In India, power distribution companies (discoms) play the crucial role of providing last-mile connectivity and power supply to consumers. However, they are also the weakest link in the electricity supply chain due to their poor financial health (Garg 2020). Discoms’ financial turnaround is fundamental to achieving multiple policy objectives, including the provision of uninterrupted 24×7 electricity supply to all consumers and the country’s clean energy transition (PIB 2024).

The aggregate technical and commercial (AT&C) losses in public discoms in India declined from 25.21 per cent in FY 2014–15 to 16.37 per cent in FY 2023–24 (PFC 2025).2 This followed several policy reforms and infrastructure upgrades. Yet only Kerala and Gujarat have discom losses closer to the world average of ~6 per cent, indicating significant scope for continued improvement across discoms in all states (Figure 1) (IEA 2025).

Despite previous policy interventions, discoms’ finances have often worsened after short recovery periods (A et al. 2020; Verma et al. 2020). Even after the Financial Restructuring Plan (2012) and Ujwal DISCOM Assurance Yojana (2016–20), the accumulated losses of the public discoms increased by ~50 per cent from INR 4.1 lakh crore (USD 49.5 billion) to INR 7.1 lakh crore (USD 85.3 billion) between FY 2014–15 and FY 2023–24, highlighting structural inefficiencies that remain unresolved (PFC 2018, 2025).3 These inefficiencies – such as outdated and inefficient infrastructure, poor billing quality and electricity theft, rising operational cost with reduced revenue generation, the inability to invest in infrastructure upgradation, and low technology adoption – contribute to persistently high distribution losses (FOR 2025). These losses lead to revenue shortfalls and mounting financial losses for public discoms in India (Agrawal et al. 2023; Swain 2016).

To address these gaps, in 2021, the Government of India introduced the Revamped Distribution Sector Scheme (RDSS) with the goal of reducing discoms’ AT&C losses to 12–15% by 2024–25. The scheme period has recently been extended to March 2028 (Ministry of Power 2025). The scheme has a provision for an outlay of INR 3 lakh crore (USD 34 billion) – a mix of grants and debt – with a significant contribution towards smart metering consumer connections, besides the metering of distribution assets (transformers and feeders) (PIB 2023). As of June 2025, 224 million smart meters have been sanctioned and 32 million installed across India (NSGM n.d.). As the state and its discoms begin to make large investments in the sector, India requires new strategies to address the existing on-ground challenges and the diverse factors affecting distribution losses.

We undertook this study to investigate the factors and mechanisms that underpin distribution losses in the electricity sector and to generate insights on systemic measures that can help discoms plug their energy and revenue leakages. We adopted a mixed-methods approach for an immersive research-cum-intervention pilot in a specific high-loss electricity feeder of a public discom in Uttar Pradesh (UP). We carried out an energy auditing exercise in the specific feeder and explored the following questions:

• What is the composition of distribution losses in terms of the share of observed technical and commercial losses?
• What kind of local and systemic factors influence the distribution losses, and through what mechanisms?
• What is the current status of revenue recovery and arrears in the feeder, and what are the potential factors contributing to the accumulation of arrears?
• What interventions could help reduce losses, and how could smart metering infrastructure be leveraged effectively?

This approach allowed us to develop a deep understanding of the local and systemic factors responsible for the concerned feeder’s inefficiencies, which we draw on to suggest specific and systemic measures to plug these gaps. Most of the study insights are specific to the chosen feeder, as no single feeder can represent a discom or state’s entire distribution sector. So, we apply inductive reasoning to generalise our observations and suggest interventions for discoms across states that are facing similar challenges.

Distribution loss in an electricity feeder reflects the amount of energy supplied that is not billed.15 Our analysis indicates an average monthly distribution loss of 2.25 lakh units (42 per cent) in the feeder between November 2021 and October 2022 (Figure 7). Thus, four out of every ten units supplied to the feeder were a lost commercial opportunity. The feeder has around three times the normative distribution losses of 12–15 per cent targeted under the RDSS.

Notes:

• Distribution loss is computed as 1 minus the ratio of total billed units (including unmetered connections) and feeder energy input.
• The total billed unit, including unmetered connections, captures the normative consumption for domestic, commercial, and tube well unmetered connections, as defined by the state regulator. However, total billed units excluding unmetered connections record zero consumption against unmetered consumers, as recorded in the billing data.
• We have represented the data through a stacked area graph for simplicity in visualising the key trends.

The monthly variation in distribution losses was in the range of 15–58 per cent, with higher losses during higher-demand months. When we began this study in January 2022, monthly distribution losses were 53 per cent, and they remained above 48 per cent until May 2022.17 However, average distribution losses during June–October 2022 were 35 per cent, nearly a third lower than the average observed during January–May 2022. This can be partly attributed to the discom’s efforts to address the metering and billing problems identified during the study (discussed in Section 2.6).

In the next subsection, we discuss the composition of the distribution losses in the feeder and examine specific issues that directly or indirectly affect the feeder’s losses and their accurate estimation.

Technical losses

Using the feeder and DT meter data, we estimate technical losses of ~7 per cent (of the feeder input energy) at HT levels (11 kV) or in the primary distribution network. 18,19 This is significantly higher than the 1.5–2.5 per cent prescribed by the Central Electricity Authority (CEA 2023). These high technical losses are due to ageing conductors, the longer length of the feeder (~21 km)20, the presence of agricultural, that is, high inductive loads, and the overloading of DTs and lines during the high-demand months. A low ratio between the lengths of the HT and LT lines also results in low operating voltages and high technical losses. The feeder’s estimated HT–LT line length ratio is 0.63.21 This is close to the all-India HT–LT line length ratio, which was 0.62 in March 2022, but is much lower than the prescribed unity plus levels—that is, where HT line length is at least equal to or greater than LT line length (Aggarwal 2024; CEA 2023).

Further, technical losses in the LT network are typically higher than HT line losses, implying total technical losses of at least 14 per cent in the feeder. Technical losses in the LT network (below 11 kV) are typically higher than in HT lines of the feeder due to factors such as lower voltage levels, longer network length, inadequate conductor sizes, the presence of overloaded DTs and lines, single-phase LT lines, and imbalanced three-phase loading.

In the concerned feeder, we observed the following contributors to high technical losses:

• Four out of 11 mixed DTs were loaded beyond 100 per cent during the peak demand month (May 2022).22 In addition, two DTs had a maximum loading of ~150 per cent.
• Two-thirds of the DTs displayed high imbalance during May 2022, even though their load factor was in the acceptable range. High DT imbalance increases technical losses, reduces voltage quality, and negatively impacts a DT’s lifespan.
• There were inductive loads, such as a flour mill and 48 tube wells – a cumulative load of 333 HP, or ~248 kW.
• Although capacitor banks were installed at the substation for reactive power compensation, they were not being used.

In the absence of time-block-wise data on DT loading, we could not assess the specific times of the day when DTs were overloaded and the source of the excess loads.

Non-technical (commercial) losses

Computed as the difference between total distribution and technical losses, about 28 per cent of the energy supplied to the feeder during April–May 2022 was lost to inefficiencies in metering and billing as well as electricity theft. The notable presence of non-technical distribution losses is illustrated in Figure 7, where the seasonality in feeder energy input is not reflected in the overall billed units. This was particularly true in the pre-intervention peak-demand months of January and May 2022. Unpacking these non-technical losses is an important step in addressing them.

In the feeder, as in the rest of the state, the meter reading is contracted out to third-party billing agencies that hire members of the local community as meter readers to conduct monthly spot billing.24 Meter readers are critical agents in ensuring timely and correct billing for consumers, as well as in highlighting and addressing any metering issues. We observed several gaps in metering and billing in the feeder.

Metering gaps

As per the physical verification by the survey team, around 360 consumers (14 per cent) served by the feeder were unmetered or had dysfunctional meters. While the discom’s billing data for April 2022 recorded only 39 unmetered consumers, our survey identified 282 unmetered connections, 96 per cent of which were for domestic consumers.25 Additionally, 75 consumers reported dysfunctional meters. Metering gaps directly feed into billing inefficiencies.

Our field staff highlighted several reasons for this discrepancy and the presence of unmetered and dysfunctional meters.

• Meter readers generated bills for unmetered consumers after wrongly marking their meters as identified defective, or they generated bills on MU basis, as their billing application did not allow the billing of unmetered consumers.
• In cases where meters with a defective display were taken away, they were not replaced even after several months.
• Replacement meters were unavailable when required in local discom division stores, which prioritised the allocation of new meters for new connections.
• In some localities, consumers deliberately burnt their meters and did not allow the discom staff to replace them.

Billing inefficiencies

We observed several gaps in billing in the billing data. These included a significant number of cases of provisional billing26 or billing for defective and unmetered connections (Figure 8). Consequently, billing based on actual consumption data, or MU billing, was typically low, averaging 85 per cent in 2022.

We estimate that 4.7 per cent of distribution losses (or 7.1 per cent of non-technical losses) could be attributed to metering gaps (Table 6).

Table 6. Around 5% of the distribution losses were due to metering gaps

Source: Authors’ analysis based on the discom’s billing data

Note: For this analysis, we used the master data for the count of metering gaps, which is significantly lower than the number of such cases as per the survey data. These estimates are therefore somewhat conservative.

Our field work revealed further gaps in billing; we found that meter readers play a key role in table-billing and under-billing. Table billing happens when meter readers generate consumers’ bills without visiting their premises to record the actual meter readings. With a monthly target of billing 2,500 consumers within 15 days, meter readers face considerable pressure. As their commission is based on the number of consumers billed, this creates incentives to either reduce costs or maximise their income.

In the feeder, only ~60 per cent of consumers’ bills were generated by meter readers who were officially allotted the feeder for billing purposes. The remaining 40 per cent of consumers were being billed by meter readers from outside the feeder, implying that bills were generated without physical visits to their premises. Table 7 shows that this trend is highest in May and June (the peak summer months) and in August (the peak monsoon month). This could happen because all meter readers in a sub division had full access to the consumer list (billing book) in the sub division. At the same time, the discoms’ billing data from this period shows that ~80 per cent of such bills were recorded as generated on MU basis. This indicates a lack of adequate checks and balances in the billing software and billing application used by the meter readers, which would have otherwise prevented such faulty billing on such a large scale.

Table 7 also shows the difference in average units between the consumers billed by the feeder meter readers and those billed by outside readers.28 We estimate average under-billing of ~49 MWh per month in the feeder. This is equivalent to 22 per cent of the feeder’s distribution losses and around one-third of the non-technical losses. The under-billing estimates are highest for May and June 2022 and lowest in September 2022. Across months, ~90 per cent of the consumers billed by meter readers from outside the feeder were domestic, highlighting that under-billing is disproportionately prominent in this category.

Table 7. Consumers billed by meter readers from outside the feeder had lower electricity bills than those billed by officially assigned readers

Consumer survey data on billing frequency in the year prior to the survey also validate these observations. Only ~40 per cent of consumers in the feeder reported having received bills every month; 45 per cent reported receiving bills irregularly or a few times during the year; and ~8 per cent mentioned never having received an electricity bill the previous year. The share of table billing declines slightly after June (barring the rainy month of August). This could be partly attributed to better weather and increased vigilance by discom staff, who were apprised of this issue in June 2022, after which they took action.

However, table billing was not limited to meter readers residing outside the feeder. After our on-ground interactions, we found that the meter readers who were formally allotted for billing may have also been practising table billing. There are two modes for spot bill generation. One is the online mode, where the bill generated is directly communicated and updated in the billing system. The other is the offline mode, used to avoid billing disruptions due to the unavailability of mobile networks. However, the meter readers were allegedly misusing the offline mode to carry out table billing. In some instances, consumers also colluded with the meter readers to generate lower unit bills in exchange for facilitation fees. From the discom data, we found that the cases of offline billing of consumers increased from 7–12 per cent during April–July 2022 to 32 per cent in August and ~70 per cent in October 2022. Still, it is challenging to distinguish genuine offline billing cases from table billing. The high incidence of offline billing raises valid concerns about potential revenue loss.

Presence of electricity theft/unauthorised connections and a poor distribution network 

While conducting a door-to-door consumer survey, our enumerators noted about 70 cases of electricity theft in broad daylight through katiya and meter bypassing. The discom staff mentioned that such cases of electricity theft surge during night-time. Such a high prevalence of electricity theft (by ~3 per cent of consumers) is aided by the presence of bare conductors near consumer premises. The subdivision officials also shared that neighbourhoods with low aerial bunched (AB) cabling had a high prevalence of theft, and that the discom is undertaking several measures, including ramping up vigilance drives, to curb electricity theft in these localities. During the consumer surveys, our team also found instances of connections that were disconnected on a temporary or a permanent basis in past vigilance exercises.

Factor-wise contribution to feeder’s losses

Based on our empirical research, we conclude that commercial factors, such as gaps in metering, under-billing, and electricity theft, are the predominant drivers of high distribution losses in the concerned feeder. Table 8 shows our estimated break-up of distribution losses across key technical and commercial factors, as well as their share as feeder input energy. Due to the lack of robust, end-to-end metering data, our estimates rely on several assumptions (as discussed earlier) and should be used as broad indicators of key issues requiring the discom’s attention.

Table 8. Commercial factors emerged as the key drivers of distribution losses in the feeder

Using the discom’s administrative data and primary data from the surveys and field observations, we attempt to explain the composition of distribution losses in a specific high-loss feeder in UP. Technical and non-technical losses comprised around one-third and two-thirds of the feeder losses, respectively, on account of diverse network-related planning, execution, governance, and socio-political issues.

At 42 per cent in 2022, the distribution losses in the studied feeder are considerably higher than the state-wide distribution losses in UP, which are at 16.39 per cent (FY 2023–24). Yet our findings offer several learnings for the state discoms. Most of the local problems contributing to non-technical losses were resolved during the study period (see Section 2.6), but many of the aggravating factors are systemic issues. These include gaps in consumer indexing and feeder tagging; the metering status not being fully reflected in the master billing data; design gaps in the billing application and assignment of consumers to meter readers; and the lack of a robust real-time metering and network monitoring system.

Systemic issues require innovative and scalable solutions. The ongoing smart metering drives under the RDSS present a significant opportunity for discoms in UP and other states to reduce their losses. Smart prepaid meters, with good infrastructure to support them, can help improve billing efficiency and ensure timely revenue collection (Agrawal, Mani, Ganesan, et al. 2020; Agrawal, Mani, Kalra, et al. 2023). Following our assessments, the discom initiated smart metering in the concerned feeder. As of 31 August 2025, nearly half the consumers served by the feeder have smart meters installed.

However, the success of smart metering initiatives will depend on conducting robust consumer indexing, judiciously deploying strategies with frequent consumer engagement, enhancing the capacity of discom staff across levels, and upgrading the distribution network. In the next subsection, we present our recommendations for discoms to address the key factors underpinning distribution losses in UP and other states.

Ensure robust consumer indexing and energy auditing across all feeders

The Bureau of Energy Efficiency (Manner and Intervals for Conduct of Energy Audit in Electricity Distribution Companies) Regulations, 2021 mandate discoms to submit both quarterly and annual energy audit reports. However, very few audit reports are publicly available. Further, the audit report from a UP discom highlights gaps in consumer indexing as a systemic challenge underpinning the high losses across many feeders (MVVNL 2024).

Consumer indexing before smart meter installation is also mandatory under the RDSS, and must be pursued diligently to ensure the success of smart meters. However, our immersive research shows that correct consumer indexing requires the direct involvement of field staff and a system to periodically update the indexing. Discoms must leverage digital tools (instead of manual registers) and third-party audits to maintain a reliable and updated record of their consumers. Further, end-to-end and time-bound consumer and DT metering in every feeder is critical to enable energy auditing and accounting. Additionally, as the audit report suggests, there is a need to build the capacity of the audit cell and the IT team to allow for accurate, division-wise energy audits (MVVNL 2024).

Prioritise smart metering in feeders and subdivisions with high distribution losses

Commercial factors accounted for most of the losses in the studied feeder. Despite the efforts of the sub division staff, meter readers continued to generate bills without visiting the consumer’s premises; the reasons for this were flexibility of the billing software, limited ground-level vigilance, and socio-political factors. Technological interventions such as smart metering is essential in such circumstances. Discoms should prioritise smart metering in high loss areas, as they can help ensure timely and accurate billing, early redressal of metering issues, and targeted detection of electricity pilferage. This approach would help discoms reduce losses in a targeted manner and ensure short payback periods on initial smart meter investments. This, in turn, would bolster the confidence of discom staff and regulators in this new technology and facilitate a rapid scale-up.

Educate consumers about the benefits of smart meters and timely bill payment

Areas with high provisional billing or under-billing are likely to witness consumer resistance to smart metering drives. When consumers with lower bill amounts due to under-billing switch to smart meters (which will accurately record their consumption levels), they will experience a rise in electricity dues. This would create a negative perception among consumers about smart meters. Discoms must pre-empt such issues by raising consumer awareness and rolling out engagement drives well before deploying smart meters. Discoms could also retain the old meters along with smart meters for a few months to instil confidence in consumers about the technology.

Enhance the capacity of discom staff in technology use and data analytics

With the ongoing smart meter rollout, discoms have the opportunity to use technology and digital portals for system monitoring, vigilance drives, and quick grievance resolution. The discom management should ensure that all discom officials, particularly the field staff, get intensive and frequent training in the new technology; this is needed for the competent use of digital tools and portals, such as smart meter dashboards, GIS tools, and portals for DT-level data. The discom should also develop adequate institutional capabilities in data analytics to utilise the smart meter data efficiently and generate actionable field insights. Discoms can use budgets available for training under the RDSS to build such capabilities.

Invest in network upgrades to reduce energy losses and detect pilferage

Our assessment attributes one-third of the distribution losses to technical factors, including low HT–LT ratio, ageing conductors, overloaded transformers, and large inductive loads. A state-wide assessment of such issues is needed to plan, procure, and deploy asset and network upgrades. The UP discoms’ proposed project for replacing bare and aged conductors with AB cables has already been approved under the RDSS. Discoms must also prioritise feeder segregation for effective management of agricultural loads and deploy adequate capacitor banks and other innovative strategies for reactive power management during peak periods of agricultural demand.

Our study findings are highly relevant to other discoms in the region, or in other geographies, facing similar challenges. We emphasise that no single intervention can resolve all the challenges faced by the sector. Instead, a comprehensive approach combining policy reforms, technology adoption, and capacity building at the ground level is essential to ensure the operational turnaround of the feeder.