Is the greater “statistical tragedy” in Africa (Devarajan 2013) the scarcity of census
data, or the lack of access to the existing data? Is the problem “Poor Numbers” (Jerven
2013) or inaccessible numbers? In each decade since the 1970s, at least 80% of the
continent’s population was censused, yet much of the microdata are not available for
scientific or policy research. From the late 1980s, twenty five countries entrusted
microdata to the African Census Analysis Project (ACAP), amassing a stock of
microdata for 47 censuses. Nonetheless, in recent years, the project seems moribund
with no published research, nor cosmetic touch-ups to the ACAP website since 2007.
The director, Dr. Tukufu Zuberi, no longer permits access, even to researchers wishing
to study their own country nor does he respond to requests to repatriate copies to the
official statistical office-owner of the microdata.
In 1999, the Minnesota Population center began a global initiative, IPUMSInternational,
offering free, internet access (www.ipum.org/international) to integrated
census microdata for researchers world-wide under a single license agreement with
National Statistical Office partners. Microdata for 69 countries, totaling 480 million
person records (212 samples), are accessible for research. The June 2013 release will
increase the number of countries to 74 with 238 samples and over 540 million person
records. IPUMS-International disseminates microdata encompassing 80% of the
world’s population, but the coverage for Africa is barely half that, at 42%. Africa is
under-represented in the database, not only due to a slow start and ACAP’s refusal to
cooperate but also because the African statistical offices are exceedingly reluctant to
allow outsiders access to the data. Nonetheless, microdata for fifteen African countries
(29 censuses, 55 million person records) are currently available and Africa has become
a top priority as more African census data are entrusted. Integration work is underway
for another fifteen countries, but some very important nations—Nigeria, Algeria, Zimbabwe, etc.—are not yet participating (see Appendix, table A1). Microdata are
inaccessible for one-third of the population of Africa.1

The African Development Bank, seeking to promote open access to census microdata
recently commissioned a technical expert to visit statistical offices that have been slow
to open their doors to assemble the data on the spot. Success was achieved within
months in five countries—Benin, Cameroon, Ethiopia, Liberia, and Mozambique.
Funding remains on the table, awaiting a signal to proceed, for 17—Algeria, Burundi,
Central African Republic, Comoros, Congo-Republic, Cote d'Ivoire, Equatorial Guinea,
Eritrea, Gabon, The Gambia, Libya, Mauritania, Namibia, Nigeria (National Population
Commission), Swaziland, Tunisia, and Zimbabwe. Hopefully the Bank’s support will
bear fruit in the not too distant future so that the vast major of African statistical offices
make their census microdata available.
This paper analyzes 24 African census samples (13 countries) available from the
IPUMS website to illustrate how microdata may be used to assess development and
pinpoint basic human needs at local administrative levels over time. We calculate a
Human Development Index-like measure for small areas (typically municipalities,
henceforth denoted as MHDI), recently proposed by Permanyer (2013). Unlike the
United Nations Development Program’s classic HDI, Permanyer’s measure is computed
solely from census microdata and therefore, when the data are accessible, may be easily
calculated for small administrative areas, where much of the responsibility lies for
executing policies related to health, education and general well-being. Summarizing the
UNDP’s HDI at the national level has its attractions, but the MHDI exposes inequalities
exist within country at the same time that it offers a summary statistic for the entire
country, although somewhat different from the classic HDI. In this respect, the MHDI is
one of the latest attempts to construct human development indicators defined below the
country level2
.
One of the most attractive features of the use of complete census data is the possibility
of disaggregating national-level averages and exploring the distribution of human
development and its components with unprecedented geographical detail. In particular,
the availability of complete census microdata allows pinpointing those administrative
units leaping ahead or lagging behind in the pace of well-being progress. Therefore, the
MHDI methodology can be particularly useful for policy-makers in need of highly
detailed data. The MHDI is a composite with three components: health (proportion surviving of liveborn
children), education (a composite of literacy and primary education completion),
and standard of living (assets, such as potable water, waste disposal and electricity).
For countries with two or more suitable sets of census microdata, we compare change
over time. For all countries with at least one set we offer cross-national comparisons
and calibrate the national census-based measure against the conventional HDI.
The paper is structured as follows. In section 2 we present the definitions, the data and
the methodology that has been used to construct the MHDI for 24 census samples in 13
African countries. The empirical results of our analysis are shown in section 3. We
discuss the implications of our results in section 4. We conclude with a discussion of
methodological, theoretical, and policy implications as well as an appeal to African
statistical agencies that have not yet done so to open access to census microdata.
Despite the pessimism in the epigraph, we argue that Africa is the continent to benefit
the most from the MHDI, when African census agencies adopt twenty-first century
principles of access to microdata.