With a doctorate in demography from Cornell University, USA Dr. Mehtab is a social scientist with more than four decades of experience in the US, UK and in Pakistan. In the US, he has been associated with Princeton John Hopkins and George Mason Universities, whereas in Pakistan he has been previously associated with the Aga Khan University and Pakistan Institute of Development Economics (PIDE) and served on various committees constituted by the Government of Pakistan. He has also served as a consultant to the United Nations and the World Bank.
In this interview Dr. Mehtab reflects on his experience with Pakistan’s census while shedding light on the various reasons why the population census of 2017 does not necessarily accurately represent the country’s population. We also discuss the issue of Karachi’s population and the reasons why urban population is seriously under represented. Below are the edited transcripts.
BR Research: Let’s begin with your reflections on your experience with the census in Pakistan?
Mehtab Karim: My first interaction with Pakistan’s census organisation was soon after the 1981 census. The staff was well trained and experienced, because the census was being conducted every ten years. However due to the postponement of the 1991 census, things started going wrong. When they conducted the housing census in 1991, which always precedes population census, the number of reported housing units and population in Sindh had increased sharply, which was considered not reflecting the accurate picture, and therefore the population census exercise was postponed. Whether it was intentional or unintentional, but the census had become political.
Eventually the census was conducted in 1998 after pressure from multilaterals institutions. However, it wasn’t without drawbacks, for instance, they didn’t do the post enumeration survey (PES) because of which estimations of over or under count could not be determined, though according to my estimates, there was an under count of 6 million in the 1998 census. At the time of first census in 1951, 17.8 percent of West Pakistan’s population was living in Sindh. That increased to 22.6 percent in 1981 census. Thus, when census was being conducted after every ten years, the share of Sindh’s population was increasing by over 1.5 percentage points, mainly because of migration to Karachi from other provinces. Interestingly, in the 1998 census, Sindh’s share increased to 23 percent, an increase of only 0.4 percent in 17 years. Given that there was no variation in birth and death rates in the provinces, it was very strange that Sindh’s share did not increase much in 1998. The only way this could have happened was that migration to Sindh, specially Karachi had stopped or Sindh’s population was undercounted.
BRR: What was the story in Karachi back then?
MK: According to a media statement given by the then census commissioner, when he was asked that why only 9.5 million people were officially reported in Karachi in the 1998 census, whereas the estimated number was much more; he said that they had counted about 2 million more people in Karachi but they were not included in the official count, since they were illegal aliens, being Bengalis or Afghan refugees. So even in 1998 census, Karachi’s population was officially under reported.
BRR: Well, did it stop?
MK: I looked at 1998 census data, and I noted in one of my publications that migration to Sindh from other provinces was still prevalent. On the basis of questions asked about migration, of the internal or domestic migrants, 23 percent ended up in Karachi, whereas Lahore, Faisalabad, Rawalpindi and Multan, together have had half of Karachi’s share in migration. One wonders then how come the share of Sindh’s population in 1998 didn’t increase as per the trend.
Another issue was that the average household size in Sindh as reported by 1998 census went down tremendously from 6.5 persons per household in 1981 to 5.5 per household in 1998. Sindh was the only province where it went down. Was it due to decline in birth rate? There was no evidence of that. In 1997, according to Pakistan Demographic Survey the average household size in Sindh was 6.5 persons per household, whereas the 1998 census reported it was 5.5 persons. Surely something must have gone wrong.
BRR: What was the official government stance to the critique by academics such as yourself?
MK: I raised the issue many times but every time I was told that census numbers must be right because it was done in the presence of the military. It is conveniently ignored that in all the demographic surveys conducted in Pakistan after 1998, Sindh has been reporting an average of 6.5 persons per household. That’s why I have been a staunch advocate of the post enumeration survey (PES).
The same thing happened in the 2017 census, where Sindh’s average household size was reported at 5.6 persons. PBS, however, again refused to pay heed to the demands for a PES against the recommendations by the professional advisory committee consisting of senior demographers.
BRR: Why can’t the presence of military ensure flawless count?
MK: The presence of military is not a substitute for statistical methodology and due procedure. There are always risks of under or over enumeration everywhere even in countries where there is no politicization of census or otherwise a need for the army to be present.
In the 2001 census of Australia, two percent people were undercounted; in USA in 1990, there was undercount of 2 percent which now has been reduced to less than one percent. In India, there was 2.5 percent under enumeration in 2011; in Bangladesh it was 4.5 percent in 2011. If you don’t do the PES, you have no way to learn from your mistakes to improve the quality of data in the next census. Pakistan is the only country where military is involved in the census. Perhaps, military personnel could have accompanied the census enumerators only in sensitive areas. However, their presence is no guarantee that everybody will be counted.
BRR: What are some of the other things wrong with the 2017 census?
MK: There are many. For instance, a census was being conducted after 19 years but without any pilot test; the decision that each adult will show his/her National Identity Card (NIC) at the time of the census and conducting the census on de jure basis. These decisions were taken by PBS despite the fact the census advisory committee vehemently opposed it because, those people who don’t have the NIC could be missed and those who had migrated could be counted at their original place of residence. Although it was decided that everybody was to be counted, but by insisting on checking NIC might have resulted in most illegal aliens being left out, because perhaps most aliens (even those with a NIC) did not come forward and be counted, thinking that they could be apprehended especially in the presence of a soldier, who was accompanying census officials.
Secondly, there are two ways you can conduct the census; on de facto basis or de jure basis. The 2017 census was conducted on de jure basis, which was a mistake. What it means that in the census those who had migrated from one province to another and specially to urban areas were counted at their original place of residence. Besides, for the first time in Pakistan’s history of census taking, questions on migration was dropped.
This was also the first time the census was conducted in a period of two months, whereas all previous censuses were completed in fifteen days.
BRR: Do you think the fact that they checked the NICs is also the reason why Sindh’s population in general and Karachi’s in particular count is much below than what was expected?
MK: Yes, it is likely that those people who had migrated were not counted at their present residence since people don’t change the permanent address on their NICs even though they may have been living at their present residence for over one or two decades.
As a result of this, urban areas have been highly underrepresented in the latest census. Even if you take the government’s definition of urban areas and if you don’t count migration, urban areas will be under represented.
In Karachi, for example where migration has been a continuous phenomenon, population may have been undercounted due to this reason. It is rather unusual for a city like Karachi that its population growth rate is the same as national growth.
In Karachi, during the past two decades in most neighbourhoods, there has been mushrooming of multiple story buildings. However, unlike previous censuses, which conducted a housing census three to four months prior to the population census, a housing census was not conducted. Thus, most of the new high-rise buildings were not registered. For example, if a plot previously had one dwelling unit, now has a high rise building having ten or 30 residential units. I have looked at the provisional results of sub-divisions of Karachi, where multiple high-rise buildings have been constructed on small plots, which have recorded either negative or very slow growth rate of population since 1998, which means population has not been properly recorded in those sub-divisions.
BRR: That Karachi’s population has been underestimated seems to be something everyone in Karachi is sure of. But what is the most realistic or reliable estimate, that we don’t know. What are your estimate and the basis of it?
MK: PBS used to do a Pakistan Demographic Survey (PDS) every year. But in 2008 it was discontinued, so we don’t know the natural growth rate of Pakistan and its provinces. Therefore, there is no basis for any estimates for population estimates of Karachi or other major cities. During 1981-1998 censuses, population growth rate of Karachi city was 3.5 percent (it would be over 4 percent if 2 million people who were left were included), which was about one percent higher than the country’s growth rate. During 1998-2017 censuses it the same (2.4 percent) as of the country. It is highly unlikely that in Karachi both the birth rate and migration has declined substantially and death rate has remained the same. Though I think that natural growth rate (difference between birth and death rates) has remained the same, still there has been substantial migration to Karachi from KPK, Northern areas, Southern Punjab as well as interior Sindh.
Apparently, 2017 census figures suggest that people are leaving Karachi. In fact, if you look at central districts of Karachi, its annual growth rate in the last 18 years is reported to be 1 percent per annum, which means the area has reached the same level of population growth as many European countries or Iran where they have had a very successful family planning programme, but that is not true for Karachi, where like the rest of the country birth rate is still too high. Based on these factors amongst others, my own estimate for Karachi city is about 20 million populations, instead of 15 million, as reported in the 2017 census.
BRR: Shouldn’t Pakistan move towards having a census every five years instead of ten years considering that mobility has increased in the modern world. And is the cost of technology for census too prohibitive?
MK: The advisory committee had urged the PBS to use tablets in census 2017 but they ignored it. The population census of Bangladesh in 2011 was done through tablets. Egypt also used a tablet for their 2017 census.
Some countries for example Iran and Australia are already doing their census every five years. In some other countries where they don’t do it every five years, they already have a very good database system, which provides very accurate estimate of population for every year.
When doubts emerged over the reliability of the 2017 census, all political parties which are represented in the Senate signed an agreement that 5 percent validation survey should be done under the supervision by an independent population commission consisting of three demographers, before the results of 2017 census could be accepted. They said up to 2 percent error is acceptable but anything beyond is not.
The commission was notified in February 2018, but it only met once, and since then has never met again. But now it is too late to even do that validation survey. The census results cannot be made official until the Council of Common Interest approves it, and it is not likely that all provinces will agree with the census results.
BRR: Do you think the census will become a political issue once pending macroeconomic issues are settled?
MK: I think it will. The only province which has maintained its share since 1981 is Sindh, of course Punjab has reduced its share, whereas KPK and Balochistan have increased their share. We don’t have any demographic data to explain this. When we look at the fertility rate in Sindh, it is very similar to the rate in KPK and higher than Punjab. Since infant mortality rate in Sindh is lower than KPK and Punjab, the natural growth of Sindh should be very similar to KP or even higher therefore, the overall population growth rate in Sindh (due to migration from other provinces) should have been higher than reported.
These are the kind of questions that will be raised in the NFC award since the census is critical for determination of provincial quota, allocation of national assembly seats, and the NFC award. Which is why as per the constitution, the government must conduct census every ten years.
BRR: What is your estimate of Sindh’s population?
MK: I think it should be around 26 percent of the total population. I think about 8 million people were reclassified from Sindh to other provinces.
BRR: In developing countries and the developed is the census commission an autonomous and independent body ala how the Election Commission or the central bank ought to be?
MK: In Canada for example, their statistics body is independent and autonomous, and in Pakistan the PBS was envisioned to be like that. But that clearly never happened, whereas the staff at the PBS is not competent. For the first time, the PBS had members of governing council, but that governing council has no power.
Because the census was conducted after 19 years, they didn’t have qualified and experienced persons who would know the census. The person at the PBS who was running the census had background in surveys and not census, which are two different exercises altogether. There was no person who had training in planning and executing the census. And that’s why you see that the internal report that came out showed that there were lot of drawbacks in the census.
BRR: Who decides what is urban and what is not and on what basis?
MK: It is the provincial government that decides that. For example, just before the 2017 census the Punjab government announced that the entire Lahore district is urban. In the case of Karachi, 5 percent of the area of Karachi Division is still defined as rural. While the provinces are at fault for not making correct classification, the federal government should have at least advised provinces by way of policy notes to appropriately redefine their regions. The federal government should have done some kind of pilot studies to highlight which areas in various provinces are effectively urban and not rural and how getting the classification right affects public and private sector decisions.
Because they have conducted the census on de jure basis, urban areas have been underrepresented, despite advice by the UN and the advisory committee of which I was a part of. So it’s not only the definition because of which Pakistan’s official urban population is so small.
BRR: A study in early 2000s said Pakistan’s demographic window was going to end by 2040. What are the recent estimates?
MK: Total fertility rate in South Asia, which is the average number of children born per woman, was between 4 and 5 in the early 1990s. After 2000s, fertility rates started declining in all the countries of South Asia and they have all reached fertility rate of about 2.1, whereas Pakistan’s fertility rate is still 3.5. This means the demographic window, which was previously estimated to end by 2040 is likely to extend.
But while the youth bulge would remain there, unless this youth is well trained you can forget about the demographic dividends. As of 2010, only 5 percent of Pakistan’s college age youth were attending tertiary level education, which was 16 percent in India, 10 percent in Bangladesh, about 26 percent in Egypt, 38 percent in Turkey, and 78 percent in South Korea.
Secondly, the size of youth is increasing – about 65 percent of country’s population is less than 30 years old. These are going to be future parents. If they remain poor and uneducated, as is likely, high population growth will continue for another 20 years, resulting in negative impact on economic growth rate.
BRR: Who or which institution in Pakistan is thinking or ought to think about meta trends in demography, and its impact on state of society and economy?
MK: Until 1980s, PIDE used to do that, but later they decided not to do it anymore, although they still have a couple of good demographers. In 1980s, National Institute of Population Sciences (NIPS) was set up to focus on conducting research on population issues, but it hasn’t produced quality research.
Pakistan is perhaps the only country where there is no centre of demographic studies at any of its universities.
In each state in India, there is at least one university that has a centre of demographic studies. Bangladesh also has many. We don’t have a training programme for demographers. How will Pakistan deal with its demographic issues ten years later, I don’t know.
BRR: How would you rate the quality of provincial population departments especially in the context of family planning programmes?
MK: I think the provinces are now doing a good job in promoting family planning. It takes time to get results in family planning, but for the first time I am seeing an efficiently run family planning programmes by the provinces, at least in Sindh and Punjab. For instance, in Sindh they have made very wise move by giving the ministry of population and health to the same person since the delivery of family planning services has to be done by the health department. The population department neither has the infrastructure nor the capacity to deliver on family planning services.
I see a change that perhaps in the next ten years’ fertility rates would go down, if not in five years. One of the reasons why fertility rate was not going down was because places like planning commission were dominated by macroeconomists who didn’t understand the importance of demographics, and how unskilled population is a factor in retarding economic growth. But if provincial health departments take charge, as they have begun to have, then things can improve many folds.