Search interesting materials

Sunday, July 27, 2025

Examining the performance of ERCs at APTEL

by Chitrakshi Jain, Bhavin Patel, and Renuka Sane.

Introduction

The efficiency of the State Electricity Regulatory Commissions (SERC)s, the Central Electricity Regulatory Commission (CERC) and the Joint Electricity Regulatory Commission (JERC) influence investability and growth of the electricity sector. For example, it costs regulated entities time and resources to petition the relevant ERC for decisions and potentially, to challenge decisions taken by the ERC at the appellate tribunal (APTEL), which exercises supervisory control over the ERCs and reviews their decision-making.

This article studies how ERC decisions perform at APTEL. We collect information about aspects of the ERCs' functioning from the text of orders passed by APTEL. This helps us (a) identify the most-litigious areas across ERCs and (b) examine how the ERCs' decisions perform in appeal. We use sub-national comparative analysis to understand the variation in the functioning of the different ERCs and the litigiousness of issues in different states.

We ask the following questions:

  1. Which ERCs contribute the most appeals at APTEL?
  2. What are the most litigious issues at APTEL?
  3. In how many appeals was the ERC's decision:
    1. Upheld, i.e. appeal was dismissed by APTEL?
    2. Partially upheld, i.e. appeal was partly allowed at APTEL?
    3. Overturned, i.e. appeal was fully allowed at APTEL?
  4. How often were the ERCs ordered to reconsider their decisions, i.e. the matter was remanded?

Our results suggest that issues related to tariff determination and restructuring are the most litigated issues at APTEL across ERCs, with the exception of Maharashtra. ERCs are differently situated in their ability to defend their decisions at APTEL and in the quality and clarity of their orders. We argue that such assessments, if regularised, can assist ERCs in improving the quality and form of their decision-making by creating a feedback loop, and can also assist in identifying areas for policy reform at the sub-national level.

Methods: Data

We obtained the orders passed by APTEL between the years 2013-2022 where the ERCs are a party to the challenge before APTEL. We excluded interim orders given that they do not include the outcome of the case. We focused on ten states, including Andhra Pradesh, Karnataka, Madhya Pradesh, Maharashtra, Odisha, Punjab, Rajasthan, Tamil Nadu, Uttar Pradesh, and West Bengal. These were selected keeping in mind geographical coverage, size of the state, and installed renewable energy (RE) capacity in the state.

We collected information on 26 indicators from the orders, related to the following categories:

  1. Time-related: e.g., date of orders, date of impugned order
  2. Party-related: names of appellants, respondents
  3. Bench-related: e.g., quorum, members' names
  4. Subject-matter related: e.g., prayers, issues
  5. Outcome-related: e.g., disposition and remand

We processed the text of the orders through LLMs, which we prompted to collect the information by placing reliance on explicit language in the text. After collecting the information for the relevant indicators, we ran verifications based on rules of legal consistency and logic to ensure that the collected information is accurate and reliable. For indicators related to outcome, we have made subjective inferences when the explicit information regarding its outcome was not articulated in the order. We have relied on individual appeals as the unit of analysis, given that outcomes are typically uniform for all parties in an order. We have integrated human verification at every stage of data collection to ensure reliability. Our final dataset consists of 513 orders and 919 appeals. The data is available here.

Methods: Issue categorisation

In order to study the issues that were being agitated before APTEL, we identified themes from the statement of issues in appeals which explicitly articulated them. There were 318 appeals (out of 919) that did not include a statement of issues. After classifying the issues thematically, we decided upon the final categories presented in Table 1 in consultation with practitioners. We ran keyword searches to sort the statement of issues into identified categories and verified the classification by reading the statements when they yielded unclear results for accuracy and reliability.

Table 1: Categorisation of Issues

Issue Category Coverage
Tariff determination and restructuring Challenges to tariff determination and adoption under Sections 62 and 63; inadequate attention to principles in arriving at tariff; revision of tariff and truing up.
Contractual disputes Liquidated damages, outstanding payments, renegotiation or termination of contracts, excluding change in law and force majeure.
Change in law and force majeure Subset of contractual disputes, relating to change in law and force majeure clauses in the contracts.
Procedural and jurisdictional Procedural lapses, violation of principles of natural justice, challenges to ERC's jurisdiction.
Open access consumers Wheeling and banking charges, and issues relevant to open access consumers.
Transmission and grid-related Connectivity, ISTS and grid-related issues, including compliance with grid code.
Specific compliance with regulations Mandatory non-tariff-related requirements for obligated entities, such as RPOs and RECs.
Captive status Captive status of power plants or group captive power plants.
Others Issues not falling under previous categories, e.g., distribution licensing.

Methods: Limitations

For the categorisation of issues we have relied on the statement of issues as determined by APTEL, in the instances it was explicitly identified in the order. Understandably, analysing the full text of the order will give deeper insights on the litigated questions. The outcomes, such as appeal allowed or dismissed, also do not provide information about outcomes on specific issues. This would entail reading the full orders and making subjective inferences. While the outcomes at APTEL have been used to assess the performance of ERCs, they do not measure the functioning of ERCs holistically, especially because studying the performance of APTEL is beyond the scope of this research.

Results

1. Distribution of litigation

Between the ERCs under study, Maharashtra followed by Karnataka, contribute to the most litigation at APTEL, as represented in Figure 1. These results indicate that the two states have a relatively larger private industry. However our analysis excludes writ proceedings, which are also used as a way to challenge ERC decisions. The total number of orders passed by ERCs is also not available uniformly across the ERCs to accurately calculate the rate of appeal.

Figure 1: Distribution of litigation at APTEL

2. Most litigious issues

ERCs are empowered to decide a wide range of issues. As a consequence, the issues dealt with by the APTEL in appeals are also varied. An appeal may involve more than one issue, and hence the number of issues involved is more than the number of appeals. As represented in Table 2, tariff determination and restructuring are the most litigated upon issues at APTEL across ERCs with the exception of Maharashtra. In Maharashtra, procedural and jurisdictional issues emerge as the most litigious. While the high incidence of such issues is concerning, issues of the procedural and jurisdictional variety can be resolved easily if ERCs invest in capacity building and follow the procedure under the law faithfully.

Our results corroborate the findings of an earlier study (Prayas 2018) which had found that a third of the issues being litigated before APTEL were concerned with tariff. Pertinently, the ERCs have enacted specific regulations related to tariff determination and made the calculation of tariff an exercise which is assisted by detailed delegated legislation. In this context, it is worrying that tariff continues to be the predominant category with regard to appellate litigation.

Table 2: Issue portfolio at different ERCs (Values in percentages)

State Total Issues Tariff related Procedural & Jurisdictional Contractual disputes Specific Compliance Change in law & force majeure Open access Trans-mission & grid Captive status Other
Maharashtra 248 26.61 32.66 2.82 3.63 3.63 6.85 4.44 16.53 2.82
Karnataka 171 30.41 26.90 22.81 1.75 2.34 6.43 6.43 0.00 2.92
Tamil Nadu 138 28.99 12.32 7.25 20.29 0.72 5.80 7.25 10.87 6.52
Punjab 96 33.33 11.46 15.62 14.58 8.33 7.29 4.17 1.04 4.17
Rajasthan 76 25.00 11.84 19.74 10.53 13.16 5.26 7.89 1.32 5.26
Madhya Pradesh 75 41.33 17.33 13.33 0.00 1.33 9.33 4.00 9.33 4.00
Andhra Pradesh 65 47.69 26.15 9.23 3.08 1.54 1.54 7.69 0.00 3.08
Uttar Pradesh 59 33.90 16.95 20.34 8.47 5.08 1.69 13.56 0.00 0.00
Odisha 49 34.69 16.33 6.12 16.33 0.00 8.16 10.20 6.12 2.04
West Bengal 29 62.07 13.79 3.45 13.79 3.45 0.00 3.45 0.00 0.00

In addition to the most litigated issues, we could also identify the states which contributed most to the litigation of a particular issue at APTEL. Maharashtra contributes the most to issues related to tariff, and procedure and jurisdiction. This outcome is also a function of Maharashtra being involved in the highest number of appeals in our dataset. The findings are presented in Table 3 below.

Table 3: Distribution of Issues

ERC which contributes most to the litigation of a particular issue at APTEL and the percentage share of their contribution
ERC Issue Contribution (%)
Maharashtra Tariff related 20.1
Maharashtra Procedural and jurisdictional 37.5
Karnataka Contractual disputes 33
Tamil Nadu Specific compliance with regulations 34.5

3. Outcomes

We focus on indicators related to the 'disposition' of the appeal from the information we had collected. We scored the performance of the ERCs relative to each other by making the number of decisions that were upheld, overturned or modified by APTEL as the basis of comparison. The results have been compiled in Table 4.

At an outcome level, if an ERC succeeds in defending its decisions, then it would indicate that the orders are well-reasoned, and the ERC follows the procedure under the law. A high overturn rate would indicate weak decision-making capacity.

Table 4: Dispositions at APTEL across ERCs

ERC Allowed Dismissed Partly Allowed Other Remanded Total
Andhra Pradesh 27 32 7 4 11 70
Karnataka 103 49 14 4 67 171
Maharashtra 92 61 31 62 35 246
Madhya Pradesh 22 17 11 8 13 58
Odisha 17 15 17 2 6 51
Punjab 18 28 22 6 13 74
Rajasthan 32 45 7 3 20 87
Tamil Nadu 22 33 19 9 20 83
Uttar Pradesh 15 26 11 5 10 57
West Bengal 4 8 6 4 5 22
Total 352 314 145 108 200 919

We find that ERCs are differently situated in their ability to defend their decisions at APTEL and the quality and clarity of their orders. Rajasthan found the most success at APTEL, Maharashtra had the least.

Typically, matters are remanded when APTEL is of the opinion that the relevant ERC did not, amongst other things, follow the procedure or frame the issues or determine question of facts sufficiently well. A high remand rate is worrying since it implies that either the ERCs in question are ill-equipped to resolve disputes in the first instance or that APTEL, unless it has insufficient evidence to make the decision, is abdicating its mandate.

Remands lengthen the resolution of disputes and burden regulated entities with legal and compliance costs. This can stymie the growth of the electricity sector, especially in states like Karnataka, for KERC has been asked to reconsider most number of its decisions when compared to other ERCs.

Recommendations

Both APTEL and ERCs are empowered to implement these recommendations.

1. Regularise assessments through use of emerging technologies

We recommend that such comparative assessment exercises be regularised through the use of emerging technologies. The composition of ERCs is constantly changing, and members would benefit from information about the performance of their decisions at APTEL closer to the date of the decisions. This can be made possible by creating a customised tool that leverages LLMs and the competence of researchers and practitioners familiar with the sector.

2. Publish granular statistics

APTEL can improve upon the collection and publication of litigation statistics and include the subject matter of litigation and the relevant laws that are under litigation, amongst other categories, in this exercise. Similarly, while some ERCs publish the number of orders they hear and decide annually, they can include more relevant details in this publication and also publish these at shorter intervals. Collecting this data at source would make the identification of litigious issues, which are often proxies for policy problems, easier.

3. Identify areas for policy reform

ERCs should study the precise reasons for disputes that correspond with the litigious issue categories in their states and respond by changing and adapting their regulations to minimise them. The persistence of tariff as the most litigious category is concerning, given that detailed regulations on calculation and imposition of tariff have been enacted by the regulators.

Conclusion

In summary, we find that:

  • Between the ERCs under study, Maharashtra, followed by Karnataka, together contribute to the most litigation at APTEL.
  • Issues related to tariff determination and restructuring are the most litigated issues at APTEL across ERCs. This is worrisome given the detailed subordinate legislation that govern the regulation of retail and other categories of tariff.
  • ERCs are differently situated in their ability to defend their decisions at APTEL and the quality and clarity of their orders. We find that Rajasthan found the most success at APTEL, while Maharashtra had the least.
  • Remands lengthen the resolution of disputes and burden regulated entities with legal and compliance costs. This can stymie the growth of the electricity sector, especially in states like Karnataka, since KERC has been asked to reconsider the most number of its decisions when compared to other ERCs.

References

Amicus Populi? A public interest review of the Appellate Tribunal for Electricity , by Vaishnava S, Chitnis A and Dixit S, 2018, Prayas Energy Group

The authors are researchers at TrustBridge Rule of Law Foundation. They would like to acknowledge and thank Natasha Aggarwal, Madhav Goel, Abhinav Hansaraman, Amol Kulkarni, Praduta Singh, Aparna Jha, Varun Soni, and Gaurav Aswani for compiling, collecting, and verifying the data used in our analysis. We would also like to thank Upasa Borah for helping with verifying, cleaning, and consolidating the dataset.


The authors are researchers at the TrustBridge Rule of Law Foundation.

Friday, July 11, 2025

Households that live within their means in India

by Jay Kulkarni and Susan Thomas.

The economic well-being of households is primarily about their ability to spend on consumption. Household consumption is dominated by what the income of the household is, but not limited by it. Households that spend less than they earn, build their savings. Households that spend more than they earn either borrow or draw down on earlier savings. There is a big difference in the life-cycle possibilities between households that manage to save versus those that do not. In this article, we analyse a panel dataset of Indian households to understand what differentiates households who live within, or beyond, their means.

An often discussed measure of the household's income-consumption dynamic is the `marginal propensity to consume' or MPC, which is the marginal change in consumption for a marginal change in income. The MPC is a valuable part of the toolkit of macroeconomics. An equally important measure is the 'average propensity to consume' (which is abbreviated as APC). This is the fraction of disposable income that the household consumes. The APC shows the income-consumption dynamics of a household in a stated time period. When the APC is below 1, the household is saving, and on average, building up its wealth. There is a clear line between low APC households (i.e. those with APC below 1), who are building up wealth, vs. the households that are not.

In an advanced economy, we think of the APC as a part of life cycle optimisations. When an affluent and financial unconstrained household is young, it builds up savings (i.e. low APC), and then it dis-saves in old age (i.e. high APC). In a poor country, we see many households who are dis-saving even when they are young. Building up wealth versus drawing down wealth takes on a different character in the context of a low middle income economy (Badarinza et al, 2019).

Aggregate facts about household APC, and its covariates, are an important element of understanding India. This article aims to establish such facts. What is the average household APC in India? What fraction of households have a low APC? Do higher income households have a low APC? Do low APC households have lower income volatility? Are low APC households systematically older households? What is the connection between financial inclusion and household APC?

Data and Methodology

The measurement of consumption is an important feature of many government statistical systems. Aggregative statements are derived from the national accounts. The best information about households is found in advanced economies such as the US (Consumer Expenditure Surveys) and the UK (Family Resources Survey), which are observed at annual frequencies.

Less is known about Indian households. In recent times, better measurement of households has commenced in India. One such dataset is the Consumer Pyramids Household Survey (CPHS) published by CMIE, which began in 2014 and now surveys about 200,000 households every year, thrice a year. For this article, we focus on their 2023 and 2024 data.

Computing an APC can be done at different frequencies. In this article, we compute the APC at both monthly and annual data.

Average household APC in India using annual data

We start at the annual APC and establish basic facts. Household data is hard to measure, given difficulties in survey administration, in the interest of the household in offering information, and in the correct recollection by the household. Hence we show a robust estimator of the mean APC across the values obtained for each household. Table 1 reports this value along with other summary statistics. The big fact that we take away is that the (robust mean of the) APC was 0.64 in 2024. If (1 - APC) is the savings rate, this implies a savings rate of 0.33 percent in 2023 and 0.36 in 2024.

Table 1: Distribution of annual household APC in India, 2023, 2024

          25th       50th       75th       Mean       Std.Dev.    Fraction






with APC<1
2024       0.50 0.65 0.80 0.64 0.33 94.23
2023       0.54 0.69 0.84 0.67 0.74 91.79

 

Going deeper into cross-sectional variation and higher frequency observation

Households may have an annual APC < 1, while having some months where APC > 1. For example, a farming family may be above the water when viewed at the level of the year, but it may earn income only at the Kharif harvest, and run with APC > 1 for all other months. We now define a `Low APC household' as one which lives strictly within its means, where every monthly APC (and therefore the annual APC) is less than 1. Using this definition, we partition the data into Low vs. High APC households.

Table 1 shows that 94.23 percent of Indian households in 2024 are at an annual APC < 1. But when we switch to this modified view of the APC within the year, the picture changes. In this perspective, 54 per cent of Indian households in 2024 are low APC. For 2023, this value was 48 per cent.

We also observe the age of the household head, as well as other household features such as the fraction of members who are dependents and the fraction who are employed. To make numbers comparable, we adjust prices for inflation using an all-India series re-based to December 2024, and use per-capita numbers to account for different household sizes in the sample.

We construct a measure of household income volatility, as the standard deviation of the percentage monthly changes in household income. We construct a financial participation score as in Palta et al (2022). This is the fraction of the number of financial assets households own, out of the 10 that dataset records. The debt status of each household is also separately observed.

We then explore cross-sectional variation by estimating a probit model to predict a low APC household based on the annual data. All explanatory variables are contemporaneous. Figure 1 presents the estimated coefficients from this regression. In this figure, the vertical dashed line indicates the 0 value of the null hypothesis. Any coefficient on the right is positive, and to the left the coefficient which can be used to answer the questions raised above. The distance of the error bars from the 0 value line shows that the coefficient is statistically significant, and influential in the probability of the household being a low-APC household.

Figure 1: Factors affecting the probability of being classified as a low APC household 

 

What do we see here?

  • Do low APC households have higher income?
    The coefficient of log income is positive and significant. The higher the income, the higher the probability that the household is low APC.
  • Do low APC households have higher income volatility?
    Income volatility has a negative and significant coefficient. This means that higher the volatility of income, the lower the chance of the household being low APC.
  • Are low APC households older?
    The age of the head of the household is a proxy for the age of the household. The coefficient for this is close to zero (value of 0.0052) but is positive and significant. Households with older heads tend to be low APC. This income-consumption pattern is consistent with the life-cycle hypothesis of Modigliani and Brumberg (1950), or the permanent income hypothesis of Friedman (1957).
  • Do low APC households have a better financial participation score?
    The household financial participation score is a useful way to think about the asset side of the household balance sheet (Ghosh and Thomas, 2022). This coefficient is positive and significant. In addition, the presence of borrowing tends to run in the opposite direction (borrower households are more likely to be high APC).

Discussion

We have a new fact about Indian households: About half of these have at least one month a year where they live beyond their means. Many of the results that we see here are consistent with empirical findings in other countries (Goodman and Webb, 1995, Blundell and Preston, 1998, Gorbachev, 2011, Fisher et al, 2020). Higher income, higher fraction of members employed, higher financial participation score, older households, lower income volatility, lower fraction of dependents, and not having debt, correlate with being a low APC household.

The trajectory of income, savings and wealth by an affluent, financially unconstrained household, operating in a well functioning macroeconomic and financial system is well-established. We expect households to save when they are young, and dis-save when they are old. In the Indian setting, such behaviour is perhaps the privilege of a small number of households who face more complex financial planning problems within the year.

In thinking about households in India, the distinction between households that are adding to their savings versus the households that are not, seems fundamental. It has far-reaching consequences for the life of a household. From the viewpoint of governments and firms, this is an interesting distinction which can be applied when thinking about households. This article is a first look, based on novel mechanisms of measurement, covering two years of data only. More research is needed to obtain insights into the causes and consequences of these phenomena. What is the dynamics of low APC across time? What kinds of households are able to achieve low APC on a sustained basis? How does the build-up of household wealth reshape the decisions of a low APC household?

References

  1. Cristian Badarinza, Vimal Balasubramaniam and Tarun Ramadorai, The household finance landscape in emerging economies, Annual Review of Financial Economics, Volume 11, pages 109-129, 2019.
  2. Richard Blundell and Ian Preston, Consumption Inequality and Income Uncertainty, The Quarterly Journal of Economics, Volume 113, Number 3, May 1998, pages 603-640.
  3. Jonathan D. Fisher, David S. Johnson, Timothy M. Smeeding and Jeffrey P. Thompson, Estimating the marginal propensity to consume using distributions of income, consumption and wealthJournal of Macroeconomics, Volume 65, 2020.
  4. Indradeep Ghosh and Susan Thomas, Financial inclusion measurement: Deepening the evidence, Chapter 9, Inclusive Finance India Report 2022, 17th edition, pages 117-125, January 2023.
  5. Alissa Goodman and Steven Webb, The distribution of UK household expenditure, 1972-1992, Fiscal Studies, Volume 16, Number 3, pages 55-80, 1995.
  6. Olga Gorbachev, Did household consumption become more volatile?, American Economic Review, Volume 101, Number 5, August 2011, pages 2248-70.
  7. Geetika Palta, Mithila A. Sarah and Susan Thomas, Measuring financial inclusion: how much do households participate in the formal financial system?, The Leap Blog, 3 July 2022.

Acknowledgments

Jay has just wrapped up his masters in economics from Università Bocconi. Susan is senior research fellow at XKDR Forum. We thank Geetika Palta for help on working with CPHS, and Ajay Shah for positioning and inputs.