Aadhaar savings numbers for the perplexed

Numbers might be hard to crunch, but even reputed financial journalists and publications find Aadhaar savings numbers hard to understand because of the notorious “lemons problem”.

This guide is written to redress the information asymmetry so that readers can understand the savings numbers put out by the government objectively without relying on intermediaries (including this writer).

1. All databases have inactive users

The absolute number of users in any database is a meaningless statistic. What really matters is the active count. Internet businesses report a metric called Monthly Active Users (MAU) and not Total Users in the database. The active user metric is important for social welfare schemes because it impacts fund outflow from the consolidated fund of India.

For instance take LPG, PDS or NREGA database. They might have 100 crore database entries but only 20 crore active accounts. Only active accounts affect fund outflow.

2. Inactive users get suspended after some time

When a beneficiary doesn’t avail the service for a certain period (6 to 12 months), their account gets suspended and is marked as inactive.

3. Few schemes have auto-expiration of entries and aliveness checks

4. You don’t need Aadhaar to eliminate inactive entries in a beneficiary database

Inactive entries are identified by non-drawal or non-renewal and can be suspended and eventually eliminated, they are typically discounted from calculating Aadhaar savings, since their contribution to the outflow from consolidated fund of india is Zero.

5. Aadhaar only helps identifying potential duplicates and ghosts

Duplicates are those entries in the DB which have the same Aadhaar number, but they are only potential duplicates as we cannot discount manual errors caused by the operator or by the inorganic seeding tool. (Discussed here.)

If a person in a database cannot be determined to actually exist (by attaching their Aadhaar number), they are labeled a “ghost”. But one cannot truly differentiate between a person who does not have Aadhaar and a ghost entry. Any ghosts or duplicate entries that are found in a DB are only “potential” and need further investigation to ascertain if they actually do not exist. Otherwise this only causes intentional exclusion.

6. True success is when duplicates and ghosts are deleted after verification

Potential duplicates and ghosts needs to be investigated through other means such as door to door checks, social audits and when strong evidence is available that they were indeed fraudulent, are removed from the beneficiary database.

Hence only confirmed duplicates detected through Aadhaar and who are then deleted from the beneficiary database can be counted towards the “Aadhaar effect”.

The ratio of such confirmed deletions against the beneficiary database size is called deduplication rate.

7. What are the confirmed deduplication rates of Aadhaar across schemes?

On LPG (PAHAL) it is 0.6%. (Source: Cabinet Secretary Note)

On NREGA, it is 0.2% (Source: MediaNama).

On PDS, there are two data points, 0.32% in Orissa and 1.8% in AP (see #10 below).

8. Did not Aadhaar delete 3.56 crore ghost LPG connections?

No, it did not. The number was an epic exaggeration, as it included deleted connections from 2012 (when the PAHAL scheme was not active), inactive accounts which were suspended and were not drawing any LPG cylinders, and connections deleted through other means. This is also reiterated by the CAG report (on page 46).

Relevant Passage in the CAG Audit.

9. Did not Aadhaar delete 1 crore NREGA job cards?

No, it did not. As per the RTI reply to Jean Dreze, the total number of deleted job cards is 94,71,350 for various reasons. Fraudulent and fake cards were 12,06,559. So how many of these were deleted because of Aadhaar seeding?

The answer as per Business Standard is zero, because all these 9.47 million (94.71 lakhs) job cards were deleted without the use of Aadhaar:

This is consistent with UIDAI’s own study on NREGA deduplication, which had 67,637 duplicates on 74.72 lakh seeded accounts (almost all of it coming from Tripura).

The number is also very consistent with LS UQ 1826 on 05.03.2015, where 63,943 duplicates were removed and eliminated (but curiously did not specify the total number of seeded cards).

10. Did not Aadhaar delete 2.33 crore fake ration cards?

The source for this 2.33 crore deleted ration cards figure is from LS SQ 93 on 22.11.2016.

Deleted Ration Cards

Let us expand the caveats described in the footnotes in this image.

  • These are cards that were deleted for a variety of reasons (not just Aadhaar seeding), such as ineligibility, migration and death from 2013 onwards.
  • Of the 2.33 crore deleted cards, 66 lakhs are from West Bengal which — unlike other states — issues individual ration cards to every person, while others states issue one ration card per household. Hence mixing these two distinct categories paints an inaccurate picture of deleted cards.

Exactly how many of these were deleted using Aadhaar seeding? We can use Aadhaar seeding of ration cards from LS UQ 4289 on 28.03.2017.

Seeding Ratio of PDS Cards

When seeding figures are just 76.83%, how is it possible for all the deleted ration cards to be because of Aadhaar? Assam had 0% seeding and yet 72,746 cards were deleted. That means methods not involving Aadhaar were also used.

Further LS UQ 2756 on 10.05.2016 also provided the total number of seeded cards.

Seeded Ration Cards

Bihar had only 9,000 seeded cards on 2016. However, the total number of deleted cards in 2014 and 2015 was 38,113. That means most of the deletions were not based on Aadhaar.

More data is needed to assert the above for other states. Hindustan Times filed an RTI request and published their analysis of the response.

A few key findings from this article:

  • Andhra Pradesh said 8.54 lakh BPL cards were removed after Aadhaar seeding, out of a total 9.67 lakh deleted cards. It had 4.71 crore PDS cards in August 2014 (source), making this a deduplication rate of 1.8%.
  • Tamil Nadu said that 547,000 bogus cards have been eliminated through 100% door-to-door verification, which does not match the 3,70,727 deleted cards figure specified in the Parliament answer.
  • Andaman and Nicobar Islands responded that after the introduction of e-PDS, the Directorate of Civil Supplies and Consumer Affairs found 7,004 inoperative cards. No cards have been declared as bogus.
  • Haryana’s response is perplexing because it had 25.35 lakh seeded cards out of a total of 29.41 lakh cards and 1.92 lakh deleted cards.
Haryana said it had 283,635 ration cards, “and 1,149,988 beneficiaries having duplicate/repeated Aadhaar numbers in the ePDS. Therefore no information regarding bogus ration cards is available as per record available in PDS data base.”

Orissa also published data on the contribution of Aadhaar towards removal of PDS cards:

Only 27,000 out of 6.6 lakh deleted cards were because of Aadhaar. The total number of PDS cards in Orissa in 2014 was 85.39 lakhs (source). We can now surmise the following:

  • Aadhaar’s fraud detection rate is only 0.32% (27,000 / 85.39 lakh cards).
  • Fraud detection by other means is 7.41% (6.33 L / 85.39 lakh cards).
  • Fraud detection without Aadhaar is 27 times (7.41 / 0.32) better than with Aadhaar.

11. If Andhra had only 8.54 lakh BPL cards removed in 2017, why did the Union claim to the Supreme Court in 2015 that it had 69 lakh duplicate cards?

Most of PDS savings that is claimed by the Union can be thought of as spurious numbers, generated to convince the gullible about Aadhaar’s utility. The SC Affidavit anomaly has been extensively dissected here:

12. But surely it is a tall claim that the government is putting out spurious numbers?

Exhibit 1:

Exhibit 2:

Before 2015, deleted cards were not counted as savings by Aadhaar, but 2015 onwards, even previous year deletions were counted towards it.

Exhibit 3:

The total number of ration cards in Rajasthan is only 1 crore as per LS UQ 4289 on 28.03.2017. Hence, according to The Financial Express, Rajasthan deleted every ration card in it’s state, 2.33 times!

Exhibit 4:

Crores became lakhs within two months.

13. Andhra did remove 1.8% of the ration cards using Aadhaar, so does that not count for something?

(a) Duplicates by themselves does not mean the PDS card is not genuine, since seeding errors are quite common.

(b) Not uploading the photo of the beneficiary resulted in removal of the beneficiary from the ration card list.

(c) AP government’s own social audit showed that genuine beneficiaries were struck off rolls because their cards were not seeded, or because of mismatches such as in the name and age.

(d) The cost of setting up an Aadhaar ePOS machine is 67,000 rupees (192 crores for 28,599 shops, on page 18 of the above document).

The total expenditure involved in the implementation of FPS Automation and installation of 28,599 ePoS Devices — cum — Electronic Weighing machines and 28,599 Iris readers is around Rs.192 crore.

And a reduction in PDS grain outflow is equated as “Savings”, even though subsequent social audits revealed that in fact, exclusion is shown as savings.

Sale of PDS commodities to BPL beneficiaries has declined from 100% to 82–84%

If Andhra is the best example for Aadhaar savings in PDS, then there exists no savings in PDS.