Unfortunately, the accuracy of population data counted by censuses varies enormously across the world. The problem is particularly serious in Africa where GDP data is also of extremely poor quality making GDP per capita estimates of dubious value for most purposes. A paper by Deborah Potts published in World Economics in 2011 cast serious doubts on the accuracy of urban population estimates across Africa.
Only a census can provide accurate population data. The United States was the first country in the modern age to undertake a census in 1790, but this was closely followed by European countries including the United Kingdom which introduced its first census in 1801. Now most countries conduct regular census counts, but the regularity and the accuracy varies, especially in Africa and Asia.
Currently, the UN World Population and Housing Census Programme recognises Population and Housing Censuses (PHCs) as a main data source needed for policies and programmes aimed at inclusive socioeconomic development and environmental sustainability. They are also an important source for supplying disaggregated data for the measurement of progress of the Sustainable Development Goals (SDGs), especially in the context of assessing the situation of people by income, sex, age, race, ethnicity, migratory status, disability and geographic location, or other characteristics.
According to the 2019 African Governance Report by the Mo Ibrahim Foundation over the last decade only 39 African countries, home to 711.1 million citizens, have conducted a PHC. This means that only just over half (54.4%) of the continent's population live in a country where a census has been conducted during the period 2009 to 2018. In DRC, Eritrea and Somalia, the last PHC was conducted before 1990, almost 30 years ago. An additional 15 African countries are planning to conduct their population and housing census before the end of 2019.
Censuses are resource intensive, however, and strain the statistics budgets of many countries. In the United States, according to the Government Accountability Office, the cost of administering its decennial counts have increased in constant prices from US$16 per household in 1970 to an estimated US$92 per household for 2020. For many emerging markets the costs of carrying out a census often leads to problems in the extent of the coverage of the population and the census accuracy. In many countries a census can be distorted by political manipulation and violence. For example, there has been uncertainty about the size of the population of Nigeria and many other countries for some time and each census is a highly politicised event. The last Nigerian census in 2006 calculated that the total population was 140 million, but current estimates using data on population replacement place it as lying between 180 and 200 million. The differences imply large divergences in estimates of GDP per capita levels in the country. No budget has been agreed for a subsequent census, but a United Nations report published in 2018 projected that, by 2050, Nigeria will become the world’s third largest country by population and one of the six nations with a population of over 300 million.
Even in the United States, which holds a census every 10 years, and where the Census Bureau has a permanent staff of over 10,000, errors can creep in to the census process. It has been estimated by Freedman and Wachter (2003) that the 2000 census produced errors varying between an undercount of 0.7% to an overcount of 1.2%, or a difference of 5.1 million people. The 2020 US census is also already controversial as a result of a question on citizenship introduced by the administration which it is feared will lead to a reduction in the accuracy of the poll due to under-reporting.
In the absence of regular census data in many countries population estimates have to be made between censuses by reference to data on official births and deaths registration and net migration data produced by government statistical authorities whose reliability varies across countries. Unfortunately, an independent report by UNICEF found that between 2012 and 2016 30% of children under the age of 5 in Nigeria had never had their births registered. The recent Mo Ibrahim Foundation report found that only eight countries in Africa – register more than 90 per cent of births.
UNICEF estimated that 250 million children worldwide have not been registered and World Bank data on completeness of birth registration shows that in some countries the estimated level of registrations of live births is low lying below half of all births, for example, Chad 12%; Bangladesh, 20%; Nepal 46%, Pakistan 42%. The problem of raising the proportion of births registered in many emerging markets is affected by the sheer size of the informal economy, poor institutions and the size of large undocumented communities that have migrated from rural to urban communities. In Nigeria, registration is also reduced by payments changed by the National Population Commission.
In conclusion, there are serious questions of the reliability of population data across the world. Even in the developed world census data, the most accurate method of measuring population size and demographic characteristics, suffers from biases. In many emerging markets in Africa censuses occur infrequently and population estimates between these irregular counts, are subject to unreliable estimates of the birth rate combined with the low statistical capacity of local institutions. Population estimates should be used with caution.