Forone thing,thereferencecategoriesare the level found in any other Indian states, Given the close association between real
not the same. For another,there is no but that this does not translate into a wages and rural poverty, the growth rates
reasonto believe thatthe CSO estimates correspondingly low level of ruralpoverty, of real wages over time provide useful
aremoreaccuratethantheNSSestimates; possibly because high wages are partly supplementaryevidence on poverty trends.
indeedthecross-validation exerciseraised offset by high unemployment, or because According to recent estimates based on
seriousquestionsabouta numberof the other determinantsof ruralpoverty are also AWI data, real agricultural wages were
consumptioncategoriesin theCSOdata.25 at work. In 1999-2000, the correlation growing at about 5 per cent per year in the
coefficient between real wages and eighties and 2.5 per cent per year in the
11.2Agricultural Wages
Table 4: Growth and the Headcount Ratio, 1993-94 to 1999-2000
Agriculturalwages provide an impor- HCR50 Derivativewith Six Years Change in Change in
tant source of further information on Respect to Growth Growth HCR55 HCR55,
InequalityFixed Actual
poverty.Thereare, in fact, two ways of
thinkingaboutthe relevanceof this infor- Rural
mation.First,real agriculturalwages are AndhraPradesh 29.2 -0.90 2.8 -2.5 -3.0
Assam 35.4 -1.27 0.9 -1.4 0.1
highly correlatedwith standardpoverty Bihar 48.6 -1.06 6.9 -8.2 -7.4
indexes such as headcountratios:where Gujarat 32.5 -0.91 15.1 -12.1 -12.4
povertyis higher,wagestendto be lower, Haryana 17.0 -0.63 31.0 -12.9 -11.3
and vice versa. Based on this statistical HimachalPradesh 17.1 -0.75 16.2 -8.3 -7.3
Jammuand Kashmir 10.1 -0.50 5.4 -2.6 -4.0
association,real wages can be used to Karnataka 37.9 -0.91 9.5 -9.0 -7.2
provide some informationabout other Kerala 19.5 -0.62 19.6 -10.3 -9.5
povertyindexes.Second,it is alsopossible MadhyaPradesh 36.6 -0.93 6.6 -6.5 -5.3
Maharashtra 42.9 -0.81 14.1 -10.9 -11.0
to think aboutthe real wage as a rough Orissa 43.5 -1.04 1.4 -1.2 -0.5
povertyindicatorin its ownright.Theidea Punjab 6.2 -0.34 20.2 -4.0 -3.8
is that,if the labourmarketis competitive Rajasthan 23.0 -0.78 7.0 -5.5 -5.7
(at least on the supplyside), thenthe real TamilNadu 38.5 -0.90 15.7 -13.3 -14.1
UttarPradesh 28.6 -0.79 8.3 -6.6 -7.2
wagemeasuresthe 'reservationwage', i e, West Bengal 25.1 -0.97 2.1 -2.0 -3.2
thelowest wage at which labourersare All-India 33.0 -0.88 8.7 -6.8 -6.7
preparedto work.This has directeviden- Urban
tial valueas an indicationof the deprived AndhraPradesh 17.8 -0.62 18.5 -9.0 -6.9
Assam 13.0 -0.64 8.8 -3.1 -1.2
circumstancesin which people live (the Bihar 26.7 -0.79 4.8 -4.0 -2.0
moredesperatepeople are, the lower the Gujarat 14.7 -0.55 20.9 -8.7 -8.3
reservationwage), independentlyof the Haryana 10.5 -0.47 23.0 -6.3 -6.0
HimachalPradesh 3.6 -0.26 28.5 -2.9 -2.4
indirectevidentialvalue arisingfrom the Jammuand Kashmir 3.1 -0.15 8.0 -0.4 -1.8
statisticalassociationbetweenrealwages Karnataka 21.4 -0.60 26.5 -12.9 -10.6
and standardpovertyindexessuch as the Kerala 13.9 -0.46 18.2 -7.1 -4.2
headcountratio. MadhyaPradesh 18.5 -0.63 14.1 -8.0 -4.6
Maharashtra 18.2 -0.45 16.7 -6.1 -6.2
Detailed informationon agricultural Orissa 15.2 -0.54 0.0 0.1 0.4
wages is availablefromAgriculturalWages Punjab 7.8 -0.38 17.9 -4.9 -4.4
in India(AWI),an annualpublicationof Rajasthan 18.3 -0.59 15.4 -8.4 -7.5
the Directorateof Economicsand Statis- TamilNadu 20.8 -0.66 25.1 -12.9 -9.6
UttarPradesh 21.7 -0.59 10.1 -6.0 -4.4
tics, Ministryof Agriculture.The data West Bengal 15.5 -0.56 11.5 -5.8 -4.3
initially come in the form of district- Delhi 8.8 -0.26 30.7 -5.7 -6.4
specificmoney wages.26These are typi- All-India 17.8 -0.56 16.6 -7.4 -5.9
callyaggregatedusingthenumbersof agri- Source: Authors'calculationsfromthe unitrecorddata of the 43rd , 50th , and 55th Roundsof the NSS.
culturallabourersin differentdistrictsas Note thatthe hypotheticalall-Indiafiguresare calculatedon the counterfactualasssumptionthat
each household received the state growthrate. They thereforedo not show what would have
weights,anddeflatedusingthe Consumer happened had growthbeen more equally distributedacross the states: see the text for this
Price Index for AgriculturalLabourers alternativecalculation.

nineties.28 Thus, real agricultural wages Figure 4: Agricultural Wages and Rural Poverty, 1999-2000
were growing considerably faster in the 50 -
eighties than in the nineties. But even the
reduced growth rate of agricultural wages OR
in the nineties, at 2.5 per cent per year, *
A
BI
40
points to significant growth of per capita
expenditure among the poorer sections of AS
^A
the population and reinforces our earlier |
MP
findings on poverty reduction. In fact, this ' 30 A MA
reduced growth rate is a little higher than Ca KA
the growth rate of average per capita .o A INDIA
AP A
expenditure (1.5 per cent per year) that c I~ A
TN
AT
sustains our estimated declines of rural 20- UP WB A
-o GU
headcount ratios and headcount indexes 0 A
between 1993-94 and 1999-2000. .1 RA
ICu
The data on real wages also provide
some independent corroboration of the 10
state-specific patterns of poverty decline. A
This is illustrated in Figure 5, where we PU HA
plot state-specific estimates of the growth

rate of real agriculturalwages in the nine- 0 L
ties against the estimated proportionate 0 2 3 4 5 6 7 8
decline in the headcountratio(a very similar Real agriculturalwage
patternapplies to the poverty-gap index). Source: Dreze and Sen (2002), StatisticalAppendix,Table A 3, and Table 2a of this paper. The "real
Here the two main outliers are Punjab and agriculturalwage"is a three-yearaverage ending in 1999-2000.
Haryana, where the headcount ratio has
declined sharply without a correspond- estimates, based as they are on data for two frequency items, for which the reporting
ingly sharp increase in real wages (indeed years only. Yet it is reassuring to find that period was 30 days in the 50th Round and
without any such increase, in the case of they are consistent with the AWI-based 365 days in the 55th Round. As we have
Punjab). Leaving out these two outliers, estimates. already noted, the 365-day reporting pe-
the association between the two series is riod for these items pulls up the lower tail
remarkably close (with a correlation co- 11.3The 'Employment- of the consumption distribution, and thus
efficient of 0.88). Unemployment Surveys' biases down the headcount ratio compared
An interesting sidelight emerging from with earlier methods. However, Sundaram
Figure 5 is that a healthy growth of real The National Sample Survey's 1993-94 and Tendulkar note that the 50th Round
agricultural wages appear to be a 'suffi- and 1999-2000 employment-unemploy- contained both 30-day and 365-day report-
cient' condition for substantial poverty ment surveys (EUS) also include consumer ing periods for the low frequency items.
decline in rural areas: all the states where expenditure data. These can be used for Hence, by recalculating the 50th Round
real wages have grown at more than, say, further scrutiny of poverty trends. This headcounts using the 365-day responses,
2.5 per cent per year in the nineties have task has been undertakenin a recent paper they can put the 50th and 55th Rounds on
experienced a comparatively sharp reduc- by Sundaramand Tendulkar(2002). They a roughly comparable basis. When they do
tion of the rural headcount ratio. Con- note that the consumption survey in the this, they find that, in both ruraland urban
versely, in states with low rates of reduc- 1999-2000 EUS uses the traditional 30- sectors, they can confirm a little more than
tion of the headcountratio (say, 15 percent day reporting period, but differs from the three-quartersof the official decline in the
or less over six years), real wages have standard questionnaire by only asking an headcount ratios between the two rounds
invariably grown at less than 2 per cent abbreviatedset of questions. However, the [Sundaram and Tendulkar 2002:
per year. This applies in particular to the authors find that, in those cases where the Table III.8]. These calculations are not
entireeasternregion (Assam, Orissa, West questions have comparable coverage, the identical to our first-step adjustments (see
Bengal and Bihar) and also to Andhra means from the EUS, using the traditional Table la), but they are close enough to
Pradesh and Madhya Pradesh. 30-day reportingperiod, aretypically close inspire some confidence that both sets-of
Independentevidence on the growthrates to those from the 30-day questionnaire in results are in the right range.
of real wages has recently been presented the main consumption survey. Based on To sum up, the all-India poverty indexes
by K Sundaram(2001 a, 2001 b), based on this correspondence, they argue that the presented earlier in this paper are broadly
the 'employment-unemployment surveys' 30-day questions in the main 1999-2000 consistent with independentevidence from
(EUS) of the National Sample Survey for survey were not much distorted by the the national accounts statistics and the
1993-94 and 1999-2000. For the present seven-day questions thatwere asked along- employment-unemployment surveys, as
purpose, these surveys are comparable. side them. In this version of events, the well as with related information on agri-
Sundaramestimates that the real earnings major source of incomparability between cultural wages. There is also some congru-
of agricultural labourers have grown at the 55th and 50th Rounds is not the con- ence between the inter-state contrasts
about2.5 percent peryearbetween 1993-94 tamination of the 30-day questions, but emerging from NSS data and independent
and 1999-2000. These are tentative rather the revised treatment of the low information on state-specific growth rates
3738 Economicand PoliticalWeekly September7, 2002

Figure 5: Wage Growth and Poverty Decline, 1993-94 to 1999-2000 between these last two columns is the
change in the headcount ratio that is at-
80 tributable to changes in the shape of the
consumption distribution.
It is important to note that the last two
A columns are highly correlated. The corre-
HA lation coefficients across the states are
60 PU
cc PU 0.97 (rural)and0.93 (urban),so thatgrowth
0
I alone can predict much of the cross-state
-
A
KKE
patternof reductionin HCRs. Nevertheless,
the estimates are far from identical. In
440- TN
A
particular,the all-India calculations show
GU that 'growthalone' would have reduced the
MA UP poverty rate by more than actually hap-
c
o
A A A pened, implying that there was an increase
t:r A in inequality that offset some of the effects
20- A
MP INDIA KA of growth, or put differently, that APCE
growth among the poor was less than the
0|
AP
AAP average. These inequality effects vary
somewhat from state to state and are much
0 -A *AS
A
weaker in rural than in urban areas. In
-4 -2 0 2 4 6 8 10 urban India, increasing inequality moder-
Growthrate of real agriculturalwage
ated the decline in the headcount ratio in
Source:See Figure4. all states except Delhi, Maharashtra,and
Jammu and Kashmir. In some cases, such
of 'state domestic product' and real agri- distribution, would have led to a decline as urban Kerala and Madhya Pradesh, the
cultural wages. The combined evidence of 0.9 percentage points in the rural 'moderating effect' is pronounced, with
from these differentsources is fairly strong, headcount ratio.29This derivative depends actualratesof reductiononly a littleover half
even though each individual source has positively on the fraction of people who those predictedby the growth in the mean.
significant limitations. are at or near the poverty line, which is For the urbansector as a whole (the last
typically larger in the poorer states. The row of the table), the actual decline in the
III figures in column 2 vary from -1.27 in HCR is one and a half points lower (5.9
inthe
EconomicInequality rural Assam to -0.15 in urbanJammu and versus 7.4 per cent) than would have been
Nineties Kashmir. Column 3 reproduces the total the case had growth been equally distrib-
percentage growth between 1993-94 and uted within each state.This estimate, which
111.1Growth, Poverty 1999-2000 from Table 3. is the population-weighted average of the
and Inequality If we multiply the second column (the corresponding numbers for each state,
derivative with respect to growth) by the calculates what would have happened if
It is possible to think about poverty third column (the amount of growth), we each household in each state had experi-
decline, as captured by standard poverty get an estimate of the amount of poverty enced the average growth for that state. An
indexes, in terms of two distinct compo- reductionthatwe would expect fromgrowth alternative,andequally interesting,counter-
nents: a growth component and a distri- alone, in the absence of any change in the factual is what would have happened if,
bution component. The growth compo- shape of the distribution. This is an ap- between 1993-94 and 1999-2000, each
nent reflects the increase of average per proximation,becausethe derivativeis likely household in the country had experienced
capita expenditure. The distribution com- to change as the headcount ratio falls. In the countrywide growth rate of 10.9 per
ponent captures any change that may take column 4, we report a more precise cal- cent. Such a calculation yields an all-India
place in the distribution of per capita ex- culation: an estimate of what the headcount HCR of 21.4 per cent (for ruraland urban
penditure over households. ratio would have been in 1999-2000 if the areas combined), compared with an actual
This decomposition exercise is pursued distributions of consumption in each state all-India HCR of 22.7 per cent based on
in Table 4, with reference to the headcount were identical to those in 1993-94, but had the 55th Round. In other words, the all-
ratio (very similar results apply to the been shifted upwards by the amount of India HCR in 1999-2000 was 1.3 percent-
poverty-gap index). The first column re- growth in real per capita expenditure that age points higher than it would have been
peats the headcount ratio for 1993-94 actually took place. This can be readily (with the same growth rate of APCE) in
from Table 2. The second column (labelled calculated by reducing the 1993-94 pov- the absence of any increase in inequality.
'derivative with respect to growth') shows erty lines by the amount of growth, and
our estimate of the percentage-point re- re-estimating the headcount ratios from 111.2Aspects of Rising Inequality
duction in HCR associated with a distri- these adjusted lines and the 1993-94 ex-
bution-neutral,1 percent increase in APCE pendituredata.These hypotheticalchanges Three aspects of rising economic in-
in the relevant state. To illustrate, in rural can then be compared with the actual equality in the nineties have come up so
Andhra Pradesh a 1 per cent increase in reductions in the headcount ratios, shown far in our story. First, we found strong
APCE in 1993-94, with no change in in the final column. The difference evidence of 'divergence' in per capita

consumption across states. Second, our our findings on rising economic inequality occupation groups. Since agricultural
estimates of the growth rates of per capita within the urbansector are consistent with labourers and public sector employees
expenditure between 1993-94 and 1999- recentworkby BanerjeeandPiketty(2001), typically reside in rural and urban areas,
2000 (Table 3) point to a significant in- who use income tax records to document respectively, this finding may just be
crease in rural-urbaninequalities at the all- very large increases in income among the another side of the coin of rising rural-
India level, and also in most individual very highest income earners. They show urban disparities. Even then, it streng-
states. Third, the decomposition exercise that, in the 1990s, real incomes among the thens the evidence presented earlier on
in the preceding section shows that rising top one per cent of income earners in- aspects of rising economic inequality in
inequality within states, particularlyin the creased by a half in real terms, while those the nineties.
urbansector, has moderated the effects of of the top 1 per cent of 1 per cent increased To sum up, except for the absence of
growth on poverty reduction. by a factor of three in real terms. clear evidence of rising intra-rural in-
Table 5 provides more systematic evi- Second, it is interesting to compare the equality within states, we find strong indi-
dence on recent changes in consumption growth rate of real wages for agricultural cations of a pervasive increase in eco-
inequality within each sector of each state labourers with that of public sector sala- nomic inequality in the nineties. This is
using two different measures of inequality. ries. As we saw earlier, real agricultural a new development in the Indianeconomy:
We show the logarithm of the difference wages have grown at 2.5 per cent or so until 1993-94, the all-India Gini coeffi-
of the arithmetic and geometric means in the nineties. Public sector salaries, for cients of per capita consumer expenditure
(approximately the fraction by which the their part, have grown at almost 5 per cent in ruralandurbanareaswere fairly stable.31
arithmetic mean exceeds the geometric per year during the same period.30 Given Further, it is worth noting that the rate of
mean), as well as the variance of the that public sector employees tend to be increase of economic inequality in the
logarithm of per capita expenditure. much betteroff thanagriculturallabourers, nineties is far from negligible. For in-
The table shows that the correction for this can be taken as an instance of rising stance, the compounding of inter-state
questionnaire design is critical for under- economic disparities between different 'divergence' and rising rural-urban
standing what has been happening. (Note
that the correction for prices has no effect Table 5: Inequality Measures
within sectors and states.) The direct use
of the unit record data in the 55th Round, logAMYlogGMa Varianceof Logs
50th Round 55th Round 55th Round 50th Round 55th Round 55th Round
with no adjustment, shows a substantial Adjusted
Adjusted
reduction in inequality within the rural
AndhraPradesh 0.14 0.09 0.13 0.24 0.17 0.22
sectors of most states, with little or no 0.05 0.06 0.10 0.13 0.11
Assam 0.07
increase in the urban sectors. With the Bihar 0.08 0.07 0.08 0.16 0.13 0.16
correction, we see that within-state rural Gujarat 0.10 0.09 0.11 0.17 0.18 0.18
0.16 0.10 0.23 0.28 0.19 0.31
inequality has not fallen, and that there Haryana
0.13 0.10 0.14 0.22 0.17 0.24
HimachalPradesh
have been markedincreases in within-state Jammuand Kashmir 0.10 0.06 0.07 0.16 0.12 0.14
urbaninequality. We suspect that the main Karnataka 0.12 0.10 0.12 0.21 0.18 0.22
reason why the unadjusted data are so Kerala 0.15 0.14 0.16 0.26 0.24 0.27
MadhyaPradesh 0.13 0.10 0.12 0.22 0.18 0.22
misleading in this context is the change Maharashtra 0.16 0.11 0.16 0.27 0.20 0.28
from 30 to 365 days in the reportingperiod Orissa 0.10 0.10 0.12 0.18 0.18 0.21
for the low frequencyitems (durablegoods, Punjab 0.13 0.10 0.14 0.22 0.19 0.24
clothing and footwear, and institutional Rajasthan 0.12 0.07 0.10 0.20 0.14 0.18
medicalandeducationalexpenditures).The TamilNadu 0.16 0.14 0.15 0.27 0.23 0.24
UttarPradesh 0.13 0.10 0.12 0.23 0.18 0.21
longer reporting period actually reduces West Bengal 0.11 0.09 0.08 0.17 0.15 0.15
the mean expenditures on those items, but All-IndiaRural 0.14 0.11 0.14 0.23 0.21 0.24
because a much larger fraction of people AndhraPradesh 0.17 0.16 0.17 0.30 0.29 0.33
Assam 0.13 0.16 0.14 0.25 0.30 0.27
reportsomething over the longer reporting Bihar 0.15 0.17 0.17 0.27 0.30 0.30
period, the bottom tail of the consumption Gujarat 0.14 0.14 0.14 0.25 0.25 0.26
distributionis pulled up, and both inequal- Haryana 0.13 0.14 0.15 0.24 0.27 0.28
HimachalPradesh 0.38 0.16 0.42 0.37 0.29 0.40
ity and poverty are reduced. Whether 365- Jammuand Kashmir 0.13 0.09 0.12 0.24 0.16 0.21
days are a better or worse reporting period Karnataka 0.16 0.18 0.17 0.31 0.32 0.34
than 30-days could be argued either way, Kerala 0.20 0.17 0.22 0.31 0.32 0.37
but the main point here is that the 55th and MadhyaPradesh 0.18 0.17 0.18 0.29 0.29 0.33
50th Rounds are not comparable, and that Maharashtra 0.21 0.21 0.21 0.40 0.36 0.40
Orissa 0.15 0.14 0.16 0.29 0.26 0.29
the former artificially shows too little 0.13 0.14 0.14 0.23 0.25 0.25
Punjab
inequality compared with the latter. Once Rajasthan 0.14 0.13 0.14 0.25 0.23 0.26
the corrections are made, we see that, in TamilNadu 0.21 0.27 0.20 0.39 0.34 0.35
UttarPradesh 0.17 0.18 0.19 0.31 0.31 0.34
addition to increasing inequality between 0.34 0.31 0.35
West Bengal 0.19 0.20 0.19
states, there has been a marked increase Delhi 0.29 0.21 0.30 0.43 0.39 0.46
in consumption inequality within the All-IndiaUrban 0.19 0.20 0.21 0.34 0.34 0.37
urban sector of nearly all states. All-India 0.17 0.18 0.19 0.29 0.29 0.32
Two further pieces of evidence are Note:a AMis the arithmeticmean and GMis the geometricmean:the differencein theirlogarithmsis the
worth mentioning in this context. First, mean relativedeviation,a measure of inequality.

Figure 6: Food Intake for Different Per Capita Income Groups, as a Proportion seventies and eighties, when poverty was
(Per Cent) of Average Intake (1996-97)
.
certainly declining. Hanchate and Dyson's
250.0'-----------.......' --- - .- -.-- (2000) recent comparison of rural food
consumption patterns in 1973-74 and
1993-94 sheds some useful light on this
200.0-
matter. As the authors show, during this
period per capita cereal consumption in
150.0 rural areas declined quite sharply on av-
erage (from 15.8 to 13.6 kgs per person
0.0. per month), but rose among the poorest
households. The decline in the average is
driven by reduced consumption among the
higher expenditure groups.35
The average decline is unlikely to be
driven by changes in relative prices;
indeed, there has been little change in food
Cereals and millets Fats and oils and Milkand milk prices, relative to other prices, in the inter-
Sugar Jaggery products
Per capita income groups (Rs/cap/month) .' less than 90 [3 90-150 vening period. Instead, this pattern ap-
pears to reflect a substitution away from
E[ 150-300 LC 300-600 U morethan 600
Source: CalculatedfromNationalNutritionMonitoringBureau(1999), Table 6.9. The data relateto rural
cereals to other food items as incomes rise
areas of eight sample states. (at least beyond a certain threshold). The
consumption of 'superior' food items such
disparities produces very sharp contrasts real agriculturalwages slowed down con- as vegetables, milk, fruit, fish and meat did
in APCE growth between the ruralsectors siderably. And cereal production barely rise quite sharply over the same period,
of the slow-growing states and the urban kept pace with population growth.33 across all expenditure groups. Seen in this
sectorsof the fast-growing states (Table 3). The virtualstagnationof percapitacereal light, the decline of average cereal con-
This is furthercompounded by the accen- production in the nineties has been accom- sumption may not be a matter of concern
tuation of intra-urbaninequality, which is paniedby a gradualswitch from net imports per se. Indeed, average cereal consump-
itself quite substantial, bearing in mind to net exports, and also by a massive tion is inversely related to per capita in-
that the change is measured over a short accumulationof public stocks. Correspond- come across countries (e g, it is lower in
period of six years (Table 5). ingly, there has been no increase in esti- China than in India, and even lower in the
It might be argued that a temporary matedpercapita 'net availability' of cereals United States), and the same applies across
increase in economic inequality is to be (Table 6). If anything, net availability states within India (e g, cereal consump-
expected in a liberalising economy, and declined a little, from a peak of about 450 tion is higher in Bihar or Orissa than in
that this trend is likely to be short-lived. grams per person per day in 1990 to 420 Punjab or Haryana).
Proponents of the 'Kuznets curve' may grams or so at the end of the nineties. This Food intake data collected by the Na-
even expect it to be reversed in due course. is consistent with independent evidence, tional Nutrition Monitoring Bureau
However, China's experience of sharpand from National Sample Survey data, of a (NNMB) shed further light on this issue.
sustained increase in economic inequality decline in per capita cereal consumption Aside from detailed information on food
over a period of more than 20 years, after in the nineties. Between 1993-94 and 1999- intake, the NNMB surveys include rough
market-oriented economic reforms were 2000, for instance, average cereal con- estimates of household incomes. These are
initiated in the late 1970s, does not inspire sumption per capita declined from 13.5 kg used in Figure 6 to display the relation
much confidence in this prognosis.32 It is, per month to 12.7 kg per month in rural between per-capita income and food in-
in fact, an important pointer to the pos- areas, and from 10.6 to 10.4 kg per month take, for different types of food. The sub-
sibility of furtheraccentuationof economic in urban areas.34 This comparison is stitution from cereals towards other food
disparities in India in the near future. based on the 'uncorrected' 55th Round items with risingper-capitaincome emerges
data, and the 'true' decline may be larger quite clearly.36 This pattern,if confirmed,
IV still, given the changes in questionnaire would fit quite well with the dataon change
Qualifications andConcerns design (Section I.1). over time.37It also implies that the decline
The reduction of cereal consumption in of average cereal consumption in the nine-
IV.1Food Consumption the nineties may seem inconsistent with ties is not inconsistent with our earlier
the notion thatpoverty has declined during findings on poverty decline.38
Therehave been majorchanges in India's the same period. Indeed, this pattern has
food economy in the nineties. The eighties been widely invoked as evidence of 'im- IV.2 Localised Impoverishment
were a period of healthy growth in agri- poverishment' in the nineties. If cereal con- and Hidden Costs
cultural output, food production, and real sumption is declining, how can poverty
agricultural wages. During the nineties, be declining? The overall decline of poverty in the
however, productivity increases slowed It is worth noting, however, that the nineties does not rule out the possibility
down in many states. The quantity index decline of cereal consumption is not new. of impoverishment among specific regions
of agriculturalproduction grew at a lame A similar decline took place (according to or social groups. Thatpossibility, of course,
2 per cent per year or so. The growth of National Sample Survey data) during the is not new, but it is worth asking whether

Figure 7:Progress of Selected Social Indicators in the 1980s and 1990s are poorlycaptured,if at all, in standard
(Per Cent Per Year)
povertyindexes or for that matterin the
other social indicatorsexaminedin this
paper. Examples of such costs include
irregularschool attendance,the spreadof
HIV/AIDS,the disruptionof family life,
and rising urbancongestion.42Similarly,
4. involuntarydisplacementof personsaf-
' " e"- ....' , ' i " fectedby largedevelopmentprojectssuch
': ' ,< - 5
6.0 ....~'. . ..... ;."' .~'.';~,~-- -,*:.-*
? .**;*"*..28 "-'. as damsandminestendsto haveenormous
human costs. These, again, are largely
hiddenfromview in income-basedanaly-
ses of poverty. In fact, the incomes of
-' A..
:40
I
XL:-? I
tl.'
I : d : .-' - -0 : displaced persons often rise (with 'cash
**'
1
""^ -
--
' * * '"
:
.
* 1 *
- *:.-'' :
'
compensation') evenastheirlivesarebeing
M
, *~ _ ^ ,.'''..,,,~% ' . ,, . *'''. *-*'
;'''i
*
'00 .. .-',..' ^ - '.*:..,-.':..*."*..**.?-.-.-....

:'.- -
.^ :.* shattered.43The'informalisation' of labour
marketsis anotherexampleof economic
l . , . . , * *.. . . . . : . , . ; r. -,2. '-
-
.' changewith substantialhiddencosts (e g,
agticultural wages mortara fertilityrate longer workinghours,higherinsecurity,
ircymat. lower status,anddeterioratingworkcon-
Sowwi Drae andSen (2002),chapt': : ditions).44These issues are not new, but
it is importantto acknowledgethe possi-
its scope has expandedduring the last considerable disruption of earlier liveli- bility that the hiddencosts of economic
decade.As theeconomygivesgreaterroom hood patterns. Examples include a deep growth have intensifiedin the nineties.
to marketforces,uncertaintyandinequal- recession in the powerloom sector, a se- This acknowledgementhelps to recon-
ity often increase, possibly leading to rious crisis in the edible oil industry after cile the survey-basedevidence reviewed
enhanced economicinsecurityamongthose importtariffs were slashed, periodic waves earlierwith widespreadmediareports,in
who are not in a positionto benefitfrom of bankruptcyamong cotton growers, the recentyears, of sectoraleconomiccrises
the new opportunities,or whose liveli- displacement of traditional fishing by and localised impoverishment.45This
hoodsarethreatenedby thechangesin the commercial shrimp farms, and a number issue calls for furtherscrutiny,based on
economy.The increaseof economic in- of sectoral crises associated with the abrupt more focused analysis of survey data as
equality in the nineties, noted earlier, lifting of quantitative restrictions on im- well as on micro-studies.
suggeststhattendenciesof this kind may ports in mid-2001.40 The destruction of
well be at work in Indiatoday. Adverse local environmental resources is another IV.3 The 'Thin' Rounds: An
trendsinlivingstandards couldtakeseveral common cause of disrupted livelihoods in Unresolved Puzzle?
distinctforms,including:(1) impoverish- many areas.
ment among specific regions or social A relatedissue is thepossibility of 'hidden We have so farsaidverylittleaboutthe
groups, (2) heightened uncertainty in hardships' associated with recent patterns 'thin' rounds,and the povertyestimates
general,and(3) growing'hiddencosts' of of economic development. To illustrate, that can be calculatedfrom them. Yet
economicdevelopment. there is much evidence that, in many of Figure1 showsthattherecentthinrounds,
In connectionwith the first point, we the poorer regions of India, furtherimpov- from the 51st throughthe 54th Round,
havealreadynotedthatsomeof thepoorer erishmenthas been avoided mainly through generatepovertyestimatesthatarehardto
states,notablyOrissaandAssam,havenot seasonal labour migration.41 The latter reconcile with the quinquennial'thick'
faredwell at all in the nineties.It is quite often entails significant social costs that rounds.If we were to connect up these
possible that the poorer regions within
thesestateshavedone even worse, to the Table 6: Cereal Availability in the Nineties
point of absolute impoverishmentfor (Gramsper person per day)
substantialsections of the population.In Net Production Net Imports Net Change in 'NetAvailability'
thecase of Orissa,thereis some indepen- PublicStocks (1+2-3)
dentevidenceof localisedimpoverishment 1985-89 422.7 2.0 -5.3 430.1
in thepoorerdistricts,due interalia to the 1990 456.9 0.3 5.0 452.1
destruction of thelocalenvironmental base 1992 1991 447.9 -1.4 0.4 446.1
446.8 1.3 4.2 443.8
and to the dismal failure of state-spon- 1993 446.4 2.5 16.6 432.3
sored developmentprogrammes[Dreze 1994 456.3 0.2 16.6 439.9
2001].39 1995 448.6 -5.9 -2.3 445.1
1996 451.7 -6.9 -11.7 456.4
Similarly,the overall improvementof 1997 445.5 -6.7 -4.2 443.0
living standardsmay hide instances of 1998 455.3 -4.7 11.0 439.6
impoverishment among specific occupa- 1999 456.3 -5.4 25.3 425.6
tiongroups.Theninetieshavebeenaperiod 2000 452.7 -5.2 30.7 416.8
of rapidstructuralchange in the Indian Note: Allfigures(exceptfirstrow)arethree-yearaverages centredat the yearspecifiedinthe firstcolumn.
economy, leading in some cases to Source: CalculatedfromGovernmentof India(2002), p-S-21.

points with the official HCR estimates, we not bias, and in any case, the thin round is very hard to reconcile with other evi-
would get a series in which poverty rose sample sizes are perfectly adequate to dence. Having said this, we have not been
between 1993-94 and 1994-95, fell from generate accurate estimates for the all- able to identify any 'smoking gun' that
1994-95 to the end of 1997, rose very India HCRs. The discrepancies in Figure 1 would point to a specific problem with any
sharply in the first half of 1998, and then cannot be explained by inadequate of these rounds and explain their appar-
fell with extraordinary rapidity in 1999- sample sizes. ently anomalous poverty estimates. Until
2000. As we have seen, the official esti- Thereareotherdifferences between thick thatpuzzle is resolved, we see the evidence
mate for 1999-2000 is too low, and the last and thin rounds. For example, the sam- from the thin rounds as casting a shadow
thin round, the 54th Round, ran for only pling frame for the 51st, 53rd, and 54th of doubt over the interpretation of the
the first six months of 1998, and may Rounds was not the census of population, poverty estimates presented earlier in this
therefore not be fully comparable with but the 'economic' census. In the popu- paper. Perhaps the thin rounds in the next
other rounds. Even so, and with due al- lation census, each household is asked if five years will offer some useful clues.
lowance for corrections, it is very hard to it has a family business or enterprise, and
integrate the poverty estimates based on only such households are included in the V
thethinroundswith the picturethatemerges first-stage sampling from the economic Indexes
from the thick roundsas well as from other census when 'first-stage units' are drawn
Beyond Poverty
sources surveyed in this paper. with probability proportionalto size. This The decline of poverty in the nineties,
The story is further complicated by the means that a village with few or no such as captured in the indicators examined so
fact that these thin rounds were run in two households has only a small or no chance far, can be seen as an example of continued
versions, one of which resembled the of being selected as a first-stage unit. Even progress during that period. Whether the
standardquestionnaireup to and including so, when the team reaches the village, all rate of progress has been faster or slower
the 50th Round, and one of which - the households are listed and have a chance than in the eighties is difficult to say, and
experimentalquestionnaire- had different of being in the sample, so it is unclear that the answer is likely to depend on how the
reporting periods for different goods. this choice of frame makes much differ- rate of progress is measured. There is, at
Headcount ratios based on the experimen- ence. Indeed, comparison of various socio- any rate, no obvious pattern of "accelera-
tal questionnaire (not shown in Figure 1) economic indicators (e g, literacy rates, tion" or 'slowdown' in this respect.
are lower than those from the standard years of schooling, landholding, or family It is important to supplement the evi-
questionnaire, because the experimental size) from the surveys suggests no obvious dence reviewed so far, which essentially
questionnaire generated higher reports of breaks between the 51st and 53rd Rounds relates to purchasing power, With other
percapitaexpenditure.However, they also on the one hand, and the 52nd Round indicators of well-being relating, for in-
show rising HCRs from the 52nd through (which used the population census) on the stance to educational achievements, life
the 54th Rounds, and the increase contin- other. Conversations with NSS and Plan- expectancy, nutritionallevels, crime rates,
ues into the 55th Round if we use com- ning Commission staff sometimes suggest and various aspects of social inequality.
parablereportingperiods from that round. that there may be other (non-documented) This broader perspective reveals that so-
Based on the experimental questionnaire, differences in the sampling structureof the cial progress in the nineties has followed
a case could be made that the all-India thin rounds. Certainly, a tabulation of the very diverse patterns, ranging from accel-
HCR has been rising since 1995-96 [Sen population sizes of the first-stage units erated progress in some fields to slow-
2000]. As we have seen, there are good shows that the 52nd Round contained down and even regression in other re-
grounds for distrusting the experimental relatively few large units compared with spects. The point is illustrated in Figure 7,
questionnaire in the 55th Round, because the 51st, 53rd, 54th, and 55th rounds; this where simple measures of the progress of
of thejuxtaposition of the seven-day recall is a different issue from the use of the different social indicators in the nineties
and 30-day recall data for food-pan and economic rather than population census are compared with the corresponding
tobacco. Quite likely, the 'reconciliation (both the 52nd and 55th Rounds use the achievements in the eighties.
effect' (see Section I.1) pulled down the latter), and the finding suggests that the Elementary-educationprovides an inter-
estimates of per capita expenditure from first-stage units in the 52nd Round were esting example of accelerated progress in
the experimentalquestionnaire, thus exag- selected differently from other rounds in the nineties.46 This trend is evident not
gerating poverty by this count. Even so, some way that is not documented. More- only from census data on literacy rates, but
if poverty were genuinely falling, there is over, the measurement of consumption is also from National Family Health Survey
no obvious explanation why the experi- not the main purpose of any of these thin data on school participation. To illustrate,
mental questionnaire should show a rise rounds, all of which have some other school participationamong girls aged 6-14
in poverty from 1995 through 1998. objective, so it is possible that consump- jumped from 59 per cent to 74 per cent
The Planning Commission has never tion is not so fully or carefully collected between 1992-93 and 1998-99.47 The re-
endorsed poverty counts from the thin as in the quinquennial rounds. gional patterns are also instructive. It is
rounds. In part, this has been because of In short, there aregrounds for scepticism particularlyinteresting to note evidence of
the smaller sample sizes. The Planning about the validity of the thin rounds for rapid progress in Madhya Pradesh and
Commission needs estimates of HCRs, not poverty estimation purposes, and this is all Rajasthan,demarcating them clearly from
just for all-India, but for individual states, the more so if we rememberthataside from Bihar and Uttar Pradesh, the other two
and the thin rounds are not large enough to indicating no poverty decline in the late members of the so-called BIMARU set.48
supportaccurateestimates for some of the nineties, the thin rounds also suggest that There is an important pointer here to the
smaller (of the major) states. But in- average per capita expenditure was stag- relation between public action and social
adequate sample size generates variance, nating during that period - something that achievements. Indeed, Madhya Pradesh

and Rajasthan are two states where there Delhi) thatarerelatively well-off economi- based poverty indexes tends to foster a
have been many interesting initiatives in cally, and have also experienced compara- simplistic view of the relation between
the field of elementary education in the tively high rates of growth of per capita economic growth and poverty decline.
nineties (on the partof government as well expenditure in the nineties (Table 3).51 Third, it is also interesting to re-examine
as non-government institutions), in con- A detailed assessment of the progress of the issue of trends in inequality, in the light
trast with Bihar and Uttar Pradesh where development indicators in the nineties is of this broader perspective. As discussed
schooling matters continue to be highly beyond the scope of this paper. However, in Section III, there is much evidence of
neglected. The fact the literacy rates and a few general observations can be made rising economic inequality in the nineties,
school participation have surged in the on the basis of these illustrations. First, as in the form of a widening rural-urbangap,
more 'active' states is an encouraging noted already, poverty is not unidimen- enhanced inter-state disparities, and also
indication of the possibility of effective sional. The poverty indexes used in the growing inequality within urban areas in
public intervention in this field. first partof this paper are useful indicators most states. What about other types of
Turning to instances of 'slowdown', we of inadequate purchasing power, but on social inequality, involving other dimen-
have already referred to the slackening of their own do not do justice to the range sions of well-being (e g, educational levels
the growth rate of real agricultural wages of deprivations we ought to be concerned or life expectancy) and other bases of
in the nineties. Anotherimportantexample with. Following on this, it is importantto disadvantage (e g, gender or caste)? The
is the slowdown of infant mortality de- acknowledge thatrecent progress in elimi- decline of the female-male ratio among
cline. During the eighties, India achieved nating poverty and deprivation has been children illustrates the fact that the pheno-
a reduction of 30 per cent in the infant quite uneven in different fields. The debate menon of rising inequality in the nineties
mortality rate- from 114 deaths per 1,000 on 'poverty in the nineties' has often is not confined to standard economic
live births in 1980 to 80 per 1,000 in 1990. overlooked this basic point. inequalities: 'natality inequality' between
During the nineties, however, the infant Second, this recognition is also impor- males and females is also rising.53But this
mortality rate declined by only 12.5 per tant in assessing the relation between is not to say thatinequality has risen across
cent - from 80 to 70.49 In fact, in the poverty decline and economic growth. As the board. Even within the field of gender
second half of the nineties, India's infant noted earlier, the decline of poverty in the inequality, there are changes in the other
mortality rate has remained virtually un- nineties, as captured by conventional in- direction, such as the emergence of a
changed. In some states,notably Rajasthan, dexes such as the headcount ratio or the substantial gender gap in life expectancy
the infant mortality rate has stagnated for poverty-gap index, has been overwhelm- in favour of women, overturning India's
as long as 10 years. These worrying trends ingly driven by the growth of average per long history of female disadvantage in this
have received astonishingly little attention capita expenditure. From this it may seem respect. Similarly, it is interesting to note
in policy debates, and even in the debate that the reduction of poverty is mainly a that while economic disparities between
on 'poverty in the nineties'. question of economic growth. However, rural and urban areas have sharply risen
Finally, there have also been some areas there is an element of circularity in this in the nineties, there are trends in the
of 'regression' in the nineties. The in- argument: if poverty is defined as lack of opposite direction as well. The rural-urban
crease of economic inequality, discussed income, it is not surprisingthat the growth gap in life expectancy, for instance, has
earlier, can be seen in those terms. Given of income plays a key role in reducing it. declined from 10 years or so in the late
the adverse social consequences of eco- When the multidimensional nature of 1970s to seven years or so today, and rural-
nomic inequality(rangingfromelitist biases poverty is acknowledged, this relation urban differentials in school participation
in public policy to the reinforcement of appears in a different light. To illustrate, have also narrowed.54 Here again, the
other types of inequality), this accentua- consider child mortality as an aspect of the picture is more diverse (and more inter-
tion of economic disparities is not a trivial deprivationsassociatedwith poverty.There esting) than it appears on the basis of
matter. Another example of adverse is, of course, a significant (negative) re- purchasing-power indicators alone.
development in the nineties is the decline lation between child mortality and pur- Fourth,the broadapproachexplored here
in the female-male ratio among children, chasing power. But child mortality is also calls for a correspondingly broad reading
from 945 girls per 1,000 boys (in the 0-6 strongly influenced by other factors such of the causal influences underlying the
age group) in 1991 to 927 girls per 1,000 as educational levels, fertility rates, public identified changes. In the debate on 'pov-
boys in 2001.50 This decline appears to be health provisions (including clean water erty in the nineties', there has been a
driven by the spread of prenatal sex- and vaccinations), and various aspects of tendency not only to view development
determination technology and sex- gender relations. Looking at inter-state trends in unidimensional terms, but also
selective abortion, but this does not mean contrasts in India, the correlation between to attribute these trends in a somewhat
that it is a 'technological' phenomenon, child mortality and average per capita mechanical manner to the economic re-
unrelated to other recent economic and expenditure (or even poverty indexes) is forms initiated around 1991. At one end
social trends. Economic growth, in parti- actually quite weak. Otherfactors, particu- of the spectrum, it has been claimed that
cular, may facilitate the spread of sex- larly female literacy, are often more im- the last decade has been a period of un-
selective abortion, by making the use of portant.52Similar comments apply in the precedented improvement in living stan-
sex-determinationtechnology more afford- context of elementary education: the nine- dards, thanks to liberalisation.55 At the
able. In this connection, it is worth noting ties have demonstrated the possibility of other end, the nineties have been described
thatthe largest declines of the female-male rapid progress in this field through public as a period of widespread 'impoverish-
ratio among children between 1991 and intervention, with or without rapid eco- ment', attributedto liberalisation.56Clearly,
2001 occurred in five states (Gujarat, nomic growth. In short, the standardfocus these readings fail to do justice to the
Haryana, Himachal Pradesh, Punjab and on headcountratios and otherexpenditure- diversity of recent trends. But in addition,

they ignorethe diversityof causal influ- on the basis of headcount ratios. For the Sixth, the case for going beyond expen-
ences thathavea bearingon thesetrends. purpose of the poverty comparisons exam- diture-based indicators applies also to the
The acceleratedprogressof elementary ined in this paper,the headcountratio turns assessment of inequality. While expendi-
educationin thenineties,for instance,has out to be no less informative than the ture-based data suggest rising disparities
littleto dowithliberalisation,
andthesame poverty-gap index. Yet it was important in the nineties, the same need not apply
appliesto the slowdownof infantmortal- to calculate the PGIs, if only to discover to other social indicators. For instance,
ity decline,not to speakof the declineof that this refinement does not, after all, while economic disparities between rural
the female-male ratio among children. make much difference in this particular and urban areas have increased in the
Muchelsethanliberalisation hashappened context. nineties, there has been some narrowing
in the nineties,and while issues of eco- Third, growth patternsin the nineties are of the rural-urban gap in terms of life
nomic reform are of course extremely characterisedby majorregionalimbalances. expectancy and school participation.
important,so are other aspects of eco- Broadly speaking, the western and south- Finally, we have argued against reading
nomic and social policy. ern states (AndhraPradeshexcluded) have these trends simply as evidence of the
tended to do comparatively well. The low impact (positive or negative) of 'liberalis-
VI growth states, for their part, form a large ation'. Forone thing, the impactof liberalis-
ConcludingRemarks contiguous region in the north and east.
This is a matter of concern, since the
ation is a 'counterfactual' question, and
much depends on how the alternatives are
A numberof usefullessonsemergefrom northern and eastern regions were poorer specified. For another, much else has hap-
this reexaminationof the evidence on to start with. Indeed, National Sample pened in the nineties, other than liberalis-
povertyandinequalityinthenineties.First, Survey data suggest a strong pattern of ation. The evidence we have reviewed is
thereis consistentevidenceof continuing inter-regional 'divergence' in average per of much interest in its own right, indepen-
povertydeclinein the nineties,in termsof capita expenditure (APCE): states that dently of the liberalisation debate. Much
the 'headcountratio'. The extent of the started off with higher APCE levels also work remains to be done in terms of identi-
decline, however, remains somewhat had higher growth rates of APCE between fying the causal relations underlying the
uncertainat this time. Given the method- 1993-94 and 1999-2000. In some of the trends we have identified. 313
ologicalchangesthattook place between poorer states, notably Assam and Orissa,
the 50th and55th Roundsof the National there has been virtually zero growth of Notes
SampleSurvey,theofficialfigures(imply- average per capita expenditure (and very
ing a decline from 36 per cent to 26 per little reduction, if any, in rural poverty) [We are grateful to SureshTendulkarfor helpful
cent in the all-Indiaheadcountratio be- between 1993-94 and 1999-2000. These comments]
tween1993-94and1999-2000)are,strictly regional patterns are at least broadly con- 1 See Datt(1999a), Gupta(1999),Bhalla(2000a,
speaking, invalid. We have discussed sistent with independent estimates of the 2000b), Deaton and Tarozzi (2000), Dreze
alternativeestimates,basedoncomparable (2000), Lal, Mohan and Natarajan(2001),
growth rates of state domestic product
datafromthetwo surveys.As it turnsout, Nagaraj(2000), Ravallion(2000), Sen (2000),
(SDP). Sundaramand Tendulkar(2000, 2001, 2002),
these adjustedestimates suggest that a Fourth, the intensification of regional Visaria (2000), Sundaram (2001a, 2001b,
largepartof thepovertydeclineassociated disparities is only one aspect of a broader 200 c), Chandrasekharand Ghosh (2002),
with official figuresis 'real', ratherthan pattern of increasing economic inequality Datt and Ravallion (2002), among others.
drivenby methodologicalchanges.While in the nineties. Two otheraspects are rising 2 On the first position, see, e g, Bhalla (2000a),
furthercorroboration andinvestigationof rural-urbandisparities in per capita expen- Bhagwati (2001), Das (2000). On the other
side, see Mehta (2001), Sainath (2001a,
the adjustmentprocedureis required,the diture, and rising inequality of per capita 2001b), Shiva (2001a), among others.
resultshave been supportedby one inde- expenditure within urban areas in most 3 Theseestimatesbuildonearlierworkby Deaton
pendentstudyusing an entirelydifferent states. Further, the real wages of agricul- and Tarozzi (2000), Deaton (2001a, 2001b)
and Tarozzi (2001).
methodology[Sundaramand Tendulkar turallabourershave increased more slowly
4 In the 50th Round,the correlationbetween the
2002]. Further,the adjustedfigures fit than per capita GDP, and conversely with
logarithm of total household per capita
reasonablywellwithrelatedevidencefrom public sector employees, suggesting some expenditure and the logarithmof per capita
the nationalaccountsstatistics,the em- intensification of economic inequality expenditureon this subset of commodities is
ployment-unemployment surveys,anddata between occupation groups. 0.79 and 0.86 in the ruraland urbansectors,
on agriculturalwages. Fifth, we have argued for assessing respectively.
5 Moreprecisely,andsomewhatless restrictively,
Second,we havediscussedsomeimpor- changes in living standards in a broader we require that the probability of being
tantlimitationsof the headcountratioas perspective, going beyond the standard poor, given the amount of a household's
an indexof poverty(even withinthe stan- focus on expenditure-based indicators. In expenditure on these intermediate goods,
dard expenditure-basedapproach),and that broader perspective, a more diverse remainsthe same in the 55th Round as it was
in the 50th. We requirethis on a state by state
arguedfor wider adoptionof alternative picture emerges, with areas of accelerated basis, one sector at a time, which allows the
povertyindexes such as the poverty-gap progress in the nineties as well as slow- conditionalprobabilityto vary by stateandby
index. The main argument for using down in other fields. For instance, there sector.
headcountratios is that they have good is much evidence of rapid progress in the 6 In DeatonandTarozzi(2000), it is shown that,
'communicationvalue', in so far as they field of elementary education, but the rate between the 43rd and 50th Rounds, the
are relativelyeasy to understandand in- of decline of infant mortality has slowed component of the CPIAL for the uncovered
items grew somewhat less rapidly than the
terpret.However,this transparencyis to down. These and related trends deserve
some extentdeceptive,andmuchcaution component for the covered items. In conse-
greaterattentionthanthey have received so quence, if we were to supplement our price
is requiredin interpretingpovertytrends farin the debateon 'povertyin the nineties'. indexes for uncovereditems from the CPIAL,

the estimated rate of increase of consumer 15 See Governmentof India (1993b), p 16. ratiowith respectto the logarithmof meanper
priceswouldcome down,andcorrespondingly, 16 In Table 2 and elsewhere in this paper, the capita expenditure.
therewould be a faster decline in the poverty terms 'Bihar', 'Madhya Pradesh' and 'Uttar 30 Calculatedfrom Governmentof India(2002),
indexes, at least for the periodfalling between Pradesh' refer to these states as they existed p S-5 1.Therehavebeenfurthermajorincreases
these two rounds (i e, 1987-88 to 1993-94). priorto theformationofJharkhand,Chattisgarh in public sector salariesafter 1999-2000, with
7 The Tornqvist price index is a weighted and Uttaranchalin late 2000. the gradual implementation of the recom-
geometric index with weights that are the 17 In the first sub-period,the estimates suggest mendationsof the Fifth Pay Commission by
averageof the expenditureshares in the base some increase in poverty in rural Haryana many state governments.
and comparison periods. It is a superlative and Himachal Pradesh, and also in urban 31 See DrezeandSen (2002), StatisticalAppendix,
index in the sense of Diewert (1976). Himachal Pradesh,Punjaband Delhi. These Table A.6; also Datt (1999a, 1999b).
8 For further details, see EPW Research patterns,however, should be interpretedwith 32 On rising income inequality in China in the
Foundation (1993). On the conceptual and caution, given the relatively small sample post-reform period, see Bramall and Jones
practicalproblemsinvolvedindefining'calorie sizes for these states and the possibility of (1993), Griffinand Zhao Renwei (1993), Yao
norms', see Dasgupta and Ray (1990) and transient fluctuations in poverty levels in Shujie(1999), KhanandRiskin(2001), among
Osmani(1990), and the literaturecited there. specific years. others.
Note also that, if the calorie norms were to 18 Here and elsewhere, it is useful to remember 33 On these andrelatedtrends,see Dreze and Sen
be reappliedtoday, they would not generate that the period between 1993-94 and 1999- (2002), chapter9. On the growthof foodgrain
the samepovertylines. Updatingcalorie-norm 2000 was one of 'peak' economic growth production,see Governmentof India (2002),
based poverty lines for inflation does not for the Indian economy, with per capita pp S-21 and S-22.
preserve their calorie-norm status. GDP growing at a healthy 4.4 per cent per 34 See Shariff and Mallick (1999), Table 5, and
9 Similar issues arise, of course, in the context year. NationalSample Survey Organisation(2001),
of inter-statecomparisons,especially between 19 While the relative growth rates of APCE in pp A-101 and A-134.
states (e g, Kerala and Uttar Pradesh) with different states are consistent with the 35 Forsimilarobservationsbasedon acomparison
radically different consumption patternsand corresponding relative growth rates of per of 1.972-73and 1993-94 NSS data, see Rao
social environments. In both cases, com- capitaSDP, thelevels of percapitaSDP growth (2000).
parisons,of living standardscall for supple- tend to be higherthanthose of APCE growth. 36 Forfurtherevidence,see also NationalInstitute
mentingexpendituredata with other types of We shall returnto this issue in the next section, of Nutrition (1997). This pattern,sometimes
information, relating for instance to public with reference to the all-India figures. known as the 'nutritiontransition',is familiar
amenities, health achievements, educational 20 On the growing divergenceof per capita SDP to nutritionists[Drewnoski 1999]. It is worth
levels, etc. in the nineties, see also Ahluwalia(2000) and noting that its implicationsfor health are not
10 The official line is actually 115.20. We use Dreze and Sen (2002). uncontroversial;some nutritionexperts have
115.70 because this is the figure yielded by 21 For inequality indexes, see Table 5 in apparently"pointed to the beneficial health
theofficial methodologywhen thecalculations Section III. effectsof direpoverty,poordiets,andstrenuous
are based on the unit recorddata, as opposed 22 See particularlyBhalla (2000a), Kulsheshtra manual labour",presumablyreferringto the
to the interpolations used by the Planning and Kar(2002), Ravallion(2000), Sen (2000), benefits of a low fat, low sugar,and high fibre
Commission. See notes to Table 2. Sundaramand Tendulkar (2001). diet, ratherthan low quantities. (Drewnoski,
11 The case of Delhi is handled differently. 23 Calculated from Central Statistical Organis- 1999, p 195).
Because there are few sample households in ation (2001), p xxxii. 37 Unlike NNMB data, National Sample Survey
ruralDelhi, it is not advisableto use the price 24 In nominalterms,between 1993-94 and 1999- datasuggestthatpercapitacerealconsumption
indexfor ruralDelhi as partof the calculations. 2000, consumerexpenditurehas been growing rises monotonically with per capita expen-
The povertyline for urbanDelhi is calculated at about 11.5 per cent per year according to diture. The contradictionbetween nutritional
from the all-India urban poverty line by CSO data, and 10 per cent per year according food intakeand,expendituresurveys is neither
multiplying it by the price index for urban to our NSS-based estimates. Both the CSO's uncommon nor fully understood; for two
Delhi relative to urban India. implicitpricedeflatorandourTornqvistindex differentinterpretationsof the Indiancase, see
12 Note thatthis is not the only way of using the have been growing at 8 per cent per year or Subramanian and Deaton (1996) and
indexes;another(butonly one other)possibility so during this period. Thus, differences in Subramanian(2001).
would be to updatethe poverty line for each price deflatorsdo not seem to help to resolve 38 Also worth noting in this context is tentative
sector of each state by its own inflation rate. the CSO-NSS discrepancy in this case, even evidence of recentimprovementin nutritional
Becausewe aredealingwith priceindexes, not though price-index differences may have indicators based on anthropometric
prices, the different alternatives will give played a role in enhancing that discrepancy measurements.Accordingto NNMB data,the
different answers. in earlier periods (see Sen, 2000). proportionof adults with a low "body mass
13 More precisely, the poverty-gapindex (PGI) 25 TheNSS surveys,fortheirpart,almostcertainly index" has declined in the nineties [see
calculates the total shortfall of consumption disproportionatelymiss wealthyhouseholdsat Vaidyanathan 2002]. The National Family
below the poverty line, per capita of the total the verytopof thedistribution,andas Banerjee HealthSurveysalso suggestthattheproportion
population,and expressed as a percentageof andPiketty(2001) have shown, therehas been of undernourished children has declined
the povertyline: PGI (l/z)[(X (z-y,)/n] where a marked rise in incomes among the very between 1992-93 and 1998-99 [see Inter-
z is the poverty line, n is the populationsize, highest earners. Even so, they show that the nationalInstituteforPopulationSciences 1995:
and yJis the consumptionlevel of the ith poor total amountof these earnings is not enough xxxiii, andInternationalInstituteforPopulation
person. to explainthe increasingdisparitybetweenthe Sciences, 2000: 267 and 443].
14 The poverty-gapindex, however, retainsone NSS and the CSO estimates of consumption 39 It should also be noted, however, that Orissa
limitation of the headcount ratio: it is not expenditure. was hit by a devastatingcyclone in October
sensitive to the distributionof per capita ex- 26 For details, see, e g, Acharya (1989). 1999, around the middle of the 55th Round
penditurebelow the poverty line. This limita- 27 See, e g, Jose (1988) andSarmah(2000). Note survey period. The 1999-2000 poverty
tion is overcome by higher-ordermembersof thatthe 'realwage' estimatesused hereignore estimates for Orissa are thereforelikely to be
the FGT class, such as the 'squaredpoverty- inter-statedifferences in price levels. somewhat 'above trend'.
gap index' (SPGI), and also by the Sen index. 28 See Dreze and Sen (2002), p 328; on the 40 For insightful case studies of localised
While we have calculatedthe SPGIestimates, slowdownof thegrowthrateof realagricultural economic crises in the nineties, see e g, Roy
we confine ourselves here to the poverty-gap wages in the nineties (compared with the (1999), Breman (2001a), Krishna (2001),
index, for two reasons. First, it is easier to eighties), see also Sarmah (2000, 2001). Jhabvalaand Sinha(2002), Samal (2002), and
interpret.Second, SPGIs are highly sensitive 29 Note that these derivativesare not elasticities Dabir-Alai (2002).
to measurementerrorsat the bottomof the per in the usual sense, and are not the same as 41 See Rodgers and Rodgers (2000), Rogaly et
capita expenditurescale, and their reliability the elasticities sometimes quoted, which are al. (2001), Sharma(2001), Institutefor Human
calls for furtherscrutiny. thederivativesof thelogarithmof theheadcount Development (2002), among others.

42 On the other hand, labour migration can 56 To illustrate: "Both under the World Bank sitions', mimeo, World Bank, Washington,
also have positive roles, such as facilitating structuraladjustment,and from the finance DC.
the diffusion of knowledge [Maharatna2001] ministry- it's feet might be in India, but it's Datt, Gaurav, and Ravallion, Martin (2002): 'Is
and enabling the disadvantaged castes to head is in Washington- and then under the India's Economic Growth Leaving the Poor
"escape from the clutches of the prevailing World Trade Organisationobligations, we're Behind?', mimeo, World Bank, Washington,
caste discriminationin the village" [Sharma basically getting a fundamentaldestructionof DC.
2001: 18]. notions of the rights of citizens... Very vital Deaton, Angus (2001a): 'AdjustedIndianPoverty
43 For a telling case study of the human costs resourceswe need both for survival- drinking Estimatesfor 1999-2000', Princeton,Research
of involuntarydisplacement,see Bhatia(1997). water, all the resources people need for ProgrammeinDevelopmentStudies,processed.
44 On this, see particularly Breman (2001a, livelihoods - arejust disappearingso rapidly Available at http://www.wws.princeton.edu/
2001b). that life is becoming impossible...we really -rpds.
45 Seee g,Sainath(2001a,2001b,2001c,2001d), have a very, very major crisis of survival at - (2001b): 'ComputingPrices and Poverty Rates
Breman(2001b), Dreze (2001), Mehta(2001). hand...." [Shiva 2001b]. For further contri- in India, 1999-2000', Princeton, Research
We are not referringhere to media reportsof butions on both,sides of the debate, see the Programmein Development Studies, proces-
short-termhardshipassociatedwith the recent literaturecited in footnote 2. sed. Available at http://www.wws.princeton.
drought(in 2000 and 2001), but to stories of edu/-rpds.
sustained impoverishment.
46 For further discussion, see Dreze and Sen
References Deaton, Angus and Tarozzi, Alessandro (2000):
'Prices and Poverty in India', Princeton,Re-
(2002), chapter 5. Acharya, Sarthi (1989): 'AgriculturalWages in search Programmein Development Studies,
47 See Dreze and Sen (2002), p 148; also India: A Disaggregated Analysis', Indian processed. Available at http://www.wws.
InternationalInstitutefor PopulationSciences Journal of Agricultural Economics, 44. princeton.edu/-rpds.
(1995, 2000), for furtherdetails. Ahluwalia, Montek S (2000): 'Economic Diewert, W Erwin(1976): 'Exact and Superlative
48 To illustrate, looking at the percentage Performanceof Statesin Post-reformsPeriod', Index Numbers', Journal of Econometrics, 4.
reductionin illiteracybetween 1991 and2001, Economic and Political Weekly, May 6. Drewnowski,Adam (1999): 'Fat and Sugarin the
MadhyaPradeshandRajasthandid betterthan Banerjee,A and Piketty,T (2001): 'Are the Rich Global Diet: DietaryDiversity in the Nutrition
anyothermajorstateexceptHimachalPradesh; Growing Richer: Evidence from Indian Tax Transition' in Grew, R (ed), Food in Global
Uttar Pradesh is very close to the all-India Data',MIT,CambridgeMA,andCEPREMAP, History, Westview, Boulder.
average; and Bihar ranks second from the Paris, processed. Available at http:// Dreze, Jean (2000): 'Poverty:Beyond Headcount
bottom (calculatedfrom census data given in www.worldbank.org/indiapovertyworkshop. Ratios', The Hindu, September 9.
Governmentof India,200 lb). The comparison Bhagwati, Jagdish (2001): 'Growth,Poverty and - (2001): 'No More Lifelines: Political Economy
is particularly instructive because all four Reforms', Economic and Political Weekly, of Hunger in Orissa', The Times of India,
BIMARU states startedoff with similar (very March 10. September, 17.
low) levelsof literacyin 1991.Onthe 'schooling Bhalla, Surjit (2000a): 'Growth and Poverty in Dr6ze,Jean,andMurthi,Mamta(2001): 'Fertility,
revolution'in HimachalPradesh(aneven more India: Myth and Reality', mimeo, Oxus Education,and Development: Evidence from
remarkableexample of successful promotion Research and Investments, New Delhi. India', Population and DevelopmentReview,
of elementaryeducation), see PROBE Team - (2000b): 'FAQ's on Poverty in India', mimeo, 27.
(1999), chapter 9. Oxus Researchand Investments,New Delhi. Dreze, Jean and Sen, Amartya(eds) (I990): The
49 See Governmentof India (1999a), Table 1, Bhatia, Bela (1997): 'Forced Evictions in the Political Economy of Hunger, Clarendon,
andSampleRegistrationBulletin,April 2001; NarmadaValley' in Dreze, J P, Samson, M, Oxford.
also UNICEF (2001) and Mari Bhat (2002). and Singh, S (eds) (1997), The Dam and the - (2002): India: Development and Participation
The comparison with Bangladesh is also Nation: Displacementand Resettlementin the Oxford University Press, New Delhi.
instructive:Bangladesh'sinfantmortalityrate Narmada Valley, Oxford University Press, EPWResearchFoundation(1993): 'PovertyLevels
was muchhigherthanIndia's in 1990 (91 and New Delhi. in India: Norms, Estimates and Trends',
80 per 1,000, respectively),but by 1999 it had Bramall,Chris, and Jones, Marion(1993): 'Rural Economic and Political Weekly,August 21.
come down to 61 as against India's 71. See Income Inequality in China since 1978', Governmentof India( 1993a):Reportof the Expert
WorldDevelopment Indicators2001, pp 16-17. Journal of Peasant Studies, 21(1). Group on Estimation of Proportion and
50 On this and relatedissues, see Dreze and Sen Breman, Jan (2001a): 'An InformalisedLabour Numberof Poor, PlanningCommission, New
(2002), chapter 7. System: End of Labour Market Dualism', Delhi.
51 The relevantfemale-male ratios are given in Economic and Political Weekly, 35(52). - (1993b): Sample RegistrationSystem: Fertility
Governmentof India (2001b), pp 92-94. - (2001b): 'The State of Poverty in Urban India and MortalityIndicators 1990, Office of the
52 Across states, the correlationcoefficient bet- at the Beginning of the 21st Century', RegistrarGeneral, New Delhi.
weenchildmortalityandpercapitaexpenditure Sukhamoy ChakravartyMemorial Lecture, - (1999): Compendiumof India's Fertility and
is 0.4 (in absolute value), comparedwith 0.8 Delhi School of Economics, November r6. Mortality,Indicators 1971-1997, Office of the
for the correlationcoefficient between child CentralStatisticalOrganisation(2001): National RegistrarGeneral, New Delhi.
mortalityandfemaleliteracy.Aftercontrolling Accounts Statistics 2001, CSO, New Delhi. - (2001a): EconomicSurvey2000-2001, Ministry
for female literacy, per capita expenditure Chandrasekhar,C P, and Ghosh, J (2002): The of Finance, New Delhi.
bears no significant association with child Marketthat Failed: A Decade of Neoliberal - (2001b): 'ProvisionalPopulationTotals',Census
mortalityat the state level. For furtherdis- EconomicReformsin India, LeftwordBooks, of India2001, Series 1(India),Paper1 of 2001,
cussion, see Dreze and Sen (2002), pp 87-89. New Delhi. Office of the RegistrarGeneral, New Delhi.
On the determinantsof mortalityand fertility Dabir-Alai,Parviz(2002): 'The Socio-Economics -(2002): Economic Survey 2001-2002. Ministry
in India, see also Murthi, Guio and Dreze of Street Vending: A Case Study of of Finance, New Delhi.
(1995) and Dreze and Murthi (2001), and Vulnerabilityfor Delhi in the 1990s', mimeo, Griffin, K, and Zhao Renwei (eds) (1993): The
furtherstudies cited there. Richmond School of Business, London. Distributionof Income in China, Macmillan,
53 On 'natality inequality' and its significance, Das, Gurcharan(2000): India Unbound,Penguin, London.
see Sen (2001). New Delhi. Gupta,S P ( 1999): 'TricleDown TheoryRevisited:
54 See e g, Governmentof India (1999), p 16, Dasgupta, Partha, and Ray, Debraj (1990): The Role of Employmentand Poverty', V B
and Dreze and Sen (2002), p 148. 'Adapting to Undernourishment: The Singh Memorial Lecture, 41st Annual Con-
55 To illustrate:"Economic reforms initiated in Biological Evidence and Its Implications' in ference of the Indian Society of Labour
1991 have led to a radical transformationin Dreze and Sen (1990), vol i. Economics, November 18-20, Mumbai.
the well-beingof thebottomhalf of the popula- Datt,Gaurav(1999a): 'HasPovertyDeclinedsince Hanchate,A and Dyson, T (2000): 'Trendsin the
tion. Froman approximatelevel of 38 percent EconomicReforms?',Economicand Political Composition of Food Consumptionand their
in 1987, poverty level in India in 1998 was Weekly, December 11-17. Impact on Nutrition and Poverty in India',
close to 12 per cent" [Bhalla 2000b, p 7]. - (1999b): 'Povertyin India:TrendsandDecompo- mimeo, London School of Economics.

Institute for Human Development (2002): Rao,C H Hanumantha(2000): 'DecliningDemand Sundaram, K (2001a): 'Employment-
'Dynamics of Poverty, Employment and for Foodgrains in Rural India: Causes and Unemployment Situation in Nineties: Some
HumanDevelopmentin Bihar',projectreport; Implications',Economicand Political Weekly, Results from NSS 55th Round Survey',
to be published as a monograph. January22. Economic and Political Weekly,March, 17.
InternationalInstitute for Population Sciences Ravallion, Martin (2000): 'Should Poverty - (2001b): 'Employmentand Poverty in 1990s:
(1995): National Family Health Survey Measures be Anchored to the National- Further Results from NSS 55th Round
1992-93: India, IIPS, Mumbai. Accounts?', mimeo, ARQADE, Universityof Employment-UnemploymentSurvey, 1999-
- (2000): NationalFamilyHealth Survey1998-99 Toulouse. 2000', Economic and Political Weekly,
(NFHS-2): India, IIPS, Mumbai. Rodgers,GerryandRodgers,Janine(2000): 'Semi- August 11.
Jhabvala,R and Sinha, S (2002): 'Liberalisation FeudalismMeets the Market:A Reportfrom - (2001c): 'Poverty in India: Some Issues in
and the Woman Worker', Economic and Purnia',WorkingPaper6, Institutefor Human Measurementand Database', mimeo, Delhi
Political Weekly, May 25. Development, New Delhi. School of Economics, Delhi.
Jose, A V (1988): 'AgriculturalWages in India', Rogaly, B, Biswas, J, Coppard, D, Rafique, Sundaram,K and Tendulkar,S (2000): 'Poverty
Economic and Political Weekly,June, 25. A, Rana,K andSengupta,A (2001): 'Seasonal in India:An Assessment and Analysis', draft
Khan,A R and Riskin, C (2001): Inequalityand Migration,Social Change and Migrants'Rights: reportforthe Asia DevelopmentBank;mimeo,
Povertyin China in the Age of Globalisation, Lessons from West Bengal', Economic and Delhi School of Economics, Delhi.
Oxford University Press, New York. Political Weekly, December 8. - (2001): 'NAS-NSS Estimates of Private
Krishna,Sridhar(2001): 'Phasing Out of Import
Roy, Tirthankar(1999): 'Growth and Recession Consumptionfor Poverty Estimation:A Dis-
Licensing:Impacton Small-scale Industries', in Small-Scale Industry:A Study of Tamil aggregated Comparison for 1993-94',
Economic and Political Weekly,July 7. Nadu Powerlooms', Economic and Political Economic and Political Weekly,January 13.
Kulshreshtha,A C and Kar,A (2002): 'Estimates - (2002): 'Recent Debates on Data Base for
of Food Consumption Expenditure from Weekly,October 30.
Sainath, P (2001a): 'Will the Real National Measurement of Poverty in India', Delhi
Household Surveys and National Accounts',
Calamity Please Stand Up?', The Hindu, School of Economics, processed.Presentedat
paper presented at the joint Planning Com-
mission/World Bank workshop on poverty February25. joint GOI/World Bank poverty workshop,
- (2001b): 'It's the Policy, Stupid, Not Imple- Delhi, January 2002. Available at http://
measurement, Delhi, January. Available at
mentation' (2 parts), The Hindu, September. www.worldbank.org/indiapovertyworkshop.
http://www.worldbank.org/
-(2001c): 'Mass Media:DisconnectedfromMass Tarozzi, Alessandro (2001): 'Estimating Com-
indiapovertyworkshop.
Lal, D, Mohan, R and Natarajan, I (2001): Reality', The Hindu, September 2. parable Poverty Counts from Incomparable
'EconomicReformsand Poverty Alleviation: - (2001d): 'WhereStomachachesAre Terminal', Surveys:MeasuringPovertyinIndia',Research
A Tale of Two Surveys', Economic and The Hindu, April 29. Programmein DevelopmentStudies,Princeton
Political Weekly, March, 24. Samal, K C (2002): 'Shrimp Culture in Chilika University, processed. Available at http://
Maharatna,Arup (2001): 'On the 'BrighterSide' Lake: Case of OccupationalDisplacementof www.wws.princeton.edu/-rpds.
of SeasonalMigration:A Case Study of Ferti- Fishermen',Economic and Political Weekly, UNICEF (2001): 'GOI-UNICEF Country
lity Transitionin Tribal Population in Rural May 4. Programme of Cooperation 2003-2007',
West Bengal,India',mimeo,GokhaleInstitute Sarmah, Sasanka (2000): 'AgriculturalWages: mimeo, UNICEF, New Delhi.
of Politics and Economics, Pune. TrendsandDeterminants',M Philthesis,Delhi Vaidyanathan,A (2002): 'Food Consumptionand
MariBhat, P N (2002): 'Has the Decline in Infant School of Economics. NutritionStatus:A Re-examinationBased on
MortalityRate Slowed Down? A Review of -(2001): 'AgriculturalWagesinIndia:An Analysis IndianEvidence', mimeo, MadrasInstituteof
SRS Evidence', paperpresentedat a national of Regions and States', mimeo, Centre for Development Studies, Chennai.
workshopon 'InfantMortality:Levels, Trends Development Economics at the Delhi School Visaria, Pravin (2000): 'Alternative Estimates
and Interventions' held at New Delhi on of Economics;forthcomingin IndianJournal of Poverty in India', The Economic Times,
April 11-12. of Labour Economics. June 27.
Mehta,Jaya(2001): 'Give Povertya Face Please!', Sen, Abhijit (2000): 'Estimates of Consumer Yao Shujie (1999): 'Economic Growth, Income
in Alternative Economic Survey 2000-2001, Expenditureand its Distribution',Economic Inequality and Poverty in China under
Rainbow Publishers, Delhi. and Political Weekly, December 16. Economic Reforms',Journal of Development
Murthi, M, Guio, A C and Dreze, J P (1995): Sen, Amartya (2001): 'Many Faces of Gender Studies, 35.
'Mortality,FertilityandGenderBias in India', Inequality', Frontline, October 27, 2001.
Population and Development Review, 21. Shariff, A and Mallick, A C (1999): 'Dynamics
Nagaraj,R (2000): 'IndianEconomy since 1980: of Food Intakeand Nutritionby Expenditure
VirtuousGrowthor Polarisation?',Economic Classin India',EconomicandPolitical Weekly,
and Political Weekly, August 5. July 3. CONTEMPORARYEUROPE
National Instituteof Nutrition (1997): 25 Years Sharma, Alakh (2001): 'AgrarianRelations and
of NationalNutritionMonitoringBureau,NIN, Socio-Economic Change in Bihar', paper AND SOUTH ASIA
presented at a conference on 'Labour and
Hyderabad.
National Nutrition Monitoring Bureau (1999): CapitalistTransformationin Asia' held at the
Editedby D.K.Giriand
Centre for Development Studies, Thiru-
B. Vivekanandan
Report of Second Repeat Survey - Rural
(1996- 97), National Institute of Nutrition, vananthapuram,December 13-15.
Hyderabad. Shiva, Vandana(2001a): 'India:Corporatisation Eleveneminentscholars drawnfrom
National Sample Survey Organisation (200-la): of Agriculture Disastrous', The Guardian,
'Level and Patternof ConsumerExpenditure March 7. Europeand South Asia have exam-
in India 1999-2000', Report 457, National -(2001b): 'Is Globalisation Killing the Environ- betweencontem-
inedthe relationship
Sample Survey Organisation,New Delhi.
- (2001b): 'Consumption of Some Important
ment?', interview with Frederick Noronha;
available at http://www.onweb.org/archive/
poraryEuropeand SouthAsia inthis
features/shiva/shiva.html.
book.
Commodities",Report461, National Sample
Survey Organisation,New Delhi. Subramanian, A (2001): 'Are Income-Calorie 2001 254pp. Rs.450
Osmani, Siddiq R (1990): 'Nutrition and the Elasticities Really High in Developing
Economics of Food: Implications of Some Countries? Some Implications for Nutrition Concept Publishing Company
RecentControversies'in DrezeandSen (1990), and Income', mimeo, National Council of A/15-16, Commercial Block,
vol i. Applied Economic Research, New Delhi. Mohan Garden, New Delhi-59
PROBE Team (1999): Public Report on Subramanian,ShankarandAngus Deaton (1996): Ph.5351460,5351794 Fax:091-11-5357103
Basic Education in India, Oxford University 'The Demandfor Food andCalories',Journal
Email : publishing@conceptpub.com
Press, New Delhi. of Political Economy, 104, 133-62.


Economic and Political Weekly

Transféré par

Informations du document

Titre original

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Economic and Political Weekly

Transféré par

Droits d'auteur :

Formats disponibles

Poverty and Inequality in India: A Re-Examination

Author(s): Angus Deaton and Jean Dreze

This content downloaded from 147.26.11.80 on Fri, 9 Aug 2013 15:53:20 PM

Poverty and Inequality in India