Ghost savings: Understanding the fiscal impacts of India’s LPG subsidy

By Kieran Clarke, Shruti Sharma, Damon Vis-Dunbar, Kieran Clarke, Damon Vis-Dunbar on September 30, 2015

Since April 1st 2015, India’s cooking gas subsidies have been distributed solely by electronic transfer through the Direct Benefit Transfer for Liquefied Petroleum Gas scheme (otherwise known as DBTL or PAHAL).

It’s the largest unconditional cash transfer program in history. Under the DBTL, which has replaced the direct sale of cooking gas cylinders at subsidized prices, households place an order for LPG with their gas distributor, receive an amount equivalent to the current subsidy amount via electronic transfer to their bank account, then pay the full unsubsidized price for the cylinder in cash on collection or delivery. As such, DBTL does not remove the LPG subsidy, but simply changes the mechanism by which it is delivered.

Despite this, the government has maintained that the introduction of DBTL will improve the operational efficiency of the LPG subsidy system, and result in significant savings in overall subsidy expenditure. Throughout the process of introducing DBTL there has, however, been a worrying lack of official clarity regarding the scheme’s actual fiscal effects.

On July 2nd 2015, the government’s Chief Economic Advisor stated that “[b]ased on sales and subsidy levels for 2014-15, savings of Rs 12,700 crore [US $1.98bn] are estimated from [DBTL]”. This was the latest in a series of statements and media briefings by government and oil company representatives in the previous nine months on the size of fiscal savings attributable to DBTL. In October 2014, a prominent business daily reported that “[t]he National Democratic Alliance government is expecting to save Rs 10,000 crore [US $1.56bn] through cash transfer of the cooking gas subsidy”. By April 2015, the Press Trust of India was reporting that “[t]he government has so far saved Rs. 8,000 crore [US $1.25bn] due to subsidy transfer through DBT”. In early June 2015, Petroleum Minister Dharmendra Pradhan was quoted as claiming that "[t]he scheme [DBTL] has helped us identify and eliminate 40 million ghost connections,” with a senior executive at Indian Oil Corporation suggesting that "[t]he government's savings could easily be about Rs 10,000 crore linked to these 40 million connections”, leading the same business daily to conclude that “the Narendra Modi-led central government is set to save a little over Rs 10,000 crore in petroleum subsidy…in the current financial year [2015-16] thanks to the successful nationwide rollout of [DBTL]”.

The government has not so far provided calculations regarding either its claimed savings from DBTL in FY 2014-15 or projected savings for FY 2015-16. The much-publicized fiscal savings figure of Rs 12,700 crore, in particular, deserves scrutiny: based on an analysis of publically available data, this figure seems to be a large overestimate, for the following reasons.

The current government’s implementation plan for DBTL saw the program initially reintroduced (having previously been partially implemented and subsequently suspended by the earlier UPA administration) in 54 districts in mid-November 2014, with nationwide rollout to all districts from January 2015.

Households possessing LPG connections were given three months from the initial introduction of the scheme in their district to register their bank account details for direct payment, during which time they were still able to purchase subsidized cylinders within their quota. Following this period, households would then be able to purchase only non-subsidized domestic cylinders; however for a further period of three months the equivalent per cylinder subsidy would be recorded and released to the consumer upon DBTL registration. For a period of seven and a half months from April 1st 2014 until November 15th 2014, the scheme therefore had no direct effect on total subsidy expenditure. DBTL only began to formally restrict access to subsidized LPG for non DBTL-registered households in mid-February 2015, and then only in the 54 districts selected in Phase 1 (representing 8 per cent of total districts). In the remaining Phase 2 districts (constituting 92 per cent of total districts), non-registered households retained formal access to directly subsidized LPG until March 31st 2015.

Assuming program implementation on the basis formally announced, the theoretical maximum reduction in subsidy expenditure directly attributable to DBTL in FY 2014-15 can therefore be roughly calculated by applying the initial reduction in subsidized consumption (reported as 25 per cent) to the total potential consumption available to existing connections (reported as 23.3 million), then applying the relevant data on monthly under-recoveries and fiscal subsidy to calculate the estimated ‘avoided’ expenditure figure. Note that this figure will likely represent an overestimate of the total reduction in direct subsidy expenditure achieved, as it assumes that: a) all connections unregistered as of mid-February were operational; b) that all of these connections would have used their maximum allowable allocation of 12 subsidized cylinders (rather than the national average of 6-7 cylinders) on a monthly pro-rata basis; and, c) that none of these connections would subsequently register with DBTL and receive subsidy compensation for the period from mid-February to end March. Importantly, this figure represents total ‘avoided’ subsidy expenditure, and does not take account of the profile of those connections losing subsidized access (including the percentage of total ‘avoided’ expenditure related to previously valid household consumption). Calculated on this basis, the maximum potential saving on direct subsidy expenditure from introducing DBTL was approximately Rs. 46.9 crore (US$ 7.3m) in February 2015 and Rs. 96.4 crore (US$ 15.1m) in March 2015, leading to a maximum potential saving from restricting access of Rs 143.4 crore (US $22.4m) for the financial year—approximately Rs. 12,557 crore (US $1.96bn) less than the government’s most recent stated estimate.

In short, regardless of the potential impact of introducing DBTL on total subsidised LPG consumption, it would be very difficult for the scheme to have delivered significant savings in total subsidy expenditure in FY 2014-15, as the vast majority of connected households retained formal access to subsidised LPG under the previous subsidy distribution mechanism.

In addition, any calculation of savings resulting from DBTL implementation in FY 2014-15 should also include the initial and recurring costs of introducing the program to the government (including central and district-level administrations, oil marketing companies and public sector banks), program beneficiaries, and the wider economy. While some of these costs are widely distributed and challenging to record precisely, information on key areas of operational expenditure—including commission on subsidy payments and advance payments to beneficiaries—is readily available.

For example, the cost of commission on all subsidy payments made through direct transfer (initially set at 1 per cent of total transaction size and proposed to be raised to 2 per cent—a rate that is nevertheless estimated to remain un-remunerative) was in itself largely sufficient to offset any direct saving on subsidy expenditure in FY 2014-15. In the case of the ‘permanent advance’ payments provided to all newly-registered DBTL customers upon initial cylinder booking—the costs of which are currently borne by the public sector oil marketing companies—households that had previously registered under the initial direct transfer scheme in FY 2013-14 received an advance of Rs. 435 per connection, constituting a total payment of Rs. 1,469 crore (US $229.5m) in FY 2013/14. Under the revised scheme introduced by the NDA administration, this permanent advance was increased to Rs. 568 for the period to end March 2015, with payments provided as permanent advance in FY 2014-15 alone amounting to Rs. 5,234 crore (US $817.8m), according to reliable sources.

These savings and expenditure estimates, based on publically available data, suggest that the government’s figures for net fiscal savings resulting from DBTL are likely to be incorrect—even without accounting for the various administrative costs of program implementation – and may even have been negative. In order to have an informed discussion regarding the impact of DBTL, both as a program itself and as a proposed model for the reform of kerosene and food subsidies, it is crucial that the government provides detailed and accurate data on the way that subsidy savings are calculated.

Full calculations for all figures provided are available on request.