When remedies become regulation: The Karnataka High Court's intervention in food licensing and street vending

by Prashant Narang, Aryan Pandey and Indira Unninayar.

I. When public health litigation expands into regulatory governance

On 19 September 2025, the Karnataka High Court delivered its decision in Karnataka Pradesh Hotel & Restaurants Association v. Union of India. The case began as a routine industry challenge to the Food Safety and Standards Act, 2006. The judgment oversteps statutory adjudication to engineer regulatory design. It answers a real public-health worry. But litigation like this rarely stays within the parties before the court. The Court encroached into the executive territory with no consideration of whether the state is actually capable of implementing what it now directs. Such directions tend to produce selective enforcement and compliance costs that fall hardest on those least able to bear them.

The petition arose from a 2012 directive on licensing enforcement. The judgment was delivered nearly a decade and a half later by which time, the regulatory landscape and the affected ecosystem had evolved substantially. Street vending, food delivery and the law on informal work had all changed and all bore directly on what the Court now ordered.

II. What the petition sought, and what the Court ultimately directed

Hotel and restaurant associations had challenged orders to enforce the FSS Act and its regulations. The trigger was a letter dated 13 March 2012 issued by the State Food Safety Commissioner, acting on the Union instructions, requiring all States to enforce the Food Safety and Standards Authority of India's (FSSAI) licensing and registration regime. Every Food Business Operator' ("FBOs") had to obtain a licence or registration as a condition for continuing their business.

The petitioners contended that this requirement was impractical and arbitrary, especially applied uniformly to establishments of vastly different scale and capacity. The burden, they said, fell hardest on smaller operators. They went further, asking the Court to strike down swathes of the Act and its regulations as unconstitutional.

The Court rejected these constitutional challenges in their entirety and upheld the validity of both the Act and the Regulations, noting that the Supreme Court had already affirmed the Act. It restated food safety as a public-health aim and accepted the State's claim that the rules rested on scientific and international standards.

It then issued two directions with implications beyond the immediate dispute.

1. It directed the Union Government to classify restaurants into small, medium, and large categories and to enact separate laws or frame separate guidelines for each, observing that reliance on turnover-based thresholds alone, was impractical and insufficiently responsive to differences in size and operational capacity.
2. The Court directed the State government to introduce health and safety rules specifically for street vendors and food trucks, and to establish a mechanism to ensure strict oversight of their implementation.

These directions are what give the judgment its broader regulatory significance.

III. Expanded prescriptions sans diagnosis risk over-regulation, arbitrary discretion, and regulatory incoherence.

A. New rules directed without a policy diagnosis -

The judgment's biggest gap is that it never finds that existing regulation has failed. Nor does it explain why new, vendor-specific rules are required, and whether existing processes for licensing, inspection, and enforcement have failed. It even concedes that the licensing rules already impose hygiene standards on every operator.

The FSS Act already establishes a comprehensive enforcement architecture. Section 30 vests primary responsibility in the State Commissioner of Food Safety, while Sections 36 and 38 operationalise enforcement through prescribed methods and designated officers at the district level within municipal and local jurisdictions.

The Court should have asked two questions: were existing standards inadequate, and had enforcement failed? However, the judgment neither raises nor answers these questions.

The reasoning moves from a general observation about the informality of street vending directly to remedial directions that materially reshape regulatory obligations. It does so without identifying any institutional deficiency that might have justified such an expansive remedy.

B. The Street Vendors Act framework was overlooked entirely -

The Court acts as if street vendors operate in a regulatory vacuum. They do not.

The Street Vendors (Protection of Livelihood and Regulation of Street Vending) Act, 2014 ("SVA") was specifically enacted to balance livelihoods against congestion, public health and urban order. It overrides inconsistent municipal laws and works through town vending committees ("TVCs"), surveys, and certificates of vending. The SVA is not merely a procedural architecture; it embodies a considered normative choice by Parliament, that street vendors are rights-holders, entitled to livelihood protection, meaningful participation through TVCs, and procedural safeguards before any restriction on their vending.

By directing new health and safety rules for vendors without engaging with this framework, the Court implicitly undoes that normative settlement. It treats vendors not as participants with protected rights but as subjects of fresh regulation – inverting the very premise of the statute Parliament enacted for them.

The result is regulatory incoherence and it is worth being specific about what that means in practice. Under the SVA, a vendor acquires a certificate of vending through a TVC process that must include vendor representation; this certificate is her legal entitlement to occupy a designated vending zone. Under the FSS Act, she must separately obtain a licence or registration from FSSAI, subject to turnover thresholds and hygiene standards. The Court's direction would now superimpose a third layer: vendor-specific health and safety rules with a fresh enforcement mechanism. Each of these three regimes carries its own authority, its own compliance requirements, and its own enforcement officer.

C. The Court's directions assume state capacity that does not exist -

As far back as 2020, only 47% of town vending committees had any vendor representation; seven states had not notified schemes under the SVA, and in four states no compliant TVC had been constituted at all (Narang et al., 2020).

The enforcement machinery under the FSS Act tells a similar story. As of 2021, there were only 2,531 Food Safety Officers nationally for roughly one crore street vendors, with vacancy rates between 33% and 90% across states (Mishra & Khattar, 2025). Between 2018 and 2021, fewer than 1% of food adulteration cases ended in conviction. None of this means enforcement has stopped. It means enforcement has changed. When an inspector cannot police everyone, he polices whomever he likes – and scarcity only raises the price of his goodwill.

Piling fresh directions onto this will not help; it will hurt. Pritchett, Woolcock and Andrews (2010) examined three well-funded reforms (schooling in India, budgeting in Mozambique, land titling in Cambodia) that all failed for one reason: each demanded transaction-intensive implementation, millions of scattered discretionary acts no centre can supervise. Street-food safety is the same kind of task. It is transaction-intensive (a crore of vendors, countless daily sales), discretionary (each inspector judges hygiene on the spot), high-stakes (a failed check can end a livelihood) and opaque (the encounter leaves no record). On all four counts, the very dimensions Kelkar and Shah (2022) name as the hardest for any state to master, it scores about as badly as a task can.

The sequencing is backwards, too. Early state-building, Kelkar and Shah argue, should begin with low-stakes, high-visibility tasks, short feedback loops, correctable errors – and reach for hard ones only once capacity exists. The order to keep "strict vigil" over vendors does the opposite: it escalates coercion before building the institutions that would restrain it.

This dynamic has become characteristic of the Indian regulatory ecosystem. Shah's account of the history of Indian finance documents a pattern of regulatory agencies consistently engaging in micro-management whilst lacking the state capacity to enforce their own frameworks.

High discretion combined with low capacity does not produce zero enforcement; it produces selective, rent-seeking enforcement. When inspectors are too few to visit every vendor, they must choose whom to visit and a shortage of inspectors does not dilute that discretionary power, it concentrates and rations it. The fewer the officers relative to a crore of vendors, the more valuable each discretionary decision becomes, and the higher the payment it can command.

As Rai and Shah (2015) observe, the Indian state is too often strong as in scary but not strong as in capable: it commands coercive reach without the institutional depth to convert that reach into governance outcomes. Ordering strict vigil onto a system with 90% officer vacancies in some states therefore does not produce better public-health outcomes; it produces more rent-seeking. Inspectors arrive not on a fixed schedule but whenever they are short of cash, and vague, subjective standards give them the pretext to do so (The Seen and the Unseen, Ep 18). The Court's directions thus simply widen the regulatory perimeter within which this behaviour can operate.

D. Cross-jurisdiction comparisons are persuasive only when capacity is comparable-

The judgment leans hard on foreign examples to justify a strong licensing and enforcement regime. It cites international norms to rebut the claim that the regime is impractical.

But it ignores the conditions that make those systems work. Licensing does not work in the abstract. It needs capacity, trained inspectors, predictable procedure and firm limits on discretion.

The judgment itself notes that regulators such as the United States Food and Drug Administration recognise wide variation in the size and capacity of food establishments, and that enforcement is typically carried out by local health authorities. These details matter. They determine whether regulation produces overall compliance or its very opposite by way of uneven and discretionary enforcement.

This is where the comparison breaks down. The FSLRC (2013) treats foreign models as inputs to adapt, warning against any bid to "mechanically transplant ideas from elsewhere". International standards inform; they do not, on their own, justify a domestic enforcement regime. The court inverted this. It used the FDA comparison as the justification itself, without asking whether the administrative architecture that makes those powers function exists here.

Pritchett, Woolcock and Andrews (2010) show why that architecture cannot simply be assumed to exist. As per them when governments copy institutional forms from higher-capacity settings, the laws, the agencies, the enforcement powers, without first building the administrative foundations that make those forms function, the result is the appearance of reform without its substance. It is, in their words, no reform at all.

The FDA comparison does not establish that India's enforcement regime should be intensified. It shows only that the FDA works within machinery that makes its powers function. Transplant the powers without that architecture and you import the coercion while leaving behind the restraint.

E. Non-parties bear the burden of directions issued without participation -

The High Court has not abided by one of the basic principles of natural justice, audi alteram partem – the 'right to be heard' before any orders are passed against a person, as it has not 'impleaded' and 'heard' street vendors and pliers of food trucks, before proceeding to pass directions concerning them. Yet it ordered the State to write new health-and-safety rules for them and to keep 'strict vigil' over them, without studying who they are or what they face.

The regulatory burden falls on informal workers operating under constrained economic conditions. The court treats informality as a regulatory gap to be closed, vendors operate outside the system, so the system must be extended to capture them. Shah (2026) inverts this reading. Where the state's enforcement is slow and unreliable, operating informally is not evasion of good rules but a rational adaptation to bad institutions. Vendors build workarounds precisely because formal compliance offers little protection and predictable harassment. The state then misreads the adaptation as defiance and tightens the rules, which raises the cost of formality further and entrenches the informality it set out to cure. A direction to bring a crore of vendors under "strict vigil" is the next turn of exactly this cycle.

IV. Food safety is a compelling goal, but cannot justify prescription without basis

The strongest defence of the Court's approach lies in the public interest at stake. Food safety directly impacts public health and the FSS Act itself emphasises risk management, consumer protection, and preventive regulation. The Court did not draft rules itself; it told the executive to. . Read this way, the judgment can perhaps be seen as an attempt to prompt more effective implementation of an existing legal framework.

However, that defence, has limited force if any, as the Court does not explain the reasons why such directions pertaining to street vendors and food trucks were required in the first place, and how the existing enforcement mechanisms under the FSS Act were inadequate. Without a demonstrated failure, intervention at the level of design has little to stand on.

The promise of later consultation cures nothing. Consultation after an order to make rules is not consultation about whether the rules are needed at all. Once the outcome is predetermined, the space for meaningful policy deliberation is confined to that predetermined outcome.

The Court unfortunately moved too quickly from concern to prescription, and in doing so, blurred the line between ensuring lawful administration and reshaping the regulatory architecture itself.

V. Conclusion: Prescriptions must stay focused and relevant

The judgment reflects a growing tendency: courts shifting from reviewing validity to supervising regulation, especially under the banner of public health or public interest. Such interventions may be well-intentioned. But good intentions do not substitute for institutional competence. In this case, the Court's directions go beyond correcting unlawful administration to enter the terrain of regulatory design, without any demonstrated failure of the existing framework and without hearing those most affected by the outcome.

This tendency is not confined to any single domain. As Jain and Reddy T (2025) observe, reform through judicial diktat characteristically bypasses public consultation on questions that carry complex second-order effects. The adversarial courtroom is not designed for the stakeholder deliberation that sound policymaking requires. When it substitutes for that process, the people most affected, here, street vendors and food truck operators, bear consequences that were never examined.

Lon Fuller, in The Forms and Limits of Adjudication (1978), offers a useful framework for understanding why. Fuller identified a class of problems he termed "polycentric", those where the disposition of any single issue carries implications for every other, such that pulling one strand "will distribute tensions after a complicated pattern throughout the web as a whole". In such contexts, he argued, adjudication becomes institutionally incapable, because the affected party's participation through proofs and reasoned arguments loses all meaning when no advocate "could possibly present to the tribunal the grounds that must be taken into account in the decision".

The Karnataka High Court's directions bear precisely this character. A judicial mandate to introduce new health and safety rules for street vendors does not resolve a discrete regulatory question, it simultaneously displaces an existing framework under the Street Vendors Act, imposes fresh compliance burdens on informal workers already operating at the economic margin, adds enforcement obligations to a system strained by Food Safety Officer vacancy rates and multiplies points of regulatory contact where discretion can be monetised. Each of these consequences shapes the others, and that interdependence is exactly what Fuller's framework identifies as lying beyond the proper limits of adjudication.

The cost is not only procedural. Compliance burdens imposed without the capacity to administer them do not produce better governance; they tax the everyday enterprise of people operating at the margin and dampen the very economic activity the state should want to encourage. As Shah (2026) puts it, this is the "effervescence of creativity and invention that a poor country cannot afford to extinguish."

The lesson is that remedial ambition must be matched by remedial discipline. Prescription without diagnosis, and supervision without capacity, do not produce better governance. They produce the illusion of it.

References

Bedi J. and Narang P., 2020. Progress Report 2020: Implementing the Street Vendors Act. Centre for Civil Society.

Mishra G. and Khattar J., 2025. FSS Act: Need for enforcement and accountability in India's food safety regime. Bar and Bench. 26 June 2025.

Pritchett L., Woolcock M. and Andrews M., 2010. Capability Traps? The Mechanisms of Persistent Implementation Failure. Center for Global Development.

Kelkar V. and Shah A., 2022. In Service of the Republic: The Art and Science of Economic Policy. Penguin Allen Lane.

Varma A. and Menon M., 2017. Restaurant Regulations in India. The Seen and the Unseen. 15 May 2017.

Financial Sector Legislative Reforms Commission, 2013. Report of the Financial Sector Legislative Reforms Commission. Ministry of Finance, Government of India. 22 March 2013.

Jain C. and Reddy T P., 2025. Why reform through judicial diktat is fraught with perils. Times of India. 8 November 2025.

Fuller L. and Winston K I., 1978. The Forms and Limits of Adjudication. Harvard Law Review, Vol. 92, No. 2.

Rai S. and Shah A., 2015. Going from strong as in scary to strong as in capable. The Leap Blog. 25 February 2015.

Shah A. and Varma A., 2026. Why Freedom Matters | Episode 10 | Everything is Everything. Everything is Everything. 1 September 2026.

Ahluwalia R. and Shah A., 2026. Why Firms Build Economies Ft. Ajay Shah | Growth is Good | Ep 25. Foundation for Economic Development. 27 March 2026.

Prashant Narang and Aryan Pandey are researchers at TrustBridge Rule of Law Foundation. Indira Unninayar is an Advocate-on-Record, Supreme Court of India.

What happens when arbitration deadlines are missed

by Prashant Narang and Renuka Sane.

Section 29A of the Arbitration and Conciliation Act 1996 was introduced to deal with delays in arbitration. It sets a time limit for making an award. If that time runs out, parties have to go to court to extend it. The court can also impose consequences for delay, such as reducing fees, awarding costs, or replacing the arbitrator.

Our new working paper studies how this works in practice. It looks at 202 reported orders of the Delhi High Court between 2015 and 2024.

It finds that the Court almost always grants extensions and almost never imposes sanctions.

What the data shows

Out of 202 cases, the court granted extensions in 198 (98%). Only 4 cases were dismissed, and those were on technical grounds. Sanctions were rarely imposed.

• Fee reduction: 0 out of 202 cases
• Adverse costs: 6 out of 202 cases (about 3%)
• Replacement of arbitrators: 4 out of 202 cases (about 2%)

Repeat extensions are not unusual. There are 30 cases where parties came back for a second or later extension. The court granted 29 of them (96.7%). There are no sanctions in these repeat cases.

These petitions also move quickly.

• Median time to decide: 3 days
• Median number of hearings: 1
• About 63% of cases are decided in a single hearing

So the delay is not in the court process. Courts dispose of these matters quickly. But they usually extend time without imposing any consequence.

Why extensions are common

Part of the answer lies in how Section 29A is structured.

For the Court, giving an extension is easy if both parties agree. The court can dispose of the case quickly.

Imposing a penalty is harder as the Court has to find out who caused the delay. It may have to look at the record in detail. It also has to hear the arbitrator before cutting fees. All this is likely to take more time and effort.

It is not surprising that consensual extensions are more common.

What this means for the law

Over time, this pattern shapes how the law works.

Section 29A was meant to push arbitrations to finish on time. It often works as a way to formally extend time after the deadline has passed.

If parties expect that extensions will be granted without much difficulty, the deadline may lose its force.

This does not mean the provision has no value. But it suggests that deadlines work best when consequences are easy to apply.

Looking ahead

If deadlines are not backed by predictable consequences, do they change behaviour?

The paper does not answer this fully. It focuses on what courts do once parties come for an extension. But the pattern is clear. Extensions are routine and sanctions are exceptional.

That may matter for how arbitration timelines are taken in practice.

You can read the working paper here.

The authors are researchers at TrustBridge Rule of Law Foundation.

Establishing the baseline for the Kollam district court reform

by Siddarth Raman.

An important milestone in Indian legal system reform has been underway in Kollam district for the last year. The High Court of Kerala has launched a court for cheque dishonour cases under Section 138 of the Negotiable Instruments Act as its first pilot in the 24x7 ONCourts initiative. PUCAR - of which XKDR Forum is a part - is a knowledge partner to the High Court of Kerala in this.

24x7 ONCourts aims to make the litigant experience efficient, predictable and seamless. Before assessing the intervention's impact, we need a baseline. We estimate five measures of court performance using a random sample of 100 disposed cases filed in 2015-2024. In Kollam, the median case takes 609 days over 12 hearings. Only 19% of hearings are substantive. The first substantive hearing occurs after 240 days. The gap between hearings is 61 days. Stage-wise analysis shows that the most time is spent in getting the litigants to court, and most cases do not reach trial. We compute the same metrics for Thrissur to enable future difference-in-differences analysis.

Motivation

Changing a system is hard. Public policy interventions in India are often guided by conjecture, anecdote and assumption. Too many reforms fail because they are not grounded in data, and their outcomes are not measured. A data-driven process with clear hypotheses on the problem, the solution and the expected outcomes, backed by rigorous measurement and empirical evidence, is needed to distinguish success from failure.

Early in the 24x7 ONCourts journey, we planned for continuous assessment. The first step is to establish a baseline. It fixes the pre-intervention state so later changes can be attributed to the intervention rather than unrelated trends. Publishing a compact, replicable baseline also lets other researchers repeat the measurement after go-live and compare the site of implementation against unaffected locations.

Approach to measurement

Courts and court processes should be judged by their impact on litigants; they are the central stakeholders in the system. In this, there are two types of measures one can develop. There are coarse black-box metrics that give a broad picture of outcomes - like the time to disposal. Alongside this, there can be fine-grained process metrics like the time taken to make payments, extent of delay attributable to postal services or police processes, or the time taken by an accused person to file for bail. These are necessary for continuous improvement of processes. Existing court systems do not capture or publish this type of process data.

We use coarse metrics for our baseline, relying purely on public data, so that anyone can reproduce the numbers without privileged access to court systems. As courts modernize and publish more data in the public domain, we may be able to develop more sophisticated metrics.

Using this approach, we pick five measures of court performance:

1. Time to disposal is the number of days from filing to the court's final order.
2. Hearings to disposal is the number of hearings a case goes through before it ends.
3. The share of substantive hearings is the fraction of hearings that move the case forward towards resolution. For instance, a hearing where the judge is on leave or a summons is re-issued is non-substantive; hearings where evidence is recorded or arguments are heard count as substantive.
4. Time to the first substantive hearing is the wait before meaningful progress begins.
5. Time between hearings is the within-case median gap between consecutive hearings. Lower the time, quicker the next hearing. The inter-quartile range (IQR) across cases is a measure of scheduling predictability. A lower IQR implies less variation in the time between hearings across cases.

A more detailed discussion of these measures and how to use them in evaluating court reforms is available here.

We conducted data collection and analysis in November 2024. Our approach, methodology and data are public, so that others can repeat the measurement for different samples, periods, or districts using e-Courts records.

Methodology

This article reports the pre-reform numerical values for Kollam, using a simple empirical strategy. We construct a baseline for Kollam before the 24x7 ONCourts intervention and compare it with Thrissur, which serves as the control district for future difference-in-differences analysis. We selected Thrissur as the control district after eliminating regions with existing special NI courts. Thrissur mirrors Kollam in judicial capacity (approx. 2,500-2,900 pending cases per judge), economic activity (GDVA per capita ratio of 0.9), and administrative composition (both have municipal corporations).

While Thrissur shares structural similarities with Kollam, direct cross-sectional comparisons of absolute performance levels between the two districts are methodologically unsound. Districts differ in idiosyncratic ways that are difficult to quantify -caseload volume, caseload mix, allocation rules. Consequently, our evaluation strategy avoids asking 'Is Kollam faster than Thrissur?' Instead, we focus on within-district changes over time.

We draw a random sample of 100 disposed Section 138 matters filed in 2015–2024 from Kollam and Thrissur using the public e-Courts database. Cheque dishonour cases are filed before the magistrate's court. In Kerala, these are filed as private complaints (criminal miscellaneous petitions) before becoming criminal cases. For each case, we reconstruct the case lifecycle from hearing dates and order text, tagging hearings by stage and classifying them as substantive or non-substantive.

For the baseline, we report medians and means. Medians describe the typical case; means indicate the influence of long tails. By sampling only disposed cases, we measure the speed of the 'successful' cohort. We acknowledge this introduces survival bias: this metric underestimates the time for difficult cases still stuck in pendency. We aim to address this in future work by using survival analysis to estimate the time to disposal for ongoing cases.

Results

The tables below report the pre-reform baseline values for Kollam and, for comparison, Thrissur. They describe how long cases took, how many hearings they required, and how much of that time and effort was substantive.

1. Time to Disposal (days)

   District	Median	Mean
   Kollam	609	782
   Thrissur	792	1003

2. Hearings to disposal

   District	Median	Mean
   Kollam	12	15
   Thrissur	7	10

3. Share of Substantive Hearings

   District	Median	Mean
   Kollam	19%	26%
   Thrissur	25%	28%

   Note: The dataset spans 2015-2024; COVID-19 lockdowns (2020-2021) introduce exogenous shocks to the metric; it has not been adjusted for here.

4. Time to First Substantive Hearing

   District	Median	Mean
   Kollam	240	446
   Thrissur	276	447

   Note: Changes in procedure resulting from the shift from Criminal Procedure Code (CrPC) to the Bharatiya Nagarik Suraksha Sanhita (BNSS) may affect this metric.

5. Time Between Hearings

   District	Median	Mean	IQR
   Kollam	61	74	48
   Thrissur	123	121	62

Stage-wise Analysis

We analyse how time and hearings are distributed across case stages, as part of understanding the pre-reform baseline. In S138 cases, the case trajectory is:

• A complainant files a case.
• Post scrutiny and error correction, the case is registered as a miscellaneous petition.
• The judge takes cognizance and the case is now a criminal case.
• On cognizance, a summons is issued to the accused, followed by warrants and other methods to compel appearance of the accused.
• On appearance, the accused pleads not guilty and takes bail.
• Trial follows with evidence and arguments.
• The court delivers a judgement.

Not all cases go through all stages. For cases that reach each stage, we compute the median and mean time and hearings spent at that stage.

Kollam	Time		Hearings		Cases
	Median	Mean	Median	Mean
Filing - Registration	0	4	-	-	99*
Cognizance	149	268	2	4	63
Appearance	312	533	5	7	74
Trial	337	454	11	15	41
Judgement	27	38	1	1	100

Note: *An error on e-Courts for one case reports the registration before the filing, which has been excluded.

Thrissur	Time		Hearings		Cases
	Median	Mean	Median	Mean
Filing - Registration	0	1	-	-	100
Cognizance	235	317	2	2	52
Appearance	721	878	5	8	81
Trial	335	451	8	12	13
Judgement	42	69	1	1	100

In both districts, getting the litigants to court takes the most time. We also observe that most cases do not reach trial. This reaffirms earlier observations in other district courts.

Way Forward

24x7 ONCourts has been running for about a year. It is too early to compare outcomes - only a fifth of cases filed have been disposed. But this benchmark will enable comparison when the court reaches its full load. We aim to track the court through this journey and continue to share updates on the court's performance.

We will continue to refine our methodology. The baseline analysis provides simple statistics from a set of 100 disposed cases. This suffers from the problem of censoring. We are only observing cases where the event of interest has occurred and we do not take into account ongoing cases. In systems like the court, where pendency is common, we should be using methods like survival analysis to account for ongoing cases. Other methods like time series analysis could identify trends. With Thrissur as a control, a difference-in-differences design can attribute average changes to the 24x7 ONCourts intervention. The present benchmark is the starting point; as post-intervention data accumulate, we can apply more demanding econometric checks using these tools.

Data and metrics form the backbone of evidence-based reform. The 24x7 ONCourts dashboard reports case statistics on a public dashboard in near real-time. This openness to measurement and public accountability sets a precedent for judicial reform initiatives across India. Public reporting supports continuous monitoring, mid-course correction and iterative refinement. A steady flow of transparent data, paired with simple, repeatable measures, is the practical route to learning what works.

References

24x7 ONCourts website.

Vision, ONCourts 24x7.

Evolution and Implementation of the 24x7 ONCourts: A Journey of Innovation Through Collaboration, PUCAR, 2024.

In Service of the Republic: The art and science of public policy, Vijay Kelkar and Ajay Shah, Penguin Allen Lane, 2022.

Approach to evaluate court reforms, Siddarth Raman, Seminar #9 of the series 'Indian Legal System Reform' by XKDR Forum, 2024.

Evaluating contract enforcement by courts in India: a litigant's lens, Pavithra Manivannan, Susan Thomas, and Bhargavi Zaveri-Shah, XKDR Working Paper No. 16, 2022.

How substantial are non-substantive hearings in Indian courts: some estimates from Bombay, Pavithra Manivannan, Karthik Suresh, Susan Thomas, and Bhargavi Zaveri-Shah, The Leap Blog, 6 December 2023.

Get them to the court on time: bumps in the road to justice, Mugdha Mohapatra, Siddarth Raman, and Susan Thomas, The Leap Blog, 12 June 2025.

Understanding Judicial Delays in Debt Tribunals, Prasanth Regy and Shubho Roy, Working Paper 195, National Institute of Public Finance and Policy, 2017.

Making courts transparent: What every litigant should know before filing a case, Pavithra Manivannan, Siddarth Raman, Gokul Sunoj, and Bhargavi Zaveri-Shah, Bar and Bench, 2025.

A litigant's guide to courts: How efficient is your court?, Pavithra Manivannan, Siddarth Raman, Gokul Sunoj, and Bhargavi Zaveri-Shah, Bar and Bench, 2025.

A litigant's guide to courts: Understanding predictability in India's courts, Pavithra Manivannan, Siddarth Raman, Gokul Sunoj, and Bhargavi Zaveri-Shah, Bar and Bench, 2025.

A litigant's guide to courts: Towards a more measurable future, Pavithra Manivannan, Siddarth Raman, Gokul Sunoj, and Bhargavi Zaveri-Shah, Bar and Bench, 2025.

The Long Road to Change, Ajay Shah and Amit Varma, Episode #36 of the podcast 'Everything is Everything', 1 March 2024.

Siddarth Raman is senior research lead at XKDR Forum. XKDR Forum is part of PUCAR - Public Collective for Avoidance and Resolution of Disputes. PUCAR is a knowledge partner to the High Court of Kerala in their 24x7 ONCourts initiative.

From Statute to Zero-Cost: Section 31A and the Bombay High Court's Zero-Cost Culture

by Prashant Narang and Vishnu Suresh.

Why this matters now

In 2015, Parliament rewrote Section 31A of the Arbitration and Conciliation Act to make loser-pays the default i.e., compensate the winner and deter abuse. A decade on, our mixed-methods study of the Bombay High Court's arbitration docket(2023-24) shows a stark gap between that legislative design and courtroom reality. Out of 102 decisions under Sections 11 (appointment) and 34 (set-aside), costs were imposed in only four (3.9%). Even then, costs were framed as exceptional sanctions for egregious conduct - not as routine reimbursement that follows the event.

Parliament's 2015 insertion of Section 31A followed the 246th Law Commission Report (2014), which condemned India's token-costs culture under the Code of Civil Procedure and urged a structured, outcome-linked "costs-follow-the-event" rule for arbitration. Section 31A sought to import the English loser-pays norm to ensure both indemnity and deterrence. Yet, a decade later, the provision functions largely as a dead letter.

This reluctance is not harmless. When weak Section 11 or Section 34 petitions carry no real downside, the expected cost of delay collapses onto the opponent and the taxpayer. The result is a deterrence gap: fewer incentives to screen out speculative filings, more tactical adjournments, and a credibility problem for India's "arbitration-friendly" promise.

In our new paper (publicly available), we build the first mixed-methods baseline for any Indian High Court on post-2015 costs in arbitration-related litigation. We combine a complete corpus of BHC decisions in 2023-24 with 22 confidential stakeholder interviews (judges, arbitrators, and counsel) to explain the pattern and to propose targeted reforms.

What we did

Prior commentary (and court exhortations in Salem and Uflex) diagnosed India’s token-cost culture but largely at the Supreme Court or in international arbitration. So, we chose to focus on a major High Court post-2015. We identified every Bombay High Court decision under Sections 11, 34 (and related Section 37 appeals where relevant) in calendar years 2023–24; after deduplication and exclusions, 102 reasoned decisions remained (54 in 2023; 48 in 2024). We coded each decision for whether costs were awarded, quantum, reasons, and whether judges engaged with Section 31A’s statutory factors. Interviews (22 conducted) supplied explanatory mechanisms behind judicial practice.

What we find

3.9% of decisions imposed costs; 96.1% did not. Where imposed, costs were justified as punishment for delay/illegality - not as routine indemnity. In 60 decisions, the court itself recorded conduct that maps onto Section 31A triggers. Costs were imposed in 4; in the remaining 56, no cost order followed. This selective enforcement blunts deterrence. With no articulated scale or factors applied, lawyers cannot advise clients on realistic exposure.

Why the default drifts to "zero-cost"

One, Section 31A(1) gives courts/tribunals discretion to determine costs. Section 31A(2) supplies a loser-pays presumption, but only if the decision-maker chooses to engage. Written reasons are required for departing from the presumption - not for declining to consider costs at all. This creates a one-way gate.

Two, interviewees repeatedly flagged a cultural aversion to "appearing punitive". For judges approaching retirement, the post-retirement arbitration market sharpens this caution: visible toughness on costs can be perceived as party-unfriendly, potentially affecting future appointments. That behavioural preference often persists into arbitral roles.

Third, Section 31A lists factors but lacks practical scales. In our four cost-imposing cases, sums ranged from Rs. 1 lakh to Rs. 40 lakh - with thin links to documented outlay. Practitioners consequently cannot price risk ex ante; costs become a lottery, not a calculable exposure.

How reasons are (not) written

Most orders end with a formulaic "no order as to costs", even when the judgment itself records behaviour that Section 31A treats as cost-triggering: obstruction, delay, frivolous claims, wasted hearings, or refusal of reasonable settlement.

Two of the four cost-imposing judgments offer only a one-line figure; two provide a brief narrative (e.g., "taking a chance") and sometimes cite one cost figure from the record, but none works through Section 31A's checklist or ties quantum to documented expense and deterrence. The resulting spread (Rs. 1-40 lakh) looks unanchored. For users, this reads like gesture, not governance.

What needs to be done

Our proposal aligns with the 246th Law Commission (2014) and the T.K.V. Committee (2024). One, create a rebuttable default: award reasonable, receipted costs to the successful party. Require brief reasons (even 1-2 sentences) when withholding costs. This small change collapses the one-way gate that currently rewards inaction. Two, keep indemnity as the baseline and add calibrated uplifts for documented delay tactics, non-disclosure, wasted hearings, or unreasonably rejected settlement offers - all already listed in Section 31A(3). Three, consider importing a narrow, high-threshold wasted-costs power (on the UK model) against improper or negligent conduct that needlessly increases expense. This targets the source of delay when it lies with representatives rather than clients.

Limitations

We included only one jurisdiction (Bombay), a two-year window, and interviews that exclude sitting judges and in-house counsel. Arbitral awards and proceedings were out of scope (confidentiality), though interviews illuminate arbitral practice.

Conclusion

Section 31A promised efficient, fair, and predictable cost allocation. In the Bombay High Court's arbitration-related litigation, it has largely delivered neither indemnity nor deterrence - except at the margins of overt misconduct. A rebuttable default in favour of reasonable costs, brief reasons for exceptions, receipt-anchored scales, and conduct-linked uplifts would realign daily practice with Parliament's indemnificatory intent and close the deterrence gap that now fuels speculative petitions.

Read the full working paper "From Statute to Zero-Cost: Section 31A and the High Court's Zero-Cost Culture", here.

The authors are researchers affiliated with TrustBridge Rule of Law Foundation.

Beyond Pendency: Counting Cases Correctly

by Pavithra Manivannan, Siddarth Raman and Bhargavi Zaveri-Shah.

The discourse on Indian judicial reform is dominated by questions of pendency and the workload of courts. However, official sources of caseload estimates in India have been found to be deficient in terms of both the methodology used, and the quality of underlying data (Jain and Reddy, 2025; Damle and Anand 2020). This leads to miscalculation of the caseload of courts and renders it unamenable for comparison across courts. In this article, we propose a new approach for estimating the caseload at Indian courts. We apply this to analyse the caseload at the Original Side of the Bombay High Court, which accounts for 35% of the total caseload of the Court. Our analysis yields three main findings. First, the caseload at the Original Side of the Bombay High Court (the Court) is being overcounted by 66%. Second, the caseload composition of the Court has remained largely stable over the 7-year period of our study with two case-types, namely, inheritance cases, and writ petitions filed against the government, accounting for half the cases filed in the Court. Third, suits as a case-type generates the most number of sub or interim cases.

In India, there are two official sources that publish information on caseloads - the annual report of the Supreme Court and the National Judicial Data Grid (NJDG). Apart from the quality of the data, there are two specific problems with the estimation methodology used by these sources. First, as a case progresses in a court of law, it generates multiple sub-cases. For instance, if a case is filed as a "Suit" for recovery of money, several interlocutory applications may be filed through which the main money suit (the 'main case') progresses. Such sub-cases could range from simple applications seeking the addition of a new party to the proceedings to an interim injunction seeking a stay on the transfer of assets of the respondent. Currently, the NJDG counts such sub-cases as distinct cases. This leads to overestimation of the caseload, inflating pendency and disposal rates. This is because the hearings for sub-cases are held as part of the main case proceedings. Further, a reading of the orders of cases suggests that more often than not, the final disposal order is common for both the main case and its sub-cases. Second, the taxonomy for case-type categorisation is inconsistent across official sources. The Bombay High Court's website lists 142 case-types on its Original Side. On the other hand, the NJDG reports only 19 case-types for the Original Side of the Bombay High Court. This includes an 'Original' and an 'Other' category, which provides little to no information on the case-type filed in the court. Further, the annual report of the Supreme Court has an altogether different classification system, which cannot be readily mapped to the other two official sources. It has a large bucket under 'Other' which does not have a clear definition. Our approach attempts to address these problems.

We count sub-cases as part of its corresponding main case. That is, we adopt the 'family of cases' as the unit of analysis for caseload estimation. This involves collapsing the 142 case-types on the Original Side of the Court into 17 main case-types and two sub case-types, based on their subject. For example, of the 142 case-types, 80 case-types are in the nature of sub-cases such as "Interim Applications", "Leave Petitions", "Chamber Order Lodging" and "Notice of Motion". We classify these as "Interim Applications" and count these as sub-cases. Similarly, "Arbitration Petitions" and "Arbitration Applications" are categorised as "Arbitration cases". This standardisation of case-types makes the caseload estimation exercise scalable and amenable to comparison across similar courts. The list of the 142 case-types and the classification assigned by us can be accessed here.

Data and Methodology

We collect the life-cycle data of 2,36,953 cases filed at the Original Side of the Court between the period January 2017 to December 2024 (Study Period). The Bombay High Court exercises original jurisdiction or jurisdiction over first time civil cases, and appellate jurisdiction or jurisdiction over cases that come before it as appeals from lower courts. We source the information on the life-cycle of cases filed at the Court's original jurisdiction from its website, and it is comprehensive to the extent the Court has made the data available.

As on the date of our data collection exercise (February 2025), the Court's website reported 1,43,514 (61%) cases as disposed of and 93,254 (39%) cases as pending. If the status of a case was unknown or marked as transferred, we classify it into the 'Other' category (185 cases).

We tag each case in our dataset as a main case or a sub-case. Next, we create a family of cases using the CNR number assigned by the Court as the unique identifier. This family of cases becomes our unit of analysis. Finally, each family of case is classified into one of the 17 case types.

Finding 1: Official sources overestimate caseload

The Court's website shows that about 2.5 lakh cases are filed before its Original Side during our Study Period. That is, on an average, about 30,000 cases are filed every year. However, we find that about 40% of these cases are sub-cases (Table 1 below). Viewed in this light, the 30,000 new cases per year can be understood as an overestimate. The annual average of new main cases filed before the Original Side of the Court is about 18,000, almost half of the original estimate.

Table 1: No. of filings

Nature	Count	Average per-year	% of total
Main cases	1,41,608	17,850	60
Sub-cases	95,435	11,769	40
Total	2,36,953	29,619	100

Finding 2: Six case-types dominate caseload

On applying our categorisation framework, we find that six case-types contribute to about 95% of the caseload at the Original Side of the Court (Table 2). In that, Inheritance cases and Writ petitions constitute half the caseload. We also find that the share of case filings across years for these categories do not vary significantly.

Table 2: No. of filings per case-type

Case Category	Count	% of Total
Writs	36,145	25.5
Inheritance and Succession cases	32,979	23.3
Execution cases	19,779	14.0
Tax cases	19,665	13.9
Arbitration cases	16,529	11.7
Suits	8,652	6.1
Other	7,859	5.6
Total	1,41,608	100.0

Finding 3: Suits generate the most sub-cases

We take a closer look at the number of sub-cases per main case in Table 4 for the top six case-types. We find that, while Inheritance cases and Writ petitions are the highest contributor to the caseload of the Court, Suits that is at the bottom of Table 2, has the highest number of sub-cases per main case. 50% of Suits have upto two sub-cases, suggesting that on a per-case basis, Suits may generate more workload for judges compared to Writ petitions and Inheritance cases.

Table 3: Sub-cases per case-type

Case category	Sub-cases per main case (in %)
	0	1-2	3-5	6-10	>10
Writs	86	13	1	0	0
Inheritance and Succession cases	77	21	2	0	0
Execution cases	86	13	1	0	0
Tax cases	84	16	0	0	0
Arbitration cases	84	15	1	0	0
Suits	31	52	14	3	0

Conclusion

Our finding that the caseload of the Bombay High Court is overestimated by about 66% likely means other courts across India are overreporting caseloads as well. When official sources like the NJDG count sub-cases as distinct new filings, it exaggerates the problem of pendency. This prompts the policymakers to focus on solutions like increasing the number of judges, and creating more courts or courtrooms. Such a sole focus on this metric not only neglects the underlying data quality issues leading to inefficient resource allocation but also ignores the unique challenges that each type of case filed in the court face.

Measures of the economy such as GDP, inflation, and employment rate, took decades to be built and continue to be challenged and improved, by researchers and policy-makers alike. Similar sound systems for the measurement of court metrics, of which caseload is only one part, need to be developed. Such systems are imperative for any meaningful discussion on court reform.

References

Chitrakshi Jain and Prashant Reddy T. Tareekh Pe Justice: Reforms for India's District Courts. Simon and Schuster India, 2025.

Devendra Damle and Tushar Anand. Problems with the e-Courts data. NIPFP WP Series, 314, 2020.

Mugdha Mohapatra, Siddarth Raman and Susan Thomas. Get them to the court on time: bumps in the road to justice. The Leap Blog, 2025.

The authors are researchers at XKDR Forum, Bombay.

Get them to the court on time: bumps in the road to justice

by Mugdha Mohapatra, Siddarth Raman and Susan Thomas.

India's district courts currently face a staggering backlog of 4.6 crore pending cases (as of May 2025): 3.5 crore criminal and 1.1 crore civil. Proposals to solve this are familiar: hire more judges, build special courts, adopt new technology. But before rushing to solutions, it is important to understand where cases get stuck in their journey through courts. We hand-collect and analyse the life-cycle of a sample of cases from district courts, with some surprising observations. First, between 50-70 percent of cases are disposed before they get to trial, which is before the judge hears the substantive matter of the dispute. The time spent waiting for parties to appear is over a year. While criminal cases necessarily require strict adherence to due process, even civil cases face delays. These findings challenge conventional wisdom about judicial delays and point to a unexpected bottleneck. If getting people to show up in court is the core source of delays and pendency, strengthening the administrative processes of the court rather than the size of the bench, could lead to more speedy justice delivery from our courts.

The objective of building judicial capacity to achieve judicial efficiency requires an understanding of how cases move through courts, not just tracking pendency rates. This is because the journey of a case moves through deterministic stages, which vary in duration, and imposes varying resource demands from judges and staff. There have been few systematic studies of how a case moves through court. While some studies examine the total time for case disposal, few break this down by stage.

This study analyses stages for two common types of cases that represent a significant portion of the workload of courts: cheque bouncing cases (criminal matters under Section 138 of the Negotiable Instruments Act) and motor accident claims (civil matters under the Motor Vehicles Act).

Cheque bouncing cases account for 10-15% of criminal court workloads, while motor accident claims constitute over 10% of pending civil cases. Cheque bouncing cases happen when a cheque issued does not deliver payment as expected. Motor accident claims are filed to claim compensation for damages caused in the accident against the owner of the vehicle involved, with the vehicle insurance company as a co-respondent. Cheque bouncing cases are filed in a magistrate court, and the motor accident claims at the Motor Accident Claims Tribunals (MACT). Across these two types of cases, there are differences in procedures: whether it is for criminal and civil cases, and for different types of courts.

We use this analysis to answer the following questions:

1. What fraction of the cases go through the whole life-cycle?
2. How much time is spent in different stages of the case life-cycle? Is this different for civil and for criminal cases?

Methodology

The analysis examined 200 disposed cases randomly sampled from from the e-courts database for district courts from courts across Maharashtra, Kerala, Karnataka, Tamil Nadu, Delhi, Telangana and Rajasthan, filed between 2018-2022. After excluding transfers and circumstances where cases never went a court process, the final sample included 147 cases - 77 cheque bouncing cases and 70 motor accident claims cases.

Each case was tracked through its entire journey by analysing court orders and hearings. Cases go through different stages - filing, admission, summons, warrants, bail, written statements, framing of issues, evidence and others. We classify these different stages of 'Pre-Trial' and 'Trial'. Trial begins after both parties appear before the judge - in cheque bouncing cases, after the accused files for bail; in motor accident claims, after written statements are filed and issues are framed.

Results

1. The first finding relates to the stage at which the cases are disposed. Table 1 shows the number of cases disposed at each stage.

Table 1: Where cases end their journey

Stage	Case type: MV		Case type: S138
	No. of cases	Percentage	No. of cases	Percentage
Pre-trial	38	54 %	55	71 %
Trial	32	46 %	22	29 %

More than half of the cases analysed never reached trial. This is higher for the (criminal) cheque bouncing cases, where 70% of cases are disposed before they reach trial. For the (civil) motor accident claims cases, 54% of the cases are disposed before reaching trial.

2. The second finding relates to the time spent in the two stages

Table 2: Time taken by stage

Case type	Total no. of cases	Pre-trial	Trial
MV	70 (32 reached trial)	9.5 months	4 months
S138	77 (22 reached trial)	12 months	3.5 months

Once all parties are in present in court, cases resolve quickly - usually in 3-4 months. Most of the delay in matters is in the pre-trial stage where the court is waiting for parties to appear (usually the respondent). This takes between 9 months to a year.

These findings align with broader patterns visible in the National Judicial Data Grid for district courts (NJDG). The data shows that 72% of pending cases are stuck before trial: 48% are at the appearance stage, 14% are awaiting service of summons, and 10% are awaiting service of warrants. While the data from the NJDG is useful to know where cases are placed within the judicial system, it does not provide insights on the time spent in different stages. Our analysis quantifies the extent of the bottleneck.

Discussion

The analysis points to two key observations: Most cases that are filed in court do not reach trial, where judicial mind is applied to decide issues of the case. Further, the bulk of the time is spent in getting the parties to court. Once all parties are present, the time to resolution is much lower. The puzzle is in understanding what shapes these features, and how this understanding can be used to improve court efficiency in dealing with case workload.

• Are the delays in court cases inevitable?

The analysis points to the paradox of procedural protections for some cases. Cheque bouncing cases and other criminal matters demand the presence of the accused. This creates an inherent tension between speedy resolution of the matter and judicial procedure. The accused, facing potential imprisonment, has every incentive to delay appearing in court until forced by warrant. The very protections meant to ensure fair process become tools for delay.

In India, these procedures continue to evolve. Under Section 223 of the newly introduced Bharatiya Nagarik Suraksha Sanhita, magistrates must now offer the the accused an opportunity to be heard before admitting a complaint as a criminal case. This involves sending a notice by post, a process not unlike the current summons process. While intended to enhance due process, this additional step could further extend the timeline for cheque bouncing cases. The new code also allows for trial 'in absentia' under Section 356. If a person is declared as a 'proclaimed offender', and if the judge thinks that they are absconding to evade trial, the court can proceed without the accused. How these practices are implemented remains to be seen.

• Administrative and judicial functions of the court

The findings expose a fundamental blind spot in how courts actually work. The popular image of justice - a judge hearing arguments, weighing evidence and delivering verdicts - represents only one aspect of the judicial system. Behind every courtroom drama lies an extensive administrative operation of filing documents, scheduling hearings, maintaining records, and getting parties to court. These two systems complement each other, but our understanding of the administrative aspects of the court system is limited, because it is behind the scenes.

Current reform proposals focus heavily on expanding judicial capacity: hiring more judges, creating specialised courts, and implementing new technologies for case management. While these interventions have merit, they miss the core issue revealed by this analysis. The judicial system extends far beyond judges and courtrooms. Delivering summons and notices typically involves police officers, postal services, or process servers. When the simple act of getting parties to court becomes the biggest bottleneck, the solution requires rethinking the entire administrative infrastructure supporting the courts.

What does imply for potential solutions for institutional reforms of the judiciary? Some approaches that could address the summons/notices bottleneck include:

1. Digital service of summons and notices could reduce delays, though this requires updated legal frameworks and reliable technology infrastructure.
2. Police-court integration might improve warrant execution, though this raises questions about optimal resource allocation - should a capacity constrained police forces pursue cheque defaulters or focus on serious crimes?
3. Quicker escalation to warrants may secure attendance faster, but wielding state power to restrict liberty demands careful consideration. A judge's decision to issue an arrest warrant carries real consequences.
4. Penalties for non-appearance could be introduced to create stronger incentives for timely court attendance.
5. Private process servers, as used in U.S. courts, offer another model worth exploring.

Conclusion

The clamour for court reform has been dominated by traditional solutions: more judges, rewritten procedures, and new technology. But when the relatively simple task of getting parties to court becomes the system's biggest bottleneck, a more nuanced approach is essential. Court reform must recognise that efficient justice delivery requires strengthening both judicial and administrative capacity in parallel. Separating court administration from judicial functions, as some countries have done, could allow specialised focus on each component while maintaining their complementary relationship.

The invisible administrative machinery of courts deserves as much attention as the visible judicial functions. Until administrative capacity matches judicial capacity, Indian courts will continue struggling with delays that have less to do with complex legal reasoning and more to do with basic case management. The path to speedier justice may lie not in the courtroom, but in the clerk's office, the process server's route, and the administrative systems that bring cases to life. Only by addressing both aspects of the judicial system can India's courts deliver the swift justice that 4.6 crore pending cases demand.

Siddarth and Susan are senior research lead and senior research fellow at XKDR Forum. Mugdha was a research associate at XKDR Forum. We thank Pavithra Manivannan for insights, Shubho Roy for help with the interpretation, and Ajay Shah for inputs.

Evaluating India's Customs Authority for Advance Rulings (CAAR) and charting a path for reform

by Vijay Singh Chauhan, Prashant Narang, and Monika Yadav.

Advance rulings are critical for trade facilitation - they offer clarity on tariff classifications, customs duties, and valuation, enabling importers and exporters to navigate complex regulatory environments with confidence.

India's journey with advance rulings began in 1999 with the establishment of the Authority for Advance Rulings (AAR), which handled both direct and indirect tax matters. However, the AAR faced severe criticism for its procedural inefficiencies and delays. As one senior customs consultant quoted in the paper noted, "We had cases pending for 4-5 years, forcing many businesses to abandon their plans entirely." The centralised structure, with its single Delhi office, created substantial logistical challenges for businesses across India.

In response to these shortcomings, the Customs Authority for Advance Rulings (CAAR) was introduced in 2018 under Chapter VB of the Customs Act, transforming India's framework from a judicial model to a quasi-judicial one led by senior customs officers. This reform aimed to leverage domain-specific expertise and decentralise operations with benches in Delhi and Mumbai.

However, has CAAR succeeded in delivering timely and consistent rulings, and how does its performance measure up against international benchmarks?

In our recent paper, “Decoding CAAR: Insights, Challenges, and Pathways for Reforms”, we critically assess CAAR's performance between January 2021 and August 2024. Our mixed-methods analysis combining stakeholder interviews with quantitative evaluation of 414 advance rulings uncovers systemic inefficiencies impeding CAAR's effectiveness, notably delays beyond the statutory 90-day timeframe and inconsistencies from limited nationwide applicability.

Despite improvements over its predecessor (AAR), CAAR remains burdened by procedural bottlenecks - chiefly, dependence on port commissioners for technical inputs, uneven workload distribution, and outdated manual processes. Drawing comparisons with jurisdictions like the U.S., Canada, and Australia, we propose actionable reforms: establishing dedicated in-house technical expertise, adopting AI-driven case management systems, and ensuring the nationwide and indefinite applicability of rulings.

By identifying critical gaps and presenting pathways for reform, our research seeks to align CAAR with global standards -essential for strengthening India's role as a reliable global trade partner.

Measuring CAAR's performance: The 90-Day challenge

A central finding of the research is that CAAR struggles to meet its statutory obligation to issue rulings within 90 days. The analysis of rulings issued between January 2021 and August 2024 reveals that only 46.2% of decisions were delivered within this mandated timeframe. This compliance rate varies dramatically among officers, with one achieving 86.1% compliance while another managed just 2.2%.

The primary bottleneck identified is CAAR's dependence on port commissioners for technical inputs. Although regulations allow commissioners just two weeks to provide comments, these responses are often delayed, extending the ruling process by months. As one CAAR presiding officer acknowledged in an interview, delays frequently occur when "comments from jurisdictional commissioners are not received on time," leaving officers with "no option but to delay further".

Some CAAR officers have developed informal practices to mitigate these delays, including sending reminders, making personal phone calls, and issuing demi-official letters. However, these efforts reflect systemic inefficiencies rather than sustainable solutions. The research also highlights the CAAR's reluctance to issue ex parte rulings (without port commissioner input), despite having the authority to do so under Regulation 8(8) of the CAAR Regulations, 2021.

Port-specific applicability: A self-imposed limitation

Another significant limitation is the port-specific applicability of rulings. Unlike systems in the United States, Canada, and Australia- where advance rulings apply nationwide - CAAR rulings are binding only at the specific port where they're issued. This creates inconsistent enforcement across India's customs jurisdictions, forcing businesses that import through multiple ports to seek separate rulings for identical goods.

One respondent articulated this frustration: "Rulings should be consistent across all ports. My classification should not fall under X at one port and Y at another". This limitation not only increases administrative burdens but also undermines the predictability that advance rulings are designed to provide.

The temporal restriction of rulings to a three-year validity period further compounds these challenges. Globally, countries adopt more flexible approaches - Australia's rulings remain valid for five years, while those in Canada and the U.S. have indefinite validity unless there are changes in law or circumstances. As one participant noted, "Unless there is a change in the product or technology, limiting advance rulings to three years seems unnecessary".

Workload imbalance: The Mumbai-Delhi divide

The research reveals significant disparities in workload distribution between CAAR's two benches. The Mumbai bench handles substantially more cases (256) than Delhi (158), with Maharashtra alone accounting for approximately 37.11% of Mumbai's workload. This concentration of cases in Mumbai is followed by Tamil Nadu (31 rulings, 12.11%) and Karnataka (27 rulings, 10.55%), with these three states collectively accounting for about 59.77% of Mumbai's workload.

In contrast, Delhi's jurisdiction shows a different distribution pattern, with Delhi (NCT) itself accounting for 54 rulings (34.18%), followed by Haryana (23 rulings, 14.56%) and Uttar Pradesh (9 rulings, 5.70%). These regions together contribute approximately 54.43% of Delhi's total caseload. The Mumbai bench also faces the additional challenge of 75 orders lacking specified applicant addresses, which further complicates case management.

While both benches experience procedural bottlenecks - such as delays in receiving feedback from jurisdictional commissioners - the Mumbai bench appears disproportionately burdened, given its coverage of the economically significant regions of Western and Southern India. The paper acknowledges this workload imbalance but, rather than recommending additional benches, focuses on process-oriented solutions discussed below.

A path forward: Recommendations for reform

The paper proposes several actionable reforms to enhance CAAR's efficiency and alignment with global best practices:

1. Transition to a Technical Unit Model - Establish in-house technical expertise through dedicated classification specialists and valuation analysts, modeled after systems in Australia, Canada, and the U.S. Pilot at one bench first, with performance measured through turnaround times and stakeholder feedback.
2. Digital Process Optimisation - Implement AI-driven case management using Large Language Models (LLMs) to auto-generate case briefs and identify classification issues. Develop long-term AI solutions integrating HS codes, trade agreements, and global tariff jurisprudence.
3. Nationwide Applicability of Rulings - Amend Section 28J(1)(c) of Customs Act to mandate uniform enforcement across all Indian ports, eliminating jurisdiction-specific inconsistencies.
4. Extending Ruling Validity - Introduce auto-renewal mechanism maintaining rulings' validity unless material facts or trade laws change, reducing business compliance burdens.
5. Enhanced Transparency and Accountability - Create real-time performance dashboard tracking case disposal rates, 90-day compliance, appeal rates, and ruling consistency while maintaining necessary confidentiality.

Implications for India's trade ecosystem

The study's findings have significant implications for India's position in global trade networks. While CAAR represents progress compared to its predecessor, systemic inefficiencies continue to hinder its full potential. Addressing these challenges is crucial not only for domestic traders but also for strengthening India's reputation as a reliable trade partner internationally.

The research highlights an encouraging statistic: more than two-thirds of CAAR rulings align with the applicant's proposed position. This suggests that when the system functions effectively, it provides valuable certainty to businesses. However, the procedural bottlenecks identified in the study prevent this benefit from being fully realised.

As global trade regulations evolve and become increasingly complex, ensuring that CAAR remains agile and responsive is critical to sustaining India's economic growth. The reforms proposed in this paper offer a roadmap for enhancing the efficiency and relevance of advance rulings within India's broader trade facilitation framework.

Conclusion

This process audit of India's Customs Authority for Advance Rulings (CAAR) provides a rigorous assessment of its strengths and limitations. The study effectively documents progress since transitioning from AAR while identifying persistent operational inefficiencies, particularly the 90-day timeline compliance challenge, port-specific applicability constraints, and the technical expertise gap compared to global benchmarks.

For policymakers and trade stakeholders, this research offers a clear roadmap to transform CAAR. The evidence-based recommendations target critical friction points in CAAR's workflow: establishing in-house technical expertise to reduce dependence on port commissioners, implementing AI-driven case management, expanding nationwide ruling applicability, and extending validity periods. These practical reforms align with international best practices observed in jurisdictions like the United States, Canada, and Australia.

Here is the link to the paper.

Vijay Singh Chauhan is a Executive Director at Deloitte Touche Tohmatsu India LLP, Prashant Narang and Monika Yadav are researchers at the TrustBridge Rule of Law Foundation.

The Blind Spot in Indian Arbitration: Fees, Power, and Structural Oversights

by Prashant Narang and Vishnu Suresh.

In India, when parties fail to agree on the composition of an arbitral tribunal, courts intervene and appoint retired judges as arbitrators, who unilaterally determine their own fees-without the consent of both parties. This process, known as "ad hoc" arbitration, has led to concerns about excessive charges. While no comprehensive dataset proves a systemic pattern of exorbitant fees, recurring judicial and committee observations suggest that the issue is widespread enough to warrant closer scrutiny. The Indian policy response has been to implement some form of fee regulation for such arbitration.

This article presents a history of the Indian policy thinking on arbitrator fees and presents an argument about why fee regulation alone may not remedy the structural inefficiencies in ad hoc arbitrations dominated by retired judges. Judges who design (or are expected to design) and implement arbitration appointment rules often later serve as arbitrators themselves, benefiting from these same rules - or the lack thereof - post-retirement. Even when they do not directly benefit, enforcing such rules against fellow judges, particularly their seniors in the profession, is challenging given the inherently hierarchical nature of the legal fraternity.

The article argues that the current fee regulation approach further entrenches judicial control over arbitration rather than reducing it. By deepening the judicialisation of the arbitration process, it raises further concerns about perpetuating systemic inefficiencies. At the same time, we explore whether a more fundamental shift towards institutional arbitration - centred on dejudicialisation and the decoupling of the judiciary from arbitration - is necessary to create a cost-effective, competitive, and independent arbitration ecosystem in India.

The evolution of the debate on arbitration fees

Concerns about high fees in arbitration were explicitly raised by the Supreme Court in Union of India v M/s Singh Builders Syndicate (2009) 4 SCC 523. The Supreme Court reiterated its concerns in Sanjeev Kumar Jain v Raghubir Saran Charitable Trust (2012) 1 SCC 455, acknowledging that high arbitration costs discouraged parties from opting for arbitration.

This focus on high fees has meant that Indian policy response has also relied on mandating "fee schedules" for tackling the problem. This is consistent with other jurisdictions as well. For example, Germany prohibits arbitrators from unilaterally deciding their own fees on the ground that it violates the prohibition on in rem suam decisions (i.e., ruling in one's own cause). Austria and Switzerland likewise disallow arbitrators to issue binding and enforceable orders regarding their own remuneration. Italy permits arbitrators to fix fees in the absence of explicit party agreement, but these fees only become binding after the parties themselves consent. Singapore, lacking a written fee agreement, lets a disputant seek assessment of fees by the Registrar of the Supreme Court under the Supreme Court of Judicature Act, 1969.

The Indian policy response

The key elements of the Indian response are as follows:

1. The Fourth Schedule under the 2015 Amendment: The 246th Report of the Law Commission of India (2014) recommended a structured fee schedule to bring uniformity to arbitration costs. This led to the introduction of the Fourth Schedule under the Arbitration and Conciliation (Amendment) Act, 2015, which provided a model fee framework for arbitrators in ad hoc arbitrations. In addition, it also inserted a provision empowering high courts to make rules for fee determination in case of domestic ad hoc arbitration.
2. Shifting towards institutional arbitration under the 2019 Amendment: The Arbitration and Conciliation (Amendment) Act, 2019 introduced a framework that shifted appointment powers from courts to arbitral institutions. The amendment required the Supreme Court and High Courts to designate arbitral institutions for making appointments under Section 11, rather than appointing arbitrators directly. The amendment goes a step further and creates a fallback mechanism for jurisdictions where graded arbitral institutions are not available. In such cases, the High Court Chief Justice can maintain a panel of arbitrators who effectively function as an arbitral institution. These empanelled arbitrators must follow the Fourth Schedule's fee structure, creating a hybrid between institutional and ad hoc arbitration. However, this part of the 2019 amendment is not notified yet.
3. Alternative fee arrangements by the TKV Report, 2024: In June 2023, the Ministry of Law and Justice constituted an expert committee, chaired by former Law Secretary T.K. Vishwanathan, to review arbitration costs and propose amendments to the Arbitration and Conciliation Act, 1996. The T.K. Vishwanathan Committee Report, 2024 identified multiple shortcomings in the existing Fourth-Schedule fee framework, most notably the reliance on "claim quantum" as the primary basis for calculating arbitrator fees. Such a simplistic approach, the Report argued, neglected case complexity and procedural variations: for instance, an ostensibly small claim requiring extensive oral evidence or expert testimony can command more arbitrator time than a large claim resolved on documents alone. In response, the TKV Report advocated Alternative Fee Arrangements (AFA), emphasising value-based pricing that accounts for factors like complexity, time, and potential cost savings. Most notably, the TKV Report suggested eliminating Section 11A and the Fourth Schedule entirely, replacing them with a more flexible framework in which the Central Government would prescribe fee structures through rules.
4. The Draft Arbitration and Conciliation (Amendment) Bill, 2024: This draft bill empowers the Arbitration Council of India (ACI) as a proxy for the Union Government to specify arbitrator fees. Under the Bill, the Fourth Schedule would be deleted, and Section 11A would be revised so that the ACI could determine fees, except where parties have explicitly negotiated their own fee arrangement or are using an arbitral institution with its own fee rules. Another significant change is the removal of the Chief Justice's consultative power in the appointment of ACI's governing board, shifting oversight from judicial control to greater executive control of the arbitration regulatory body.

Why the fee regulation approach has fallen short

Before we analyse the reasons for the failure of the 2015 amendment and the Fourth Schedule, it is useful to describe the political economy that confronts any policy change on arbitration: namely, the near-monopolistic environment created by a small group of retired judges who often command premium fees and face minimal accountability. Courts retain the ultimate power to appoint arbitrators under Section 11, and this process frequently involves the same cadre of retired judges who benefit from the laxity of fee caps. The entire appointment and fee determination process is still largely vested in the judiciary. Given that many judges become arbitrators upon retirement, they have little incentive to enforce rigorous fee caps that might constrain their own future earnings. This fundamental public choice problem has been frequently documented, including by the Vice-President's observation that "nowhere in the world is arbitration in such tight fist control as in our country".

The ONGC v Afcons Gunanusa JV (2022) provides a telling example of how court-appointed arbitrators can exploit their position. Despite initially accepting a contractual fee cap of Rs. 10 lakh per arbitrator, the tribunal - composed of retired Supreme Court and High Court judges - unilaterally enhanced their fees multiple times. They first sought adoption of the Fourth Schedule's more generous framework, then further increased their fees citing case complexity, and even attempted to apply these increases retrospectively. When ONGC, a public sector enterprise subject to audit scrutiny, refused to pay the enhanced fees, the arbitrators recused themselves, forcing the matter back to court. The Supreme Court ultimately had to terminate the tribunal's mandate, highlighting how the current system enables arbitrators to leverage their position to demand higher fees with limited accountability.

This reality was not addressed by the 2015 amendment. While the Fourth Schedule was introduced precisely to limit excessive fees, it was neither made mandatory nor accompanied by a robust enforcement mechanism. As a result, it did little to disrupt the underlying political economy that sustains high-cost ad hoc arbitration. In fact, it risked consolidating judicial influence rather than attenuating it, especially since it granted High Courts the discretion to frame their own fee rules, ultimately placing regulatory power over arbitrator remuneration in the hands of those who may later serve as arbitrators themselves.

Moreover, this one-size-fits-all imposition overlooked regional variations and pre-existing institutional successes. The Karnataka Arbitration Centre, for instance, already offered a more economical schedule capped at around Rs. 12 lakhs for disputes above Rs. 20 crores, whereas the Fourth Schedule ceiling reaches Rs. 30 lakhs based on the thresholds set by the Delhi International Arbitration Centre. Rather than drawing on such local expertise and diversity to foster competitive discipline, the reforms proceeded on a centralised model that did little to leverage market discovery or locally tailored fee structures. The Law Commission's proposals were more concerned with containing arbitrator fees than with dismantling the structural conditions (judicial appointments, confined arbitrator pools, discretionary rule-making by courts) that perpetuate high costs.

The subsequent 2019 amendment intends to reduce judicial intervention and promote institutional arbitration. It revised Section 11 so that courts could "designate" arbitral institutions for appointments. "Fallback" arrangements enable High Court panels of arbitrators - often the same retired judges or those close to the judiciary - to retain effective control over the process, with fee structures mandated by the Fourth Schedule.

Recent developments, including the constitution of a new committee chaired by T.K. Vishwanathan in 2023, reflect growing discontent with the rigid claim-quantum basis that underlies the Fourth Schedule. The TKV Report contends that arbitrator fees should account more flexibly for complexity, time, and the overall resources required. While the proposed reforms contemplate eliminating the Fourth Schedule, transferring fee-setting authority to the Arbitration Council of India, and moving towards executive rather than judicial oversight of arbitration rule-making, they too risk replicating hierarchical models unless accompanied by genuine plurality and transparency in the appointment of arbitrators and the choice of fee structures.

Ultimately, each successive round of reform, from the 2015 amendment and the introduction of the Fourth Schedule to the latest proposals from the TKV Report, has prioritised adjusting fee schedules over reducing systemic reliance on a narrow circle of retired judges. The 2019 amendment and its stillborn promise of institutional appointments is an exception. As a result, what begins as a nominal attempt at "dejudicialisation" typically ends in reaffirming the dominance of court-nominated arbitrators, with little recourse for parties subjected to escalating costs. The persistent gap between nominal regulatory interventions and the practical realities of enforcement serves as a stark reminder that fee caps and model schedules, however laudable, are unlikely to produce fundamental change unless the structural incentives and entrenched hierarchies that govern Indian arbitration are addressed in earnest. Indeed, the recurring inclination to concentrate power - first in the High Courts, now potentially in the central government which is also the largest litigant - overlooks the fundamentally decentralised ethos of arbitration, which thrives on party autonomy and market-driven checks.

The next chapter in arbitration reform: Evidence-based vs. assumption-driven reforms

India's ongoing journey toward arbitration reform reveals a classic illustration of the "knowledge problem" that arises when policymakers attempt top-down interventions without robust, localised information. Observations from courts and committees certainly highlight inefficiencies - especially in court-appointed arbitrations that often lean on retired judges. Yet the absence of systematic, comparative data on whether these inefficiencies truly amount to a widespread market failure should give us pause before imposing sweeping fee controls or rigid schedules.

We must ask: do we need price caps because parties stuck in deadlock are unable to negotiate with court-appointed arbitrators? Or because retirees form a monopoly and pose barriers to entry? Excessive or poorly calibrated regulation can distort incentives and stifle innovation in arbitration services - problems that often follow when market-based processes are replaced by bureaucratic mandates. Fee ceilings, in particular, risk becoming a blunt tool that overrides local knowledge and decentralised experimentation. If parties truly had meaningful alternatives - like institutional forums or specialised arbitrators - they would naturally gravitate toward more cost-effective options, compelling fee discipline through competition rather than imposed caps.

Likewise, the unilateral fee determination by certain court-appointed arbitrators raises critical questions about capture - what might be called a narrowly "clubby" arrangement favouring a select group. But imposing top-down reforms in the absence of clear data on how widespread or severe this dynamic is invites "presumptive regulation". Such policy-by-assumption can inadvertently lead to higher costs, reduced choice, and entrenched favouritism - precisely the path we want to avoid.

By contrast, implementing the 2019 amendments and cultivating robust institutional arbitration offers a more polycentric and evidence-driven approach. This would expand the pool of competent arbitrators, reduce dependence on judge-led ad hoc appointments, and ultimately let competition, reputation, and local knowledge discipline fees. Notably, India's largest litigant - its own government - has already started shifting away from ad hoc arbitration, indicating that when parties sense an overcharge or imbalance, they do respond by seeking out better alternatives.

Before erecting rigid structures such as a universal Fourth Schedule, policymakers should verify that the alleged market failures cannot be resolved through the competitive process. Empirical, comparative research - analysing cost differentials between judge-led ad hoc arbitration and institutional arbitration - would illuminate whether exorbitant fees reflect a systemic shortcoming or isolated pockets of inefficiency. Only when we ground policy in such evidence can we ensure that reforms address real problems and do not accidentally lock in the very system they aim to correct.

Prashant Narang and Vishnu Suresh are researchers at the TrustBridge Rule of Law Foundation. We thank our colleagues Renuka Sane, Bhavin Patel, as well as two anonymous reviewers, for their comments.