Home / Columns / Can pensions impact fertility in Africa?
2,15
π Society

Can pensions impact fertility in Africa?

Pauline Rossi
Pauline Rossi
Professor of Economics at CREST* (IP Paris)

Tra­di­tion­al­ly, the issues of social pro­tec­tion and pop­u­la­tion growth have been stud­ied sep­a­rate­ly. Where does this idea that there is a link between fer­til­i­ty and the pen­sion sys­tem come from?

This is not a new idea. Since the 1950s, soci­ol­o­gists and demog­ra­phers have high­light­ed the role of social pro­tec­tion sys­tems, and more specif­i­cal­ly pen­sions, in fer­til­i­ty. By car­ry­ing out opin­ion sur­veys in the field, they have found that eco­nom­ic rea­sons appear to be very well placed in the moti­va­tions of respon­dents for hav­ing chil­dren: peo­ple repro­duce to keep the fam­i­ly busi­ness going, to mul­ti­ply the fam­i­ly’s income and to pro­vide for their old age. All in all, this phe­nom­e­non is quite logical.

The fact remains that these qual­i­ta­tive stud­ies were not able to mea­sure the quan­ti­ta­tive impact of this or that mea­sure on fer­til­i­ty: when a gov­ern­ment sets up a pen­sion sys­tem, does fer­til­i­ty fall by one, two or three chil­dren per woman, for exam­ple? The study I con­duct­ed in Namib­ia showed that the intro­duc­tion of a pen­sion sys­tem reduced the coun­try’s birth rate by at least one child per woman.

Why did you choose Namib­ia as a case study?

A few coun­tries in Africa have set up non-con­trib­u­to­ry pen­sion sys­tems, i.e. they do not depend on the con­tri­bu­tions of mem­bers, and there­fore ben­e­fit the entire elder­ly pop­u­la­tion. Among them was South Africa which, at the end of apartheid, set up a gen­er­ous pen­sion scheme, thus becom­ing a kind of world lab­o­ra­to­ry for the study of the socio-eco­nom­ic impli­ca­tions of the basic pen­sion. The prob­lem is that there was not enough usable data before this reform to be able to exam­ine the con­se­quences of pen­sions on fer­til­i­ty in detail.

So, I turned to Namib­ia – a for­mer South African province – which intro­duced an equiv­a­lent sys­tem after inde­pen­dence in 1990. In the mid-1990s, all Namib­ians over the age of 60 start­ed receiv­ing a kind of old-age min­i­mum equiv­a­lent to three times the pover­ty line ($1 a day at the time), or $90 a month in pur­chas­ing pow­er par­i­ty. The effect on the birth rate was immediate.

There­fore, since they were antic­i­pat­ing that they would receive this mon­ey in the future, women start­ed hav­ing few­er children?

Yes, that is what the study shows. In a decade, the birth rate in my sam­ple dropped from almost 6 chil­dren per woman to 4. Not all of this is due to the pen­sion sys­tem, as we shall see, but the study has shown that the intro­duc­tion of a uni­ver­sal old-age pen­sion reduces fer­til­i­ty by at least one child per woman. This phe­nom­e­non is more vis­i­ble among old­er women (over 30), prob­a­bly because they antic­i­pate their future retire­ment more.

How can we be sure that oth­er fac­tors did not influ­ence fertility?

This was the dif­fi­cul­ty of the study. Many fac­tors can influ­ence a coun­try’s birth rate: school­ing, urban­i­sa­tion, but also the reduc­tion in pover­ty, access to health­care, the reduc­tion in infant mor­tal­i­ty, or the intro­duc­tion of oth­er social expen­di­ture, such as fam­i­ly allowances (which tend to work in the oppo­site direc­tion). In Namib­ia, pen­sions con­sti­tute 90% of the coun­try’s social expen­di­ture, so there was dan­ger of con­fu­sion. To neu­tralise the oth­er fac­tors, it was nec­es­sary to iso­late the “vari­able of inter­est” (pen­sions), by con­sti­tut­ing two groups of peo­ple, more or less affect­ed by the reform.

In this sense, Namib­ia was an ide­al exam­ple, as the reform was imple­ment­ed grad­u­al­ly, so it was easy to mea­sure its effects on dif­fer­ent study groups. Before 1992, some eth­nic groups already had access to a pen­sion sys­tem, but the actu­al cov­er­age var­ied from 30 to 80% depend­ing on the region. It was not until the late 1990s that full cov­er­age was achieved. From then on, it was pos­si­ble to iso­late pen­sions from oth­er phe­nom­e­na that applied uni­form­ly to all groups. This led to the con­clu­sion that pen­sions could be attrib­uted to a decline in the num­ber of chil­dren of at least one child per woman. And that, over­all, the actu­al amount of the ben­e­fit had less of an impact on fer­til­i­ty than the imple­men­ta­tion of the sys­tem itself.

Today, devel­op­ment aid focus­es more on fam­i­ly plan­ning and con­tra­cep­tion. Why is this so?

Because it is too often assumed that the num­ber of chil­dren does not depend on wom­en’s choic­es, but on poor preg­nan­cy con­trol and lim­it­ed access to con­tra­cep­tion. In the past, this assump­tion has been explored with the estab­lish­ment of fam­i­ly plan­ning and con­tra­cep­tive pro­grammes in sev­er­al Asian coun­tries. In the 1970s, an influ­en­tial study in Bangladesh, for exam­ple, found that these mea­sures reduced births by about one child per woman. But today econ­o­mists are find­ing it hard­er to estab­lish the effec­tive­ness of these plan­ning poli­cies on the birth rate, espe­cial­ly in Africa. Every year, bil­lions are spent by OECD gov­ern­ments and large inter­na­tion­al NGOs to improve access to con­tra­cep­tion. Even if this is not their only objec­tive, the impact on the birth rate of African coun­tries is not very visible.

Interview by Julie de la Brosse

Contributors

Pauline Rossi

Pauline Rossi

Professor of Economics at CREST* (IP Paris)

Pauline Rossi's research focuses on applied microeconomics, development economics and family economics. She published in 2021 with Mathilde Godard (CNRS) "Children as a stick of old age: quantifying the impact of the extension of public pensions on fertility in Namibia", in the American Economic Journal: Economic Policy. In 2016, she received the special mention of the AFSE (French Association of Economic Science) for her thesis "The determinants of fertility choices in Africa: gender preferences, insurance strategies and reproductive rivalries".

*CREST: a joint research unit CNRS, ENSAE Paris, École Polytechnique - Institut Polytechnique de Paris, GENES