For credit and risk teams at banks. The rating agencies publish which banks lend to a rated borrower. TatvaRatings indexes that relationship from the other end, so the query can start with the bank.
All figures as at 30 August 2026.
How do you find out which companies a bank lends to?
Rating agencies publish a lender-wise annexure naming the banks behind each rated facility. TatvaRatings parses those annexures out of press releases from seven SEBI-registered agencies and indexes them from the lender’s side, so a query can start with the bank and return the rated borrowers that name it. Figures are floors, as at 30 August 2026.
The question that has no home
A credit analyst can answer “what is this borrower rated, and who lends to it?” in an afternoon. The rating press release prints it. Pull the PDF, find the annexure, read the lender names.
The question that has no home is the other direction:
Which rated borrowers name this bank on a facility — and what do those borrowers look like in aggregate, by sector, by rating band, by state, by facility type?
Nothing in the market is organised to answer it. Rating agency portals are entity-search archives: you type a company name, you get that company’s history, one agency at a time. Commercial bureau reports are sold one entity at a time and start from a borrower you already name. Portfolio analytics tools operate on a book you upload — which presupposes you already know who your borrowers are, and says nothing about anyone else’s. The difference between that kind of tool and a lender-side index over published disclosure is set out in the Crisil Quantix comparison.
The rows to answer the lender-side question have been published all along. They sit inside PDFs, in an annexure table, one borrower per document, across seven agencies with seven layouts.
What TatvaRatings does with them
Every ingested press release is parsed into the same schema, and the annexure is parsed row by row — which lender, which facility, how much, on what date, under which rating. Lender identities are then resolved: “State Bank of India”, “SBI”, branch-suffixed variants and the legacy names of merged banks all reconcile to one lender.
The result, as at 30 August 2026:
| Measure | Figure |
|---|---|
| Lender-exposure rows | 179,857 |
| Distinct lenders named | 1,501 |
| Rating actions parsed | 70,337 |
| Borrowers | 45,528 |
| Rated facilities | 194,471 |
| Agencies | 7 |
One worked figure, for calibration: State Bank of India is named in 5,476 borrower annexures in the corpus, as at 30 August 2026. That is a count of ingested annexures naming SBI. It is a floor, not a total. Why →
Without lender-name resolution, a reverse query returns a fraction of what it should. That step is the difference between a search box and an index.
What a credit team does with it
See where a prospective borrower already banks
A name comes in for a working capital line. Before the first meeting, you can see which lenders are named on that borrower’s rated facilities, what each facility type is, what each was sanctioned at, and how the set has changed across successive rating actions. Where a lender that used to appear no longer does, that is a question worth asking the borrower directly.
Read a competitor’s rated book in aggregate
Pick a bank. Read the sector composition, the rating band distribution and the facility mix of the rated borrowers that name it. This is a cross-sectional read that no single-borrower report can produce, and it is built entirely from disclosures that bank’s own borrowers consented to publish.
Size your own position against the peer set
Take a sector and a rating band. See which lenders are present in it and at what published sanctioned amounts. A concentration you thought was distinctive may be crowded; one you avoided may be thin.
Look at a consortium before you join it
A rated borrower’s annexure names the other lenders on the facility. Read who else is in, at what facility types, and what the rating history has done since each of them appeared.
Read rating migration across a named borrower set
Define a set of borrowers — your own names, a sector, a lender’s book — and read rating actions across seven agencies in one place, rather than checking seven portals. Migration here means the published rating transitions the agencies have already recorded; it is a historical read, not an alerting layer.
An illustrative reading
Query: lender = a public sector bank. Three rows from that bank’s rated borrower book, sorted by published amount.
| Borrower | Rating | Facility | Amount | Action date |
|---|---|---|---|---|
| Meridian Textiles Ltd | BBB+ | Cash credit | ₹85 cr | 12 Mar 2026 |
| Kaveri Agro Processors Pvt Ltd | BB | Term loan | ₹32 cr | 04 Feb 2026 |
| Northline Logistics Ltd | A− | Bank guarantee | ₹18 cr | 21 Jan 2026 |
Read down a column for a book. Read across a row for a facility. Read the action dates for how old each fact is.
What you are actually reading
This is the part a credit team will test in the first hour, so it is stated here rather than found later.
Each row is a named lender on a rated facility at a point in time. Not a confirmed outstanding, not a drawn balance, not a current position. The annexure gives a sanctioned amount as at the rating action date.
Every count is a floor. Two filters sit between the number and reality. Only 46.2% of rating actions publish a lender-wise annexure at all — 32,498 of 70,337, as at 30 August 2026 — because disclosure of lender details depends on the issuer’s consent — where consent is absent, no annexure publishes and there is no row to hold. And the rated book is a minority of any bank’s borrowers, because most lending relationships never involve a public credit rating. The true position is larger, by a margin that varies by bank and by sector and that is not estimated here.
Every record is dated, and most are not recent. A cooperating issuer’s rating is reviewed periodically rather than continuously, broadly once a year, so a typical annexure record describes a position a year or more old. A facility repaid, refinanced or taken over the day after a rating action still reads as sanctioned until the next review publishes. A new facility sanctioned between reviews does not appear until then.
Proposed and unallocated limits are flagged. Releases routinely carry proposed or unallocated facilities, sometimes with no named lender. A lender total that silently includes them overstates the position; the flag exists so it does not have to.
Nothing here is a credit opinion of ours. No score, no rank. The ratings held are the agencies’ opinions, attributed to the agency that issued them.
What it is not
Not exposure monitoring. There is no continuous observation of any lender’s book and no mechanism by which a change is detected before the next rating action publishes. The review cycle rules that out.
Not a system of record. It should not be relied on as a control in any process requiring a current or complete exposure position, nor for regulatory reporting or capital computation.
Not a substitute for verification on a named borrower. Use it to see the shape of the market, find counterparties, size a sector and generate questions. Then confirm a specific name through your own channels before you act on it.
Not a supervisory or bureau dataset. CRILC is an RBI supervisory return with its own reporting and access rules. Bureau credit information is available to specified users under the Credit Information Companies (Regulation) Act, 2005. TatvaRatings is built from public rating disclosures, carries no regulatory status, and is a different thing on a different basis. What CRILC is →
Getting to a first answer
Tell us the first lender you would look up, and the question you would ask of that book. We will come back with what the current corpus can and cannot answer for it — including where it cannot, which is a useful reply either way.
Common questions
Can I see a bank's whole lending book?
No, and nothing built from public disclosure can show one. Only 46.2% of rating actions publish a lender-wise annexure (as at 30 August 2026), and the rated book is a minority of any bank's borrowers. Every count here is a floor — "at least this much, from public record" — never a total. Data limits specifies both filters.
How current is the lender data?
Each row reflects the sanctioned position as at its rating action date — not today, and not an outstanding balance. A cooperating issuer's rating is reviewed broadly once a year, so a typical record is a year or more old, and every row carries its date so you can read the age alongside the figure.
Where does the data come from?
Credit rating press releases published by seven SEBI-registered rating agencies on their own websites — public, regulatorily required disclosures. The lender-wise annexure in each release is parsed row by row, and every record keeps its source agency, document and date. Methodology walks the pipeline; Sources links each archive.
Related
- Syndication and advisory — the same index, read before a mandate
- NBFC portfolio review — co-lender and consortium views
- Coverage — the seven agencies and the measured counts
- Data limits — the full specification
- Methodology — how the annexure becomes a row