Sto­ries about Africa num­bers can’t tell

Lesotho Times - - Leader - morten Jer­ven ven

IN NOVEM­BER 2010, Ghana Sta­tis­ti­cal Ser­vices an­nounced new and re­vised gross do­mes­tic prod­uct (GDP) es­ti­mates. As a re­sult, the es­ti­mated size of the econ­omy was ad­justed up­ward by more than 60 per­cent, sug­gest­ing that in pre­vi­ous GDP es­ti­mates eco­nomic ac­tiv­i­ties worth about $13bn had been missed.

While this change in GDP was ex­cep­tion­ally large, it did not turn out to be an iso­lated case. In April 2014, the Nige­rian Bureau of Sta­tis­tics de­clared new GDP es­ti­mates. GDP was re­vised up­ward to $510bn, an 89 per­cent in­crease from the old es­ti­mate.

Th­ese well pub­li­cised sta­tis­ti­cal events have led to an in­crease in the at­ten­tion be­ing paid to the qual­ity of macroe­co­nomic sta­tis­tics in low-in­come coun­tries, es­pe­cially in African coun­tries. The nu­mer­i­cal ba­sis used to study African economies is poorer than we would like to think and needs to im­prove in or­der to close Africa’s knowl­edge gap. This is to say we know less about growth and poverty pat­terns on the con­ti­nent than many be­lieve.

The prob­lem of study­ing Africa by num­bers is in part a sim­ple knowl­edge prob­lem: ig­no­rance through num­bers. There are ac­cu­racy prob­lems. The data are just very weak guesses and have large er­rors at­tached to them. There are avail­abil­ity prob­lems. Some coun­tries con­sis­tently go miss­ing in some of the data sets.and there are sim­ple prob­lems of cap­tur­ing com­plex so­cial and eco­nomic re­al­i­ties and trans­lat­ing th­ese into num­bers. This mat­ters more as we move from eco­nomic con­cepts to po­lit­i­cal and so­cial con­cepts.

For re­searchers and data users, the mes­sage is that study­ing Africa by num­bers can be mis­lead­ing. Many num­ber-based stud­ies suf­fer from a disconnect with re­al­ how do schol­ars and pol­icy mak­ers do good re­search with num­bers? How can they shed light on what they’ve got to work with? In my re­search pa­per I set out to an­swer th­ese ques­tions.

Faulty GDP es­ti­mates The con­cur­rent large and seem­ingly abrupt changes in GDP have led to a re­con­sid­er­a­tion of the qual­ity of the data needed to eval­u­ate ba­sic trends in growth and poverty. At the same time, ac­cord­ing to the African De­vel­op­ment Bank, such sub­stan­tial re­vi­sions have un­der­stand­ably alarmed many ob­servers. The World Bank’s chief econ­o­mist for Africa has writ­ten about “Africa’s sta­tis­ti­cal tragedy”.

The per­sis­tent doubts about African coun­tries’ abil­ity to pro­vide valid sta­tis­tics may in part be a true re­flec­tion of the data. But it may also be a per­cep­tion and cred­i­bil­ity prob­lem. The examples above are telling us that many sta­tis­ti­cal sys­tems in low-in­come coun­tries are be­ing up­dated after years of rel­a­tive ne- glect. Yet the re­sult­ing im­prove­ments in the ac­cu­racy of the GDP data have of­ten been met more with bewil­der­ment than recog­ni­tion of the GDP re­vi­sions as good news. They are good news. Not only are the coun­tries richer than pre­vi­ously thought, but also the up­dat­ing of bench­marks is a tan­gi­ble symp­tom of sta­tis­ti­cal sys­tems that are be­ing im­proved. I think that they are read as con­fus­ing or even bad news, be­cause it com­pli­cates our knowl­edge and raises ques­tions about what we ac­tu­ally know about in­come and growth in African economies. In turn, those are good doubts to have, be­cause our nu­mer­i­cal ba­sis to study African economies is poorer than we would like to think.

Be­yond growth: poverty, pop­u­la­tion and so­cial sta­tis­tics The mal­leabil­ity of the num­bers does raise ba­sic re­search ques­tions about states’ abil­ity to gather and dis­sem­i­nate sta­tis­tics. Out­dated meth­ods and gaps in the un­der­ly­ing sta­tis­tics are not some­thing that per­tains only to — nor to all — coun­tries in Africa. Nor are the is­sue of sta­tis­tics and prob­lems of data qual­ity unique to low-in­come coun­tries.

All other things be­ing equal, there are a pri­ori grounds to be­lieve that poorer economies will have lower-qual­ity sta­tis­tics. A poorer econ­omy will have rel­a­tively fewer avail­able re­sources to fund the func­tions of an of­fi­cial sta­tis­tics of­fice.

The para­dox here is that in th­ese very coun­tries, num­bers may have a greater im­por­tance. Th­ese coun­tries are more de­pen­dent, both po­lit­i­cally and fi­nan­cially, on in­ter­na­tional or­gan­i­sa­tions and global gover­nance — are­nas where num­bers are key mo­ti­va­tors of the pol­icy de­bate.

One of the strik­ing prob­lems in as­sess­ing the re­cent growth in sub-sa­ha­ran Africa is not only the ac­tual rate of eco­nomic growth but also how this growth is dis­trib­uted and whether the growth is in­clu­sive. The big ques­tion is: what hap­pens to poverty dur­ing growth?the most im­por­tant met­ric of poverty, judg­ing purely in terms of in­flu­ence, is the head­count in poverty. But then our knowl­edge based on num­bers is dou­bly bi­ased: we know lit­tle about poor coun­tries and even less about the poor peo­ple who live in th­ese coun­tries.

Th­ese prob­lems emerge from a va­ri­ety of sources. At the de­sign level, an in­com­pat­i­bil­ity ex­ists be­tween sta­tis­ti­cal cat­e­gories that were con­ceived for in­dus­tri­alised so­ci­eties with clearly de­fined prop­erty rights and for­mal em­ploy­ment re­la­tion­ships, and the de­vel­op­ing con­texts to which they are ap­plied.

At the im­ple­men­ta­tion level, a lack of ca­pac­ity and record keep­ing at of­fi­cial sta­tis­ti­cal of­fices is made worse by the chal­lenge of in­ac­ces­si­bil­ity as­so­ci­ated with poor and re­mote ar­eas. Thus, num­bers and in­di­ca­tors are es­pe­cially in­ad­e­quate in less de­vel­oped coun­tries.

Num­bers need to be in­ter­ro­gated There are ad­di­tional char­ac­ter­is­tics per­tain­ing specif­i­cally to African states that may jus­tify a spe­cial re­gional fo­cus. Study­ing sta­tis­tics is of course sim­i­lar to study­ing states. Just as one would ques­tion whether one should ex­pect the emer­gence of a sim­i­lar form to the Ja­panese or Nor­we­gian state in sub-sa­ha­ran Africa, one would also be care­ful in treat­ing sta­tis­tics from dif­fer­ent state sys­tems as fac­tual equiv­a­lents.

Some sta­tis­ti­cal sys­tems are built on reg­is­ter­ing and tax­ing land while oth­ers do not con­form to this foun­da­tion. His­tor­i­cally, African poli­ties were typ­i­cally land abun­dant and labour was rel­a­tively scarce. This has im­pli­ca­tions for the prop­erty rights regime. Land has typ­i­cally not been sub­ject to pri­vate prop­erty rights, and states have not col­lected taxes on land hold­ings. This also has di­rect im­pli­ca­tions for the power of the state.

The deeper point here is that num­bers need to be in­ter­ro­gated metic­u­lously. Con­fronted with sec­ondary data in the in­ter­na­tional data­bases, users need to con­duct ba­sic source crit­i­cism and ask, “Who made this ob­ser­va­tion?”, “Un­der what con­di­tions was this ob­ser­va­tion made?” and, “Is there any rea­son to think that the ob­ser­va­tion is bi­ased?” Fail­ure to do so in­creases the dis­tance be­tween the ob­server and the ob­served, and may lead to a disconnect be­tween re­al­ity and the num­bers. l This is an edited ex­tract from a re­search note ti­tled, “Africa by Num­bers: Re­view­ing the data­base ap­proach to study­ing African economies”.

Newspapers in English

Newspapers from Lesotho

© PressReader. All rights reserved.