The Labour Supply Effects of the South African State Old Age Pension: Theory, Evidence and Implications¶

Literature › Works

Alex Sienaert (2008). The Labour Supply Effects of the South African State Old Age Pension: Theory, Evidence and Implications. SALDRU Working Paper Series, Number 20. Southern Africa Labour and Development Research Unit, University of Cape Town.


Type	working paper
Scope	Post-1994 South African economy
Method	econometric
Period studied	2002–2004
Regions	South Africa (national), Eastern Cape, KwaZulu-Natal, Limpopo
Themes	labour markets wages, inequality poverty unemployment, transition post apartheid, demography urbanisation, mining migrant labour, methods data sources, comparative theory
Citation key	`sienaert2008`

Argument¶

Using September 2002-2004 Labour Force Surveys, the paper tests whether South Africa's non-contributory old age pension changes the labour supply of prime-age (19-50) black members of pension-receiving households. Parametric cross-section, regression-discontinuity and household-panel difference-in-differences estimates show modest negative effects on participation and larger, more robust negative effects on employment, mainly when pensioners are women, alongside a positive migration response. Defining the household to include non-resident migrants shrinks the negative estimates, reconciling Bertrand et al. (2003) with Posel et al. (2006). The paper argues that the gender-differentiated effects need not reject income pooling, since the female pension has a roughly 44% higher expected annuity value. It concludes that the pension reduces job-search intensity but that these effects are of little policy concern given high involuntary unemployment.

Evidence and method¶

Statistics South Africa Labour Force Surveys (September 2002, 2003, 2004; 85,663 black individuals in 21,847 households in 2004); a balanced 2002-2003 household panel of 12,212 households; a pooled discontinuity sample of 11,142 households; 2001 Census-based mortality estimates; secondary literature on the pension.

Findings¶

The pension is large relative to household resources and originated as a racially discriminatory scheme that was equalised during the transition. (p. 4)

payouts (at 870 rand, or about 120 U.S. dollars, a month) averaging more than twice black median income per capita
Parity was reached just before the 1994 election because the government raised black benefits rather than cutting white ones. (p. 4)

By 1993, just before the ﬁrst democratic elections in April 1994, and after large increases in access and beneﬁts for black pensioners, parity was essentially achieved.
Pension-receiving households are poorer in labour-market terms, with more unemployed residents and more migrants. (p. 20)

the probability that a recipient household contains migrants is 33%, compared to 21% for a non-recipient household
Cross-section probits show a sizeable negative association between pension receipt and employment. (p. 24)

in the pooled speciﬁcation, the pension is associated with an 11% lower probability of employment, all else equal
Including migrants in the household definition reduces the estimated negative employment association. (p. 25)

The OAP is associated with considerably lower residential household employment – -14% in the pooled OLS speciﬁcation
The discontinuity design finds little aggregate effect on participation or employment, and no influx of working-age people into pension households. (p. 30)

There are no bandwidth- and speciﬁcation-robust responses to the OAP threshold, except for the female labour participation rate, which is consistently, signiﬁcantly reduced across speciﬁcations.
The panel difference-in-differences finds a large negative employment effect for first-time recipient households, especially for women. (p. 34)

Prime-age gender-disaggregation indicates the magnitude of the effect for women is large (-11%), while the coefﬁcient on male employment narrowly misses statistical signiﬁcance
Women account for most first-time pension receipt, helping explain why female pension effects are stronger. (p. 36)

Of the 569 switches to OAP receipt observed in the LFS panel constructed for Section 6, 63% are associated with women becoming age-eligible, and 23% with men
The expected present value of the pension differs sharply by gender because of eligibility age and longevity. (p. 37)

the approximate expected present value of the pension is 62,400 rand for men and 89,300 rand for women
Employment elasticities with respect to pension income are small, much lower than Bertrand et al.'s estimate. (p. 42)

Both labour participation and employment look distinctly inelastic, and are much smaller than those obtained by Bertrand et al. (2003)
Labour supply responses crowd out part of the pension's income value, but labour supply is not the binding constraint on employment. (p. 41)

to a consumption-maximising policymaker, the labour supply effects of the OAP reduce its efﬁcacy by close to two-ﬁfths at the mean