The PTI’s inability to read statistics

Party Chairman Imran Khan has claimed in a recent speech that his party’s government in KP has reduced poverty in the province by half in only two years.

The claim was a repetition of an earlier tweet and a presser in Islamabad before that. The figure has also been flaunted by KP’s Chief Minister and other senior ministers of the government in the province. Shah Farman, KP’s information minister, did a dedicated press conference on 10^thJuly, 2017 where he announced this ‘massive reduction’ in poverty. Sadly, the claims are based on the faulty analysis of an article published in The News on 7^th July, 2017.

Yasir Cheema, the writer of the article and a PTI activist, made his own conclusions in the article titled ‘Behind The numbers’ by using statistics from two summary reports of Pakistan Bureau of Statistics (PBS) for the year 2013-14 & 2015-16.

The author did a comparative analysis of the population share in five different quantiles (there are 5 quintiles based on income with poorest population in the lowest quintile and richest in the highest quintile) across the two time periods for the four provinces. In his conclusions, the author found, based on PBS summary reports, which population in the poorest quintile in KP reduced from 23.86 % in 2013-14 to 12.27% in 2015-16. The author further conclude that population share in the poorest quintile has increased in Punjab and Sindh while falling slightly in Balochistan. The author, sadly, did not really go ‘behind the numbers’ and used statistics from the summary reports for his final analysis.

To put things in context, the PBS disseminates its datasets online through three different forms. First, it publishes a summary descriptive report, which the author used for his analysis, with some selected concise data tables to show the trends across provinces in various socio-economic indicators. Second, it publishes a list of 25 tables, which gives detailed statistics on various socio-economic aspects of the data set with separate data tables for provinces and Rural/Urban divide. Finally, uploads the micro data sets of all surveyed households/individuals in STATA & SPSS statistical software format The last one is a great initiative from PBS as previously the micro data sets were available only on demand and with some costs attached to it.

First pointed out by Shahid Saeed (@niharifan) on twitter, the author of the article used the data as reported in the summary report of PBS, which does not corresponded with the detail tables published by the same PBS. A friend and former colleague, Dr Adnan Haider, further investigated and found that the population in five different quintiles, as reported in the detail tables, do correspond with the micro data set of the PBS while the summary reports contain a major compilation error.

To explain this, the readers are directed to the two tables in this article. Table A shows the quintile distribution of population for the two time periods across four provinces as reported in the PBS summary reports. Table B, on the other hand, shows the quintile distribution of population across four provinces as reported in PBS detailed statistical tables and microdata sets. One can clearly see that, for the year 2013-14, PBS staff mistakenly put KP’s figures in Punjab, Punjab’s figures in Sindh and Sindh’s figures in KP while Balochistan figures are at the same place in both summary reports and detail tables. This findings also corresponds when one calculates these different quintiles from the microdata sets as provided by PBS and can be downloaded from the PBS website.

The results of these surveys show that the KP share of population in the first quintile is reduced from 15.43% in 2013-14 to a 12.27% in 2015-16. This is a reduction of 3.16 percentage point as opposed to Yasir Cheema’s reduction of 11.59 percentage points from 23.86 % in 2013-14 to 12.27% in 2015-16.

The final analysis shows that over the two time periods, share of population in the poorest quintile marginally increased in Punjab while slightly reduced in Sindh, KP and Balochistan. The reduction in KP is, however, significantly better than Sindh & Balochistan. When one really ‘goes behind the numbers’, it is clear that the reduction in KP is due to two reasons. One, the number of earners in the poorest quintile in KP rose from 2.37 earners per household in 2013-14 to 2.72 earners in 2015-16. Only detailed analysis can reveal the real factors as this could even mean more children put to work due to economic hardships. Second, the poorest quintiles in KP received more income as transfers in the forms of Gifts and Assistance in 2015-16 than in 2013-14. In 2015-16, for example, the poorest quintile in KP received 8.47% of their income from Gifts and Assistance while in 2013-14 it was half this amount at 4.91%. Gifts and assistance mainly includes BISP, Zakat and other charities. Both these indictors have nothing to do with policies of the KP government, which the PTI leadership is boasting about. A more careful and detailed analysis can find more caveats of these trends.

Poverty is a complex phenomenon and it need serious research to correctly diagnose its different causes so that the right kind of policies and programs can be initiated. Poverty reporting is thus an important & serious task and claims on the basis of faulty analysis harm the efforts to tackle it.

The PBS needs to be extra vigilant in reporting its data and minimize the human errors as the data collection methodology employed by PBS is otherwise very robust & rigorous. Politicians also need to be careful in accepting the data reported in the newspapers. All provinces have their own Planning & Development Departments and they are capable enough of producing provincial and district-wise poverty figures. These authentic and correct poverty figures for districts can serve to achieve an equitable distribution of resources among different provinces/districts. Finally, it is Federal Government’s responsibility to officially report poverty figures for each year with provincial, regional and district wise divisions so that to avoid such confusions in the future. Sadly, reporting poverty numbers in Pakistan is marred by controversy and only last year official poverty figures were announced after a gap of 10 years.