The first article last week on the census of 2017 focused on the provisional results with regard to the overall size and growth rate of the population of Pakistan. This article analyses the estimates of population in different regions of Pakistan, especially the Provinces, and the distribution of the national population between urban and rural areas.
Population distribution According to the census, the 207.8 million people of Pakistan are distributed as follows: 52.9 percent in Punjab; 23.1 percent in Sindh; 14.7 percent in Khyber Pakhtunkhwa; 5.9 percent in Balochistan; 2.4 percent in FATA and 1 percent in Islamabad, the Federal capital.
There are many surprises in these estimates. The Federal Ministry of Planning and Development and the Pakistan Bureau of Statistics (PBS), which conducted the census, had projected somewhat different shares in 2017. The share of Punjab in the national population was estimated with higher share of 54 percent, Sindh also with a larger share of 24 percent and K-PK and Balochistan both with significantly lower shares.
The Provincial Bureaus of Statistics had also made projections in their respective publication, Development Statistics. In particular, the Punjab Bureau was conservative and estimated the inter-censal growth rate at only 1.9 percent. The actual growth rate revealed by the census is 2.1 percent.
Similarly, the Balochistan Bureau projected a lower growth rate at 2.9 percent. This would have implied one million less population in 2017 than with the census estimate of growth rate of 3.3 percent.
The Sindh Bureau expected that the population of the Province was growing at a significantly faster rate of 2.8 percent than the growth rate revealed by the census of 2.4 percent. As such, the Bureau's projected population of Sindh in 2017 was 3.5 million higher than the census estimate. It is not surprising that Sindh has shown strong dissent on the provisional results, especially since in the previous census of 1988 it showed the highest growth rate. The only Bureau with a remarkably accurate projection was that of Khyber-Pakhtunkhwa, with a growth rate of 2.9 percent, the same as in the census.
The question that arises is why there is variation in the population growth rate of the Provinces from a low of 2.1 percent in Punjab to a high of 3.3 percent in Balochistan. This implies that the cumulative growth of population from 1998 to 2017 was the lowest in Punjab at 49 percent as compared to a high of over 86 percent in Balochistan.
First, there is the possibility of under-enumeration in the previous census of 1998. This is likely particularly in a province like Balochistan, which is very sparsely populated. Consequently, with perhaps better coverage in 2017 the growth rate is artificially higher.
Second, natural growth rates of population differ among Provinces. There is evidence that the birth rate and the total fertility rate are smaller in Punjab, followed by Sindh, especially in the urban areas. Resort to family planning is also more prevalent in these Provinces. However, the rate of in-migration into these Provinces is higher, thereby compensating partially for the lower underlying population growth rate.
Interestingly, although Khyber-Pakhtunkhwa has traditionally experienced significant out-migration, especially to Karachi, it has begun to receive relatively large numbers of migrants from FATA. These findings have emerged from the Labor Force Surveys of PBS, which include a migration module.
Third, the de-facto approach used in the Census may have had an impact. In particular, the disproportionate location of refugees in Khyber-Pakhtunkhwa and Balochistan has probably affected the Provincial shares of population.
The population distribution among different Provinces has significant implications on the sharing among Provinces of revenue transfers from the Federal Government as per the NFC awards, on seats in the National Assembly and on quotas of government employment. Therefore, there is a strong motivation for each Provincial Government to get higher population share from the census.
The implications of the change in population shares of Provinces due to the 2017 census are significant. First, due to the decline in share of Punjab it will receive almost 4 percent less according to the horizontal sharing formula used in the on-going 7th NFC Award. This is equivalent to a reduction in transfer of over Rs 40 billion. There is little change in the case of Sindh. However, Khyber-Pakhtunkhwa could get an extra transfer of Rs 22 billion and Balochistan an extra Rs 18 billion respectively. This will positively contribute to the process of fiscal equalization in the country with the rise in the revenue share of the two relatively less developed Provinces of Pakistan.
Second, Punjab could lose four seats in the National Assembly if the provisional census estimates are used as the basis for allocation of seats. If current trends continue Punjab will cease to have the major share of the national population within the next fifteen years.
However, there is an upside for Punjab. With a lower rate of population growth, Punjab will tend to have faster per capita income growth. Research on the change in size of the Provincial economies reveals that Punjab had the fastest growth in per capita income from 1998 to 2017 of 2.5 percent per annum, as compared to just over 2 percent in the country as a whole. Cumulatively, the real per capita income has grown by 60 percent in Punjab since 1998, 15 percent higher than in Pakistan as a whole. Also, the Province will be in a better position to expand coverage to the incremental population, given the lower rate of population growth. This will mean a relatively better quality of life.
Extent and rate of urbanization The main issue here is the approach adopted in the 2017 Census to define 'urban areas'. The pragmatic method used has been to follow the areas covered by urban local governments as specified by Provincial Governments. However, this can lead to under-coverage in two ways. First, urban metropolitan boundaries may not have been expanded to adequately reflect residential development at the urban-rural periphery. This is probably the case especially with Karachi.
Second, rural settlements may have grown in population and acquired access to basic services, thereby qualifying for treatment as part of urban areas. Interestingly, the last census in India of 2011 used the some definition as in Pakistan. This led to an estimate of the share of urban population at 31 percent. However, when towns with population size of 5000 or more were also included the share of urban population increased substantially to 47 percent.
The estimates of the share of urban population in total population of a region in the provisional census results are as follows: 36.7 percent in Punjab; 52 percent in Sindh; 18.8 percent in Khyber-Pakhtunkhwa; Balochistan , 27.6 percent and FATA at 2.8 percent. For the country as a whole, the share of urban population in the total population is 36.7 percent.
There is a surprise here also. The Ministry of Planning and Development and PBS estimated the urban share at 40.5 percent in 2017. This is almost four percentage points higher than the census. Similarly, the World Bank estimate was higher at 39.2 percent in 2016. Also, earlier estimates of the rate of urbanization were higher. The urban population growth rate ranged from 3.2 percent by the World Bank to 3.3 percent by PBS as compared to 3 percent by the census.
However, despite lower estimate of the extent of urbanization in Pakistan by the census at 36.7 percent, it is perhaps a big surprise to know that Pakistan is the most urbanized country among the relatively large countries of South Asia. The share of urban population ranges from 18.4 percent in Sri Lanka to 19 percent in Nepal, 33.1 percent in India and 35.1 percent in Bangladesh, as of 2016. Can this be attributed at least partly to the GT road which has over 60 percent of the national population living in settlements in close proximity?
We await the final results of the population census. These will either confirm the many surprises in the provisional estimates or come closer to earlier estimates made prior to the census.
(The writer is Professor Emeritus and a former Federal Minister)