{
  "version": "https://jsonfeed.org/version/1.1",
  "title": "LiquiLens daily institution-risk articles",
  "home_page_url": "https://liquilens.in/articles/",
  "feed_url": "https://liquilens.in/articles/feed.json",
  "description": "Evidence-led bank and lender analysis, with historical failure replays when the current board is quiet.",
  "authors": [
    {
      "name": "LiquiLens",
      "url": "https://liquilens.in/"
    }
  ],
  "items": [
    {
      "id": "liquilens:article:2026-08-15-esaf-s-warning-is-a-dated-filing-signal-meeting-a-live-market-price-signal-n",
      "url": "https://liquilens.in/articles/2026-08-15-esaf-s-warning-is-a-dated-filing-signal-meeting-a-live-market-price-signal-n/",
      "title": "ESAF’s warning is a dated filing signal meeting a live market-price signal—not evidence of a funding run",
      "summary": "ESAF Small Finance Bank is a yellow-tier monitoring case because an elevated GNPA reading sits beside a market distance-to-default below the screen threshold. The filing, price and model signals should not be treated as the same thing.",
      "content_text": "## The finding\n\nESAF Small Finance Bank is the clearest case in this eligible comparison set where a dated balance-sheet warning and a current market-price warning point in the same direction. That is not the same as saying it will fail first, that funding stress has begun, or that the market has measured a default probability. The case is narrower: ESAF’s filing shows GNPA of 8.5%, while its market distance-to-default is 1.927, below the yellow-screen trigger of 2.\n\nThe timing is the discipline. ESAF’s filing record is as of 2025-09-30, its market layer is as of 2026-08-11, and this analysis is dated 2026-08-15. A filing-period signal describes reported conditions at a past reporting date. A market-price signal is a current repricing input. A model derivation transforms disclosed and market inputs into a monitoring measure. None can simply be substituted for another.\n\nThe screen assigns ESAF a score of 66.0, a BBB grade and a yellow tier. These are corpus-monitoring outputs, not a credit rating or a failure forecast. The [failure-radar board](https://api.liquilens.in/api/failure-radar/board) places Utkarsh Small Finance Bank in orange, although that row is older; Belstar Microfinance Limited and IndusInd Bank Ltd are yellow. ESAF is not therefore established as the universally weakest institution. It is the most analytically interesting fresh-enough case for this specific collision of signals.\n\n## The mechanism\n\nThe proposed pressure chain begins with impaired assets, not with an observed liability flight. GNPA of 8.5% is the reported asset-quality fact in the dossier. Impaired assets can demand earnings, provisions or capital attention. That may reduce the room available to absorb a further shock. If equity investors reprice that capacity, the market signal can deteriorate before a disclosed deposit outflow appears.\n\nThe second link is not itself a funding result. ESAF’s naive Bharath-Shumway distance-to-default is 1.927. Its associated Merton-form one-year figure is 2.696%. The dossier is explicit: this is a market repricing and ranking signal, not a measure calibrated to Indian failure frequencies. It uses market capitalisation of Rs 2,056 cr, deposits of Rs 23,276 cr as the barrier, realized equity volatility of 42.9%, and prior-one-year return of +27.7%.\n\nThe barrier is imperfect by construction. Other liabilities are not in the supplied schema, so the barrier is slightly understated. That limitation cuts against turning a below-threshold distance-to-default into a claim about imminent insolvency. At most, it says the equity-market configuration merits scrutiny alongside the asset-quality reading.\n\n## What the filings say\n\nThe filing-based case is meaningful but old. GNPA of 8.5% is both the stated score basis and the stated basis for the hazard model. ESAF’s reported 12-month monitoring probability is 0.42%, versus 0.34% in FY25Q2. The change is 0.08 percentage points. The probability is a model derivation, however, not a filing line and not a market price.\n\nThe yellow rule can be met by deterioration, a level at or above 0.25%, a funding flag, a forensic indicator, or market distance-to-default below 2. ESAF’s listed fired signal is `market_dd_below_2`. The GNPA figure supplies economic context, but the live trigger is market-based.\n\nThere is no PCA breach finding. Small finance banks are outside the 2021 bank-PCA framework in this dossier. CRAR is not assessed, and the stated reference point is the 15% licensing CRAR floor. Absence of a PCA finding is not evidence of capital comfort; an unassessed measure is not a pass.\n\nThe filing clock also limits the inference. Where an explicit publication time is unavailable, the historical construction uses period-end plus 60 days. It is construction-PIT, not a complete public-availability and revision archive. Overwritten amendments cannot be fully reconstructed, and the dossier says lead times are optimistic. Read the [validation record](https://api.liquilens.in/api/failure-radar/validation) as a limits record as well as a performance record.\n\n## What the market says\n\nThe market layer is newer than the filing layer and is marked not stale. It relies on Yahoo Finance daily auto-adjusted closes. ESAF’s 1.927 distance-to-default fired its sole listed signal. The [market evidence index](https://api.liquilens.in/api/evidence/markets) sets out the source boundary.\n\nYet the market does not offer a one-way distress narrative. Prior-one-year return was +27.7%. The conformal alarm is also not an independent confirmation: its gate is closed, its wiring is suspended pending revalidation, and it has no score or tier authority. The published tier rests on published components and rules, not on the suspended alarm.\n\n## The strongest counter-case\n\nThe counter-case can defeat the thesis. ESAF may be a stale-filing, model-sensitive yellow alert rather than an institution moving into financial stress. Funding is labelled stable. The recorded funding basis is worst deposit quarter-on-quarter growth of +5.3%, not a withdrawal. No deposit-run flag fired. No forensic indicator fired.\n\nThe market model is naive, uncalibrated to Indian failure frequencies, and built on a barrier that slightly understates liabilities. A distance-to-default below 2 may be useful for ranking, while still failing to estimate loss, regulatory action, funding pressure or failure. The positive equity return further resists a simplistic account of persistent market rejection.\n\nUtkarsh’s orange tier is another constraint. ESAF is not the highest-tiered board concern. The strongest alternative interpretation is that a dated GNPA snapshot is being combined with a modelled market signal while deposits and unobserved buffers remain resilient. That explanation remains fully compatible with the supplied evidence.\n\n## The evidence that is dark\n\nThe missing evidence is material. Wholesale reliance, certificate-of-deposit strain and LCR headroom are dark funding lenses. CRAR is unassessed. The dossier does not quantify current provisioning, capital absorption, liability mix or liquidity-buffer headroom. Missing data are not evidence of calm.\n\nThe universe is incomplete by design. Institutions without vetted dossiers are absent, and 21 stale cases were excluded. The board has 0 red names, 1 orange, 3 yellow and 15 green, but those are eligible dossier counts, not a census of institutional risk.\n\nSeiche reports a guarded system read: 45.2, STRAIN, with modelled or slow-moving structure leading while current market plumbing has not broadly confirmed it. That is system context, not confirmation of ESAF. See the [Seiche overview](https://api.seiche.info/api/overview).\n\n## What would change the call\n\nThe call would weaken if a newer vetted filing showed improving asset quality alongside demonstrably resilient deposits, liquidity headroom and capital measures. It would weaken further if market distance-to-default moved above 2 and remained there without a funding or forensic flag.\n\nIt would strengthen if a newer filing preserved or worsened the GNPA pressure and disclosed strain in a currently dark funding lens, or if market distance-to-default deteriorated further. A disclosed capital measure would matter because CRAR is presently unassessed, not because this dossier establishes a capital shortfall. This is a falsifiable monitoring call, not investment advice.\n\n## Follow the pressure chain\n\nWatch the sequence rather than a single metric: impaired assets; capacity to absorb their cost; equity-market repricing; funding behaviour; disclosed liquidity headroom; and institutional response. The key test is whether a later filing links the 8.5% GNPA reading to currently unlit funding or capital channels.\n\nLiquiLens covers **institution and lender balance-sheet risk**. Seiche covers **system dollar-funding capacity**. Undertow covers **market liquidity and executable exit capacity**. These are distinct research boundaries, not mutually confirming products. The [Undertow board](https://api.seiche.info/undertow/board.json) contains several partial segment reads; partial coverage should not be mistaken for a broad exit-stress conclusion.\n\n## Sources, method, and limits\n\nThis article uses the supplied [failure-radar board](https://api.liquilens.in/api/failure-radar/board), [validation endpoint](https://api.liquilens.in/api/failure-radar/validation), [market evidence endpoint](https://api.liquilens.in/api/evidence/markets), [Seiche overview](https://api.seiche.info/api/overview), and [Undertow board](https://api.seiche.info/undertow/board.json). Further background is available through [LiquiLens research](https://liquilens.in/research/) and the [replay archive](https://liquilens.in/replay/).\n\nThe historical evidence is PERIOD_END_PROXY_CONSTRUCTION_PIT. It is not validated-backtest eligible, not real-money eligible, and lacks a bitemporal input contract. The hazard temporal diagnostic is 0.645 against a 0.65 threshold, so it did not pass its stated promotion gate. Historical replay can illustrate mechanisms, not validate a forward call on ESAF. The conformal tier wiring remains suspended. Missing evidence remains missing; it is not evidence of calm.\n",
      "date_published": "2026-08-15T11:12:50Z",
      "tags": [
        "current_analysis",
        "institution risk"
      ],
      "_liquidity_lab": {
        "schema": "liquidity-lab.editorial-item.v1",
        "product": "liquilens",
        "article_type": "current_analysis",
        "evidence_as_of": "2026-08-15",
        "word_count": 1339,
        "evidence_fingerprint": "25b700c1a13149b9fa4d6fe3521f0b12afcdc29b6e5ae37dd0ce61c302c87082",
        "generation_mode": "model_assisted",
        "quality_gate": {
          "status": "PASS",
          "checks": [
            "depth",
            "structure",
            "lede",
            "countercase",
            "sources",
            "funnel",
            "numeric_grounding",
            "link_grounding",
            "institution_boundary",
            "evidence_status"
          ]
        },
        "authority": {
          "factual_authority": "published_article_only",
          "training_allowed": false
        }
      }
    }
  ]
}