CAM in Banking: What a Credit Appraisal Memorandum Should Actually Contain

CAM stands for Credit Appraisal Memorandum: the document a credit analyst writes to recommend a lending decision to the sanctioning authority. It sets out who the borrower is, what they are asking for, what the analysis found, what could go wrong, and why the analyst believes the exposure is sound. Banks also call it a credit appraisal note, a credit proposal or a sanction note.

That is the definition. The more useful question is why most CAMs are worse than they need to be, and the answer is structural: they are written as compilations rather than as arguments.

CAM, CMA and the other CAM

Three terms get confused in Indian credit, and searching for one usually returns the others.

The rest of this post is about the first: the memorandum.

A CAM is an argument, not a compilation

The person reading your CAM has fifteen other files. They are looking for four things:

1. What is being proposed?
2. What could go wrong?
3. Why do we believe it will not?
4. What are we doing about the parts that might?

Everything else is appendix. A twenty-page memorandum that reproduces every extracted figure and attaches a recommendation at the end makes the committee do the analyst's work, and committees that have to do the work either defer the file or approve it on the relationship manager's word, neither of which is the intended outcome.

Front-load. The first page should carry the proposal, the recommendation, the principal risks and the conditions. Everything after it is evidence for a reader who wants to check.

CAM format: the section structure

A workable order, with what each section is actually for.

1. Proposal summary. Facility type, amount, tenor, pricing, security, purpose. Five lines. A reader should know what they are deciding before they know anything else.

2. Borrower and promoter background. Constitution, vintage, ownership, promoter experience, group structure, and the results of bureau and defaulter list checks on the entity and its promoters. Group structure matters more than it is usually given space: it determines whether related-party flows in the banking are expected or anomalous.

3. Business and industry. What the business actually does, who its customers are, what drives its cash cycle. Brief. Industry commentary copied from a sector report adds length and no information.

4. Financial assessment. Turnover, profitability, leverage, coverage, over three years, with the ratios the institution's credit policy specifies. For working capital proposals the projections usually arrive as CMA data. State the basis of every figure: audited, provisional or management-certified. A provisional figure presented without that label is a common and quiet misrepresentation in Indian credit files, and banks have long treated it seriously. One public sector bank instructed its branches in 2013 that where a borrower's provisional and audited statements differed by more than 20%, that borrower's provisional figures were not to be relied on from the following year.

5. Banking conduct assessment. Covered below. This is the section that deserves the most attention and usually gets the least.

6. Security and collateral. Nature, valuation, valuation date, who valued it, and the realisable rather than market figure. RBI expects banks to use professionally qualified, independent valuers under a board-approved valuation policy, with two independent valuations for properties worth ₹50 crore or more.

7. Risk assessment and mitigants. Covered below.

8. Policy deviations. Stated plainly, in their own section, near the front. Deviations buried on page fourteen are how approvals get reopened.

9. Recommendation and conditions. Precedent conditions, subsequent conditions, covenants and monitoring triggers, each one specific enough to be tested.

One rule sits underneath all nine sections. RBI has long required banks to make their own independent credit appraisal rather than rely on reports prepared by outside consultants, especially consultants retained by the borrower. A CAM that paraphrases a consultant's project report fails that test however well it is formatted.

The ratios, and where the benchmarks come from

There is no single regulatory benchmark for most CAM ratios. RBI withdrew the Tandon-era 1.33 minimum current ratio along with the mandatory MPBF in 1997, and as its own circulars put it, banks are now free to decide on the minimum current ratio, so the thresholds a CAM tests against come from the institution's own board-approved credit policy. Published examples show how much they vary.

An annual DSCR that clears the threshold can still hide months that do not. The case for computing it monthly from banking data is set out in our post on DSCR from banking data.

The banking section is the weakest and should be the strongest

In most CAMs the banking conduct section reads approximately: average bank balance ₹X lakh, credit summations ₹Y crore, no major bounces observed.

Three lines, describing the only document in the entire file the borrower did not write.

Financials are prepared. Projections are argued. Declarations are asserted. The bank statement is a record produced by a third party of what actually happened to the money, and in the MSME segment it is frequently more reliable than the audited accounts sitting above it in the same file.

A banking section that earns its place covers the following, each as a monthly series rather than a period total.

Balances. Average, highest and lowest end-of-day balance by month, and the count of zero-balance days. The lowest-balance trend matters more than the average: a floor dropping while the mean holds means the borrower is running closer to empty between inflows.

Throughput against declared turnover. Monthly credit value and transaction count, with non-revenue credits removed, compared against what the borrower declared and filed. The variance between GST turnover and banked credits needs normalising before it means anything, and a naive gross comparison frequently points in the wrong direction.

Facility utilisation. For borrowers with existing limits: days above 90%, overdrawn days, lowest drawn balance as a share of limit, and whether the facility swings. A cash credit account permanently at the ceiling has stopped being working capital and become a term loan with no amortisation schedule.

Bounce analysis. Not a count. Indian banks name cheque returns from their own side of the clearing, and the two directions mean different things. An inward return is the borrower's own cheque or debit dishonoured: a conduct signal, to be split by reason code into technical failures, genuine shortfalls and deliberate stops. An outward return is a cheque the borrower deposited that came back unpaid: their customer failed to pay, which is a receivables signal rather than a conduct one. Narrations vary between banks, so classify by direction rather than by the label alone.

EMI conduct. Missed and delayed separately, by counterparty, month-wise. Delay days lengthening is a signal that fires before any failure and is routinely discarded by systems that record whether an EMI was paid rather than when.

Obligations found in banking but absent from the bureau. Recently disbursed loans, informal and small-ticket lenders, BNPL, revolving card balances. The bureau establishes what is owed on record; the account establishes what is actually being debited. Our note on obligation mapping covers the method.

Related-party and self-transfer net flow. Gross movement is noise. Net direction tells you whether the promoter is funding the business or extracting from it, and a change in direction is one of the stronger pre-default signals in owner-managed lending.

Counterparty concentration. Top buyers and suppliers by share of throughput, with related parties excluded from the ranking. A group entity sitting at the top of a concentration table reports diversification that does not exist.

Every item above is derived from the same statements already in the file. The difference between three lines and a real section is not more data collection. It is whether the analysis was performed.

What the difference looks like

An illustrative MSME file, the same twelve months of statements written up two ways.

The left column describes an account. The right column describes a borrower whose liquidity is tightening, whose largest receivable is slowing, and whose promoter is quietly covering the gap. Only one of them gives the committee something to decide on.

Risks and mitigants: write the real ones

Most risk sections are written defensively: the analyst lists the risks they expect the committee to raise, each paired with a reassuring sentence. The result is ten generic risks and ten generic mitigants, and a committee that learns nothing.

A CAM that names the two or three risks the analyst actually believes in, and proposes a structural response to each, gets approved faster than one that lists everything. Structural responses are things like an escrow on a concentrated receivable, a repayment schedule weighted to the borrower's observed cash cycle rather than a flat EMI, a covenant on related-party outflows with a defined test, or a monitoring trigger tied to a specific banking metric.

"Promoter has agreed to infuse funds if required" is not a mitigant. "Facility sized to the observed peak deficit of ₹42 lakh against a calculated requirement of ₹60 lakh, with drawing power tested monthly" is.

The same applies to deviations. A deviation stated plainly on page one, with the compensating factor named, is a decision the committee can make. The same deviation surfacing during review is a credibility problem for every other number in the document.

Conditions that can actually be tested

A condition nobody can verify is a condition nobody will enforce.

Compare: "Borrower to maintain satisfactory account conduct" against "Average monthly credit throughput in the operating account not to fall below ₹35 lakh for two consecutive months, tested on the 10th of each month."

The second can be monitored automatically, triggers a defined action, and tells the borrower exactly what is expected. The first is decorative.

Monitoring triggers written this way turn the CAM into a live document rather than an artefact filed at sanction. The metrics worth attaching triggers to are the same ones that appear earliest in a deterioration sequence (facility utilisation, credit throughput against baseline, related-party inflow share, and EMI delay days), all of which precede DPD by months, as set out in our post on early warning signals. They also feed the bank's own early warning system: RBI's fraud risk management directions, reissued on 31 July 2026, expect early warning indicators drawn from the transactional data of accounts, with alerts examined preferably within 30 days.

The recommendation has an audience beyond the committee, too. Banks must give the main reasons for rejecting a loan application in writing, and retail and MSME term loans sanctioned from 1 October 2024 carry a Key Facts Statement setting out the all-in cost. A CAM with explicit reasoning and testable conditions makes both straightforward to produce.

How a CAM differs by segment

The structure holds across segments. The weight shifts.

The segment differences in reading the statement itself run deeper than the CAM format, and a template built for one segment produces systematic gaps in the others.

Why CAM preparation takes as long as it does

The common complaint is that analysts spend most of their time assembling rather than assessing: pulling figures out of statements, reclassifying misclassified transactions, consolidating accounts by hand, and reconciling totals before any judgment can begin.

That assembly burden is largely a function of how much correction the upstream output requires. Per-case reclassification of 30 to 45 minutes on a multi-account business file is common, and it is driven by misclassification volume rather than by the analysis itself. We measured that error rate directly across 200,001 transactions.

The practical consequence for CAM quality is that the banking section is thin not because analysts think it unimportant, but because by the time the figures are trustworthy there is no time left to interpret them.

Frequently asked questions

What is the full form of CAM in banking? Credit Appraisal Memorandum: the internal document in which a credit analyst sets out the proposal, analysis, risks and recommendation for a sanctioning authority to decide on. Banks also call it a credit appraisal note, credit proposal or sanction note. Public sector banks also use CAM for their digital Credit Assessment Model for MSME loans, which is a different thing.

What is the difference between CAM and CMA in banking? A CAM is the credit analyst's written appraisal and recommendation. CMA data, the Credit Monitoring Arrangement format, is a standard set of historical and projected financial statements and working capital calculations prepared by or for the borrower. CMA data is an input to the financial section of a business CAM.

What are the main sections of a credit appraisal memorandum? Proposal summary, borrower and promoter background, business and industry, financial assessment, banking conduct assessment, security and collateral, risk assessment and mitigants, policy deviations, and recommendation with conditions.

What should the banking section of a CAM include? Monthly end-of-day balances and zero-balance days, credit throughput against declared and filed turnover, facility utilisation and swing behaviour, bounce analysis separating the borrower's own dishonours (inward returns) from returned deposits (outward returns) and technical from non-technical failures, EMI conduct showing both missed and delayed payments, obligations not visible on the bureau, net related-party flow, and counterparty concentration.

Which ratios should a CAM include? Usually DSCR, current ratio, TOL/TNW, interest coverage and fixed asset coverage, tested against the thresholds in the institution's own board-approved credit policy. RBI does not set a single benchmark for most of them, and withdrew the 1.33 minimum current ratio in 1997.

Who prepares and who approves a CAM? A credit analyst or credit manager prepares it, usually with input from the relationship or business team. Approval sits with the sanctioning authority appropriate to the exposure size under the institution's delegation matrix.

How can CAM preparation time be reduced? Most of the time goes to assembling and correcting figures rather than assessing them. Reducing upstream misclassification and consolidating multi-account cases automatically removes the manual reclassification pass, which is where the bulk of per-case analyst time is spent.

---

Fiscus produces output structured for direct CAM use: income, obligations, FOIR, conduct, counterparty and related-party analysis, consolidated across accounts and presented in analyst-readable form. Book a parallel evaluation on cases your team has already written up.