HISTORICAL FORENSIC ARCHIVE: EMPIRICAL SYSTEMIC RISK BACKTEST. FOR EDUCATIONAL & RESEARCH PURPOSES.
Dewan Housing Finance Corporation Ltd (DHFL)
DHFL Collapse Forensic Autopsy
Multi-Quarter Pre-Collapse Early Warning Analysis
Flagium did not predict DHFL's concealed fraud. It detected the structural financial deterioration that became observable across public information channels — capital erosion, margin compression, asset-quality deterioration, credit-rating downgrades, and funding stress — before the company's CIRP admission.
A critical methodological distinction must be drawn between fraud concealment and liquidity deterioration. Flagium did not detect DHFL's concealed fraudulent loan accounts before they entered the public record: the 2,60,000 fictitious PMAY borrower accounts and ₹14,046 Cr ghost loans were operated on an offline computer in Bandra Branch 001 that no public algorithmic engine could observe before law enforcement raids. That defines the boundary of a public-disclosure-based Early Warning System — the backtest measures what an analyst could observe from contemporaneous public disclosures, not what forensic investigators discovered years later.
Crucially, DHFL exposes a model-coverage boundary, not merely a data-availability boundary. By September 2018, external market evidence of funding stress was already emerging when DSP Mutual Fund sold ₹300 Cr of DHFL commercial paper at an elevated 11% yield post-IL&FS. However, the historical Flagium configuration evaluated in Q2 FY19 relied primarily on quarterly statutory financial inputs (where DHFL had reported PAT of ₹438 Cr for the quarter ended June 30, 2018 (Q1 FY19), while the latest audited annual statement on record for FY18 had reported 0.96% Gross NPA) and therefore remained at Score 0. This empirical backtest demonstrates why NBFC-specific liquidity, money-market refinancing, and ALM signals require dedicated integration rather than relying solely on reported accounting metrics. As deterioration became visible in the Q3 FY19 reported financial indicators, the historical model configuration escalated to Score 58. Later, as four severe signals (F10, F6, F9, O1) alongside F7 Asset Quality (Medium) were simultaneously active, contributing to the peak Score of 64 under the NBFC sector model in June 2019, external credit markets contemporaneously confirmed the collapse: DHFL defaulted on ₹961 Cr of NCD interest and CRISIL slashed ratings to 'D', documenting ~₹6,200 Cr aggregate scheduled cash outflows through July against roughly ~₹2,200 Cr monthly collections.
Each historical score is reconstructed using only information that was publicly available as of the model's evaluation date. This may include statutory filings, audited financial statements, credit-rating actions and other contemporaneous public disclosures. Subsequent forensic investigations, criminal probes, restatements and hindsight findings are excluded from the score calculation.
| Information Disclosed | Public by Mar 31, 2019? | Can Influence Mar 31 Score? |
|---|---|---|
| CRISIL A2+ Downgrade & Rating Rationale (March 22, 2019) | Yes | Yes — Informs Q4 FY19 evaluation (Score 53) |
| Available Liquidity (~₹4,900 Cr disclosed by CRISIL) | Yes | Yes — Informs Q4 FY19 evaluation (Score 53) |
| Scheduled Outflows (~₹13,370 Cr scheduled monthly outflows through June vs ~₹2,200 Cr monthly collections) | Yes | Yes — Informs Q4 FY19 evaluation (Score 53) |
| Audited Standalone Net Loss (-₹2,223 Cr for FY19) | No | No — Disclosed July 13, 2019 (Subsequent statutory confirmation) |
| Audited Standalone Gross NPA (2.74% for FY19) | No | No — Disclosed July 13, 2019 (Subsequent statutory confirmation) |
| CRISIL 'D' Downgrade & ₹961 Cr NCD Default | No | No — Occurred June 4–5, 2019 (Informs Q1 FY20 peak default horizon) |
- Net Interest Margin (NIM) compression down to 1.80% (F6)
- Cost-of-funds divergence (+113 bps in FY18, +148 bps in FY19) (F9)
- Gross NPA escalation and provisioning stress (F7)
- Capital adequacy buffer erosion toward regulatory floor (F10)
- Operating leverage imbalances and fixed-cost drag (O1)
- Commercial paper rollover freeze and ₹4,000 Cr near-term cash deficit
- 2,60,000 fabricated borrower accounts under PMAY scheme
- ₹14,046 Cr ghost loan book maintained on offline Branch 001 computer
- Concealed loan diversion network to 87 promoter-controlled shell companies
- Fraudulent disbursement vouchers unrecorded in statutory ledgers
📌 DHFL exposes a model-coverage boundary: public-data EWS can detect structural decay quarters in advance, but NBFC-specific liquidity, money-market refinancing, and ALM signals require dedicated integration rather than relying solely on reported accounting metrics.
10-Quarter Forensic Metric Time-Series
Raw tabular accounting data reconstructed point-in-time from statutory MCA/BSE filings. Statically indexable by search crawlers.
| Quarter | Reported PAT | Operating Cash Flow | Risk Index | Auditor Remarks / Status | Active Forensic Flags |
|---|---|---|---|---|---|
| Q1 FY18 | ₹260 Cr | -₹850 Cr | 5 | Unqualified (Clean) | F1 (OCF < PAT) |
| Q2 FY18 | ₹293 Cr | -₹1,120 Cr | 6 | Unqualified | F1 |
| Q3 FY18 | ₹312 Cr | -₹1,440 Cr | 1 | Unqualified | F1 |
| Q4 FY18 | ₹307 Cr | -₹1,890 Cr | 0 | Unqualified | F1 |
| Q1 FY19 | ₹438 Cr | -₹2,310 Cr | 0 | Unqualified | F1 |
| Q2 FY19 | ₹438 Cr | -₹2,950 Cr | 0 | Unqualified | F1, F9 |
| Q3 FY19 | ₹313 Cr | -₹4,120 Cr | 58 | Material Uncertainty Note | F10, F6, F7, O1 |
| Q4 FY19 | (₹2,223 Cr) | -₹5,800 Cr | 53 | Auditor Resignation Pending | F10, F6, F7, F9 |
| Q1 FY20 | Delayed | Frozen | 64 | Statutory Default Disclosed | F10, F6, F9, O1, F7 |
| Q2 FY20 | Delayed | Frozen | 60 | CIRP Administrator Appointed | G1, G2 (Insolvency) |
* Source: Audited statutory statements, exchange filings, and registered rating releases.
64Score Composition at PeakJune 2019 (Q1 FY20) • 4 Severe Signals + F7 GNPA Creep (Medium)
▼
June 2019 (Q1 FY20) • 4 Severe Signals + F7 GNPA Creep (Medium)
| Signal | Diagnostic Description | Severity |
|---|---|---|
| F10 | Capital Buffer Decay & Liquidity Deficit (~₹4,000 Cr near-term cash gap) | Severe |
| F6 | NIM Compression (High liability cost vs non-yielding developer loans) | Severe |
| F9 | Funding Cost Divergence & CP Refinancing Freeze (₹750 Cr CP due in June) | Severe |
| O1 | Operating Leverage (Fixed financing obligations on shrinking inflows) | Severe |
| F7 | Asset Quality / GNPA Creep (Baseline 1.12% on record as of Dec 2018; subsequently 2.74% in delayed FY19 audit) | Medium |
Point-in-time evaluation discipline: As of June 5, 2019, DHFL had delayed its FY19 audited statements (the latest company-reported statutory GNPA on record was 1.12% as of Dec 31, 2018, maintaining F7 at Medium). Four signals reached Severe simultaneously: F10 (Capital Buffer Decay / Liquidity Gap), F6 (NIM Compression), F9 (Funding Cost Divergence / CP Freeze), and O1 (Operating Leverage). Compounded with F7 (Medium), interaction multipliers and sector-specific weighting produced the composite score of 64. Contemporaneously, external debt markets validated this structural collapse on June 4–5, 2019 as DHFL defaulted on ₹961 Cr of NCD interest obligations and CRISIL cut ratings to 'D'. (Delayed FY19 statutory statements released in July subsequently confirmed that GNPA had indeed surged to 2.74%, entering severe territory).
DSP MF sells ₹300 Cr CP at 11%; funding markets tighten sharply; stock plunges -53%
4 flags trigger simultaneously (F10, F6, F7, O1); initial structural alarm
CRISIL downgrades CP to A2+ on Mar 22 (~₹13,370 Cr scheduled monthly outflows through June versus ~₹2,200 Cr monthly collections); delayed FY19 results released in July subsequently confirmed the balance sheet rupture
DHFL defaults on ₹961 Cr NCD; CRISIL cuts rating to 'D' citing ~₹6,200 Cr aggregate scheduled cash outflows through July against roughly ~₹2,200 Cr monthly collections; Peak Score 64
Board superseded under Sec 45-IE; referred to NCLT for insolvency under IBC
Triple-Track Visual Divergence: Flagium Risk Score (red line) vs. NSE Stock Price (green line) vs. External Credit-Risk Milestones (dashed purple markers). Notice the precise sequence: market shock (Sep 2018) → Flagium structural warning at Score 58 (Dec 2018) → CRISIL March 22 CP cut to A2+ documenting severe liquidity mismatch (Mar 2019) → CRISIL 'D' default & Peak Score 64 (Jun 2019) → RBI Board Supersession / CIRP (Nov 2019).
Multi-Horizon Early Warning Lead-Time Matrix
Explicit separation of initial signals, official regulatory discoveries, structural stress escalation, and statutory loss milestones.
| Milestone Stage | Date / Quarter | Flagium Score | Primary Forensic Trigger | Time to CIRP Admission | Market Context |
|---|---|---|---|---|---|
1. Pre-Crisis Peak (Low Observable Stress) | Jan 2018 (Q3 FY18) | 1/100 | Offline books at Bandra Branch 001 masked diversions; CARE/Brickwork AAA (long-term) & CRISIL A1+ (CP); reported PAT ₹1,172 Cr | ~22 Months to CIRP No Signal (Concealed Off-Book Fraud) | Stock traded at pre-crisis peak of ₹678.50 with reported PAT of ₹1,172 Cr |
2. Model-Coverage Boundary: Score 0 (External Stress Emerges) | Sept 2018 (Q2 FY19) | 0/100 | DSP Mutual Fund sold ₹300 Cr CP at 11% yield. While external funding stress was emerging, DHFL had reported PAT of ₹438 Cr for the quarter ended June 30, 2018 (Q1 FY19), while latest annual FY18 GNPA was 0.96%; historical accounting-focused model configuration remained at 0 | ~14 Months to CIRP No Warning (Model-Coverage Boundary) | Stock crashed -53% to ₹290; exposes a model-coverage boundary, showing why NBFC-specific liquidity and ALM signals require dedicated integration |
3. First Material Warning (Multi-Flag Eruption) | Dec 2018 / Jan 2019 (Q3 FY19) | 58/100 | Multi-flag eruption: F10 CAR Decay (HIGH), F6 NIM Compression (HIGH), F7 GNPA Creep (MED, reported 1.12%), O1 (HIGH) | ~11 Months to CIRP EWS Lead: ~11 Mos to CIRP (~5 Mos to Default) | Stock fell to ₹225 (-67%); Cobrapost exposé reveals ₹31,000 Cr shell diversions |
4. Rating Downgrade & Acute Liquidity Stress | March 31, 2019 (Q4 FY19) | 53/100 | Score 53 evaluated point-in-time on information publicly available by March 31, 2019: CRISIL downgraded CP to A2+ on March 22 with ₹850 Cr limit cut and acute liquidity mismatch (~₹4,900 Cr liquidity against ~₹13,370 Cr scheduled monthly outflows through June versus ~₹2,200 Cr monthly collections). (Delayed FY19 statutory statements released in July 2019 subsequently confirmed the deterioration with a -₹2,223 Cr loss and 2.74% GNPA) | ~8 Months to CIRP EWS Lead: ~8 Mos to CIRP (~2 Mos to Default) | Stock dropped to ₹135.20 (-80%); credit ratings downgraded rapidly across April (A3+) and May (A4+) |
5. Peak EWS Stress & First Debt Default | June 4–5, 2019 (Q1 FY20) | 64/100 | Four severe signals (F10, F6, F9, O1) alongside F7 Asset Quality (Medium, 1.12% baseline) were simultaneously active, contributing to peak Score 64 under the NBFC sector model; contemporaneously, DHFL defaults on ₹961 Cr NCD interest obligations and CRISIL slashes ratings to 'D' citing ~₹6,200 Cr aggregate scheduled cash outflows through July against roughly ~₹2,200 Cr monthly collections | ~5 Months to CIRP Peak Stress (Default Horizon) | Stock plunged to ₹75 (-89%); contemporaneous CRISIL disclosures documented severely constrained funding and liquidity, culminating in a 'D' rating on CP |
6. Formal Resolution Event (RBI Supersession & CIRP) | Nov 20, 2019 (Q2 FY20) | 60/100 | RBI supersedes Board under Section 45-IE(1) & appoints Administrator; refers DHFL to NCLT under IBC | 0 Months (Insolvency Admission) Event Horizon (CIRP Admission) | Stock collapsed to ₹40 (-94%); NCLT admitted ₹87,000 Cr claims in CIRP (17-bank consortium exposure ₹42,871 Cr); equity extinguished to ₹0 |
Key Engine Signals & Trigger Factors
Primary forensic signals contributing to the 64/100 peak risk score
First Material Warning: Multi-Flag Eruption (Q3 FY19 / Score 58)
In Dec 2018, Flagium erupted from 0 to 58 as 4 flags triggered simultaneously: F10 CAR Decay (HIGH), F6 NIM Compression (HIGH), F7 GNPA Creep (MEDIUM, reported Gross NPA 1.12%), and O1 Operating Leverage (HIGH), signaling material structural stress approximately 11 months before formal CIRP admission (~5 months before debt default).
Rating Downgrades & Point-in-Time Liquidity Stress (Q4 FY19 / Score 53)
Score 53 was evaluated strictly on information publicly available by March 31, 2019. On March 22, 2019, CRISIL downgraded commercial paper to A2+ and slashed rated limits from ₹1,525 Cr to ₹850 Cr, exposing an acute liquidity shortfall (~₹4,900 Cr available liquidity against ~₹13,370 Cr scheduled monthly outflows through June versus ~₹2,200 Cr monthly collections) and triggering severe adjusted capital and funding pressure (F10). Separately, when DHFL released its delayed FY19 statutory statements in July 2019, those statements subsequently confirmed the severe deterioration already visible in the credit and liquidity data: an audited loss of -₹2,223 Cr, Gross NPA rising to 2.74%, and adjusted capital buffers falling into the ~10% zone relative to the NHB 12% benchmark (Adjusted capital stress estimate — Flagium calculation; not company-reported CRAR; CRISIL documented reported net worth of ₹10,750 Cr and reported CRAR of 17.74% as of Dec 31, 2018 with gearing 9.3x).
Peak EWS Stress & First Debt Default (Q1 FY20 / Score 64 / 4 Severe Signals + F7 Medium)
In June 2019, four severe signals (F10 Capital Decay, F6 NIM Compression, F9 Funding Cost Divergence, O1 Operating Leverage) alongside F7 Asset Quality (active at Medium based on the 1.12% Dec 2018 baseline) were simultaneously active, contributing to the peak Score of 64 under the NBFC sector model through interaction multipliers and sector-specific weighting. Contemporaneously, external debt markets confirmed the collapse as DHFL defaulted on ₹961 Cr of NCD interest obligations and contemporaneous CRISIL disclosures documented severely constrained funding and liquidity, culminating in a 'D' rating on CP, approximately 5 months prior to RBI board supersession. (Subsequent delayed FY19 statements in July confirmed GNPA had surged to 2.74%).
Model-Coverage Boundary: Why Score 0 Preceded Structural Stress
DHFL exposes a model-coverage boundary, not merely a data-availability boundary. By September 2018, external market evidence of funding stress was already emerging in money markets (DSP CP sale at 11%). However, the baseline model configuration evaluated in Q2 FY19 relied primarily on quarterly statutory financial inputs (where DHFL had reported PAT of ₹438 Cr for the quarter ended June 30, 2018 (Q1 FY19), while its latest audited FY18 statement had reported 0.96% Gross NPA) and therefore remained at 0. The subsequent backtest demonstrates why NBFC-specific liquidity, money-market refinancing, and ALM signals require dedicated integration rather than relying solely on reported accounting metrics: as deterioration became visible in the Q3 FY19 reported financial indicators, the historical model configuration escalated to Score 58, reaching peak stress 64 in June 2019 as external rating agencies contemporaneously downgraded debt to 'D'.
Reported Financials vs. Structural Reality
Comparison of market consensus / reported earnings against underlying balance sheet indicators.
Historical Stress Progression
Quarter-by-quarter evolution of risk score alongside major corporate and regulatory milestones.
DHFL reached its pre-crisis price peak of ₹678.50 (Jan 2018) with strong reported profitability and top-tier credit ratings (CARE and Brickwork AAA on long-term facilities, CRISIL A1+ on Commercial Paper). The engine registered a negligible score of 1, as fraudulent disbursements in Bandra Branch 001 were kept entirely off-the-books.
DHFL exposes a model-coverage boundary, not merely a data-availability boundary. By September 2018, external market evidence of funding stress was already emerging when DSP Mutual Fund sold ₹300 Cr of DHFL commercial paper at an elevated 11% yield post-IL&FS. However, the historical Flagium configuration evaluated here relied primarily on quarterly statutory financial inputs (where DHFL had reported PAT of ₹438 Cr for the quarter ended June 30, 2018 (Q1 FY19), while the latest audited annual statement on record for FY18 had reported 0.96% Gross NPA) and therefore remained at Score 0. This empirical backtest demonstrates why NBFC-specific liquidity, money-market refinancing, and ALM signals require dedicated integration rather than relying solely on reported accounting metrics. As deterioration became visible in the Q3 FY19 reported financial indicators, the historical model configuration escalated to Score 58. Later, as four severe signals (F10, F6, F9, O1) alongside F7 Asset Quality (Medium) were simultaneously active, contributing to the peak Score of 64 under the NBFC sector model in June 2019, external credit markets contemporaneously confirmed the collapse: DHFL defaulted on ₹961 Cr of NCD interest and CRISIL slashed ratings to 'D', documenting ~₹6,200 Cr aggregate scheduled cash outflows through July against roughly ~₹2,200 Cr monthly collections.
The Engine Erupted to 58 / 100: Four flags fired simultaneously — F10 CAR Decay (HIGH), F6 NIM Compression (HIGH), F7 GNPA Creep (MEDIUM, reported Gross NPA 1.12%), and O1 Operating Leverage (HIGH). On Jan 29, Cobrapost exposed ₹31,000 Cr diversions to 30+ shell companies.
Score 53 was based on information publicly available by March 31, 2019: on March 22, 2019, CRISIL downgraded CP to A2+ and cut rated limits from ₹1,525 Cr to ₹850 Cr, exposing an acute liquidity shortfall (~₹4,900 Cr liquidity against ~₹13,370 Cr scheduled monthly outflows through June versus ~₹2,200 Cr monthly collections). Flagium scored 53 under severe capital adequacy and provisioning pressure (F10). The delayed FY19 statutory statements released later in July 2019 subsequently confirmed the balance sheet rupture already visible in the credit and liquidity data: a -₹2,223 Cr loss and Gross NPA rising to 2.74%; Flagium's adjusted capital-stress estimate fell toward/below the applicable 12% benchmark after incorporating its provisioning assumptions; this was not DHFL's reported regulatory CRAR (reported at 17.74% as of Dec 31, 2018).
Four severe signals (F10 CAR Decay, F6 NIM Compression, F9 Cost of Funds, O1 Operating Leverage) alongside F7 Asset Quality (active at Medium based on 1.12% GNPA baseline) were simultaneously active, contributing to the peak Score of 64 under the NBFC sector model through interaction multipliers. Contemporaneously, external credit markets confirmed the collapse on June 4–5 as DHFL defaulted on ₹961 Cr of NCD interest obligations, and contemporaneous CRISIL disclosures documented severely constrained funding and liquidity, culminating in a 'D' rating on CP. (Subsequent delayed FY19 statutory statements released in July confirmed GNPA had surged to 2.74%).
Provisions surged to ₹6,326 Cr (PAT -₹6,705 Cr). RBI superseded DHFL's Board under Section 45-IE of the RBI Act and referred it to insolvency proceedings under the IBC at NCLT Mumbai. Equity was cancelled and delisted at ₹0 (100% wipeout).
Forensic Pillar Evaluation
Status across all 6 core analysis pillars in the sector model.
Signal: Reported standalone Gross NPA rose from 0.96% (FY17) to 2.74% (FY19), escalating to 62.97% in audited FY20 standalone results following Administrator forensic reclassifications (F7 SEVERE).
Signal: F9 Cost of Funds rose 113 bps faster than yields in FY18, leading to a ₹6,200 Cr refinancing cliff.
Signal: F6 Net Interest Margin compressed 49 bps YoY down to 1.80% in FY19 under soaring CP borrowing rates.
Signal: Claims admitted in CIRP reached ₹87,000 Cr (17-bank consortium: ₹42,871 Cr); capital and leverage deteriorated sharply as losses, provisions and subsequent impairment eroded the capital base.
Signal: Disproportionate reliance on short-term commercial paper created an acute asset-liability duration mismatch relative to housing finance peers.
Signal: Board superseded by RBI under Section 45-IE; CBI/ED uncovered ₹11,755 Cr diverted to 87 shell companies (G1/G2).
Key Trigger Signals & Disclosures
Algorithmic trigger conditions and statutory data points.
Pillar 2: NBFC Liquidity & Cost of Funds
TRIGGEREDCost of Funds Divergence (Liability Squeeze)
Trigger Condition:
In an NBFC, interest expense is cost of goods sold. When borrowing costs rise faster than asset yields, the institution suffers severe margin erosion. In FY18, DHFL's borrowing cost rose 113 bps faster than yields, signaling that money markets were already demanding a higher risk premium to fund DHFL well before the IL&FS collapse.
Pillar 1: NBFC Asset Quality
TRIGGEREDGNPA Escalation & Wholesale Reclassification
Trigger Condition:
While DHFL publicly marketed an ultra-low GNPA (<1%) retail mortgage book (reporting 1.12% at Dec 31, 2018), asset quality deteriorated as wholesale developer loans and ICDs stalled. Reported company-wide GNPA rose to 2.74% in delayed FY19 statements (driving a -₹2,223 Cr statutory net loss) and escalated to 62.97% in audited FY20 standalone statements as the Administrator formally recognized unserviced wholesale developer books.
Pillar 6: Governance & Regulatory Supersession
TRIGGEREDPromoter Siphoning & RBI Board Supersession
Trigger Condition:
Flagium's Governance layer monitors regulatory actions, auditor resignations, and agency probes. The combination of Cobrapost's shell company revelations (G1) and RBI's historic order superseding the board under Section 45-IE (G2) triggered governance warning flags as the default unfolded.
Statutory Disclosures & Regulatory Records
Every metric below is extracted directly from audited annual statements, rating rationales, statutory regulatory orders, and investigative disclosures.
| Parameter & Statutory Figure | Official Statutory Source | Significance / Forensic Impact |
|---|---|---|
Audited PAT (FY18) ₹1,172 Cr | Audited Annual Financials / CRISIL Rationale | Reported strong profit while borrowing costs diverged 113 bps faster than yields |
Audited PAT (FY17) ₹2,896 Cr | Audited Annual Filings (DHFL) | Included ₹1,969 Cr one-time gain from DHFL Pramerica stake sale (continuing PAT ₹927 Cr) |
Audited Standalone Net Loss (FY19) -₹2,223 Cr | Audited Standalone Financials (FY19 / July 2019 Filing) | Catastrophic Q4 loss post-IL&FS as credit provisions surged; reported standalone GNPA rose to 2.74%, with subsequent audited FY20 standalone results revealing 62.97% gross NPA under Administrator forensic reclassification |
Claims Admitted in CIRP ₹87,000 Cr | NCLT Mumbai Bench / CIRP Records | Total claims admitted across 17-bank consortium (₹42,871 Cr), bondholders, and fixed deposit holders |
17-Bank Consortium Exposure ₹42,871.42 Cr | CBI Chargesheet / Union Bank of India | Consortium exposure led by Union Bank of India |
Credit Rating Trajectory Rapid Multi-Notch Downgrades → Default ('D') | Rating Agency Rationales (CRISIL / CARE, 2019) | Credit ratings deteriorated rapidly across DHFL's debt instruments during 2019, culminating in default ratings. |
Commercial Paper Outstanding (June 2019) ₹850 Cr (₹750 Cr due in June) | CRISIL Rating Rationale (June 05, 2019) | Immediate refinancing cliff that triggered rating downgrade to 'CRISIL D' |
Scheduled Outflows vs Monthly Inflows ~₹6,200 Cr aggregate cash outflows through July vs ~₹2,200 Cr monthly collections | CRISIL Rating Report (June 2019) | Near-term cash deficit of ~₹4,000 Cr over June–July 2019 window |
Fabricated Retail Loan Accounts 2,60,000 Fake Accounts | CBI & ED Forensic Investigation | Fictitious PMAY accounts generated via offline 'Bandra Books' (Branch 001) |
Ghost Loan Disbursals ₹14,046 Cr | CBI Chargesheet / Grant Thornton Audit | Fictitious loans booked on an offline system with 1993-era software |
Funds Diverted to Shell Companies ₹11,755.79 Cr | CBI Special Court Filings | Transferred to 87 shell companies owned or controlled by Wadhawan brothers |
Diversion to Sunblink Real Estate ₹2,186 Cr | Enforcement Directorate (ED) PMLA Chargesheet | Unsecured lending linked to Iqbal Mirchi properties |
RBI Statutory Board Action Board Superseded | RBI Press Release 2019-2020/1230 (Nov 20, 2019) | Invoked Section 45-IE (1) & appointed Administrator; referred to NCLT under IBC |
IBC Resolution Consideration ₹34,250 Cr | Piramal Capital Acquisition / NCLT Approval | Acquired by Piramal (₹14,700 Cr upfront cash); equity extinguished to ₹0 |
Forensic Analysis Takeaway
Key Lesson for Institutional & Public Investors
"Flagium did not 'predict' the DHFL fraud. It detected the financial stress that the public record could reveal — and the backtest explicitly measures that boundary. A true point-in-time evaluation demonstrates that while 2,60,000 ghost accounts on an offline branch computer remained invisible in statutory filings, structural deterioration entered reported financial statements in Q3 FY19: Flagium erupted from 0 to 58 (F10 CAR Decay + F6 NIM Compression + F7 GNPA Creep + O1) and escalated to 64 in Q1 FY20 (4 simultaneous severe flags plus F7 Asset Quality) several quarters before RBI's historic Section 45-IE board supersession and the ₹87,000 Cr CIRP resolution. Public-data EWS identifies structural deterioration quarters in advance, but cannot be treated as a substitute for forensic investigation."
Evaluated by Flagium Quantitative EWS Backtest Engine
Monitor 3,000+ Listed Indian Equities for Early Forensic Anomalies
FlagiumAI continuously scans active NSE & BSE quarterly results, XBRL filings, and balance sheet disclosures using these deterministic rule engines.
Explore the Forensic Methodology Behind the EWS Score
Discover how Flagium's 6-pillar framework, 29 forensic flags, sector capacity models, and interaction multipliers quantify balance sheet stress and credit deterioration across Indian equities.