‘Leave out migrants and you undercount population’

How well is the census going, TFT asks an expert

‘Leave out migrants and you undercount population’
How has Pakistan fared when it comes to census-taking and how is it shaping up this time? The Friday Times asked Dr. Mehtab S. Karim, who did his PhD in Demography from Cornell University. He has also worked as a senior research fellow at the Pew Research Center in Washington and as a professor of Demography at the Aga Khan University in Karachi. He has also worked for the United Nations in New York on census issues. His insight on Pakistan’s demographics and government policy grew during his work with various national committees and with his appointment in 2012 to the governing council of the Pakistan Bureau of Statistics. He currently works as the vice chancellor of Malir University of Science & Technology in Karachi where he is also the executive director of the Centre for Studies in Population & Health.

TFT: Can you provide historical perspective on census taking in Pakistan? How did we conduct them in the past? 

Dr Karim: During the past 15 years, a census has been conducted in almost all UN member countries, however, only a handful among major countries have not done so. These include:  Lebanon, Somalia, Afghanistan and Pakistan. In 1881, the British colonial government established the tradition of conducting a census in the subcontinent and since then, after every 10 years, one was conducted till 1941.

Pakistan followed this tradition and conducted censuses in both its wings in 1951 and 1961. Due to civil war in East Pakistan, followed by war with India the 1971, the census was held in 1972 followed by one in 1981. In 1991, a housing census was conducted, but due to suspicion of over-reporting of individuals living in Sindh, the detailed census exercise was suspended. Under the pressure from international agencies (the UN and World Bank) a census was conducted in 1998, which was the first time in the history of any country that the census was conducted under the supervision of the military.

The next census was due to be held in 2008, but was postponed till 2011, when a housing census similar to the one held in 1991 was conducted. But again, due to similar reasons as in 1991, it was postponed indefinitely.

Eventually, last year, the Supreme Court took suo moto notice and forced the federal government to conduct the overdue census, which is currently under way. On the other hand, in India, censuses have been conducted without any interruption till 2011 and Bangladesh has followed that tradition, with the exception of delaying the 1971 census till 1974.

TFT: What did we get right, what did we get wrong? What did the census results in the past reveal about us (demographics)?

Dr Karim: Since the census is a gigantic undertaking, and in countries where a large majority of people are not educated, census results suffer from undercoverage and under-reporting of the population. Besides this, since in Pakistan (and in most other democratic countries) census results determine the allocation of seats in the National Assembly, census- taking in Pakistan has become quite politicized. This became more evident after the 1981 census—largely believed to be the most accurate census conducted in Pakistan—showed that the share of Sindh has been increasing and of Punjab declining. However, the  results of the 1998 census showed that share of each province remained the same. Since Sindh has been on the receiving end of migrants from the other three provinces, the results of the 1998 census were unanimously rejected by the then Sindh Assembly.

Although a census is supposed to count everybody, many people are missed. For example, in Pakistan after the 1961 census, when a post-enumeration survey was conducted, about 6% of an undercount in the population was discovered. Therefore, the population was later adjusted upwards.

Since the 1998 census was held with the assistance of the armed forces, it was assumed that in the presence of military personnel who had accompanied the civilian enumerator, the household head would not exaggerate the number of his family members, as was experienced in some areas during the housing census in 1991. However, since a post-enumeration survey was not conducted, results could not be validated, particularly when it came to the possible undercount in the 1998 census.

Subsequently, in 2000, a steering committee of experts was constituted by the Government of Pakistan with support from the United Nations Population Fund. I served as a member of the steering committee and did estimate the population of Pakistan as 142 million (instead of 132 reported in the census), by applying the annual growth rates derived from the Pakistan Demographic surveys, conducted between 1982 and 1997.

However, after accounting for about 3 million to 4 million people who had emigrated from Pakistan to the Gulf countries, Europe and North America, during the 1981-98 period, there was still a shortfall of about 6 million people.

In a subsequent report published in 2008 by the Pakistan Institute of Development Economics, it was pointed out that in the 1998 census, the population of urban areas and migration to large cities were under-estimated, data on education was inadequate and there were issues pertaining to data collected on economic activity.

TFT: How have you been involved with the census this time around? What have you noticed about how the census was planned? You have been in the field. Given the difficulties, how do you rate the effort so far? 

Dr Karim: My earlier involvement with the census in Pakistan was as a short-term adviser on behalf of the United Nations to help with further analysis of the 1981 census data and as a member of the steering committee set up to evaluate the 1998 census results.

This time, I have been involved prior to conducting the census. I served as a member of the governing council of the newly established Pakistan Bureau of Statistics (PBS) during 2012-14 and in that capacity, I was appointed by the finance minister as chairman of the committee to make recommendations for the forthcoming census.

Our committee had advised that the census should be conducted in 2015 with an emphasis on doing the housing and population census simultaneously, making the whole exercise transparent and conducting a post-enumeration survey immediately after the census, so that everybody could accept the results. However, the census could not be conducted because the armed forces personnel were not available. Our committee had also recommended against involving the armed forces in the census exercise as they were busy with the ongoing Zarb-e-Azab operation.

Due to the delay, for the first time, the Supreme Court notice last year and ordered that it must be conducted in March. It was expected that the PBS would be ready to do so, however, in its meeting, the governing council noticed that issues had yet to be resolved. Therefore, at the beginning of February, a committee of the governing council was notified (of which I am also a member). The committee consists of experts in demography, statistics and economics,  representing the four provinces and the Islamabad Federal Territory. The committee has so far met six times and has tried to overcome some of the issues.

The issue of asking for a CNIC number for the household head was thoroughly debated as the concern was that it may lead to undercounting of foreigners as well as those who have not been issued a CNIC, but this provision has been retained.

I have been in the field in Karachi and have noted that the exercise is going on smoothly.  However, since enumerators have not been trained about the question added at the eleventh hour related to people suffering from disabilities, information about them is not being properly collected. Besides this, all the people are well aware about the census and are very cooperative but are rarely reporting the presence of foreigners in the household. Also, households are not being visited in the evening in urban areas and so, in many cases both husband and wife are working and would not be counted.

Due to the above two factors, it is likely that foreigners and the total population of urban areas will be underreported.

TFT: In the immediate aftermath of the count what do you foresee as problems arising in the data processing, statistical analysis and on the political and bureaucratic fronts? 

Dr Karim: Data processing will be easy as forms are machine-readable and the PBS is well-equipped to do data processing. However, it has yet to devise a plan for data analysis.

We have been told that the census exercise will be completed by the end of May and within a month or two, results will be available. Perhaps we will come to know about the total population of the country, its provinces and urban and rural areas. However, the PBS has to develop a schedule of data analysis, particularly for bringing out detailed reports for each province and the federal territories, which will require a lot of effort and quite a bit of time.

Since the census is being conducted after about 19 years, most senior and experience persons who worked on writing the detailed reports of 1998 census have retired. Therefore, the PBS should think of outsourcing the analysis and report writing.

TFT:  What trends are you expecting to emerge this time round? 

Dr Karim: Very little is known about the demographics of Pakistan. Since the early 1970s, the  Federal Bureau of Statistics used to conduct a Demographic Sample Survey each year. However the last such survey was conducted in 2007. Therefore, we do not know what are our birth and death rates and consequently at what rate the country and its provinces are growing.

What we know are only guesstimates of the rates of population growth. Therefore, even the estimates of the country’s population are not accurate, which could be between 195 million (the federal government’s estimate) or even 210 million. Besides, in the absence of any demographic data, it is impossible to estimate the population of the provinces. To do so, besides information on annual birth and death rates for each province, estimates of inter-provincial migration are  also required.

TFT: What is the biggest or the most salient aspect of our demographic profile which is not being collected in this census that we are ignoring in the national debate which will have major policy implications?

Dr Karim: In the previous censuses, detailed information was collected about migration (such as district or country of birth, place of previous residence, period since living at the present residence and reason for migration), births and deaths, economic activities and nature and type of disabilities, through the long form which was administered in every 10th household. This time around, no such information is being collected in the census but will be done later, through a sample survey. This is crucial information which helps understanding important demographic, economic and health indicators for planning purposes.

Instead, efforts are being made to find Afghan refugees and illegal aliens by checking the CNICs in the house listing, which is not a census requirement. According to the UN definition everybody, irrespective of their nationality, should be counted.

This means that on one hand, most illegal aliens who do not have a CNIC will not be reported. This leads to the possibility of undercounting the population. Then, since all illegal persons are migrants, by not collecting information related to migration in the census, we will miss a very crucial demographic trend. We will miss data not only of foreigners who have moved here but also data on the trend in migration between one province to another and from rural to urban areas.