Quality and Level of Birth Registration in South Africa: 1998-2008
Mosidi S. Nhlapo
Census Analysis, Statistics South Africa, Pretoria, Gauteng, South Africa

A large body of literature exists in South Africa on levels and trends of fertility. However, these studies employ mostly indirect demographic techniques to derive estimate of fertility based on censuses and household surveys data. This is mainly due to the perceived poor state of vital registration data. Although there is general agreement about fertility trends in the country, there remains conflicting views about levels. A viable birth registration system should resolve some of the inconsistencies currently observed in fertility estimation, as it requires no adjustments. This study seeks to assess the quality and reliability of registered birth data from the South African Department of Home Affairs 1998 to 2008 and to derive and compare fertility schedules obtained from register with those based on enumerated data.

The results show improving coverage over time. However the extent of missing data related to parental age and the continued presence of births occurring to women outside the 12-49 age range points to shifting challenges from coverage issues towards quality concerns. The absence of some key variables such as race, education and marital status restricts data use to derivation of fertility schedules. Completeness estimates derived alternatively from current registration data and data updated for late registrations regardless of the time lag between registration and birth points to quality issues introduced by inclusion of late registrations in the analysis. Completeness increases from 24% in 1998 to a plausible 87% in 2008 when only data based on current registrations is used. Birth data updated for late registrations give completeness estimates of 77% in 1998 and 96% in 2008. These completeness levels seem improbably high and indicate problems associated with inclusion of late registration data.

Fertility schedules derived from completeness adjusted data compares favourably with enumerated data, but large adjustment factors in earlier years attracts some uncertainty about fertility estimates derived from the register in earlier years. Both updated and current registration data yield the same estimate of 2.6 Total Fertility Rate in 2006 due to declining levels of late registrations over time. These estimates are reasonably closer to those estimated by Statistics South Africa (2010) of 2.6 unadjusted and 2.8 adjusted TFR.

Comparisons of ASFR from the two data sources indicate that after adjustment for completeness, current registration data can be used to provide reasonable evaluation of enumerated data in 2006. However, ASFR for 2001 provide indications of under-registration of births at younger ages and possibly age misreporting. Importantly, results indicate that, current registration data could be used in direct estimation of fertility in the near future.

Keywords: Birth register; Statistics South Africa; Completeness; Total Fertility Rates (TFR)

Biography: I am Mosidi Nhlapo, I have a MA Demography degree from Makerere University, I am a employed as senior Demographer in the Demographis Analysis section at Statistics South Africa in Pretoria South Africa.I am currently seconded to the Census Analysis section. My interests include social demography, population estimations and projections.