72 Finance & economics The Economist May 21st 2022
Dodgydata
Half-measures
T
hemodernIndianstatehasaproud
statistical heritage. Soon after the
countrygainedindependencein1947,the
governmentresolvedtoachieveitsdevel
opmentthroughcomprehensivefiveyear
plans.Thestrategy,thougheconomically
inadvisable,nonethelessrequiredthecre
ationofa robustdatagatheringapparatus.
In 1950 PCMahalanobis,theleadinglight
ofIndianstatistics,designedtheNational
SampleSurvey,whichsentstafftothefar
cornersofthevastcountrytojotdowndata
regardingitsmostlyilliteratecitizens.The
survey’s complexity and scope seemed
“beyondtheboundsofpossibility”,reck
onedoneAmericanstatistician.
Oflate,however,admirationhasbeen
replacedbyalarm.India’sstatisticalservic
esareina badway.Acrosssomemeasures,
figuresaresimplynotgathered;forothers,
thedataareoftendodgy,unrepresentative,
untimely, or just wrong. The country’s
trackingofcovid19providesa grimexam
ple.AsthepandemicragedacrossIndia,
officialsstruggledtokeeptabsonitstoll.
Officially,covid hasclaimed more than
halfa millionlivesinIndia;TheEconomist’s
excessdeathstrackerputsthefigurefar
higher,between2mand9.4m.India’sgov
ernmenthasalsohamperedeffortstoas
sessthepandemic’sglobalimpact,refus
ingatfirsttosharedatawiththeWorld
HealthOrganisation(who), andcriticising
itsmethods.
Thepreferenceforflatteringbutflawed
figures is pervasive. In education, state
governmentsregularly ignore data show
ing thatIndianchildren are performing
woefullyinschooland instead cite their
ownadministrative numbers, which are
oftenwrong.InMadhya Pradesh, a state in
central India, an official assessment
showed thatallpupils had scored more
than60%ina mathstest; an independent
assessment revealed that none of them
had.Similarly,insanitation, the central
governmentsaysthat India is now free of
opendefecation,meaning that people both
haveaccesstoa toilet and consistently use
it.Anyonewhotakesa train out of Delhi at
dawnandlooksoutof the window, how
ever,mightquestionthe claim.
Whenitcomestopoverty, arguably In
dia’s biggestproblem, timely figures are
notavailable.Official estimates are based
ona povertylinederived from consump
tiondatain201112,despite the fact that
morerecentbutasyet unpublished num
bersexistfor201718. By contrast, Indone
siacalculatesitspoverty rate twice a year.
India’sgovernmentexplains its approach
bypointingtodiscrepancies between re
centlygathereddataand national accounts
statistics—butmanysuspect the true rea
son is that newer data would probably
showanincreaseinpoverty.
Insomecases,flawed data seem more a
problemofmethodology than malign in
tent.India’sgdpestimates, for instance,
havebeenmiredincontroversy ever since
thestatistics ministry introduced a new
series in 2015 (a change that was in the
works before the current government en
tered office). Arvind Subramanian, a for
mer government adviser, calculated that
the new methodology overestimated aver
age annual growth by as much as three to
four percentage points between 201112 and
201617. Although current advisers insist
that the official methodology is in line
with global standards, other studies have
also found problems with the calculations.
The erosion of India’s statistical infra
structure predates the current govern
ment, but seems to have grown worse in
recent years. Narendra Modi, the prime
minister, has previously bristled at tech
nocratic expertise and numbercrunching.
(“Hard work is more powerful than Har
vard,” he said in 2017.)
India’s data woes are also troubling for
what they suggest about the ability of the
state to provide the essential public servic
es needed to foster longrun growth. The
statistics ministry, short of staff and re
sources, is emblematic of the civil service.
Datagathering has become excessively
centralised and overpoliticised. A Nation
al Statistical Commission was set up in
2005 and tasked with fixing India’s data in
frastructure. But its work has been compli
cated by turf wars and internal politics; it is
widely considered toothless, including by
former members.
Who’s counting
The situation is not hopeless, perhaps be
cause of statisticians’ past efforts. Accord
ing to the World Bank, the quality of Indian
data is still in line with that of other devel
oping countries, even after years of ne
glect. India’s new goodsandservices tax
and digitalwelfare infrastructure are
yielding troves of data. Leading Indian
statisticians argue that an empowered reg
ulator could fix existing problems.
State governments and departments
are also doing their bit. Telangana, a south
ern state, is investing in its own household
surveys, for example. India’s ruraldevel
opment ministry recently released a data
set covering 770,000 rural public facilities,
such as schools and hospitals, inviting da
ta whizzes to peruse the figures and sug
gest improvements. Civil society is also re
sponding. During the pandemic, dozens of
volunteers cooperated to produce granu
lar, timely estimates of covid cases. New
technologies could help gather data quick
ly and cheaply, over phones and tablets.
Yet in a modern economy there is no
substitute for highquality national data
gathering. The sunlight provided by accu
rate figures is often unwelcome for an
increasingly autocratic government: tran
sparency invites accountability. But ne
glect of the statistical services also leaves
Indian policymakers flailing in the dark,
unable to quickly spot and respond to
brewing economic and social problems.n
D ELHI
India’s once-vaunted statistical infrastructure is crumbling
There’s strength in numbers