Monday May 23, 2022

Reversing Ehsaas data analytics

April 30, 2022

The government has signaled a policy intent to reverse a decision, which was taken in 2019, under the Ehsaas Governance and Integrity Policy through which we removed 850,126 undeserving people from the list of BISP beneficiaries. This reversal, which entails watering down a data analytics-driven rule-based process, is a regressive step and should be reconsidered.

Within this context, the purpose of this comment is to outline the context in which the 2019 decision was taken, reasons for the policy choice, the evidence behind it, the policy process, methodology adopted to implement this policy, and imperatives for the way forward.

In terms of the context, back in late 2018 when I joined, BISP was giving stipends based on a 10-year-old paper-based survey. The survey had many inclusion and exclusion errors. It was common knowledge that in addition to many deserving beneficiaries, pensioners, the relatively well-off, and many government servants were also benefiting. In any case, the economic situation of families evolves over time and many that may have been deserving back in 2011 may not have qualified in 2019.

We therefore decided to adopt a rule-based wealth-profiling data analytics process without space for human intervention, to clean BISP’s lists before including beneficiaries in Ehsaas. For this purpose, we used Pakistan’s ability to triangulate varied personal information using the CNIC as the peg. This capability existed for some time but was used for the first time by Ehsaas. A number of wealth proxies developed in consultation with NADRA were used as exclusion filters; these included international travel, ownership of a car (motorcycle owners were not excluded), average six-month telephone bill above Rs1000 a month (landline and mobile phone), expensive processing of passports and national identity card numbers by three or more members of the family through Executive Centres, and government employment. These wealth proxies were approved by the cabinet.

We ran these wealth proxies as exclusion filters and found out that there were 820,165 unique undeserving individuals, of which 44,438 owned cars, 525,461 had been traveling internationally, 140,313 had their six-monthly average phone bill above Rs1000; 38,216 had processed their passports and ID cards in expensive executive centers and 142,556 were government servants. Amongst government servants, 2,543 were grade 17 or above, which was shocking. These individuals did not qualify for social protection. Around five million beneficiaries were included in Ehsaas after cleaning BISP’s lists. Disciplinary action was taken and all BISP staff members whose wives had been BISP beneficiaries were dismissed from service.

In 2019, we did not have data for autonomous agencies and pensioners. After repeated requests we received partial data from these sources in 2021, based on which 29,961 employees of autonomous agencies were removed from lists. Hence in total, 850,126 were exited from the programme. It is important to appreciate that we did not include any new beneficiaries into Ehsaas in lieu of the 850,126 exited back in 2019. It was only in 2020 when data from the end-to-end digital National Socio-Economic Survey started trickling in that began adding more beneficiaries in an objective, rule-based manner and totally apolitical manner.

Two more exclusion filters were added when we ran Ehsaas Emergency Cash: income level above Rs50,000 declared by the Federal Board of Revenue; and ownership of land; the latter was only for Punjab, where digitized data existed. All the wealth proxies were discussed at length at the Design Committee and BISP Board, ECC, and cabinet levels and were approved by all. In addition, there was unanimous provincial consensus garnered through the National Coordination Committee and the National Command Operation Centre.

The wealth proxies were also validated by the Household Income and Expenditure Data (Pakistan Bureau of Statistics 2017), showing that international travel, ownership of a car, and a six-monthly average phone bill above Rs1000 implied minimal risk of excluding beneficiaries in the poorest two income quintiles. Hence there was solid evidence behind this decision and due process had also been followed.

It is important to note that the budget available for government social protection benefits does not cover the entire “poor population below the official poverty line”; hence objective criteria have to be adopted for targeting benefits and currently this includes both data from the survey as well as wealth proxies.

The government plans to open a window of appeals for those that were excluded on the basis of wealth profiling. This will further discretion. Appeals should be reserved to overcome shortcomings of the survey instrument in identifying widows, single women, and homeless abandoned elderly individuals, as we had already planned.

As for regular recipients of benefits, the Ehsaas Dynamic Socio-Economic Registry -- which has been deployed -- will ensure revision of eligible lists every three years. Wealth proxies are a double check and were introduced to overcome shortcomings of the self-reporting in the survey; together the survey and wealth profiling make the selection of beneficiaries far more robust. A discretionary window should not be opened to strike down wealth profiling. The government should instead focus on further strengthening it.

When decisions have to be made about millions of beneficiaries, it is important to use water-tight rule based objective criteria and accept their shortcomings rather than open the door for discretion, where the risk of political manipulation lurks and which inevitably results in abuse.

Ehsaas introduced far-reaching changes to promote integrity and operated in an apolitical manner; this was also evidenced in the distribution of Ehsaas Emergency Cash where 31 percent of assistance went to Sindh. The Ehsaas Building and Rebuilding Institutions Report provides details about the wide-ranging changes in BISP (which is just one of the several implementing agencies of Ehsaas) to promote integrity. Data analytics was just one facet of the many changes introduced. The new government can build further on this work to target funds and opportunities on the basis of need and merit rather than inadvertently opening to door for political influence.

The writer is a former SAPM and former chairperson BISP. She tweets @Sanianishtar


    Ayesha Zaigham commented 3 weeks ago

    It sounds like a well thought out and executed exclusion. Dare I hope that the new government will reconsider !?

    23 0

    Humayun Ghayasuddin commented 3 weeks ago

    A succinct and straightforward analysis of an econometric analysis clearly highlighting the rationale for the exclusion methodology. We pray the new government does not get enough to make their changes.

    1 0

    Zahra Ali commented 3 weeks ago

    Thank you for all the hard and genuine work. Data collection and making a plan according is the best way to build upon something and you and your team did that. Hope it can be implemented again soon.

    1 0