Time to resolve insolvencies in India

In reality, the expected time to resolution of any case reaching the NCLT depends upon many factors ranging from the size of the filing party, the amount of debt involved, already existi[r]

Time to resolve insolvencies in India

Surbhi Bhatia surbhibhatia1906@gmail.com

Manish Singh manishsingh@igidr.ac.in

Bhargavi Zaveri bhargavi@igidr.ac.in

This work was done at Finance Research Group

Indira Gandhi Institute of Development Research, Mumbai

This is a preliminary draft Please not cite or distribute without permission of the authors July 29, 2019

Abstract

Contents

1 Introduction

2 A case study of timelines under the IBC

3 A dataset of IBC cases

4 Delays in bankruptcy resolution process

5 Methodology 12

6 Findings 14

6.1 For different benches: 15

6.2 For litigant type: 16

7 Conclusion 17 List of Tables Workload of benches by litigant type

2 Time taken by insolvency cases in the study period 10

3 Case outcomes for the top four benches 10

4 Cases admitted, disposed and pending by bench 11

5 Cases admitted, disposed and pending by litigant type 11

6 Probability of case ongoing beyond the benchmarked timelines for each bench 16

7 Probability of case ongoing beyond the benchmarked timelines for each litigant type 17

List of Figures Location of NCLT benches in India

2 Life cycle of a case under the IBC

3 Timelines prescribed under the IBC

4 Case outcomes per bench

5 Caseload across all benches 12

6 Survival probability : National average as on 30th June, 2018 14

7 Bench wise case duration as on 30th June, 2018 15

1 Introduction

A commonly used regulatory tool to ensure compliance with and adherence to a mandate is prescription of timelines within the body of a law When timelines are built into legal rules, they act as regulatory mechanisms altering incentives and behaviour Apart from facilitating compliance with requirements, timelines embedded in law help in fast tracking processes such as completion of judicial proceedings or grant of regulatory approvals Many laws in India have used this tool as a means to efficiently achieve the task at hand In the domain of public service delivery, for instance, these laws have time limits written down for the process of filing, disposal and appeals in case of grievances The Insolvency and Bankruptcy Code (IBC, the Code) passed in 2016 in India is one such law in the field of finance In this paper, we take the IBC as a case study to estimate the probability of case completion within the time frame provided by the law We this using survival analysis, a technique that helps in estimating the chances of a given commercial activity or policy action happening within that timeline

As a structural reform, the IBC provided for a modern framework encouraging debtors and creditors to report default by creating an environment of speedy timelines and efficient outcomes for the value of the distressed assets The legal environment for insolvency resolution prior to the passing of the Code included debt recovery laws such as Sick Industrial Companies Act (SICA) 1985, the Recovery of Debt Due to Banks and Financial Institutions Act (RDDBFI), the Securitisation and Reconstruction of Financial Assets and Enforcement of Security Interest Act (SARFAESI), 2002 and winding up provisions of the Companies Act, 1956 Sengupta, Sharma, and Thomas 2017 document the shortcomings of each of these legal regimes in fostering a culture of credit discipline in India The Report of the Bankruptcy Law Reforms Committee (BLRC), acknowledging the “multiple contradictory elements” in the legal arrangements prior to the IBC proposed the Code as a single framework for dealing with matters of individual and corporate insolvencies

In its design, the Code brought in four essential features First, it dispossesses the management by appointing an insolvency professional (IP) who takes charge of the proceedings Second, while moving to a regime where the creditors are in control of their assets, it puts financial and operational creditors on the same footing Third, the proceedings have limited touchpoints with the judiciary, only granting it the powers to judge compliance and fairness of a process being operated by a third party Lastly, the law aims at maximising the recovery value of the enterprise by providing a time bound process In the absence of a time ceiling for resolution, restructuring or liquidation, the process could become long drawn, eroding the economic value of the enterprise

National Company Law Tribunals (NCLT) and the National Company Law Appellate Tribunals (NCLAT) were two such institutions constituted under sections 408 and 410 of the Companies Act 2013, respectively With the passing of the IBC, the NCLTs were entrusted with the power to adjudicate on matters related to the same

There are at present 15 NCLT benches across India (Figure 1), each varying in their judicial capacity For our analysis, we look at the working of benches of the NCLT with the highest workload of cases filed before them For the NCLT, there are two critical timelines The 14 day period prescribed by law as duration between filing and admission or dismissal of an insolvency petition and the timeline within which the resolution plan is approved by the NCLT The timelines under the law are described in detail in section

Figure Location of NCLT benches in India

professional (RP) is appointed and within 30 days of the RP’s appointment a committee of creditors (CoC) is set up Alternatively, it can dismiss the case leading to provisions of appeal to NCLAT being applicable to the creditor For admitted cases, a timeperiod of 37-150 days is set aside to formulate a resolution plan which if approved by the RP within 180 days, culminates into resolution and restructuring of the debtor firm In case the resolution plan gets rejected, the debtor goes into liquidation and dissolution

Figure Life cycle of a case under the IBC

The paper is divided into five sections Section defines the IBC timelines Section describes the dataset of cases filed at various NCLT benches as available to us for the purpose of un-derstanding time taken to resolve cases In section 5, the survival analysis methodology is laid down Section documents bench wise and litigant type wise results of the estimation Section concludes

2 A case study of timelines under the IBC

Under the IBC an application for insolvency resolution may be filed with one of 15 benches of the NCLT Let’s call this [T] The NCLT is bound by law to either approve or reject the application within a window of 14 days [T+14] Upon acceptance of the application, a moratorium comes into force Within the next 14 days, the NCLT appoints an interim resolution professional (IRP) [T+28] After a day cool-down period the IRP makes a public announcement for the resolution process and invites claims against the company All claims are verified and consolidated by the RP and an interim report is assembled within a period of 30 days from the day the IRP is appointed[T+44]

resolution professional (RP) The primary directives of the RP so appointed is the preparation of the information memorandum containing all relevant information pertaining to the company and a draft resolution plan The RP is also responsible for the day to day operation of the company to maintain its status as a going concern In the meanwhile, in order to come up with an agreeable resolution plan, the RP may convene several meetings of the CoC

Any member of the CoC may put forth a resolution plan to the committee requiring a 2/3rd majority for its approval within this time period The final resolution plan must be prepared by the RP in not more than a period of 100 days from the first meeting of the CoC [T+150] The final meeting of the CoC is convened twenty days from the creation of RPs draft resolution plan for its presentation to the committee and voted upon [T+170] On failure of acceptance of the plan by at least a 2/3rd strength of the committee, the company is set to undergo a liquidation process On its approval, the plan is submitted with the NCLT for review and clarification A hearing is convened for the final order 10 days from the submission of the plan with the NCLT [T+180]

Figure Timelines prescribed under the IBC

The entire IBC process is to be wound up in 180 days A leeway of additional 90 days is given for complicated cases, extending the permissible timeline to 270 days From the dataset of cases collected on this process, our analysis is based on the following three research questions:

• Once a case is admitted and the insolvency resolution process begins, how long does it take to get resolved?

• What is the probability of reaching a definite outcome within the prescribed timelines (in our case, 180 and 270 days)?

3 A dataset of IBC cases

We use the Finance Research Group (FRG) Insolvency Dataset on cases filed before the NCLT We get the case outcome data for all cases that have seen an outcome resulting in either liquidation or resolution, as published on the Insolvency and Bankruptcy Board of India (IBBI) website We merge the two datasets to get a final outcome for each of the cases, as either closed (liquidated/resolved) or ongoing

The FRG insolvency database is a dataset of all cases filed before the NCLT starting December 2016, uptil February 2018 The data provides information on unique case ID, bench at which the case was filed, who filed, type of creditor, date of filing, listing and disposal of the case, among other fields for out of the 15 consituted benches1 All cases admitted or dismissed by the NCLT in this time period are recorded in the dataset We work with only the cases which get admitted and track them till the point of completion of the IBC process The rest are either dismissed, pending or withdrawn at the level of admission itself Out of the cases filed, 68% of the cases voluntarily filed by the debtor are admitted while the rest are dismissed In case of financial and operational creditors the filed to admission percentage is 47% and 16%, respectively Data on case outcomes from the IBBI is available for cases closed till June 30th, 2018 We use this to match cases with the FRG database

Our final sample consists of data for 762 admitted cases along with information on bench, filed by, date of admission, date of outcome and the outcome itself The time period for the study is December 2016 to June 2018 We look at benchwise workload of these admitted cases and find that the four benches with the highest workload are Mumbai, Delhi, Chennai, and Ahmedabad Here, all benches of Delhi including the Principal bench are considered as having the same jurisdiction Table documents benchwise workload of these benches as filed by different parties

1

Table Workload of benches by litigant type Bench Corporate

debtor

Financial creditor

Operational creditor

Total number of cases admitted

Mumbai 44 84 76 204

New Delhi 14 61 106 181

Chennai 30 57 94

Ahmedabad 18 33 26 77

Kolkata 10 38 19 67

Chandigarh 17 19 16 52

Hyderabad 12 12 32

Allahabad 13 28

Bengaluru 17 27

Total 133 291 338 762

Figure depicts case closures across benches The benchwise performance of IBC cases reveals more cases resulting in a liquidation order as compared with resolution However, case closure is a small percentage of all cases, and more than 50% of all cases admitted are ongoing as on 30th June, 2018 This load is highest again, for Mumbai, New Delhi, Ahmedabad and Chennai benches

4 Delays in bankruptcy resolution process

As opposed to previous bankruptcy laws, the IBC was seen as a regime being brought in to strengthen the credit landscape in the country by ensuring better management of stressed assets It also provided an incentive to the debtor to move from being hesitant in declaring insolvency to wilfully approaching the tribunal and rapidly get a restructuring plan Subsequent to the vision and intended outcomes of the Code, the Indian Economic Survey 2017-18 noted India to have jumped 33 places entering the top 100 spot in the World Bank’s Ease of Doing Business (EoDB) rankings, 2018 Based on how difficult it was to wind up an insolvent entity under previous laws, this was seen as significant progress from our rank of 137 on 189 countries on the same index in 2015 The Survey in Chapter IX, however, highlighted issues in delivery of timely justice and costs of pendency and delays in the judicial infrastructure and institutional setups, as they exist today

Judicial delays have plagued earlier bankruptcy processes Regy and Roy 2017 collect evidence on 474 orders of the Debt Recovery Tribunal and investigate the reasons for delay and the incentives for the stakeholders in prolonging or closing the case beyond a certain timeframe They find that in 50 percent of the cases, time taken is more than expected on account of requirement of complete documents Chatterjee, Shaikh, and Zaveri 2018 examine these delays after the coming in of the IBC wherein they study cases filed before the NCLT for the first six months and the average time taken by each NCLT bench to dispose off petitions This study is extended to a one year time period by Zaveri et al 2018 Over a one year period, an analysis of individual cases shows that the 14 day period granted from filing to admission extends, on an average, to 34 days, thereby delaying the start of the process

Earlier studies on measurement of court performance and congestion have mostly relied on aggregate level data to arrive at pendency rates and answer questions of improving judicial capacity and efficiency Lately, availability of orders at the level of individual cases has enabled hand construction of datasets which can be used to answer more specific questions related to implementation of laws and the functioning of courts The BLRC, in this context, emphasised upon the need to build repositories of case level information to better monitor the performance of the Code

ongoing is compared with the permitted timeline of 180 and 270 days We extend this theoretical timeline to include cases being closed within 360 days of being admitted The cumulative sum of the number of cases closed in the first 180, 270 and 360 days is reported in table For those ongoing, the difference is calculated using 30th June 2018 as the date of last available information We find that only 24 cases across these benches were closed in the first 180 days, and cumulatively, 97 cases had a definite outcome by the end of 270 days However, in the same timeline, 314 cases remained ongoing for the first 180 days Bench wise, the average number of days taken is reported in table For all benches, the time exceeds 180 days

Table Time taken by insolvency cases in the study period

Time taken (in days) Cases closed Cases ongoing

<180 24 314

181-270 73 199

271-360 62 81

>360

-Table Case outcomes for the top four benches

Bench No of Resolution (a) Liquidation (b) Closed (a+b) Time to

petitions reach outcome

admitted (median)

Mumbai 204 32 40 279 days

New Delhi 181 10 14 267 days

Chennai 94 27 31 213 days

Ahmedabad 77 19 20 203 days

For the cases closed in our sample, the duration of time taken between admission of a case and its corresponding date of outcome is considered to construct a rolling window of number of admissions and completions per month We observe this for 16 months beginning March 2017 uptil June 2018 For a party filing a case in January 2017, the window will include cases admitted and closed in that month For February 2018, cases admitted and closed in January and February are considered Similarly, for each month we arrive at the cumulative number for all cases admitted and closed up until that month

Table Cases admitted, disposed and pending by bench

Mumbai Delhi

Month Admitted Closed Backlog Admitted Closed Backlog

1 17 17 2

2 28 28 9

3 37 37 16 16

4 56 56 25 25

5 70 69 35 35

6 90 88 49 49

7 103 100 61 61

8 114 111 65 64

9 119 115 81 80

10 128 120 90 87

11 137 12 125 99 95

12 144 19 125 109 101

13 156 23 133 111 103

14 170 29 141 119 110

15 188 35 153 138 11 127

16 204 40 164 181 14 167

Table Cases admitted, disposed and pending by litigant type

Financial creditors Operational creditors Corporate debtors Month Admitted Closed Backlog Admitted Closed Backlog Admitted Closed Backlog

1 9 8 23 23

2 16 16 26 26 35 35

3 27 27 42 42 46 46

4 44 44 59 59 60 60

5 66 65 77 77 69 69

6 100 99 99 97 89 86

7 135 134 125 123 96 89

8 145 144 131 127 101 10 91

9 173 168 156 151 108 15 93

10 196 187 176 168 110 25 85

11 216 12 204 202 11 191 114 34 80

12 238 21 217 231 21 210 117 42 75

13 248 28 220 239 32 207 126 52 74

14 258 32 226 260 41 219 129 57 72

15 283 39 244 283 45 238 131 63 68

16 291 47 244 338 56 282 133 66 67

Figure Caseload across all benches

While there has been a leap in India’s EoDB rankings, the time taken to resolve these insol-vencies has not seen improvement Ever since the IBC was enacted, there have been differing estimates for the time taken to resolve insolvencies under the law On the same EoDB index 2018, this duration was estimated to be 4.3 years for India

The Reserve Bank of India (RBI) referred 12 large cases to undergo the processs of resolution under the IBC in 2017 Felman, Marwah, and Sharma 2018 surveyed these cases and found that the larger cases were in resolution for more than 500 days and the smaller cases ones also took about 350 days from the date of admission

We build on this literature by estimating a survivor function to measure the time taken in insol-vency resolution processes triggered by different kinds of litigants and before different benches of the NCLT The survival estimate methodology answers the research questions outlined in section by measuring the probability with which a kind of case is expected to close

5 Methodology

In the past, survival analyses have been used to understand judicial delays in tribunals (Datta, Prakash B S., and Sane 2017) A major distinction between its application to judicial delays as opposed to IBC cases is the fact that the IBC process is led by the creditors, referring to the judiciary only for process approvals While judicial delays in the filing and admission processes are highlighted in the previous section, the estimation of case completion within 180 and 270 days is majorly for the insolvency resolution process We apply the survival model contextually to the IBC to find the duration that the entire resolution process takes

or logit i.e a generalised linear model These models answer the question of whether the case will be completed or not Survival analysis adds a time dimension to this question to answer “when”, if ever, the case would be completed It does so by calculating the probability of a case closing given the number of cases ongoing at each point in time Our dependent variable, therefore, becomes “time to event” where event is defined as “case completion”

In reality, the expected time to resolution of any case reaching the NCLT depends upon many factors ranging from the size of the filing party, the amount of debt involved, already existing backlog at the bench, complete documentation as well as the judicial capacity influencing the rate of admission and disposal of process applications at the bench at which the case is filed For instance, it is intuitive to assume that a debtor voluntarily triggering the bankruptcy process may see a speedier resolution Similarly, for benches where backlogs are higher, cases may be slow to proceed from one stage to the other

We setup our analysis by keeping these factors constant We define the survival as a function of time T i.e time to case closure One of the important features of survival time estimation is the censoring of observations At any point in time there are cases which have not been closed If the case for an entity is closed after our study period ends, it is right censored Our study period ends on 30th June Therefore, events occurred i.e cases closed after this date are not included in the study and are automatically censored

The survival function is a non parametric estimation and is defined as function of time T F (T ) is the probability that the “time taken to complete a case” denoted by T is less than or equal to the the prescribed time t

P r(T ≤ t) (1)

which follows the probability distribution dF (t)/dt

F (t) = P rob(T ≤ t) = Z t

0

dF (t)/dt (2)

Since we are interested in finding the time that ongoing cases will take to complete, our estimate is the reverse of F (T ):

S(t) = − F (T ) = P rob(T ≥ t) (3) or the likelihood that the case will not be closed within the timeline of t = 180, 270 and 360 days The survival time begins with the time at which the case first got admitted All cases remain open after admission and are theoretically expected to get completed within 180 days If the event is defined as “case completion”, then the survival function gives us the probability of case ongoing

The survival curve depicts the probability of occurrence of the event conditional on it not having occurred till then, i.e probability of case ongoing at each point in time At day 0, the function takes value since all cases remain open and ongoing However, as cases start getting closed over time, the probability of the event i.e completion of every other case is affected

Figure Survival probability : National average as on 30th June, 2018

For all benches in our analysis, the pooled KM curve is shown in figure On average, there is a less than 10% chance that a case approaching any of the benches will get completed within the 180 day timeline mandated by law Considering the 270 day timeline, the probability of a case getting closed increases, however, only to 20% with 80% of the cases on average still remaining unresolved Within 360 days of being admitted, the probability halves, to a 50% likelihood that a case would be completed in a year’s time This becomes a baseline for us to assess individual benches

6 Findings

6.1 For different benches:

Figure shows the resultant survival curves for each of the four benches The curves for Delhi and Mumbai bench lie above the national average for all three cutoffs: 180, 270 and 360 days This indicates that the probability of case closure at either of the two benches within 180 days is much lower than the national average The probability of case closure at Ahmedabad and Chennai bench moves to 30% for the 270 day mark, while it remains at 8% and 14% for Delhi and Mumbai Looking at the 360 day timeline, the outcome probabilities for Chennai, Ahmedabad, Mumbai, and Delhi are 60%, 41%, 36%, and 21%, respectively It simply suggests that not even half the total number of cases admitted at Mumbai, Delhi and Ahmedabad have a chance of being resolved within a year from the day they are admitted into the process Figure Bench wise case duration as on 30th June, 2018

Table Probability of case ongoing beyond the benchmarked timelines for each bench

Full sample Mumbai New Delhi Chennai Ahmedabad

T >180 days 0.9530 0.9856 0.9892 0.9405 0.9191

(0.9347-0.9716) (0.9659-1) (0.9685-1) (0.8913-0.9925) (0.8536-0.9898)

T >270 days 0.7818 0.8567 0.9161 0.7211 0.6981

(0.7437-0.8218) (0.7956-0.9225) (0.8577-0.9784) (0.6221-0.8359) (0.5828-0.8362)

T >360 days 0.5627 0.6410 0.7961 0.3983 0.5943

(0.5096-0.6214) (0.5483-0.7494) (0.6979-0.9080) (0.265-0.5986) (0.4666-0.7569)

6.2 For litigant type:

Apart from the bench at which a case is filed, the party that is triggering the IBC process is a major factor in defining the time that the case could take Provisions of the IBC allow the financial creditor to file under section and operational creditor to file under section of the Code A case can also be filed by a corporate debtor against itself The KM curves for litigant type are outlined in figure These estimates hold as on June 30th, 2018 The findings correspond with the hypothesis of a debtor’s case being solved more quickly than when two opposing parties are involved Survival curve for cases filed by corporate debtors lie much below the national average and see the fastest rate of completion However, even for these cases, the likelihood of completion within first 180 days remains low Looking at a longer timeline of a year, the corporate debtor filed cases see 70% chance of being resolved

For financial and operational creditors, the KM curves show a different story Cases filed by financial creditors tend to be more complicated, with higher debt amounts involved Commen-surate with this, the survival curves show that their cases are least likely to be closed in a year’s time, let alone in 180 days The curve for operational creditors revolves around the national average However, at the 360 day mark, it has a chance of closing only as low as that of a financial creditors

We corroborate this inference in detail in table There is a 70% chance of a debtor’s case being closed in a year’s time while that of both kinds of creditors sees a probability that drops to 40% In either case, the 180 and 270 day mark is missed and is rendered inadequate by the stakeholders dealing with the IBC process

Table Probability of case ongoing beyond the benchmarked timelines for each litigant type

Full sample Corporate debtor Financial creditor Operational Creditor

T >180 days 0.9530 0.9469 0.9655 0.9421

(0.9347-0.9716) (0.9063-0.9892) (0.9406-0.9910) (0.9092-0.9761)

T >270 days 0.7818 0.6874 0.8341 0.7869

(0.7437-0.8218) (0.6038-0.7825) (0.7797-0.8923) (0.7257-0.8532)

T >360 days 0.5627 0.3470 0.6343 0.6330

(0.5096-0.6214) (0.2573-0.4678) (0.5428-0.7412) (0.5546-0.7225)

7 Conclusion

Using the methodology of estimating time-to-event for a dataset of bankruptcy cases under the IBC, collected for tribunal benches of the NCLT, we find that on average:

• The probability of case completion within 180 days is less than 5% • The probability of case completion within 270 days is 22%

• The probability of case completion within 360 days is 45%

Our analysis of cases that are either admitted or undergoing a new bankruptcy resolution process in India provides for a new approach to measure the law’s effectiveness The applicability of the survival model to other laws which lay down timelines for certain actions makes it a useful tool for monitoring and evaluation studies In our case, we use the model to estimate the probability of the process ongoing across the cutoff points decided by law

legislature to rethink timelines given for compliance Second, it helps in building state capacity for efficient allocation of resources A bench wise performance analysis makes a strong case for increasing capacity at the Mumbai and Delhi benches for speedier case disposals

Most importantly, it allows stakeholders to better plan their affairs Knowledge of the chances of an insolvency case being resolved within a given time frame under each jurisdiction helps a potential litigant in making the choice between pursuing resolution versus settling out of court This analysis can also be extended If data for every phase in the resolution process is collected or made available, the framework can shift to estimate completion of those phases This can also be suggestive of the various bottlenecks delaying the process Additionally, if the analysis is conducted over a continuing time period, it can be used to evaluate institutional performance and measure the learning curve of the entire eco-system which the law has setup

References

Chatterjee, Sreyan, Gausia Shaikh, and Bhargavi Zaveri (2018) “Watching India’s insolvency reforms: a new dataset of insolvency cases” In: National Law School of India Review Datta, Pratik, Surya Prakash B S., and Renuka Sane (2017) Understanding Judicial Delay at

the Income Tax Appellate Tribunal in India National Institute of Public Finance and Policy Working Paper Series

Economic Survey 2017-18, Volume I (2018) Ministry of Finance, Govt of India

Felman, Josh, Varun Marwah, and Anjali Sharma (2018) “The RBI-12 cases under the IBC” In: Working paper on file with the authors, 2018

Ravi, Aparna (2015) The Indian Insolvency Regime in Practice-An Analysis of Insolvency and Debt Recovery Proceeding Indira Gandhi Institute of Development Research (IGIDR) Regy, Prasanth and Shubho Roy (2017) Understanding Judicial Delays in Debt Tribunals

National Institute of Public Finance and Policy Working Paper Series

Sengupta, Rajeswari, Anjali Sharma, and Susan Thomas (2017) “Evolution of the insolvency framework for non-financial firms in India” In: India Development Report, IGIDR

Shah, Ajay and Susan Thomas (2018) “The Indian bankruptcy reform: The state of the art, 2018” In: Law, Economics and Policy blog

Shumway, T (2001) “Forecasting Bankruptcy More Accurately: A Simple Hazard Model” In: The Journal of Business 74, No.1