A Summer Internship

ON

“A Study on Credit Appraisal and Risk Management at IDFC FIRST

Bank: Evaluating Lending Practices and Risk Control Mechanisms”

For

IDFC First Bank, PUNE

By

Sunil Reddy Chavan

PRN – 1062242831

MBA (Banking and Financial Services)

Batch 2024-2026

Under the guidance of

Dr. Rupali Gupta

Submitted to

MIT-WPU School of Business, Pune.

1
 TABLE OF CONTENTS

[Link] Topic
Page No.

Abstract 7

1 Chapter 1- Introduction
8

2 Chapter 2- Company Overview

15

3 Chapter 3- Literature Review

18

4 Chapter 4- Research Methodology

37

5 Chapter 5- Analysis Interpretation

39

6 Chapter 6- Key Findings

49

7 Chapter 7- Conclusion
51

8 Chapter 8- References
52

2
 Abstract
This research examines the credit appraisal and risk management framework of IDFC FIRST
Bank during the post-merger years 2019–2024, using only secondary data sources. The objec-
tive is to assess how the bank’s practices align with regulatory expectations and to identify
measures that could further strengthen asset quality. Information was drawn from the bank’s
annual reports, RBI publications, Basel Committee guidelines, rating-agency analyses, and rel-
evant academic work. The study reviews loan portfolio trends, non-performing asset ratios,
provisioning policies, and capital adequacy levels in comparison with standard approaches to
estimating probability of default (PD), loss given default (LGD), and exposure at default (EAD)
(Altman, 1968; BCBS, 2015).

Findings suggest that the bank’s shift toward a retail and MSME-oriented portfolio has reduced
concentration in large corporate borrowers, consistent with portfolio diversification theory.
However, the increased share of unsecured retail and short-tenure MSME loans also raises
vulnerability to cyclical downturns (Ghosh, 2020; Mishra & Ghosh, 2021). The appraisal
framework combining bureau scores, internal models, and cash flow analysis broadly reflects
current practices (Thomas, 2009; Jagtiani & Lemieux, 2019), but there is room to enhance
calibration of PD, refine LGD haircuts for specific product categories, and strengthen govern-
ance of overrides (BCBS, 2006; Zmijewski, 1984).

Stress-test disclosures indicate adequate buffers under mild adverse conditions, though global
evidence warns that combined shocks such as GDP decline, rising unemployment, and property
price corrections could still create sharp increases in provisions if management responses are
not pre-committed (Bank of England, 2018; ECB, 2021).

The study recommends (1) establishing clearer macro-linked risk parameters, (2) integrating
alternative and transactional data sources such as GST and digital payments in MSME scoring
while retaining relationship-based judgment (Berger & Udell, 2006; Jagtiani & Lemieux,
2019), and (3) improving model governance through regular challenger testing, population sta-
bility tracking, and greater transparency (BCBS, 2015). Overall, this study provides a policy-
relevant evaluation of IDFC FIRST Bank’s evolving risk practices and offers targeted sugges-
tions to improve resilience in retail and MSME lending

3
 Chapter 1 – Introduction
1. Introduction
The history of banking reveals how financial institutions have continually evolved to meet the
needs of economies, businesses, and individuals. From the earliest forms of lending in ancient
civilizations, where merchants advanced grain or goods against future harvests, to the modern
commercial banks that provide complex financial services, the central function has remained
the same: mobilizing savings and channelling them into productive investments. Over centu-
ries, the scope and scale of banking have expanded dramatically, but one challenge has re-
mained constant—ensuring that funds are lent prudently and are recovered in a timely manner.

As banking systems grew in size and sophistication, credit appraisal emerged as a cornerstone
of lending activity. In simple terms, credit appraisal refers to the process by which banks eval-
uate the eligibility of a borrower to receive funds. This evaluation is not limited to financial
statements but extends to repayment capacity, character, collateral, and the overall purpose of
the loan. Historically, lending was based on personal knowledge and relationships. Local bank-
ers or moneylenders often relied on trust and informal assessments of a borrower’s honesty.
With the industrial revolution and the growth of joint-stock banks, however, lending became
more formalized, requiring structured methods to analyze financial risk.

By the 20th century, the demand for credit had increased manifold with the rise of corporations,
infrastructure projects, and global trade. This expansion introduced new risks, as banks now
dealt with borrowers across geographies and industries with varying degrees of transparency.
The global financial crises of the past century—particularly the Great Depression of the 1930s
and the Global Financial Crisis of 2008—highlighted the dangers of weak credit standards and
inadequate supervision. These episodes emphasized the importance of risk management as an
institutional practice, rather than leaving lending decisions solely to individual judgment.

Credit Appraisal in Modern Banking

In today’s context, credit appraisal involves a multi-dimensional evaluation of borrowers. For

retail customers, it may include credit bureau reports, income proofs, banking transactions, and
behavioural patterns. For corporate borrowers, appraisal extends to financial ratio analysis,
cash-flow projections, debt-servicing capacity, and industry outlooks. Small and medium en-
terprises (SMEs), which often lack detailed financial records, may be assessed through GST
returns, digital payment histories, and business viability.

The objective is not only to decide whether to lend but also to determine the terms of lending,
including loan size, interest rate, and repayment structure. A well-executed credit appraisal
minimizes the risk of default while ensuring that deserving borrowers gain access to funds. For
banks, this is critical because lending constitutes the largest portion of their assets, and defaults
directly erode profitability and capital strength.

4
Emergence of Risk Management as a Discipline

While credit appraisal deals with individual lending decisions, risk management is broader,
encompassing all the uncertainties that may affect a bank’s stability. Risk management as a
formal discipline grew alongside the increasing complexity of global finance. It includes the
identification, measurement, monitoring, and control of various risks—credit risk, market risk,
liquidity risk, and operational risk. Among these, credit risk remains the most significant be-
cause it accounts for the majority of losses in banking.

Regulatory frameworks have played a crucial role in institutionalizing risk management. The
Basel Accords, introduced by the Bank for International Settlements, provided a global stand-
ard for how banks should measure credit risk and maintain adequate capital buffers. Concepts
such as Probability of Default (PD), Loss Given Default (LGD), and Exposure at Default
(EAD) have become fundamental in credit-risk modelling. In India, the Reserve Bank of India
(RBI) has aligned its supervisory guidelines with Basel principles, mandating robust credit
appraisal practices, provisioning norms, and stress testing.

Why Credit Appraisal and Risk Management Matter Today

In modern economies, banks act as the lifeline of growth by financing consumption, trade,
infrastructure, and innovation. However, the same lending activity exposes them to default risk.
Non-performing assets (NPAs) have been one of the biggest challenges for banks, especially
in emerging markets like India. Episodes such as the NPA crisis following the 2015 Asset
Quality Review demonstrated that weak credit appraisal and lax monitoring could threaten not
just individual banks but the stability of the entire financial system.

The rise of unsecured lending, rapid digitalization, and global uncertainties such as pandemics
and climate change have further heightened the need for comprehensive risk management. Tra-
ditional financial analysis alone is no longer sufficient. Banks increasingly rely on alternative
data—such as digital footprints, GST data, and payment histories—and employ advanced tools
like machine learning for credit scoring. At the same time, regulators emphasize the importance
of transparency, governance, and ethical standards to prevent excessive risk-taking.

The Link Between Credit Appraisal and Risk Management

Credit appraisal and risk management are deeply interconnected. While credit appraisal focuses
on evaluating the riskiness of each loan, risk management looks at the overall portfolio and
systemic exposures. A strong appraisal process feeds reliable data into risk models, while sound
risk management ensures that lending decisions align with the bank’s capital strength and stra-
tegic objectives. Together, they help in maintaining asset quality, sustaining profitability, and
safeguarding the interests of depositors and investors.

Contemporary Relevance

5
In the Indian context, the banking sector has witnessed significant transformation in the last
decade. Private banks, cooperative institutions, and non-banking financial companies have
adopted advanced appraisal systems to reach retail and MSME borrowers. Digital platforms
now enable instant loan approvals, but this speed also requires stronger risk controls to avoid
overexposure. With increasing regulatory expectations under Basel III and the transition to-
ward expected credit loss models under IFRS-9, credit appraisal and risk management have
become indispensable competencies for every financial institution.

1.2 Research Problem

Macroeconomic and Supervisory Context

Credit risk continues to be the most significant factor influencing banking stability. Following
episodes of global financial turbulence and India’s 2015 Asset Quality Review, regulators —
both international (Basel Committee) and domestic (RBI) — have stressed the need for for-
ward-looking provisioning, robust scenario analysis, and strong model governance. Account-
ing standards such as IFRS 9 have reinforced this by requiring banks to estimate expected credit
losses under multiple scenarios. As a result, accurate estimation, stress-testing, and disclosure
have become integral to both supervisory and financial reporting frameworks. These develop-
ments have raised expectations around transparency, governance, and the sophistication of risk
models.

Institutional Profile and Strategic Transition

In the Indian context, IDFC FIRST Bank provides a particularly relevant case for analysis.
After merging IDFC Bank’s wholesale lending franchise with Capital First’s retail-focused
NBFC, the bank deliberately shifted its portfolio toward retail and MSME borrowers. This
moves reduced concentration risk from large corporates but simultaneously increased the share
of unsecured and short-tenure exposures. Such exposures improve diversification but bring
heightened behavioural and macroeconomic sensitivity, requiring more advanced appraisal and
monitoring processes.

Core Research Problem

Despite visible progress in portfolio diversification and digital expansion, a critical question
remains: Are IDFC FIRST Bank’s publicly available disclosures on credit appraisal, model
governance, and stress-testing (2019–2024) sufficiently comprehensive, forward-looking, and
transparent to meet regulatory and market expectations? Specifically, the research investigates
whether the bank’s disclosures:

1. Capture both point-in-time deterioration and through-the-cycle risk across retail,

MSME, and corporate segments.
2. Provide credible loss estimates under severe macroeconomic stress.

6
 3. Support sound supervisory and management decisions that protect against sudden cap-
ital erosion.

Why This Problem Matters

 Confidence of stakeholders: Weak or ambiguous disclosures can reduce trust among

depositors, investors, and regulators (BCBS, 2015).
 Procyclicality concerns: If credit models are insufficiently linked to macroeconomic
conditions, provisioning may lag actual losses, amplifying downturn effects.
 SME inclusion trade-offs: Expanding MSME lending without robust alternative-data
models risks higher defaults and could undermine financial inclusion goals.
 Model risk: Increasing reliance on AI and ML techniques creates explainability and
validation challenges that require stronger governance disclosures.

Disclosure and Evidence Gaps

 Absence of macro-elasticity data: Banks describe stress scenarios but rarely publish
how changes in GDP or unemployment map into PD/LGD shifts. This study recon-
structs plausible elasticities from literature to test scenario credibility.
 Sparse detail on model governance: Disclosures seldom specify validation cycles,
PSI thresholds, override practices, or challenger testing, though literature identifies
these as critical safeguards (BCBS, 2006; Zmijewski, 1984).
 Limited alternative-data transparency: While transactional and GST data are in-
creasingly used in underwriting, few disclosures clarify governance, privacy, or ex-
plainability frameworks (Jagtiani & Lemieux, 2019).
 Ambiguous management assumptions: Stress-test outcomes often depend on as-
sumed management actions (e.g., curbing new lending), yet disclosures do not fully
describe these levers.

Forward-Looking Challenges

The study also addresses future risks shaping the bank’s credit-risk practices:

 More frequent macro shocks and tighter monetary policy cycles, which put pressure on
unsecured portfolios.
 The growing need to integrate climate and transition risks into stress-testing frame-
works.
 Regulatory emphasis on AI/ML explainability, which raises requirements for model
documentation.
 The trade-off between leveraging alternative data for better risk prediction and manag-
ing data-privacy concerns.

7
Research Questions

1. How comprehensive are IDFC FIRST Bank’s disclosures on PD, LGD, EAD method-
ologies, validation, and stability monitoring relative to Basel norms?
2. Which portfolio segments (retail unsecured, MSME, corporate) appear most vulnerable
to macroeconomic shocks when elasticities are applied?
3. How transparent and plausible are the bank’s stress-test assumptions, including stated
management actions?
4. To what extent does the bank disclose its adoption of alternative data and ML tech-
niques, and where do governance gaps remain?
5. What specific measures — in disclosure, model governance, scenario design, or product
strategy — would most effectively reduce tail losses while preserving access to credit?

Methodological Boundaries

The study relies solely on secondary evidence: audited financial statements, statutory filings,
regulatory guidance, rating-agency reviews, and academic literature. Where quantitative esti-
mates are required, conservative parameters from central bank and academic studies are applied
(Bank of England, 2018; Bellotti & Crook, 2009).

Limitations: Without access to loan-level data, the study cannot estimate borrower-level PD
or LGD, nor can it provide causal analysis. Instead, it delivers a directional assessment and a
disclosure audit useful to regulators, investors, and management.

Value: By triangulating public data with supervisory and academic benchmarks, the study pro-
duces policy-relevant recommendations and a governance checklist that IDFC FIRST Bank
can realistically implement and disclose.

1.3 Objectives of the Study:

Primary Objective

The central aim of this study is to evaluate the credit appraisal processes and risk management
framework of IDFC FIRST Bank in the years following its merger (2019–2024). The assess-
ment focuses on how the bank estimates credit risk parameters such as probability of default
(PD), loss given default (LGD), and exposure at default (EAD). In addition, it examines dis-
closures on model governance, stress-testing approaches, and portfolio resilience. Based on
this evaluation, the study develops a set of prioritized, evidence-based recommendations to
strengthen provisioning practices, reduce exposure to tail risk, and enhance transparency in line
with regulatory expectations from the Basel Committee and the Reserve Bank of India (BCBS,
2015; RBI, 2021).

8
Approach to Objectives

Since this research relies on secondary data rather than internal loan-level records, the objec-
tives are structured at multiple layers:

 Conceptual: assessing the alignment of the bank’s frameworks with supervisory stand-
ards.
 Empirical: benchmarking portfolio trends and financial indicators against peers.
 Procedural: reviewing disclosures on governance, validation, and stress-testing.
 Prescriptive: suggesting actionable steps that can be implemented within the Indian
retail and MSME banking environment (Thomas, 2009; BCBS, 2015).

This tiered approach ensures that the study remains grounded in verifiable facts, informed by
regulatory norms, and directly relevant to practical implementation.

1.4 Scope of Study

1. Credit-Appraisal Architecture
Mapping IDFC FIRST Bank’s disclosed lending workflow, including data inputs (bu-
reau scores, GST records, account statements), scorecards for different segments, and
sanctioning protocols. A well-designed appraisal process is essential to reduce infor-
mation asymmetry and improve accuracy of PD estimates (Berger & Udell, 2006; Alt-
man, 1968).
2. Portfolio Benchmarking
Comparing loan-book segmentation (retail, MSME, corporate) and asset quality indi-
cators such as GNPA, NNPA, provisioning coverage, and capital adequacy between
2019–2024 against a peer group of private banks. Portfolio structure plays a critical role
in determining risk concentration and capital requirements (Ghosh, 2020; Mishra &
Ghosh, 2021).
3. Model Governance Review
Assessing whether the bank’s disclosures mention key elements of model risk manage-
ment, such as development procedures, validation frequency, back-testing, override re-
porting, and challenger-model use. Strong model governance is a safeguard against er-
rors and enhances stakeholder trust (BCBS, 2006; Zmijewski, 1984).
4. Stress-Testing Evaluation
Examining how the bank reports stress scenarios, and whether the assumptions align
with macroeconomic-to-loss linkages identified in literature and regulatory practice.
The clarity of assumed management actions is especially important, since hidden as-
sumptions can underestimate risk exposure (Bank of England, 2018; ECB, 2021).
5. Segmental Sensitivity Analysis
Using published portfolio data and established elasticities to assess how retail unse-
cured, MSME, and corporate segments may respond to macroeconomic shocks such as
GDP decline, rising unemployment, or property market stress. This analysis highlights
which segments are most vulnerable and require enhanced monitoring (Bellotti &
Crook, 2009).
6. Alternative Data and Fintech Adoption
Reviewing disclosures and industry commentary on the use of GST and transaction
data, open banking frameworks, and machine-learning models in credit appraisal.
While alternative data can improve SME risk assessment, it also raises governance and
explainability concerns (Jagtiani & Lemieux, 2019).

9
 7. Action-Oriented Recommendations
Translating findings into a phased recommendation matrix (short-, medium-, and long-
term) that balances impact, complexity, and resource requirements. Prioritized actions
help ensure practical adoption and measurable risk reduction (BCBS, 2015).

Measurable Deliverables

 A comparative table (2019–2024) covering GNPA, NNPA, PCR, CAR, and re-
tail/MSME composition for IDFC FIRST Bank and selected private peers.
 A checklist of disclosed governance and validation practices scored on presence/ab-
sence.
 A reconstruction of stress-test outcomes under baseline, adverse, and severe scenarios
using conservative literature elasticities.
 A structured action matrix with timelines, expected impact, and suggested disclosure
templates to improve transparency.

Here are two ready-to-use sections you can insert after your Scope of Study:

1.5 Significance of the Study

This study holds significance for multiple stakeholders within the Indian banking ecosystem.
For regulators and policymakers, it provides an independent evaluation of whether IDFC
FIRST Bank’s credit appraisal and risk management disclosures (2019–2024) are sufficiently
aligned with Basel III and RBI supervisory expectations. Such an assessment is critical in light
of India’s transition toward forward-looking provisioning under IFRS 9 and the increasing reg-
ulatory emphasis on transparency, governance, and stress-testing.

For practitioners, particularly risk managers and credit officers, the research offers insights into
how a mid-sized private bank has navigated the strategic shift from wholesale to retail and
MSME lending. By highlighting areas of strength and identifying disclosure gaps, the study
delivers practical lessons that can guide model calibration, portfolio diversification, and gov-
ernance practices in other financial institutions.

For investors, rating agencies, and depositors, the findings enhance understanding of the bank’s
resilience under stress scenarios, thereby contributing to more informed decision-making. At
the academic level, the study enriches the literature on credit appraisal in emerging markets by
combining supervisory benchmarks with empirical evidence from an Indian private-sector
bank. Overall, this research bridges the gap between theory, regulatory practice, and institu-
tional disclosures, generating actionable insights that support financial stability and sustainable
credit growth.

1.6 Limitations of the Study

While the study provides a comprehensive review of IDFC FIRST Bank’s credit appraisal and
risk management framework, several limitations must be acknowledged. First, the analysis is
based solely on secondary data sources such as annual reports, regulatory filings, and rating-
agency reviews. Without access to loan-level or internal risk model data, the research cannot
estimate borrower-specific probability of default (PD), loss given default (LGD), or exposure
at default (EAD).

10
Second, the stress-test reconstructions rely on elasticities and parameters drawn from aca-
demic and regulatory literature, which may not fully capture the unique behavioural dynam-
ics of the bank’s portfolio. As a result, the stress outcomes should be interpreted as directional
rather than precise forecasts.

Third, the reliance on publicly disclosed governance details limits the ability to assess practices
such as override monitoring, challenger model use, or validation frequency, since banks
often report these in broad terms rather than operational detail.

Finally, the study is confined to the post-merger period (2019–2024), which may not fully
reflect long-term structural adjustments in the bank’s strategy. These limitations underscore
that the findings are best viewed as a disclosure audit and policy-relevant evaluation, rather
than a predictive model of future credit performance.

11
 Chapter 2 – Company Overview
2.1 Institutional Genesis and Strategic Transformation

IDFC FIRST Bank represents one of the more prominent strategic realignments within India’s
private banking sector. Its origins lie in the Infrastructure Development Finance Company
(IDFC), which had historically specialized in project and infrastructure finance after being
granted a universal banking licence (Reserve Bank of India, 2015). In December 2018, the
institution underwent a significant transformation when IDFC Bank merged with Capital First,
a non-banking financial company known for its strong retail and MSME lending platform
(Singh & Bhattacharya, 2020). This merger brought together IDFC’s wholesale banking and
funding expertise with Capital First’s established retail underwriting capabilities. The com-
bined entity deliberately shifted its focus toward retail and MSME lending, positioning itself
as a customer-oriented, technology-driven private bank.

Management articulated this repositioning as part of a broader mission to drive financial inclu-
sion, sustainable growth, and digital delivery (IDFC FIRST Bank, Annual Reports 2019–
2023). The pivot towards retail reduced the concentration risk tied to large infrastructure bor-
rowers but simultaneously expanded exposure to unsecured retail and MSME loans. Such loans
are inherently more cyclical and require more sophisticated appraisal and monitoring mecha-
nisms (Berger & Udell, 2006).

2.2 Governance, Leadership, and Organisational Structure

Following the merger, governance continuity was ensured through the leadership team that
previously steered Capital First. Strategic emphasis was placed on retail growth while main-
taining regulatory compliance and capital adequacy (IDFC FIRST Bank, Annual Report).
Structurally, the bank operates along conventional commercial banking lines, with divisions
for retail, wholesale, and treasury operations. Oversight functions are centralized through risk
and compliance units reporting to the Board Risk Committee and the Board of Directors.

Public disclosures reveal the existence of dedicated credit policy and investment committees,
as well as model validation mechanisms. These practices align with international supervisory
recommendations on credit-risk governance and model oversight (BCBS, 2015). Additionally,
the bank communicates the use of delegated approval limits, audit mechanisms, and periodic
override reporting as part of its risk culture.

2.3 Business Model and Product Mix

IDFC FIRST Bank offers a diversified product portfolio spanning deposits, consumer loans,
credit cards, mortgages, vehicle finance, MSME loans, corporate banking, and treasury ser-
vices. The bank has placed emphasis on growing its base of low-cost retail deposits through
innovative savings products and digital acquisition channels.

On the lending side, its portfolio is balanced between unsecured retail credit — such as credit
cards and personal loans, which yield higher margins but also greater behavioural risk — and
secured segments such as housing and vehicle loans, which typically carry lower loss-given-
default profiles. MSME financing has been positioned as a growth driver, though it requires
intensive monitoring due to higher informational asymmetry (Berger & Udell, 2006). This

12
strategy reflects the trade-off widely recognized in the literature: diversification reduces con-
centration risk (Markowitz, 1952), but unsecured portfolios can amplify cyclicality unless sup-
ported by advanced risk controls (Ghosh, 2020).

2.4 Credit-Appraisal Architecture and Data Sources

The bank’s credit appraisal process, as disclosed publicly, is multi-tiered. Initial screening re-
lies on bureau scores (such as CIBIL), followed by product-specific internal scorecards and
rule-based filters. For MSME and corporate borrowers, cash flow assessments and financial
statement analyses are performed, supplemented by collateral valuation where relevant.
Higher-value exposures are escalated to credit committees, with delegated sanctioning frame-
works applied depending on the risk tier (IDFC FIRST Bank, Annual Report).

In recent years, the bank has increasingly incorporated alternative datasets, including GST rec-
ords, digital payment footprints, and transactional account flows. This shift reflects an industry-
wide trend towards combining bureau data with non-traditional information to mitigate infor-
mation gaps in MSME lending (Jagtiani & Lemieux, 2019). Consistent with BCBS and IFRS-
9 guidelines, the bank has also outlined its approach to expected credit loss (ECL) provisioning,
though disclosures remain high-level, with limited transparency on elasticity mapping or de-
tailed loss curves (BCBS, 2015).

2.5 Model Governance and Validation

The bank reports having dedicated validation units responsible for periodic recalibration and
independent review of credit models. References to back-testing, sensitivity checks, and reval-
idation cycles appear in annual filings, indicating an awareness of model governance standards.
However, as with many banks, granular details such as population-stability trends, challenger-
model comparisons, or override analytics are not disclosed, which restricts external evaluation
of robustness (Zmijewski, 1984; BCBS, 2006). Publishing summarized metrics — such as PSI
movements or aggregated override rationales — could improve transparency without exposing
proprietary intellectual property.

2.6 Monitoring, Recovery, and Early-Warning Systems

IDFC FIRST Bank has disclosed investments in monitoring and recovery frameworks, includ-
ing automated alerts for delinquency patterns, portfolio-level heat maps, and early-warning
triggers derived from customer behaviour (e.g., missed payments, overdrafts) and bureau sig-
nals. Recovery mechanisms include collections, regulatory-compliant restructuring, and legal
action where necessary. Research and supervisory perspectives emphasize that robust early-
warning and recovery systems are as vital as initial underwriting for maintaining asset quality
(Mishkin, 2018; BCBS, 2006). While the bank indicates ongoing progress, the scale of auto-
mation and operational execution may vary by geography and product segment.

2.7 Capital, Provisioning, and Stress-Testing

Regulatory filings show that IDFC FIRST Bank has maintained its capital adequacy in line
with Basel III requirements, while provisioning ratios suggest a conservative approach to loan-
loss recognition. Stress-test disclosures generally include baseline and adverse scenarios with
narrative descriptions of outcomes. However, detailed mappings of scenario assumptions to

13
PD and LGD are typically omitted, a limitation that is common across industry disclosures. As
stress-testing literature highlights, the credibility of management action assumptions signifi-
cantly shapes reported resilience (Bank of England, 2018; ECB, 2021). Clearer articulation of
contingency levers, such as slowing origination or targeted run-offs, would add depth to the
bank’s disclosures.

2.8 Strategic Risks and Future Priorities

The bank’s strategic disclosures point to several priorities and emerging risks: managing unse-
cured retail growth without compromising credit discipline, scaling MSME lending through
alternative data while maintaining governance, embedding ESG and climate risk in credit
frameworks, and increasing disclosure granularity to meet market and regulatory expectations.
These align with broader debates in academic and regulatory literature, which emphasize the
challenge of balancing financial inclusion with sound risk measurement to avoid procyclical
capital impacts (Berger & Udell, 2006; Ghosh, 2020; Jagtiani & Lemieux, 2019)

2.9 Synthesis: Rationale for Case Selection

The transformation of IDFC FIRST Bank, coupled with its extensive public disclosures, makes
it a suitable case study for secondary data research. Its post-merger pivot creates a clear tem-
poral frame (2019–2024), while the availability of regulatory filings and investor presentations
enables robust analysis of credit appraisal and governance systems. Additionally, the bank’s
focus on retail and MSME segments allows for comparison with international best practices
and academic frameworks (BCBS, 2015; Jagtiani & Lemieux, 2019). Collectively, these fac-
tors allow for meaningful insights without requiring proprietary loan-level data.

14
 Chapter- 3 Literature Review
3.1 Theoretical Framework

This research is grounded in established theories, risk measurement models, and regulatory
principles that underpin contemporary credit appraisal and risk management practices. It aims
to provide a conceptual base for evaluating the public disclosures of IDFC FIRST Bank while
also deriving practical insights that can guide the secondary data analysis. The framework
brings together traditional credit theories, statistical and econometric approaches, supervisory
guidelines, and modern developments such as the use of machine learning, alternative data,
and environmental, social, and governance (ESG) considerations.

1. Core Credit-Risk Parameters: PD, LGD, and EAD

Modern approaches to credit-risk evaluation focus on three core measures: probability of de-
fault (PD), loss given default (LGD), and exposure at default (EAD). Collectively, these vari-
ables determine the expected credit loss (ECL), a framework that has been incorporated into
IFRS-9 and Basel regulatory standards (IFRS Foundation, 2014; BCBS, 2015). Risk managers
can influence each of these dimensions separately—for instance, stronger origination practices
and credit scoring can help lower PD, better collateral valuation and recovery processes can
reduce LGD, and stricter oversight of unused credit facilities can limit EAD.

In secondary-data research, it is crucial that bank disclosures explain how PD, LGD, and EAD
are calculated for different lending products and how these calculations are tied to macroeco-
nomic scenarios. When such assumptions or elasticity measures are not clearly reported, it
limits the ability of external stakeholders to judge the robustness of the bank’s credit-risk
framework (BCBS, 2015; Bank of England, 2018).

2. Asymmetric Information and Adverse Incentives

Theories of asymmetric information, such as Akerlof’s “lemons” model (1970) and the credit
rationing framework of Stiglitz and Weiss (1981), illustrate why credit markets often diverge
from efficient outcomes. Inadequate screening mechanisms can result in adverse selection,
where riskier borrowers are more likely to obtain credit, while insufficient monitoring creates
moral hazard, encouraging borrowers to take on higher risks once credit has been extended.
These issues underscore the importance of adopting comprehensive appraisal techniques,
which may include credit bureau scores, cash-flow evaluations, collateral assessments, and re-
lationship-based judgments (Berger & Udell, 2006; Jagtiani & Lemieux, 2019).

For MSME lending specifically, greater transparency in disclosures—such as reliance on GST

records, banking transactions, and qualitative assessments—offers useful evidence of how
banks attempt to reduce information asymmetry.

3. Principal–Agent Problems and Governance

Agency theory focuses on the potential conflicts of interest between owners, managers, and
loan officers (Jensen & Meckling, 1976). When governance is weak, incentives for growth may
lead to higher risk-taking. In banking, mechanisms such as model validation, monitoring of

15
overrides, delegated sanction limits, and oversight by the board are designed to address these
agency-related risks (BCBS, 2006). Therefore, transparency regarding governance practices—
like the frequency of validations, documentation of overrides, and the autonomy of credit com-
mittees—serves as an indicator of the institution’s risk culture and internal discipline (Zmijew-
ski, 1984).

4. Portfolio Diversification and Concentration Risk

Markowitz’s portfolio theory (1952) applies directly to banking portfolios. Expanding retail
and MSME books lowers single-borrower concentration but may increase systemic vulnerabil-
ity if exposures are highly correlated — for example, unsecured personal loans sensitive to
economic downturns (Ghosh, 2020). The key analytical question is whether diversification is
accompanied by strong credit scoring, monitoring, and provisioning frameworks

5. Structural vs. Statistical Credit Models

Default prediction models are commonly divided into two categories: structural models, which
define default as the point when a firm’s value falls below its liabilities (Merton, 1974), and
reduced-form or statistical models, which directly estimate the probability of default based on
relevant covariates (Ohlson, 1980; Shumway, 2001). Structural models are generally suitable
for large publicly listed firms, whereas retail and SME portfolios are more effectively assessed
using statistical or machine-learning techniques (Bellotti & Crook, 2009). Consequently, dis-
closures should clarify whether IDFC FIRST Bank applies different models for different seg-
ments and if governance practices are adapted accordingly.

6. Time Dynamics and Hazard Models

Static classifiers can miss evolving borrower risk. Hazard or survival models estimate the like-
lihood of default over time and incorporate changing borrower characteristics (Shumway,
2001; Bellotti & Crook, 2009). These approaches underpin IFRS-9 staging rules, where expo-
sures shift from 12-month to lifetime ECL when credit quality deteriorates. Evidence of vintage
analysis, roll-rate matrices, or hazard modelling in disclosures would suggest advanced risk
management practices.

7. Model Risk and Governance

Model risk arises from sampling bias, overfitting, or shifts in borrower populations (Zmijewski,
1984; Demyanyk & Hasan, 2010). Supervisory guidance emphasizes documentation, chal-
lenger testing, independent validation, and stability monitoring (BCBS, 2006; 2015). More re-
cently, machine-learning models have introduced explainability challenges, requiring tools
such as SHAP or LIME to monitor bias and drift (Lundberg & Lee, 2017).

A secondary-data audit should therefore look for disclosures about validation frequency, PSI
reporting, override governance, and ML explainability standards.

8. Alternative Data and Machine Learning

Machine-learning techniques and the use of alternative data can improve default prediction by
incorporating richer behavioural and transactional information (Khandani et al., 2010; Jagtiani

16
& Lemieux, 2019). For MSMEs, analysing GST flows and digital transaction histories provides
more accurate estimates of probability of default than relying solely on traditional balance-
sheet metrics. Nonetheless, these approaches introduce challenges related to governance, pri-
vacy, and model interpretability. Therefore, public disclosures should detail the oversight
mechanisms, bias-mitigation measures, and adherence to data protection regulations.

9. Human Judgment and Behavioural Bias

Even in predominantly quantitative credit assessment systems, human judgment remains im-
portant. Studies in behavioural finance indicate that biases such as overconfidence and anchor-
ing can influence loan officers’ decisions (Kahneman & Tversky, 1979). While relationship
lending captures valuable soft information, it may also amplify these biases. A hybrid ap-
proach—combining scorecards with relationship-based adjustments and formally documented
exceptions—can help mitigate such risks (Berger & Udell, 2006). Disclosures that include
override logs and exception governance practices serve as indicators of sound risk manage-
ment.

10. Stress Testing and Management Actions

Stress testing translates macroeconomic shocks into impacts on probability of default (PD),
loss given default (LGD), and exposure at default (EAD), with outcomes heavily influenced
by management actions such as tightening origination or enhancing collections (Bank of Eng-
land, 2018; ECB, 2021). Reverse stress testing, which determines the scenarios that could de-
plete capital buffers, is also increasingly advocated. Comprehensive disclosures should de-
scribe the scenario design, assumptions linking shocks to losses, and the management levers
considered.

11. IFRS-9 Provisioning and Integrated Risk Management

IFRS-9 embeds staging of expected credit losses (12-month vs. lifetime) and requires multiple-
scenario forward-looking estimates (IFRS Foundation, 2014). This approach links accounting
with prudential supervision and demands higher standards of governance. Effective Enterprise
Risk Management (ERM) further emphasizes integrating credit risk with liquidity, market, and
operational risks (Lam, 2014). Evidence of cross-risk dashboards or ICAAP integration in dis-
closures would reflect this holistic approach.

12. Climate and ESG Risk

Climate change introduces new risk channels, including transition risks from regulatory shifts
and physical risks from climate events. Emerging frameworks recommend integrating these
into stress-testing and portfolio evaluation (NGFS, 2019; EY, 2022). For banks like IDFC
FIRST, disclosures about green finance frameworks or climate-linked stress tests would signal
readiness for future supervisory expectations.

Conceptual Model for the Study

The study adopts a layered conceptual model:

 Inputs: public disclosures on credit architecture, portfolio mix, and governance.

17
  Analytical modules: (1) adequacy of appraisal and scoring; (2) sufficiency of govern-
ance and provisioning practices; (3) resilience of the portfolio under stress scenarios.
 Outputs: policy recommendations such as enhanced disclosure, governance strength-
ening, and stress-test recalibration.

Three guiding propositions are formulated:

 P1: Banks with more transparent model-governance disclosures (validation cycles, PSI
reporting, challenger tests) face lower unexpected volatility in provisions.
 P2: Retail expansion reduces concentration risk but increases macro sensitivity unless
matched by improved scoring and EAD controls.
 P3: Alternative data adoption enhances SME risk prediction but must be supported by
strong disclosure on data governance and model explainability.

Summary

This framework demonstrates that effective credit-risk management requires alignment across
data inputs, modelling approaches, governance structures, and macroeconomic linkages. For
secondary-data research, the task is to identify how far IDFC FIRST Bank’s disclosures reflect
these principles. Gaps — particularly in PD/LGD elasticities, validation detail, and alternative-
data governance — provide the basis for actionable recommendations (BCBS, 2015; Jagtiani
& Lemieux, 2019).

3.2 Literature Review

1. Altman, E. I. (1968). Financial ratios, discriminant analysis and the prediction of cor-
porate bankruptcy. The Journal of Finance, 23(4), 589–609

Altman’s 1968 study is regarded as a landmark in the field of credit risk, as it introduced the
Z-score model that applies discriminant analysis to a set of financial ratios. The model incor-
porates variables such as working capital relative to total assets, retained earnings, EBIT, mar-
ket value of equity to liabilities, and sales-to-assets to differentiate between firms at risk of
failure and those likely to remain solvent. Tested on U.S. manufacturing firms, the model
proved capable of predicting bankruptcy years in advance, offering both methodological inno-
vation and practical relevance.

For banking applications, the Z-score demonstrated the usefulness of ratio-based screening as
an early warning system, particularly in corporate lending where audited financial statements
are available. Though its direct application is limited in retail or MSME segments—owing to
the absence of detailed balance-sheet data—the model’s conceptual value remains significant.
Contemporary analysts, including those examining IDFC FIRST Bank, can still apply ratio-
based disclosures to assess potential distress in corporate portfolios. However, this approach
depends heavily on the consistency of financial reporting and the availability of market data.
Within a secondary-data framework, the Z-score highlights the importance of evaluating
whether banks provide adequate ratio information to allow stakeholders to assess credit risk.

18
2. Shumway, T. (2001). Forecasting bankruptcy more accurately: A simple hazard model.
Journal of Business, 74(1), 101–124.

Shumway (2001) contributed to credit risk forecasting by evaluating multiple prediction meth-
ods, including discriminant analysis, logit models, and hazard models. He found that hazard,
or survival, models generally outperform static, single-period classifiers because they account
for both the passage of time and changes in borrower characteristics. By incorporating dynamic
factors such as variations in profitability, liquidity, and macroeconomic conditions, hazard
models provide a more accurate representation of a firm’s evolving risk profile than static ratio
analysis.

A key advantage of hazard models is their alignment with regulatory frameworks like IFRS 9,
where staging requires recognition of lifetime risk as credit quality declines. Shumway also
highlighted the benefit of integrating macroeconomic indicators to enhance predictive accu-
racy, making these models particularly effective under stress conditions.

For analysts relying on secondary data, the implication is that disclosures should extend beyond
point-in-time ratios to include time-series information such as roll rates, vintages, or trend anal-
yses. Without such data, assessing the lifetime probability of default becomes limited. Shum-
way further stressed the importance of governance measures—such as cross-validation and
out-of-sample testing—to avoid overfitting, which continue to serve as standards for evaluating
the robustness of a bank’s model governance.

3. Bellotti, T., & Crook, J. (2009). Credit scoring with macroeconomic variables using
survival analysis. Journal of the Operational Research Society, 60(12), 1699–1710.

Bellotti and Crook (2009) extended survival analysis approaches by embedding macroeco-
nomic covariates into credit scoring models. Their empirical work demonstrated that variables
such as GDP growth and unemployment rates, when integrated into hazard models, signifi-
cantly improved the predictive accuracy of default forecasting compared to static models. By
capturing how borrower risk evolves alongside broader economic shocks, the model offers a
through-the-cycle perspective, making it well-suited for expected credit loss frameworks and
stress-testing applications.

The authors showed that incorporating macroeconomic data allows for more accurate forward-
looking provisioning, especially under adverse conditions. This has direct implications for reg-
ulators and banks seeking to build resilience in loan portfolios. However, the added complexity
also increases interpretability challenges, as models become less transparent to external stake-
holders.

In evaluating IDFC FIRST Bank’s disclosures, this study suggests attention should be paid to
whether stress-testing results explicitly link portfolio risk to macroeconomic drivers, or
whether references are made to scenario-based adjustments in PD or LGD estimates. The ab-
sence of such linkages would highlight a gap in the sophistication of published risk manage-
ment practices. Ultimately, the study defines good practice benchmarks for disclosure quality
and methodological soundness.

4. Khandani, A. E., Kim, A. J., & Lo, A. W. (2010). Consumer credit-risk models via
machine-learning algorithms. Journal of Banking & Finance, 34(11), 2767–2787.

19
Khandani, Kim, and Lo (2010) investigated the performance of traditional statistical ap-
proaches, like logistic regression, in comparison with advanced machine-learning (ML) algo-
rithms such as random forests, boosting, and support vector machines, applied to large-scale
consumer credit datasets. Their analysis revealed that ML methods substantially enhanced de-
fault forecasting, particularly for longer-term delinquencies where intricate behavioural pat-
terns play a significant role.

The study also highlighted the trade-offs inherent in ML adoption. Although predictive accu-
racy improved, model interpretability declined, creating challenges for transparency and regu-
latory compliance. Moreover, ML models carry risks including overfitting and high computa-
tional demands.

For secondary-data assessments of banks like IDFC FIRST, these findings suggest that claims
regarding ML usage should be accompanied by clear disclosures of the variables or features
employed, key performance metrics such as AUC or KS statistics, and the validation frame-
works applied. In the absence of this information, it is difficult to determine whether ML adop-
tion is genuinely effective or largely symbolic. The research also underscores that banks failing
to leverage alternative data or ML techniques may be at a disadvantage in managing the rising
risks of unsecured and retail credit portfolios.

5. Lessmann, S., Baesens, B., Seow, H. V., & Thomas, L. C. (2015). Benchmarking state-
of-the-art classification algorithms for credit scoring: An update of research. European
Journal of Operational Research, 247(1), 124–136.

Lessmann et al. (2015) conducted one of the most extensive benchmarking studies in credit
scoring by evaluating a variety of algorithms, including logistic regression, decision trees, ran-
dom forests, support vector machines, and neural networks. Their results indicated that ensem-
ble approaches, such as boosting and random forests, frequently offered superior discrimina-
tory power. Nonetheless, the performance gains over simpler models were not always substan-
tial, and simpler methods often retained advantages in interpretability and regulatory compli-
ance.

The study also emphasized the need to assess models on multiple dimensions—such as cali-
bration, classification error, and robustness to shifts in borrower populations. The authors ar-
gued that high predictive accuracy alone is insufficient if models cannot be effectively ex-
plained to stakeholders or fail under population changes.

For IDFC FIRST Bank, this implies that model-related disclosures should go beyond basic
accuracy metrics. Providing information on calibration plots, population stability measures,
and comparisons with simpler baseline models is essential to demonstrate sound governance.
A lack of such disclosures could indicate gaps in model validation or limited transparency re-
garding critical governance practices.

6. IFRS Foundation. (2014). IFRS 9 — Financial Instruments.

The adoption of IFRS 9 represented a significant change in how banks account for and provi-
sion against credit risk. Replacing the previous incurred-loss approach, IFRS 9 introduced a
forward-looking expected credit loss (ECL) framework, obligating financial institutions to es-
timate potential credit losses under multiple economic scenarios. Exposures are categorized
into three stages: Stage 1 (12-month ECL for assets without significant deterioration), Stage 2

20
(lifetime ECL for exposures experiencing a notable increase in credit risk), and Stage 3 (credit-
impaired assets). This staging system promotes earlier recognition of potential losses and in-
corporates macroeconomic forecasts into provisioning calculations.

For banks, IFRS 9 compliance entails robust methodologies for determining staging triggers,
constructing scenarios, and performing sensitivity analyses. The standard also bridges account-
ing and prudential oversight, making transparent disclosure of ECL methodologies essential
for maintaining investor and regulatory confidence.

Regarding IDFC FIRST Bank, secondary-data assessments should verify whether public dis-
closures adequately detail staging practices, scenario assumptions, provisioning coverage ra-
tios, and sensitivity to adverse conditions. Inadequate or absent disclosure of these elements
represents a transparency gap, limiting stakeholders’ ability to assess the soundness of credit-
risk management. IFRS 9 thus serves as the global benchmark for consistent and comparable
provisioning practices.

7. Basel Committee on Banking Supervision. (2015). Guidance on credit risk and account-
ing for expected credit losses. Bank for International Settlements.

The Basel Committee’s 2015 guidance complements IFRS 9 by setting out supervisory expec-
tations for banks implementing expected credit loss (ECL) models. It emphasizes the need for
rigorous methodologies in estimating probability of default (PD), loss given default (LGD),
and exposure at default (EAD), as well as careful scenario design and model validation. Super-
visors expect banks to ensure that provisioning is forward-looking, prudent, and responsive to
evolving macroeconomic conditions.

The guidance also underscores governance requirements, including independent model valida-
tion, oversight of management overrides, back-testing, monitoring of population stability, and
transparent disclosure of sensitivity to economic stress. Weak governance or reliance on overly
optimistic assumptions can lead to under-provisioning during downturns, posing risks to finan-
cial stability.

For IDFC FIRST Bank, this framework provides a reference point for evaluating disclosures
and risk practices. Analysts should assess whether published materials clearly present PD/LGD
assumptions, link stress tests to realistic macroeconomic scenarios, and describe model gov-
ernance. Omissions in these areas represent significant governance and transparency gaps. The
guidance also promotes peer benchmarking, allowing analysts to estimate plausible risk sensi-
tivities even if the bank does not publish them. Accordingly, BCBS 2015 functions as a super-
visory benchmark for assessing the robustness of IDFC FIRST Bank’s credit-risk framework.

8. Mishra, A. K., & Ghosh, S. (2021). Determinants of NPAs in Indian banking. Interna-
tional Journal of Finance & Economics.

Mishra and Ghosh (2021) investigate the structural factors driving non-performing assets
(NPAs) in Indian banks using a dynamic panel methodology. Their study identifies both mac-
roeconomic and bank-specific influences. At the macro level, slower GDP growth and nar-
rower interest margins are associated with higher NPA formation. Institutionally, rapid credit
expansion, poor cost efficiency, and weak governance mechanisms contribute to asset quality
deterioration.

21
The research also emphasizes portfolio composition effects: banks with larger retail and unse-
cured lending portfolios tend to experience greater default slippage during economic down-
turns. Governance and disclosure practices further influence resilience, with low transparency
heightening investor concerns.

For IDFC FIRST Bank, these findings are particularly pertinent. The bank’s post-merger strat-
egy focuses on accelerated growth in retail and MSME lending, segments that are historically
more susceptible to economic stress. Consequently, secondary-data assessments should exam-
ine whether rapid loan growth correlates with rising GNPA ratios and whether disclosures ad-
equately address governance practices related to underwriting and provisioning. Benchmarking
IDFC FIRST Bank against these empirical patterns allows analysts to evaluate whether its as-
set-quality trajectory aligns with broader industry trends.

9. Ghosh, A. (2020). Non-performing loans in emerging economies: Case of India. Inter-

national Journal of Finance & Economics, 25(1), 68–84.

Ghosh (2020) examines the macroeconomic and institutional factors influencing non-perform-
ing loans (NPLs) in India, using panel data from both public and private banks. The study finds
that slower GDP growth and higher inflation are consistently linked to increased NPL levels.
At the institutional level, banks with lower capital adequacy, rapid credit expansion, and higher
proportions of unsecured lending are particularly susceptible to asset-quality deterioration.

The research also highlights that under-provisioning often occurs during periods of mild stress,
resulting in capital shortfalls during downturns. Moreover, a lack of disclosure regarding stress-
test assumptions limits stakeholders’ ability to fully assess a bank’s vulnerability.

For IDFC FIRST Bank, these insights are especially relevant given its strategy of expanding
retail and unsecured lending to reduce reliance on corporate exposures. While this approach
mitigates concentration risk, it increases sensitivity to economic cycles. As a result, transparent
disclosures regarding capital buffers, provisioning coverage, and scenario design are essential.
The absence or incompleteness of such information may raise doubts among external stake-
holders about the bank’s resilience. Ghosh’s findings thus provide a practical framework for
evaluating both disclosure adequacy and preparedness for systemic risk.

10. Jagtiani, J., & Lemieux, C. (2019). The roles of alternative data and machine learning
in fintech lending: Evidence from the Lending Club consumer platform. Financial Man-
agement, 48(4), 1009–1029.

Jagtiani and Lemieux (2019) examine how alternative datasets and machine-learning methods
influence credit assessment in fintech lending, using Lending Club as a case study. They find
that adding nontraditional signals — for example, digital payment records, utility bill histories,
and behavioural metrics — helps shrink information gaps, especially for borrowers with sparse
credit files. Machine-learning models further boost predictive power and can broaden credit
access for previously underserved populations.

The authors also caution that these innovations carry risks: alternative data can embed biases,
create privacy vulnerabilities, and weaken model interpretability. Regulators have warned that,
absent strong governance, such approaches may undermine transparency and consumer confi-
dence.

22
For IDFC FIRST Bank, the implication is twofold. Disclosures should state whether transac-
tional sources (e.g., GST data) or other alternative inputs are used in MSME and retail scoring,
and they should describe safeguards for explainability and data quality. Without these disclo-
sures, claims of digital transformation are difficult to verify. Applying Jagtiani and Lemieux’s
framework enables analysts to judge whether the bank is managing the trade-off between ex-
panding inclusion and maintaining robust governance.

11. Thomas, L. C. (2009). Consumer credit models: Pricing, profit, and portfolios. Oxford
University Press.

Thomas (2009) delivers a comprehensive account of consumer credit modelling that merges
academic frameworks with practitioner experience. He surveys techniques such as application
scorecards, behavioural scoring, vintage and roll-rate analysis, and portfolio-level risk–return
optimisation, while arguing that models alone are insufficient. Effective consumer-credit man-
agement also requires reliable data pipelines, reject-inference approaches, regulatory compli-
ance, and incentive structures aligned with prudent risk-taking.

The book is particularly practical on validation and monitoring, discussing measures like KS,
AUC, Gini coefficients, and population stability indices (PSI). It also addresses real-world de-
ployment issues—recalibration, monitoring cadence, and governance covering pricing and col-
lections actions.

For IDFC FIRST Bank, which has accelerated retail and MSME lending, Thomas’s framework
serves as a diagnostic checklist. Secondary-data reviews should look for published segmenta-
tion logic, performance statistics, vintage/roll analyses, and PSI trends; missing disclosures
may indicate governance weaknesses, whereas their presence supports confidence in the bank’s
risk-management practices.

12. Hand, D. J., & Henley, W. E. (1997). Statistical classification methods in consumer
credit scoring: A review. Journal of the Royal Statistical Society: Series A (Statistics in
Society), 160(3), 523–541.

Hand and Henley (1997) survey statistical classification approaches used in consumer credit
scoring and provide an early, systematic appraisal of model selection. They compare traditional
methods—such as linear discriminant analysis and logistic regression—with alternatives like
decision trees and nearest-neighbour classifiers, noting that more sophisticated techniques
sometimes offer only modest gains in predictive performance while imposing costs in inter-
pretability and operational deployment.

The authors argue that predictive accuracy alone is an insufficient benchmark for credit-scoring
models. Regulators and practitioners should also weigh calibration, cost-sensitive metrics, and
the capacity to explain model outputs to decision-makers. They further emphasize the utility of
ROC/AUC analysis and cost–benefit evaluation when judging real-world effectiveness.

Applied to IDFC FIRST Bank, the study implies that any claims of advanced modelling should
be backed by disclosure of calibration evidence, performance statistics, and governance con-
trols. Absent such transparency, stakeholders cannot confidently judge whether technical com-
plexity translates into dependable, operational risk management.

23
13. Baesens, B., Van Gestel, T., Viaene, S., Stepanova, M., Suykens, J., & Vanthienen, J.
(2003). Benchmarking classification algorithms for credit scoring. Journal of the Opera-
tional Research Society, 54(6), 627–635.

Baesens et al. (2003) evaluate multiple classification techniques for credit scoring using real-
world datasets, stressing that preprocessing, feature construction, and rigorous validation (for
example, cross-validation) are as important as the choice of algorithm. Their results indicate
that while complex methods—such as neural networks and support vector machines—can out-
perform simpler approaches in specific settings, properly validated models like logistic regres-
sion frequently deliver comparable results.

The paper argues that data quality and methodological rigor often outweigh algorithmic com-
plexity. In practice, transparency, reproducibility, and consistent performance across datasets
are the key success factors for credit-scoring systems.

Applied to IDFC FIRST Bank, this means secondary-data reviewers should look beyond claims
about model types to see whether the bank discloses its validation procedures, decision thresh-
olds, and baseline comparisons. Asserting the use of advanced models without accompanying
methodological transparency may signal governance weaknesses or unproven performance
claims.

14. Blöchlinger, A., & Leippold, M. (2006). Economic benefit of powerful credit scoring.
Journal of Banking & Finance, 30(3), 851–873.

Blöchlinger and Leippold (2006) quantify how better credit-scoring translates into measurable
economic gains using portfolio simulation. More accurate estimates of probability of default
(PD) improve borrower selection, pricing decisions, and capital allocation, which in turn raise
profitability and reduce capital needs for a given risk level.

Their results indicate that the incremental value of scoring enhancements is greatest in unse-
cured, high-volume segments—for example, consumer and MSME lending—which are in-
creasingly focal for retail-oriented banks. The authors also stress the need to embed scoring
outputs into pricing frameworks, capital-allocation rules, and continuous monitoring processes
to capture these benefits.

Applied to IDFC FIRST Bank, the implication is clear: publishing scoring-performance metrics
(AUC, KS, calibration) and quantifying their economic impact (e.g., default reduction, im-
provements in net interest margin) would substantiate risk-management claims. Absent such
evidence, assertions of strong scoring capability are harder to verify. Finally, robust govern-
ance—regular recalibration and the integration of scorecards into limit and pricing systems—
is essential to realize the economic advantages described in the study.

15. Crook, J., Edelman, D., & Thomas, L. (2007). Recent developments in consumer credit
risk assessment. European Journal of Operational Research, 183(3), 1447–1465.

Crook, Edelman, and Thomas (2007) present a comprehensive review of consumer credit-risk
practice, covering application scorecards, behavioural scoring, reject-inference techniques, and
collections modelling. They argue that effective management of retail credit relies on layered
modelling — distinct tools for origination, behavioural monitoring, and collections — each
demanding tailored validation and governance.

24
A key focus of the paper is rejecting inference, which addresses the sample bias that arises
when declined applicants are omitted from model training. The authors also examine the use
of ensemble and machine-learning approaches, stressing the need to balance gains in predictive
performance with regulatory expectations for interpretability.

For IDFC FIRST Bank, the paper translates into a practical disclosure checklist: evidence of
both application and behavioural models, documentation of reject-inference methods, use of
controlled experiments (e.g., A/B tests) for scorecards, and routine vintage or roll-rate moni-
toring. Missing disclosures in these areas would raise questions about the robustness of retail
credit-risk systems, whereas their presence would indicate alignment with international best
practice.

16. Resti, A., & Sironi, A. (2007). Risk management and shareholders’ value in banking:
From risk measurement models to capital allocation policies. Wiley.

Resti and Sironi (2007) argue that credit-risk models must feed directly into strategic financial
decisions, including capital allocation and efforts to enhance shareholder value. They show
how robust estimates of probability of default (PD), loss given default (LGD), and exposure at
default (EAD) should be embedded within an economic-capital framework and aligned with
regulatory ICAAP requirements. The book supplies practical approaches for converting model
outputs into capital cushions, provisioning policies, and portfolio rebalancing actions.

Their central message is that risk models are part of a continuous managerial process—not
isolated technical artifacts—and therefore should inform choices about dividends, target CET1
levels, and growth ambitions. The authors also promote sensitivity analysis and scenario testing
as ways to quantify the trade-offs between profitability and resilience.

Applied to IDFC FIRST Bank, Resti and Sironi’s framework implies that external reviewers
should look for disclosures linking stress-test results to concrete strategic responses (for exam-
ple, adjustments to capital buffers or portfolio composition). Where model outputs appear un-
connected to management actions, technical sophistication alone may not deliver true resili-
ence. This managerial lens helps turn model audits into actionable governance recommenda-
tions.

17. Demyanyk, Y., & Hasan, I. (2010). Financial crises and bank failures: A review of
prediction methods. Omega, 38(5), 315–324.

Demyanyk and Hasan (2010) survey approaches used to forecast financial crises and bank fail-
ures, covering methods such as discriminant analysis, early-warning indicators, hazard/survival
models, and macro–financial linkage frameworks applied across historical episodes. They cau-
tion that many predictive exercises are weakened by issues like data-snooping, structural re-
gime shifts, and reliance on static model assumptions, which together reduce effectiveness once
a crisis unfolds.

The authors recommend thorough robustness testing—for example, rolling-window estima-

tion, strict out-of-sample validation, and explicit incorporation of macroeconomic shocks—to
avoid misleading confidence from calm-period results. They stress that models that perform
well in tranquil conditions can fail badly under stress, so conservative inference and clear re-
porting of validation procedures are essential.

25
Applied to IDFC FIRST Bank, this review implies caution when interpreting published model
metrics: performance claims lacking scenario-based validation or crisis-period back-testing
should be treated sceptically. Public disclosures that detail back-testing frequency, model up-
date practices, and the assumptions underlying stress scenarios increase the credibility of a
bank’s risk-modelling claims.

18. Jorion, P. (2007). Value at Risk: The new benchmark for managing financial risk (3rd
ed.). McGraw-Hill.

Jorion (2007) made a seminal contribution to risk management with his treatment of Value at
Risk (VaR). Although VaR is mainly a market-risk tool, its concepts are highly relevant to
credit-risk oversight. Jorion outlines VaR calculation approaches and their usefulness for meas-
uring tail exposures, while also warning about key limitations—model risk, heavy-tailed re-
turns, and liquidity-driven losses. He argues that VaR must be complemented by stress testing,
scenario analysis, and reverse stress tests to capture risks that historical correlations miss.

The practical takeaway is that single-number metrics cannot drive robust decisions on their
own; regular back-testing, independent model validation, and a diversity of scenarios are nec-
essary to safeguard resilience. For IDFC FIRST Bank, the lesson is clear: disclosures should
pair capital-buffer figures with the scenario assumptions and management responses that pro-
duced them. Presenting buffers without explaining the underlying scenarios or intended man-
agement levers would leave an important transparency gap in credit-risk governance.

19. Network for Greening the Financial System (NGFS). (2019). A call for action: Climate
change as a source of financial risk.

The NGFS (2019) report highlights how climate change introduces material financial risks
through both physical and transition channels. It provides scenario frameworks to translate
climate shocks into financial variables, such as collateral impairment, sectoral demand
changes, and asset price adjustments. These transmission channels affect PDs and LGDs, es-
pecially for sectors like agriculture, energy, and real estate.

The report calls on banks to embed climate scenarios into stress tests, expand disclosure on
climate-related risks, and incorporate ESG factors into credit appraisal. It stresses that ignoring
climate pathways creates blind spots in long-term portfolio resilience.

For IDFC FIRST Bank, the absence of climate-risk disclosures would represent a governance
gap compared to global peers. Conversely, adoption of NGFS-aligned stress testing or green-
lending frameworks would demonstrate forward-looking sophistication. This makes climate-
risk reporting a crucial indicator for secondary-data assessment of the bank’s resilience to
emerging risks.

20. Rajan, R. S., & Dhal, S. C. (2003); Das, A., & Ghosh, S. (2007). Representative empir-
ical studies on NPAs in Indian banking.

The NGFS (2019) report underscores that climate change poses tangible financial risks through
both physical impacts and transition dynamics. It supplies scenario toolkits that map climate
shocks onto financial variables—such as collateral impairment, sector-specific demand shifts,
and asset-price adjustments—which in turn influence PDs and LGDs, especially in vulnerable
sectors like agriculture, energy, and real estate.

26
The report urges banks to incorporate climate scenarios into their stress-testing frameworks,
broaden disclosures on climate-related exposures, and fold ESG considerations into credit ap-
praisals. Ignoring climate pathways creates significant blind spots for long-term portfolio re-
silience.

Applied to IDFC FIRST Bank, a lack of climate-risk disclosure would signal a governance
shortfall relative to international peers, whereas NGFS-aligned stress testing or green-lending
frameworks would reflect forward-looking risk management. Thus, climate-risk reporting is a
key item to check in any secondary-data review of the bank’s preparedness for emerging sys-
temic risks.

21. Ghosh, A. (2020). Non-performing assets in Indian banks: Macro and micro determi-
nants. International Journal of Finance & Economics.

According to Ghosh (2020), the issue of non-performing assets (NPAs) in Indian banks can be
explained by a mix of macroeconomic variables and institution-specific factors, assessed
through panel data covering more than a decade. The study suggests that GDP growth, infla-
tion, and changes in interest rates, alongside elements such as bank size, pace of lending, and
capital strength, all have a strong bearing on asset quality. A key insight is that rapid loan
expansion—particularly in retail and MSME portfolios—often leads to future stress in NPAs.
When credit is disbursed quickly without parallel improvements in risk monitoring, early weak-
nesses in the loan book remain hidden.

The research also stresses the role of governance structures and regulatory oversight in shaping
credit outcomes. Banks that maintain higher levels of capital adequacy and ensure greater trans-
parency in reporting are better positioned to handle financial disruptions.

These conclusions hold direct implications for IDFC FIRST Bank. While the bank’s retail- and
MSME-led growth strategy helps reduce dependence on corporate borrowers, it may also cre-
ate slippages if underwriting practices do not keep pace. Hence, an analysis of secondary data
for the bank should focus on loan portfolio growth, GNPA movements, and capital adequacy
ratios, aligning them with the evidence provided in Ghosh’s findings.

22. Mishra, A. K., & Ghosh, S. (2021). Loan restructuring, forbearance, and credit qual-
ity in India. International Journal of Finance & Economics.

Mishra and Ghosh (2021) investigate how regulatory forbearance and loan restructuring affect
credit quality in Indian banks. Their findings indicate that while such temporary measures—
like restructuring during economic downturns—can offer short-term relief, they often postpone
the recognition of problem assets and ultimately contribute to a rise in NPAs. The study high-
lights that forbearance may conceal actual credit risk, particularly in retail and SME lending
segments where borrower monitoring tends to be weaker.

The authors emphasize the need for stronger early-warning frameworks and closer borrower
surveillance instead of depending heavily on regulatory concessions. They further argue that
timely provisioning and greater transparency in disclosures are crucial for sustaining depositor
and investor trust.

For IDFC FIRST Bank, which operated under the RBI’s forbearance framework during the
COVID-19 crisis, these insights are especially relevant. The research suggests that depending

27
solely on restructuring can hide portfolio vulnerabilities. Consequently, data on delinquency
patterns, recovery approaches, and additional provisioning should be viewed as important in-
dicators for assessing the bank’s resilience.

23. Bank of England. (2018). Stress testing the UK banking system: Guidance for banks.

The Bank of England (2018) describes its stress-testing approach, focusing on evaluating
banks’ resilience against severe yet realistic macroeconomic disruptions. In this framework,
shocks such as declines in GDP, rising unemployment, and falling property prices are trans-
lated into probability of default (PD), loss given default (LGD), and exposure at default (EAD).
This enables regulators to assess whether capital buffers are sufficient under adverse condi-
tions.

An important insight from the framework is that stress testing should go beyond a formal reg-
ulatory requirement; it should be embedded as a continuous element of risk management. The
model also stresses the role of management interventions—such as curbing loan disbursements
or enhancing recovery efforts—when interpreting stress-test outcomes.

For IDFC FIRST Bank, these principles suggest the need for greater transparency in how sce-
narios are designed and how management responses are factored into stress-test disclosures.
Publishing only headline capital ratios, without linking them to assumptions or actions, limits
stakeholders’ ability to judge resilience. Incorporating practices similar to those of the Bank of
England would enhance credibility and demonstrate alignment with global supervisory expec-
tations.

24. European Central Bank (ECB). (2021). Supervisory priorities for European banks.

The European Central Bank (ECB, 2021) outlines supervisory priorities that focus on credit
risk management, loan origination practices, and governance of risk models. The report cau-
tions against excessive dependence on automated underwriting systems without adequate bor-
rower assessment and underscores the importance of transparency in provisioning and loan
classification.

It further stresses regulatory attention on model-related risks, requiring banks to provide evi-
dence of independent validation, proper documentation of overrides, and sensitivity testing
through scenario analysis. A key message is that supervisory credibility is closely tied to the
depth of disclosures, making clear public communication an essential part of resilience.

For IDFC FIRST Bank, these lessons are highly relevant. With the bank increasing its retail
loan portfolio through digital channels, maintaining a balance between operational efficiency
and strong due diligence becomes crucial. A review of secondary data should therefore con-
sider whether the bank’s disclosures reflect sound governance of credit models, consistency in
provisioning, and clarity in reporting NPAs. Any weaknesses in these aspects could echo the
vulnerabilities flagged by the ECB’s supervisory experience.

25. Zmijewski, M. E. (1984). Methodological issues related to the estimation of financial

distress prediction models. Journal of Accounting Research, 22, 59–82.

28
Zmijewski (1984) proposed a probit regression framework for forecasting bankruptcy as an
alternative to traditional discriminant techniques such as Altman’s Z-score. The model incor-
porates measures of profitability, leverage, and liquidity to estimate the likelihood of default,
while avoiding rigid distributional assumptions. This added flexibility enhances predictive per-
formance across a broader range of datasets.

The research also highlights the importance of methodological discipline, cautioning against
the risks of biased sampling and overfitting—common challenges in bankruptcy prediction
studies. Clear validation processes and strong robustness checks are emphasized as essential
for ensuring credible outcomes.

For IDFC FIRST Bank, these insights suggest the value of employing multiple modelling ap-
proaches and testing them under different scenarios. Ideally, public disclosures should reflect
whether internal risk models account for sampling bias and population shifts. Without such
governance practices, the credibility of the bank’s risk assessment framework could be ques-
tioned.

26. Berger, A. N., Klapper, L. F., & Udell, G. F. (2001). The ability of banks to lend to
SMEs: Evidence from emerging markets. Journal of Banking & Finance.

Berger, Klapper, and Udell (2001) investigate the obstacles faced by small and medium enter-
prises (SMEs) in accessing finance within emerging economies, particularly highlighting short-
comings in legal systems, collateral mechanisms, and credit information networks. Their anal-
ysis shows that weak enforcement of laws and the absence of dependable financial records
heighten the credit risks encountered by lenders.

The research underscores the importance of robust credit information frameworks and the po-
tential of non-traditional data to strengthen SME credit evaluation. In contexts where formal
systems are underdeveloped, relationship banking and digital platforms can provide partial al-
ternatives, though careful governance is required to prevent inaccurate risk assessment.

For IDFC FIRST Bank, the study reinforces the relevance of incorporating GST records, digital
transaction trails, and other unconventional data in MSME lending practices. At the same time,
the bank’s reporting should highlight governance mechanisms that maintain data integrity and
ensure transparency. Ultimately, the strength of these safeguards will decide whether such in-
novations genuinely enhance financial resilience.

27. Basel Committee on Banking Supervision (BCBS). (2006). International convergence

of capital measurement and capital standards: Basel II framework.

The Basel Committee on Banking Supervision (BCBS, 2006) introduced the Basel II frame-
work, which set out both standardized and internal ratings-based (IRB) approaches for manag-
ing credit risk. These approaches required banks to estimate probability of default (PD), loss
given default (LGD), and exposure at default (EAD), while also ensuring model validation,
supervisory approval, and alignment with capital adequacy requirements. Basel II represented
a major shift toward more risk-sensitive global capital regulation.

Its principles continue to shape later reforms, including Basel III and IFRS 9, which place
stronger emphasis on forward-looking provisioning and governance standards. Banks that

29
adopt IRB methods are expected to demonstrate independent model checks, conduct stress
tests, and provide transparent disclosures about their methodologies.

For IDFC FIRST Bank, Basel II still serves as a point of reference. Even if the bank applies
standardized methods, any secondary-data review should assess whether its reporting reflects
Basel’s governance principles—such as regular model validation, monitoring of overrides, and
consistency with ICAAP. Weaknesses in these areas could suggest that global best practices
are only partly observed.

28. Berger, A. N., & DeYoung, R. (1997). Problem loans and cost efficiency in commercial
banks. Journal of Banking & Finance, 21(6), 849–870.

Berger and DeYoung (1997) examine how loan quality, cost efficiency, and overall profitabil-
ity are connected within banks. They introduce the “bad management” hypothesis, which ar-
gues that weak governance and inadequate credit appraisal practices contribute to rising non-
performing assets (NPAs) and declining efficiency. In contrast, banks with strong management
discipline and rigorous appraisal systems tend to face lower default rates and achieve better
profitability.

The research underscores the close link between credit evaluation practices, operational effi-
ciency, and financial outcomes. Ineffective credit standards not only heighten risk exposure
but also undermine long-term stability.

For IDFC FIRST Bank, the lesson is evident: robust appraisal mechanisms function as both a
safeguard against risk and a driver of profitability. Any secondary-data review should therefore
evaluate whether the bank’s disclosures on underwriting quality, governance measures, and
monitoring processes align with observed improvements in efficiency ratios or profitability
indicators.

29. Jorion, P. (2009). Financial risk manager handbook (6th ed.). Wiley.

Jorion (2009) offers a comprehensive discussion of enterprise risk management (ERM), ad-
dressing the major categories of credit, market, liquidity, and operational risk. He argues that
credit appraisal should not be treated as a standalone activity but rather integrated with capital
planning, liquidity oversight, and stress-testing practices. The text also stresses the role of risk-
adjusted performance metrics, particularly risk-adjusted return on capital (RAROC), in align-
ing credit decisions with long-term shareholder value.

A central theme throughout the work is governance: strong validation processes, rigorous stress
testing, and cross-risk scenario evaluations are essential to achieving consistency across differ-
ent business areas.

For IDFC FIRST Bank, Jorion’s framework suggests that credit-risk reporting should be con-
nected to ICAAP, consolidated dashboards, or broader enterprise governance structures. Where
such integration is missing in disclosures, it may signal that credit risk is still being managed
in isolation—potentially weakening institutional resilience.

30. Basel Committee on Banking Supervision (BCBS). (2017). Prudential treatment of prob-
lem assets: Supervisory guidelines.

30
The Basel Committee on Banking Supervision (BCBS, 2017) outlines international stand-
ards for the recognition, provisioning, and resolution of problem loans. The framework pro-
motes a forward-looking approach to provisioning, uniform loan classification, and greater
transparency in financial reporting. It further advises that banks adopt cautious assumptions
when evaluating non-performing assets to reduce the risk of underestimating extreme loss sce-
narios.

According to the guidelines, inadequate provisioning can weaken investor trust and create sys-
temic instability. Therefore, the BCBS encourages clearer disclosure on non-performing asset
categories, provisioning coverage, and stress-test outcomes.

For IDFC FIRST Bank, integrating these principles would help build stronger confidence
among both regulators and market participants. A secondary data review should assess whether
the bank’s reporting practices reflect these supervisory benchmarks. Any deficiencies in dis-
closure or prudence in provisioning may indicate underlying weaknesses that need to be ad-
dressed to enhance financial resilience.

Synthesis and Gaps

A review of thirty studies reveals several recurring insights. First, quantitative credit-risk mod-
els continue to be the cornerstone of credit appraisal and portfolio oversight, though their rele-
vance depends on regular updates, contextual adaptation, and rigorous validation across eco-
nomic cycles (Altman, 1968; Thomas, 2009). Second, SME and retail lending require blended
approaches that combine financial data with qualitative relational and behavioural cues, help-
ing mitigate information asymmetries in these markets (Berger & Udell, 2006). Third, the ex-
pansion of fintech solutions and alternative data offers significant opportunities for innovation
in credit assessment; however, issues of bias, data protection, and model transparency must be
managed to ensure responsible and sustainable use (Jagtiani & Lemieux, 2019). Fourth, regu-
latory regimes, including the Basel accords and RBI directives, define the supervisory perim-
eter for appraisal practices, fostering prudential discipline and forward-looking provisioning
(BCBS, 2006, 2015, 2017). Finally, the scholarship on stress testing and efficiency suggests
that credit appraisal mechanisms should be viewed not as static tools, but as dynamic, strategic
levers for financial stability, operational efficiency, and profitability over the long run.

Yet, important research gaps persist. Few contributions explore how new-generation Indian
private banks—such as IDFC FIRST Bank—balance the integration of global regulatory prin-
ciples with local innovations in digital lending, retail growth, and MSME financing. Moreover,
while the theoretical discourse on fintech-driven credit scoring is mature, empirical validation
in the Indian retail and MSME segments remains sparse. This study seeks to bridge these gaps
by conducting a secondary-data analysis of IDFC FIRST Bank’s credit appraisal, stress-testing
frameworks, and governance structures, benchmarking its practices against both global stand-
ards and Indian industry experience.

31
 Chapter- 4 Research Methodology
4.1 Research Design

This study adopts a descriptive and analytical research design relying exclusively on sec-
ondary data sources. The objective is to examine credit appraisal and risk management prac-
tices at IDFC FIRST Bank by systematically synthesizing evidence from prior studies, regula-
tory documents, and industry reports. Descriptive research is appropriate for capturing the
“what” of lending practices and risk control mechanisms, while analytical design allows com-
parison across contexts and the evaluation of theoretical frameworks in relation to real-world
banking operations (Saunders, Lewis, & Thornhill, 2019).

Nature and Source of Data

The research is entirely based on secondary data. The following sources have been used:

 Academic Journals: Peer-reviewed studies on credit risk modelling, lending practices,

and risk management (e.g., Berger & Udell, 2020; Altman & Sabato, 2007).
 Books and Monographs: Standard references on financial risk management and credit
evaluation (Crouhy, Galai, & Mark, 2014).
 Industry Reports: Data from RBI publications, IDFC FIRST Bank’s annual reports,
and CRISIL credit rating reports.
 Regulatory Documents: Basel Committee on Banking Supervision (BCBS) guidelines
and Reserve Bank of India circulars on asset quality, provisioning norms, and credit
appraisal procedures.
 Web-Based Databases: Repositories such as JSTOR, Scopus, SSRN, and Re-
searchGate for access to existing literature and empirical findings.

This ensures that the analysis rests on credible and triangulated sources (Johnston, 2017).

Justification for Secondary Data Approach

The decision to employ secondary data was motivated by three reasons:

1. Feasibility and Accessibility: Banking data, particularly loan-level default infor-

mation, is confidential and not easily accessible for primary research (Bryman, 2016).
2. Reliability of Sources: Using audited reports, peer-reviewed articles, and regulatory
documents enhances the robustness and credibility of findings (Smith, 2011).
3. Comparability and Breadth: Secondary data enables comparison across different
time periods, countries, and institutions, making the analysis more generalizable
(Hinds, Vogel, & Clarke-Steffen, 1997).

32
4.2 Methodological Framework

The study applies qualitative content analysis (Mayring, 2014) and comparative review
methods:

 Content Analysis: Reviewing past studies and extracting insights on credit appraisal
procedures, risk measurement models (logit, probit, hazard models), and bank-level
practices.
 Comparative Method: Contrasting IDFC FIRST Bank’s practices with both Indian
and international banking standards to highlight strengths, weaknesses, and gaps.

Where applicable, secondary datasets (e.g., RBI reports, World Bank data on non-performing
loans) have been tabulated and interpreted to provide empirical grounding.

4.3 Limitations of the Methodology

While secondary data offers scope for breadth and historical comparison, limitations remain.
The study is dependent on the accuracy of published data and may not capture the latest
proprietary practices at IDFC FIRST Bank. Moreover, causality (e.g., whether a risk manage-
ment practice directly reduces default) cannot be firmly established without primary evidence
(Ventresca & Mohr, 2002).

33
 Chapter- 5 Analysis & Interpretation
5.1 Overview of Credit Appraisal at IDFC FIRST Bank

An examination of IDFC FIRST Bank’s annual reports (2019–2024) reveals a structural

transition from a corporate-centric portfolio to one dominated by retail lending. In 2018, whole-
sale exposures represented over 60% of total advances, whereas by 2024, retail loans accounted
for approximately 72% of the book. This reflects a deliberate strategic pivot aimed at lowering
borrower concentration risk and enhancing portfolio resilience (IDFC FIRST Bank, 2024).

These developments are consistent with Berger and Udell’s (2020) observation that a diversi-
fied retail loan base tends to stabilize default patterns, while corporate exposures remain more
vulnerable to macroeconomic downturns.

Loan Portfolio Composition

Table 1. Loan Portfolio Mix of IDFC FIRST Bank (2019–2024)

Year Retail Loans (%) Wholesale Loans (%) Source

2019 37 63 IDFC FIRST Bank AR 2019
2020 44 56 IDFC FIRST Bank AR 2020
2021 54 46 IDFC FIRST Bank AR 2021
2022 65 35 IDFC FIRST Bank AR 2022
2023 69 31 IDFC FIRST Bank AR 2023
2024 72 28 IDFC FIRST Bank AR 2024

Figure 1: Loan Portfolio Mix: IDFC FIRST Bank (Annual Reports)

34
 Figure 2: Loan Mix. Source: IDFC FIRST Bank (Annual Reports)

Interpretation of Retail and Wholesale Loan Portfolio Trends (2019–2024)

The data reveals a significant strategic transformation in IDFC FIRST Bank’s loan portfolio
over the six-year period. In 2019, wholesale lending dominated at 63%, while retail loans ac-
counted for only 37%. However, the composition shifted steadily toward retail lending,
reaching 72% in 2024. This change indicates the bank’s deliberate focus on building a granu-
lar, diversified, and relatively lower-risk retail portfolio compared to its earlier dependence
on wholesale loans.
The shift toward retail loans also reflects industry trends, where retail lending offers higher
yields, better asset quality, and reduced concentration risks. Conversely, wholesale lending
has declined from 63% to 28%, suggesting a conscious de-risking of large-ticket exposures.
This rebalancing strategy highlights the bank’s emphasis on sustainable growth, financial sta-
bility, and resilience against market volatility.

5.2 Non-Performing Assets (NPAs) and Risk Trends

Secondary data from the RBI shows that IDFC FIRST Bank’s gross NPA (GNPA) ratio stead-
ily declined despite COVID-19 disruptions. This suggests the effectiveness of revised credit
appraisal norms, restructuring policies, and early warning systems.

35
Table 2. NPA Comparison: IDFC FIRST vs. Industry

IDFC FIRST GNPA Industry GNPA Net NPA (%)

Year Source
(%) (%) IDFC
IDFC FIRST
2019 4.20 9.30 2.30
Bank AR 2019
IDFC FIRST
2020 3.65 8.20 1.90
Bank AR 2020
IDFC FIRST
2021 3.15 7.50 1.70
Bank AR 2021
IDFC FIRST
2022 2.60 6.00 1.30
Bank AR 2022
IDFC FIRST
2023 2.12 4.50 1.00
Bank AR 2023
IDFC FIRST
2024 1.88 3.00 0.85
Bank AR 2024

Figure 3: GNPA Trend. Source: IDFC FIRST Bank (Annual Reports)

Interpretation of Asset Quality Trends (2019–2024)

The data highlights a consistent improvement in the asset quality of IDFC FIRST Bank com-
pared to the broader banking industry. In 2019, the bank’s Gross NPA (GNPA) stood at 4.20%,
already significantly lower than the industry average of 9.30%. Over the six-year period, the
GNPA steadily declined to 1.88% in 2024, while the industry average reduced to 3.00%. This
demonstrates the bank’s effective credit risk management and disciplined lending practices.

36
Similarly, the Net NPA ratio showed remarkable progress, falling from 2.30% in 2019 to 0.85%
in 2024. This trend reflects strong provisioning coverage and prudent risk control measures.
The consistent outperformance of IDFC FIRST Bank relative to industry averages suggests a
strategic focus on building a high-quality, retail-oriented loan portfolio and strengthening un-
derwriting standards.
Overall, the declining GNPA and NPA ratios indicate the bank’s growing financial resilience,
stability, and ability to maintain superior asset quality in a competitive banking environment.

5.3 Capital Adequacy and Risk Absorption Capacity

Capital adequacy is a critical component of risk management. IDFC FIRST Bank has main-
tained compliance with Basel III norms but lags slightly behind top-tier peers.

Table 3. Capital Adequacy Ratios (CET1%) – FY2024

CET1 Total CAR

Bank Source
(%) (%)
IDFC FIRST
13.7 16.8 IDFC FIRST Bank AR 2024
Bank
HDFC Bank 16.5 19.1 HDFC Bank AR 2024
ICICI Bank 15.6 18.0 ICICI Bank AR 2024
Axis Bank 15.2 17.9 Axis Bank AR 2024

Figure 4: CET1 Comparison. Source: Bank’s (Annual Reports)

37
Interpretation of Capital Adequacy Ratios (2024)
The comparative analysis of capital adequacy highlights IDFC FIRST Bank’s stable but rela-
tively lower positioning against leading private sector peers. In FY 2024, IDFC FIRST Bank
reported a Common Equity Tier 1 (CET1) ratio of 13.7% and a Total Capital Adequacy Ratio
(CAR) of 16.8%. While these levels are comfortably above the regulatory requirement of 9%
for CET1 and 11.5% for CAR (including buffers), they remain modest when benchmarked
against larger peers.
For instance, HDFC Bank reported a CET1 of 16.5% and CAR of 19.1%, the highest among
the sample, reflecting strong internal accruals and capital buffers. ICICI Bank and Axis Bank
also demonstrated higher ratios, with CET1 levels of 15.6% and 15.2% respectively, and CARs
of 18.0% and 17.9%. The comparatively lower ratios of IDFC FIRST Bank suggest its ongoing
growth phase, where rapid balance sheet expansion and lending growth necessitate higher cap-
ital deployment, thereby moderating its buffers.
Nonetheless, IDFC FIRST Bank’s ratios indicate sufficient capitalization to absorb shocks and
support business expansion, albeit with relatively less headroom than its peers. This underlines
the importance of future capital planning to sustain competitiveness and ensure long-term fi-
nancial resilience.

Risk Control Mechanisms in Practice

The bank’s secondary data disclosures reveal several risk management mechanisms:

1. Quantitative Models: Use of scoring models for retail borrowers, akin to regression
and machine-learning models (Gupta & Singh, 2021).
2. Collateral Management: Strong reliance on secured lending in the SME and MSME
segment.
3. Early Warning Systems (EWS): AI-enabled monitoring of customer repayment pat-
terns and fraud detection.
4. Diversification: Portfolio spread across housing finance, MSME loans, and consumer
credit.

5.4 Comparative Credit Growth

Another insight from secondary data is credit growth performance compared to industry
peers.

Table 4. Loan Growth Rates (YoY %)

IDFC FIRST
Year HDFC Bank ICICI Bank Axis Bank
Bank

2020 25 21 14 12
2021 18 13 10 8
2022 20 15 12 11
2023 24 17 15 12
2024 26 19 17 13

38
Figure 5: IDFC Loan Growth. Source: Bank’s (Annual Reports)

Figure 6: Loan Growth Comparison. Bank’s (Annual Reports)

39
Interpretation of Loan Growth (2020–2024)

The bar chart illustrates year-on-year (YoY) loan growth of IDFC FIRST Bank in comparison
with HDFC Bank, ICICI Bank, and Axis Bank from 2020 to 2024. IDFC FIRST Bank consist-
ently exhibited the highest loan growth throughout the period, ranging from 18% in 2021 to
26% in 2024, reflecting its aggressive expansion strategy in retail and MSME segments post-
merger.

HDFC Bank maintained the second-highest growth, gradually increasing from 13% in 2021 to
19% in 2024, while ICICI Bank and Axis Bank reported relatively moderate growth rates, with
Axis Bank consistently at the lower end (8% in 2021 to 13% in 2024).

The trends suggest that IDFC FIRST Bank’s strategic focus on higher-yield retail and MSME
lending has allowed it to outperform larger peers in growth terms. However, high growth also
necessitates vigilant credit appraisal and risk management to maintain asset quality, especially
given the elevated exposure to unsecured and short-tenure loans.

Overall, the chart underscores IDFC FIRST Bank’s rapid portfolio expansion and its success
in scaling retail lending, while also highlighting the trade-off between growth and risk over-
sight that must be managed carefully.

Key Insights from Secondary Data

Analysis of secondary data reveals several important trends regarding the bank’s performance
and strategic positioning.

Portfolio Diversification: The bank has increasingly focused on retail lending, which re-
duces dependence on a small number of corporate borrowers and enhances resilience against
sector-specific shocks. However, this shift toward unsecured retail credit introduces higher
exposure to default risk, necessitating rigorous credit appraisal and monitoring mechanisms.

Declining Non-Performing Assets (NPAs): The bank’s gross NPA ratio declined to 1.88%
in 2024, outperforming industry benchmarks. This improvement reflects the effectiveness of
the bank’s credit evaluation process, proactive recovery efforts, and risk management prac-
tices, contributing to greater asset quality stability.

Capital Adequacy: With a Common Equity Tier 1 (CET1) ratio of 13.7%, the bank meets
regulatory capital requirements. While sufficient for compliance, this level is modest com-
pared to larger private sector banks, suggesting limited flexibility for absorbing unexpected
shocks or pursuing highly aggressive expansion strategies.

Growth Strategy: The bank’s loan growth has consistently exceeded peer averages, indicat-
ing a strong focus on market expansion and revenue generation. While this reflects opera-
tional strength, rapid growth necessitates careful monitoring to ensure that asset quality is not
compromised.

Technology Integration: The bank has leveraged AI-based risk models and early warning
systems to enhance its credit monitoring capabilities. These technological advancements sup-
port more accurate risk assessment, timely identification of potential defaults, and overall op-
erational efficiency, providing a competitive advantage in risk management.

40
 Chapter- 6 Key Findings
6.1 Key Finding

The findings below are derived exclusively from secondary sources: IDFC FIRST Bank annual
reports (2019–2024), Reserve Bank of India publications (Financial Stability Reports and Re-
ports on Trend & Progress), and the academic/industry literature summarized in the review.
Each item lists the finding, the supporting evidence, and the practical implication.

Finding 1 — Strategic shift: retail/MSME re-orientation

Statement: From 2019 to 2024, IDFC FIRST Bank undertook a purposeful strategic shift
from wholesale and infrastructure lending towards a predominantly retail- and MSME-oriented
franchise, with retail exposures comprising roughly 72% of the loan book in FY2024 (IDFC
FIRST Bank, 2019–2024).

Evidence: This trend is substantiated by the bank’s segmental loan disclosures across its an-
nual reports (2019–2024), and further supported by industry studies that highlight realisation
as a deliberate strategy among private-sector banks (Berger & Udell, 2006).

Implication: While the transition reduces concentration risk associated with large single bor-
rowers, it simultaneously necessitates more specialized appraisal systems—such as credit
scorecards, behavioural analytics, and early warning systems (EWS)—alongside stronger op-
erational controls to manage the scale and complexity of small-ticket retail lending (Thomas,
2009).

Finding 2 — Robust improvement in asset quality (GNPA decline)

Statement: The Gross Non-Performing Asset (GNPA) ratio fell from approximately 4.2%
in 2019 to around 1.88% in 2024, placing IDFC FIRST Bank ahead of system-wide averages
during the same period (IDFC FIRST Bank, 2019–2024; RBI, 2019–2024).

Evidence: This trajectory is documented through the bank’s annual reports (GNPA/NNPA
disclosures) and corroborated by the RBI Financial Stability Reports, as summarised in Table
1.

Implication: The decline signals improvements in credit appraisal quality, disciplined loan
origination, and stronger post-sanction monitoring. Nonetheless, prior research cautions that
such trends should be interpreted in the context of loan growth, provisioning strength, and po-
tential forbearance effects (Ghosh, 2020; Mishra & Ghosh, 2021).

Finding 3 — Rapid loan growth concentrated in retail channels

Statement: IDFC FIRST Bank recorded loan growth exceeding 20% year-on-year in multiple
periods, outpacing several peers, with the expansion largely concentrated in the retail and
MSME segments (RBI DBIE; IDFC FIRST Bank, Annual Reports).

Evidence: This trend is supported by the bank’s own disclosures on loan book expansion and
comparative data from the RBI Database on Indian Economy (DBIE) (see Table 4).

41
Implication: While rapid growth aligns with the bank’s market share ambitions, the literature
cautions that accelerated expansion can heighten underwriting and operational risks, particu-
larly if it outstrips the pace of model validation and cohort (vintage) monitoring (Demyanyk &
Hasan, 2010).

Finding 4 — Increasing share of unsecured and short-tenor retail products

Statement: The retail portfolio of IDFC FIRST Bank shows an increasing tilt toward unse-
cured lending products, notably personal loans and credit-card receivables (IDFC FIRST Bank,
Product Disclosures).

Evidence: This trend is visible in the bank’s product-mix and segmental yield disclosures re-
ported in annual filings, and aligns with empirical research indicating that unsecured retail ex-
posures exhibit higher cyclicality compared to secured credit (Altman & Sabato, 2007; Ghosh,
2020).

Implication: Given their elevated probability of default (PD) and loss-given-default (LGD)
sensitivity during macroeconomic downturns, such products demand stronger underwriting
standards, tighter controls on exposure-at-default (EAD), and more conservative assumptions
for downturn LGD in risk modelling.

Finding 5 — Capital adequacy adequate by regulatory minimums but lower than top
peers

Statement: In FY2024, IDFC FIRST Bank’s Common Equity Tier 1 (CET1) ratio stood
at approximately 13.7%, comfortably above regulatory thresholds but lower than the 15–16%
levels maintained by several large private-sector peers (IDFC FIRST Bank, 2024; RBI FSR,
2024).

Evidence: This position is documented in the bank’s Pillar III disclosures and capital ade-
quacy tables in the 2024 annual report, and is further contextualized through peer comparisons
in the RBI Financial Stability Report (Table 3).

Implication: Although regulatory compliance is ensured, the relatively thinner capital buffer
limits the bank’s loss-absorption capacity in severe stress events, underscoring the need for
cautious and forward-looking capital planning (BCBS, 2015).

Finding 6 — ECL / IFRS-9 methodology described qualitatively; macro→PD/LGD elas-

ticities not published

Statement: The bank discloses its staging/ECL approach in narrative form but does not publish
explicit macro-to-PD or macro-to-LGD elasticity tables. (IDFC FIRST Bank ARs; BCBS,
2015).
Evidence: ECL narrative sections in annual reports lack quantified elasticity mapping or vin-
tage loss curves.
Implication: Limits external assessment and benchmarking of stress-test plausibility. Best
practice recommends publishing summary elasticities or sensitivity bands to improve transpar-
ency (Bank of England, 2018).

42
Finding 7 — Model-governance exists but public granularity is limited

Statement: IDFC FIRST Bank reports model validation teams, back-testing and scorecard use,
but does not publish PSI trends, challenger-model outcomes, or override analytics in detail.
(IDFC FIRST Bank ARs; Zmijewski, 1984; BCBS, 2006).
Evidence: Governance sections in AR and disclosures about validation cadence; absence of
PSI charts or challenger-test summaries.
Implication: Presence of governance structures is positive, but lack of granular disclosure re-
duces external confidence and hampers replicability of assessments.

Finding 8 — Adoption of alternative/transactional data and AI is underway, governance

needs strengthening

Statement: The bank reports use of transactional data (bank flows, GST verification for
MSMEs) and AI for monitoring/collections, aligning with fintech literature (Jagtiani &
Lemieux, 2019).
Evidence: Technology and digital initiatives sections in ARs; product development notes.
Implication: Alternative data improves PD discrimination but requires formal ML govern-
ance: explainability (e.g., SHAP), bias testing, feature-drift monitoring and documented pri-
vacy safeguards.

Finding 9 — Stress testing disclosures are headline-level and assume management actions
without granular timelines

Statement: Stress outcomes are reported with management actions referenced qualitatively
(e.g., curtail origination, intensify collections) but detailed action triggers/timelines and their
quantitative impacts are not fully disclosed. (IDFC FIRST Bank ARs; Bank of England, 2018).
Evidence: Stress test chapters in AR; absence of scenario→PD/LGD mapping and manage-
ment action elasticity.
Implication: Without explicit management-action mapping, external stakeholders cannot as-
sess the realism of stress outcomes; this reduces interpretability of capital and provisioning
resilience.

Finding 10 — Provisioning coverage improved, but tail-risk sensitivity may be under-

stated

Statement: Provision Coverage Ratio (PCR) increased over the period, improving loss absorp-
tion. Yet literature cautions that correlated macro shocks can make downturn LGDs higher than
point estimates suggest (Bank of England, 2018).
Evidence: PCR data in ARs and RBI FSRs.
Implication: Maintain conservative overlays and scenario testing for severe, correlated shocks
to ensure PCR remains adequate.

Finding 11 — MSME segment requires hybrid appraisal and active monitoring

Statement: MSME credit remains information-opaque; IDFC FIRST uses a mix of cash-flow
assessment and transactional verification but full operationalization details are not published.
(Berger & Udell, 2006; IDFC FIRST Bank ARs).
Evidence: MSME policy narratives, product descriptions.

43
Implication: Scale MSME scorecards that combine transactional data with relationship-based
overlays and document governance/validation results publicly.

Finding 12 — Disclosure gaps reduce external assessment quality

Statement: The combination of qualitative narratives and missing quantitative mappings (PSI
trends, elasticity tables, challenger outcomes) constrains robust external evaluation. (BCBS,
2015; Zmijewski, 1984).
Evidence: Comparative disclosure audit conducted across annual reports and supervisory guid-
ance.
Implication: Enhanced, high-level disclosures (without revealing proprietary IP) — e.g., PSI
charts, sensitivity bands, summarized challenger outcomes — would materially improve stake-
holder confidence.

Finding 13 — Net assessment: progress with identifiable implementation gaps

Statement: IDFC FIRST Bank has improved asset quality, modernized risk tooling and diver-
sified its portfolio, but rapid retail growth, higher unsecured share, relative capital position and
disclosure gaps remain as principal vulnerabilities.
Evidence: Synthesis of Tables 1–4, AR narratives, RBI FSRs and the literature.
Implication: A focused program of capital strengthening, macro-linked provisioning, disci-
plined unsecured growth and improved model governance/disclosure would substantially raise
resilience.

Short, prioritized implications (actionable)

1. Capital strengthening — prioritize CET1 enhancement (equity or retained earnings).

2. Macro-linked ECL calibration — publish and use PD/LGD elasticity bands and con-
servative downturn LGDs.
3. Controlled unsecured growth — tighten underwriting, set product-level caps, and
monitor vintage performance.
4. ML governance & explainability — adopt SHAP/LIME reporting, feature-drift mon-
itoring and bias testing.
5. Transparency improvements — publish PSI trends, summary challenger outcomes
and high-level stress mapping (without disclosing IP).

6.2 Practical recommendations

Short-Term (0–12 months)

 Strengthen stress-test disclosures: Publish high-level macro→PD/LGD sensitivity

ranges and outline management actions and triggers (Bank of England, 2018).
 Tighten underwriting for new unsecured retail products: Implement product-level
exposure caps, stricter credit triggers, and enhanced affordability checks.
 Formalize ML governance: Require explainability outputs (e.g., SHAP summaries),
monthly Population Stability Index (PSI) monitoring, and documented bias assess-
ments.

44
Medium-Term (12–24 months)

 Enhance capital planning: Target capital strengthening through retained earnings,

strategic equity injections, or selective securitisation of non-core assets to bring CET1
closer to peer benchmarks.
 Institutionalize vintage and cohort monitoring: Publish aggregated vintage perfor-
mance metrics for key product lines to improve transparency and enable early risk de-
tection.

Long-Term (24+ months)

 Deepen MSME analytics: Deploy hybrid scorecards combining transactional GST

data with relationship-based overlays, with documented validation and back-testing.
 Integrate climate and transition risks: Incorporate climate-related risk channels into
portfolio stress tests, particularly for property and agricultural exposures, following
NGFS and Basel guidance.

6.3 Theoretical and managerial contributions

This study contributes to the literature and practice by:

(a) demonstrating how a retail-oriented transformation can improve observed asset quality
when accompanied by enhanced appraisal and monitoring;

(b) highlighting the dual role of digital tools as performance enhancers that simultaneously
introduce model-governance requirements; and

(c) emphasizing the continued centrality of capital adequacy as a stabilizing mechanism—even

for innovation-driven banks.

6.4 Limitations

The analysis relies solely on secondary, publicly disclosed information; proprietary loan-level
data, internal validation reports, and first-hand managerial perspectives were unavailable.
Therefore, causal attribution (e.g., precisely which policy change caused GNPA decline) is
necessarily inferential rather than definitive.

Future research avenues

 Comparative peer studies (e.g., IDFC FIRST vs. HDFC/ICICI) using harmonized pub-
lic metrics to benchmark resilience and disclosure quality.
 Primary-data research (interviews, internal model audits) to validate the governance
and operational claims made in annual reports.
 Longitudinal studies of ML adoption’s impact on NPAs across Indian banks, focusing
on explainability and fairness outcomes.

45
 Chapter- 7 Conclusion
This study examined the credit appraisal and credit-risk management practices of IDFC
FIRST Bank using secondary data sources, including annual reports, RBI publications, and
peer-reviewed literature. The research was motivated by the growing importance of resilient
risk frameworks in the Indian banking sector, particularly in the context of recurring episodes
of rising non-performing assets and sectoral stress (Ranjan & Dhal, 2020). The study specifi-
cally focused on understanding how the bank’s transition from a wholesale- and infrastruc-
ture-focused portfolio to a retail-oriented model has influenced asset quality, portfolio resili-
ence, and overall risk governance.

The findings indicate that IDFC FIRST Bank’s strategic shift toward retail and MSME lend-
ing has contributed to a measurable improvement in asset quality. The gross NPA ratio de-
clined from approximately 4.2% in FY2019 to 1.88% in FY2024, reflecting that a well-man-
aged retail portfolio can mitigate concentration risk compared to corporate-heavy exposures
(Patel, 2022). The bank’s adoption of AI and predictive analytics in credit scoring and portfo-
lio monitoring demonstrates the role of technology in enhancing fraud detection, improving
decision-making efficiency, and enabling proactive risk management (Bose & Roy, 2021). At
the same time, the expansion of unsecured retail loans and credit card exposure highlights po-
tential vulnerabilities, emphasizing that rapid growth in high-yield segments requires careful
provisioning and strong governance to prevent deterioration in asset quality (Joshi & Kul-
karni, 2020).

Although the bank maintains a CET1 ratio of around 13.7%, meeting regulatory require-
ments, this level remains modest relative to larger private banks, potentially limiting its ca-
pacity to absorb extreme stress events (Aggarwal & Singh, 2021). Sustained loan growth ex-
ceeding 20% year-on-year reinforces the bank’s competitive positioning but may introduce
underwriting and vintage risks if credit validation and monitoring do not keep pace with ex-
pansion (Banerjee, 2022). The use of AI and machine learning tools has improved operational
efficiency and early risk detection; however, their effectiveness is dependent on robust model
governance, including explainability, bias mitigation, and continuous validation (Sharma &
Bose, 2022). Overall, the research suggests that IDFC FIRST Bank’s strategic pivot and tech-
nology integration have strengthened its credit-risk management capabilities, yet continuous
improvement in governance, monitoring, and capital planning is essential to ensure sustaina-
ble resilience and long-term growth.

46
 Chapter- 8 References

8.1 Annual report links for the four banks from 2020 to 2024

HDFC Bank

 2023–24: Integrated Annual Report FY24

 2022–23: Annual Report FY23
 2021–22: Annual Report FY22
 2020–21: Annual Report FY21
 2019-20: Annual Report FY19

IDFC FIRST Bank

 2023–24: Integrated Annual Report FY24

 2022–23: Annual Report FY23
 2021–22: Annual Report FY22
 2020–21: Annual Report FY21
 2019-20: Annual Report FY19

ICICI Bank

 2023–24: Annual Report FY24

 2022–23: Annual Report FY23
 2021–22: Annual Report FY22
 2020–21: Annual Report FY21
 2019-20: Annual Report FY19

Axis Bank

 2023–24: Integrated Annual Report FY24

 2022–23: Annual Report FY23
 2021–22: Annual Report FY22
 2020–21: Annual Report FY21
 2019-20: Annual Report FY19

47
8.2 References

Aggarwal, R., & Singh, V. (2021). Capital adequacy norms and financial stability in Indian
banks: An empirical study. Journal of Banking and Finance Review, 13(2), 112–129.
[Link]

Altman, E. I. (1968). Financial ratios, discriminant analysis and the prediction of corporate
bankruptcy. The Journal of Finance, 23(4), 589–609. [Link]
6261.1968.tb00843.x

Banerjee, S. (2022). Retail credit expansion in India: Opportunities and risks. International
Journal of Emerging Markets, 17(6), 1421–1443. [Link]
0973

Basel Committee on Banking Supervision. (2019). Credit risk management: Principles and
practices. Bank for International Settlements.

Bose, S., & Roy, D. (2021). Digital innovation and credit risk management in Indian banks.
Journal of Financial Innovation, 5(3), 76–94. [Link]

Chakraborty, R. (2020). Non-performing assets and credit risk management: Lessons from In-
dian banking. Asian Journal of Economics and Finance, 12(1), 55–71.

Choudhury, P., & Jain, S. (2020). Predictive analytics for credit scoring in retail lending. De-
cision Analytics Journal, 7(2), 98–110.

Das, P., & Ghosh, S. (2021). Determinants of NPA in Indian commercial banks: Panel data
evidence. Indian Journal of Finance and Banking, 15(1), 45–62.

Dewan, R., & Kapur, M. (2022). Artificial intelligence in risk-based lending: An Indian per-
spective. Journal of Financial Technology, 8(4), 299–315.

Fernandes, A., & Joseph, T. (2021). Basel III implementation and its impact on Indian banks’
capital adequacy. Journal of International Banking Studies, 14(2), 89–104.

Ghosh, A. (2020). Credit risk management and profitability: Evidence from Indian banks. Jour-
nal of Risk and Financial Management, 13(11), 257–276.
[Link]

Greene, W. H. (2018). Econometric analysis (8th ed.). Pearson.

Gupta, A., & Sharma, R. (2021). Diversification and risk in Indian banks: Evidence from loan
portfolios. Journal of Emerging Financial Markets, 15(4), 201–220.

Hosmer, D. W., Lemeshow, S., & Sturdivant, R. X. (2013). Applied logistic regression (3rd
ed.). Wiley.

Jain, V., & Kumar, S. (2021). Loan growth and credit risk: Evidence from Indian retail banking.
International Journal of Bank Marketing, 39(7), 1129–1147.

48
Joshi, A., & Kulkarni, M. (2020). Risks of unsecured lending in emerging economies: The case
of India. Indian Banking Journal, 45(2), 134–150.

Kaur, P., & Arora, R. (2022). Role of fintech in enhancing credit appraisal processes. Journal
of Financial Innovation, 9(1), 65–83.

Kumar, N., & Sinha, R. (2020). Stress testing in Indian banks: Lessons from the NPA crisis.
Asia-Pacific Journal of Risk and Insurance, 14(3), 1–20.

Merton, R. C. (1974). On the pricing of corporate debt: The risk structure of interest rates. The
Journal of Finance, 29(2), 449–470.

Mishra, S., & Patel, R. (2021). Impact of credit risk on financial performance of private banks
in India. Asian Journal of Economics and Finance, 13(3), 221–237.

Nair, P., & Menon, K. (2020). Use of machine learning in improving credit risk models. Jour-
nal of Banking Analytics, 6(2), 45–63.

Patel, N. (2022). The resilience of retail banking models in India: Evidence from post-NPA
crisis reforms. South Asian Journal of Business Studies, 11(3), 412–430.

Rajan, R., & Dhal, S. C. (2020). Non-performing loans and terms of credit of public sector
banks in India: An empirical assessment. Reserve Bank of India Occasional Papers, 41(2), 1–
25.

Reserve Bank of India. (2019). Report on trend and progress of banking in India 2018–19. RBI
Publications.

Reserve Bank of India. (2020). Report on trend and progress of banking in India 2019–20. RBI
Publications.

Reserve Bank of India. (2021). Report on trend and progress of banking in India 2020–21. RBI
Publications.

Reserve Bank of India. (2022). Report on trend and progress of banking in India 2021–22. RBI
Publications.

Reserve Bank of India. (2023). Financial stability report: June 2023. RBI Publications.

Reserve Bank of India. (2024). Financial stability report: June 2024. RBI Publications.

Sharma, K., & Verma, A. (2021). Early warning systems in credit risk management: Evidence
from Indian banks. Journal of Banking Risk & Compliance, 10(2), 119–137.

Singh, M., & Tandon, A. (2020). Determinants of profitability in Indian private sector banks:
Role of NPAs and provisioning. Global Business Review, 21(5), 1234–1249.

Thomas, J., & Mathew, D. (2022). Financial inclusion and retail lending in India: Risk-return
trade-offs. Journal of Inclusive Finance, 8(3), 201–220.

49
World Bank. (2020). Strengthening risk management in emerging market banks. Washington,
DC: World Bank.

50