Missing people in census – a new issue in Pakistan


By Mehtab S. Karim

The provisional results of the much-delayed census in Pakistan are shrouded in mystery and ambiguity due to lack of transparency. This has led to the validity of the entire exercise being questioned by many political leaders, media personalities, members of civil society, academics and policy planners. The 2017 census recorded a total population count of 207.8 million with an annual growth rate of 2.4pc during 1998-2017.
Surprisingly, the growth rate in Sindh – the destination for millions of migrants from other provinces – is at par with the national average. Meanwhile, growth rates of 3.41pc in Balochistan and 2.89pc in Khyber Pakhtunkhwa are on the higher side due to the inclusion of Afghan refugees who have intermingled with the local population there, unlike in Sindh – particularly in Karachi – where they, along with other migrants, remain isolated and thus perhaps have not been counted. However, Punjab’s annual growth rate of 2.13pc is mainly due to reduced fertility and migration within and outside the country.
Counting in a census is never perfect, and usually some people are left out. The best way to check the validity of a census result is through the universally accepted procedure of Post Enumeration Survey (PES) which is conducted in randomly selected areas following a census, and its results are cross-checked with the census results of the same areas. A PES is routinely conducted in many countries, with varying results. For example, the PES conducted in the US after its 1990 census revealed that 1.8pc of the population was missed; in Australia’s 2001 census, 2pc was missed; in India’s 2011 census, 2.3pc was missed with wide regional variations; and in Bangladesh’s 2011 census, 4.2pc was missed.
An independent commission of inquiry should investigate and validate the 2017 census results. In Pakistan, a PES was conducted following the 1961 census, which indicated an undercount of 6.7pc and consequently the population was adjusted upwards. The PES results after the 1981 census were not released, whereas the exercise was not carried out after the 1998 census, nor is one planned following the 2017 census, where it is likely that, as elsewhere, between 2pc to 4pc people have been missed.

KARACHI: Shia leader Allama Nazim Ali Azad handing over himself to police during a rally for quick and safe recovery of missing of members of Shia community under Shia Person Release Committee’s “Jail Bharo Tehreek” (Fill the Jails Movement) at National Highway in Karachi.

Another method of cross-verification of census results is through demographic modelling, which takes into account trends in demographic indicators (yearly estimates of birth, death, and migration rates) between two censuses. I carried out such an exercise using data from the annual demographic surveys conducted in Pakistan during 1982-97, and after taking migration into account, had estimated that in the 1998 census, about 6m (4.3pc) fewer people were counted, and most were missing from Sindh. Unfortunately, the annual Pakistan Demographic Survey was discontinued after 2007. In the absence of demographic indicators, it is thus difficult to estimate the population of Pakistan and its provinces.
Due to the influx of migrants from other provinces as well as the neighbouring countries, Sindh’s share in the country’s population increased from 18pc in 1951 to 23pc in 1981. However, in the subsequent two censuses it has remained at the same level, apparently due to systematic undercounting of Sindh’s population. The evidence lies in the reported average number of persons living in a household in the three previous censuses. For example, the reported average number of persons per household declined sharply in Sindh from 6.8 in the 1981 census to 5.6 in 2017 and slightly so in Balochistan from 7.2 to seven. On the other hand, it increased sharply from 6.8 to 7.9 in KP and slightly from 6.3 to 6.4 in Punjab.
This is contrary to what is reported in the yearly household sample surveys conducted during 2000 and 2015 by the Pakistan Bureau of Statistics (PBS). These surveys indicate that the average number of persons in each household was 6.6 in Sindh; 6.2 in Punjab; 7.5 in KP and 7.5 in Balochistan. It is apparent that in the 2017 census, an average of one person has been missed in each household in Sindh and 0.5 persons in Balochistan. Thus in the 2017 census, about 9m people were possibly missed in Sindh and 0.7m in Balochistan. On the other hand, apparently about 3.5m people were over-counted in Punjab, and 1.8m in KP. Sindh’s population is perhaps about 57m constituting 27pc of the country’s population, instead of 48m which is 23pc share in the country’s population, as reported in the 2017 census due to possible undercounting.

In several meetings with senior PBS officials, demographic experts (including myself), had proposed that for a validity check a PES be conducted following the 2017 census, but unfortunately this was met with opposition, because of fears based on ‘what if results do not match?’ Earlier this month, it was reported in Dawn that PBS is not willing to entertain the Sindh government’s request to provide relevant documents in order to check the validity of the results of the 2017 census. Not only is this against the principle of transparency, but in case there is a genuine discrepancy in the numbers, it will result in misallocation of resources as well as National Assembly seats on the one hand, and also jeopardise key policy outcomes on the regional and macro levels on the other.
Census data is integral to the growth prospects of the country at large and its provinces in particular. In the absence of international standards of cross-validation that have not been followed, if the 2017 census results are accepted at face value, it could lead to discontent and deprivation in Sindh and Balochistan. Under these circumstances, only an independent commission of inquiry could investigate and validate the 2017 census results. Until the validity of the exercise is checked through a PES in randomly selected areas by a third party (eg jointly by a team of demographers and representatives of universities’ statistics departments) and the results are cross-checked, it would not be in the national interest to accept the 2017 census results at face value. Given that the census was conducted on the orders of the Supreme Court, it should consider taking suo motu notice of the matter, so that the truth may prevail.

(The writer is a professor of demography and former member of the governing council of the Pakistan Bureau of Statistics.mehtabkarim@gmail.com)