LPG Savings due to Aadhaar is only 107–203 crore

Aadhaar savings in LPG PAHAL is a fraction of 29,769 crore claimed by the Government of India, even before considering capital and running costs

1. Overview

How much did Aadhaar-based deduplication really save on the LPG (Liquefied Petroleum Gas) “PAHAL” Scheme? Did it really save ₹29,769 crore until 2016–17, as claimed by the Government of India? Is it possible to separate out UID savings from the savings generated by the entire DBT program, using a methodology that is sound and can be independently validated? This article attempts to answer these questions using publicly available data using a methodology that considers the key feature of Aadhaar: biometric deduplication.

2. Understanding the PAHAL Scheme

Most of the domestic LPG cylinders sold to households by the Oil Marketing Companies (OMCs) have an inbuilt subsidy component. The Direct Benefit Transfer (DBT) scheme does not eliminate this subsidy component, but modifies the means by which this component is delivered to the beneficiary.

In the earlier model, beneficiaries purchased LPG cylinders at the subsidized rate, and the OMCs were paid the subsidies by the government. In the PAHAL scheme, this was modified so that all beneficiaries paid the full market price of the LPG cylinder and their individual bank accounts were credited with the subsidy directly. While initially only bank accounts were required for linking with individual LPG connections, this was eventually modified to make Aadhaar mandatory for receiving the subsidy.

The motivation behind the linking is Aadhaar’s capability to provide a unique number to every individual based on biometric deduplication, with a claimed accuracy of 99.99% based on 290 million enrollments . It was assumed that the deduplication will remove duplicate connections provided to the same holder (duplicates) and connections issued to non-existent persons (ghosts). This will, in turn, reduce the overall subsidized cylinder consumption and decrease the subsidy outflow. This reduction in offtake is referred to as ‘savings’ from UID on the LPG scheme.

3. Understanding the savings claims

There had been several different savings claims produced by the Government of India and its officials. The breakup of this figure on a yearly basis (here and here) is shown in the Table below.

Table 1 (Claimed Savings)

The methodology and the data points used for computing the above figures is available for FY 2014–15, FY 2015–16 and FY 2016–17.

Table 2: Breakup of Savings for Specific years

The difference in figures for FY 2015–16 of ₹6,912 crore and ₹6,443 crore in Table 1 and Table 2 are typical. For instance, another communique contained a multiplication error and specified a lower figure for FY 2014–15 of ₹13,466 crore (3.34 crore × 12 × 336), even though it re-iterated the actual figure of ₹14,672 crore.

Also, Indian Oil Corporation, the coordinating agency of OMCs for LPG, provided a savings figure of only ₹1036 crores for FY 2015–16 because of reduction in cylinder offtake (CAG Report, Page 47). Despite the variance, the formula to compute the savings figure is:

𝑆𝑎𝑣𝑖𝑛𝑔𝑠 = 𝐵𝑙𝑜𝑐𝑘𝑒𝑑 𝑐𝑜𝑛𝑛𝑒𝑐𝑡𝑖𝑜𝑛𝑠 × 𝐷𝑢𝑟𝑎𝑡𝑖𝑜𝑛 × 𝑆𝑢𝑏𝑠𝑖𝑑𝑖𝑠𝑒𝑑 𝑐𝑦𝑙𝑖𝑛𝑑𝑒𝑟𝑠 × 𝑆𝑢𝑏𝑠𝑖𝑑𝑦 𝑝𝑒𝑟 𝑐𝑦𝑙𝑖𝑛𝑑𝑒𝑟

Every one of the above data points used by the Government of India, in Table 2, however is inaccurate, as discussed in the subsequent sections.

3.1 Blocked Connections

In the database of LPG beneficiaries, the following classifications exist by default:

• Active beneficiaries who draw subsidized cylinders (i.e., active and subsidized).
• Active beneficiaries who purchase non-subsidized cylinders (active and non-subsidized).
• Inactive beneficiaries who do not purchase any cylinders at all and have become dormant for years (inactive).

It is obvious that any scheme or method that can de-duplicate the beneficiary database can generate savings only on the list of ‘active and subsidized’ accounts. Inactive accounts by default get blocked if they don’t order LPG cylinders in 6 months and there were 1.02 crore such connections on 2012.

Out of the rest, most of the blocked connections were due to other processes that had started from 2010 and expanded on 2012, apart from Aadhaar-based deduplication. The total number of blocked connections thus identified was 98,24,732 (CAG report on DBTL scheme, Page 26, Table 8) during the period June 2012 to October 2015 across all the Oil Marketing Companies.

This figure was further reiterated by Government of India in the Parliament (LS UQ 2459, 14.12.2015) as 1,07,89,404. We can conclude that Government of India’s claim that a total of 3.34 crore connections were blocked because of Aadhaar deduplication is inaccurate, and the figure of 1,07,89,404 constitutes the upper bound of all connections blocked through all means including Aadhaar-based deduplication.

3.2 Duration

Government of India’s calculations further assume that the connections thus blocked apply for the whole year 2014–15, though the DBTL scheme itself was only operational across the entire country from 1st January, 2015.

3.3 Subsidized cylinders

The calculations further assume that blocked connections would have utilized the entire quota of 12 cylinders available per year, while the national average is only 6.27 Cylinders (CAG Report on DBTL, Page vi, Paragraph 9.1) and is further reiterated elsewhere (CAG Report on DBTL, Page 45–46).

3.4 Subsidy per cylinder

LPG subsidy is based on under recoveries(Explanation) reported by the OMCs. The under recoveries are compensated by the Government of India in the PAHAL scheme by transferring the subsidy directly to the beneficiaries. Since the under recoveries are based on international LPG prices, it is revised by the OMCs on a monthly basis (Petroleum Planning and Analysis Cell, 2017).

Further, when using Aadhaar-based deduplication to block connections, it is only possible to block a connection if it is seeded or linked with the Aadhaar number. And only subsidy outgo on such blocked connections can be counted as savings. Also, from the country wide rollout of the PAHAL scheme from 1st January, 2015 to 31st March, 2017, Aadhaar seeding was an ongoing process and hence seeded connections were reported on a monthly basis. Thus, the number of blocked accounts also varies on a monthly basis.

Using monthly figures for both blocked connections and subsidy outgo per cylinder would provide a more accurate savings figure than simply using an average annual subsidy figure.

4. Computing Aadhaar savings

It is possible to show the impact of Aadhaar on LPG savings using the same formula as used by Government of India, but with a different set of data points.

𝑆𝑎𝑣𝑖𝑛𝑔𝑠 = 𝐵𝑙𝑜𝑐𝑘𝑒𝑑 𝑐𝑜𝑛𝑛𝑒𝑐𝑡𝑖𝑜𝑛𝑠 × 𝐷𝑢𝑟𝑎𝑡𝑖𝑜𝑛 × 𝑆𝑢𝑏𝑠𝑖𝑑𝑖𝑠𝑒𝑑 𝑐𝑦𝑙𝑖𝑛𝑑𝑒𝑟𝑠 × 𝑆𝑢𝑏𝑠𝑖𝑑𝑦 𝑝𝑒𝑟 𝑐𝑦𝑙𝑖𝑛𝑑𝑒𝑟

4.1 Blocked Connections

Blocked connections can be calculated using the formula:

Blocked connections = Seeded connections × Blocking efficiency

There are four publicly available data points that allow us to compute blocking efficiency of Aadhaar-based deduplication.

1. 45,000 duplicate connections were detected out of 50 lakh seeded accounts, and one half of them were blocked, a blocking efficiency of 0.45% (22,500 / 50L).
2. 6.18 lakh duplicate connections were detected out of 4.4 crore (44 million) seeded accounts, and one half of them were blocked, a blocking efficiency of 0.7% (3.09L / 440L)
3. 8 lakh duplicate connections were detected out of 8.08 crore seeded connections, and one half of them were blocked, a blocking efficiency of 0.5% (4L / 8.08 crore).
4. 5.84 lakh duplicate connections were detected out of 9.76 crore seeded connections (Cabinet Secretary Note, Pages 3–4), a blocking efficiency of 0.6% (5.84L / 9.76 crore) as shown in Table 3 below.

We can use the average blocking efficiency of 0.60% for our calculations, while noting that using the maximum and minimum blocking efficiency will allow us to provide a range for the savings figure.

4.2 Duration

The duration of any savings attributable to PAHAL is calculated from the point in which the program was introduced nationwide from 1st January, 2015 (CAG Report on DBTL, Page 45).

Aadhaar-seeded connections however vary significantly on a monthly basis. Hence we will be using the seeded connections every month, when available (DBT Reports, 2017) and will use linear extrapolations for months, if it is not available.

4.3 Subsidized cylinders

Since the national average is only 6.27 cylinders per year (CAG Report on DBTL Scheme, Page vi, Paragraph 9.1) and is further reiterated elsewhere (CAG Report on DBTL Scheme, Page 45–46), we will be using this figure. However, since we are doing monthly tables for computing savings, we should be using the monthly average of 6.27 / 12 months = 0.5225 / month.

Ideally if monthly consumption figures were available nationwide (they are not), it would have been more accurate to use those numbers than relying on a monthly average. However, our purpose is to compute subsidy outflow and the number of subsidized cylinders is capped to 12 / year / connection. This cap limits the effect of overestimating or underestimating the usage as the variation across months always gets smoothened out to the national average at the end of the year.

4.4 Subsidy per cylinder

We will be using the appropriate monthly under recoveries reported by the OMCs.

Typically, average numbers work very well if the upper and the lower bounds are constrained. However, we do know that the LPG prices in the international market have neither an upper nor a lower bound. Also average numbers are less accurate when monthly figures are available.

4.5 FY 2014–15

Since the scheme was only operational nationwide from 1st January, 2015 (CAG Report on DBTL Scheme, Page 45), we need to only compute savings for 3 months till 31st March, 2015. However, the number of seeded connections on 1st January, 2015 is not published by Government of India and hence we need to use linear extrapolation.

Seeded connections was at 6 crore in February, 2014 and increased to 8.49 crores on 31st March, 2015. As the previous scheme was suspended in February 2014, we assume that the enrollments stayed the same till 31st December, 2014.

Thus the connections which are then UID seeded can be shown using the linear rate.

The total savings for FY 2014–15 can then be computed using table below.

The astute reader can notice the lag effect of the seeding process. Savings computation for a given month always uses the seeded connections on the last day of the previous month. This is a side effect of linear extrapolation, and also the fact that seeded connections are only available on a monthly basis.

4.6 FY 2015–16

In FY 2015–16, seeded connections were regularly published only from September 2015. We have to use linear extrapolation for the period between 1st April, 2015 to 31st August, 2015.

Seeded connections were 8.49 crore on 31st March, 2015 and 9.58 crore on 31st September, 2015. The connections which are UID seeded during the intermediate months can be shown using the linear rate.

We should then use the subsidy per cylinder figure for the relevant period to compute the savings as shown in the table below.

4.7 FY 2016–17

Using the same methodology as outlined above, we can compute the savings as shown in the table below:

5. Methodology differences don’t exist

As the above analysis indicates, the methodology used by the Government of India and the alternate approach outlined in this article do not differ. In fact, they use the same formula for computing savings. The significant difference is the data points used and the summation of monthly savings to arrive at an annual figure.

The subtler aspects of the methodology used:

1. A connection once blocked by using Aadhaar-based duplication is accounted as a blocked connection even in the succeeding period, forever. Hence savings computed on a particular year also accounts for connections blocked during the previous years.
2. All the four data points described for computing blocking efficiency reinforce each other by pointing out that it stays within a tight range (0.45%–0.7%) and is actually closer to the midpoint (0.6%) as the percentage of seeded connections increase.
3. We therefore apply this figure to even newly provided connections to account for the scenario that without Aadhaar, for every new 1000 connections, 60 additional duplicate connections would have sprung up. The methodology used also accounts for savings that could have been generated for the counterfactual scenario.
4. It is possible to incorrectly argue that all the connections blocked by other means would have come back if Aadhaar-based deduplication would not have been present, and thus they should also be accounted as savings from Aadhaar. A simple example of why this is not valid should suffice. It is possible to apply for two connections using the Aadhaar numbers of two different individuals in the same house (Different Name, Same Address). The NIC deduplication may flag it, but it can only be validated by a door-to-door check because of the many ways in which the same address can be presented. Other means of deduplication remain necessary, and duplicates identified by them cannot be counted as Aadhaar-based deduplication.
5. The methodology is mostly similar to the one published by IISD, but it uses the latest data points (Cabinet Secretary Note, Pages 3–4), which also answers questions on the impact of UID on LPG savings and other means of deduplication such as bank accounts. Further, it outlines a method to accurately identify savings based on only UID by using the most important characteristic of Aadhaar, biometric deduplication.

6. Conclusion

It is now possible to provide a range for LPG UID savings using the various deduplication rates and the described methodology:

Clearly, the government’s claims on LPG savings because of Aadhaar linking are overstated, and this is even before considering costs. The next article in this series will delve into the costs incurred to generate this savings.