Hi, responding here to @satya61229 on my two cents on evaluating banks / NBFCs. Disclosure: Used to a banking regulatory professional, currently work at a debt fund. What follows is just my personal way of looking at these companies, and I am not a seasoned equity investor / asset manager.
@fabregas has mostly nailed it, just a few things I’d add / differ on:
Revenue: This is largely interest income in the case of NBFCs. Fee income should not actually be ignored since it includes loan processing fees, which can account for a significant portion of IRR, depending on the asset class i.e. loan fees should ideally be considered part of the interest income, since lenders look at it as part of the IRR. For banks, fee income can be as much as 50% of total revenue, especially for foreign banks, who engage in a lot more proprietary trading, forex, cross border transactions, trade finance, investment banking, syndication etc - large banks like HDFC also earn a lot of CASA related fees. Non-interest, non-fee income is a v important metric for banks, since it is essentially ‘free’ money, in that it doesn’t require additional leverage and improves ROA significantly. However, income recognition practices in non-interest income can get creative for banks, as they often create backend fees, deferred bullets and other such structures, while accruing it on a monthly basis. If fees are an important metric to track for a bank, and they generally are for a large bank since it improves ROA, it is also important to understand the source and composition of fees.
Cost of funds: This is straightforward and is the equivalent of COGS since cash is the raw material for lenders. It is good to correlate this with the overall leverage as a NBFC which is not leveraging its equity will show a v high spread / NIM (due to no cost of funds) which may actually be a bad thing, since they are supposed to leverage. A NBFC would typically target a leverage of 4-5x while banks would target 8-10x. The rates for wholesale borrowing and deposits is v standardized and comparable across lenders so a higher than market COF should be scrutinized.
OPEX: A key differentiator for NBFCs is their OPEX model. It is useful to compare cost to income or cost to contribution (=NIM + fees). I don’t think OPEX to AUM is appropriate as the book size is not a good comparable - for example,consider a mortgage company and a gold loan nbfc each with a 1000 cr book and a 2% OPEX to AUM. The mortgage company’s NIM would be much thinner and its cost income ratio consequently much higher compared to the gold loan NBFC, whose spread would be higher and consequent cost to NIM would show them to be relatively more efficient. There is also the matter that AUM does not capture loans disbursed in a given period, which is the better operating metric to consider especially for shorter tenor lenders, which I will cover below separately.
Provisions: The provisioning policy of the bank / NBFC needs to be understood, as these can be v misleading indicators of financial strength or weakness. There are base regulatory prescriptions, but varying asset classes need different provisioning standards, which may not always be followed by the lender in question. Provisions also tend to move erratically depending on the regulatory environment / pressure. Also, lenders are regulatorily required to provision even standard assets, so only the portion of provisions on account of stressed assets should be considered (usually disclosed). Their main impact is on the capital adequacy of banks, by eating away at profits, and limiting future growth and necessitating future capital raises. Examining the book directly is preferred.
Liabilities: Lenders leverage, and it is useful to examine the structure of their borrowings to ascertain financial strength and also what other lenders think of their credit risk. Besides COF, the maturity and tenor of borrowings relative to the assets created by the lender i.e. the ALM mismatch is important to consider. For example, is the lender raising 3-5 month money and deploying 8-10 month loans? This will clearly create liquidity pressures on the lender. In general, longer the liabilities, the better - most lenders disclose an entire section on this. CASA, though it is on-demand in nature, is considered long term stable funding, and is the best form of leverage as it is the cheapest. It is also useful to examine the incremental liability composition and cost to see if the COF is coming down at the margin, or the lender is getting more favourable structures, which is a positive.
Assets: Straightforward, it consists of loans and advances (sometimes called trade receivables under financing), any instruments discounted, and all investments in bonds / credit instruments. Banks will typically also have a large securities / government bonds book for the purposes of maintaining statutory liquidity ratios as well as prop trading - the appropriateness of the mix between credit assets and other securities depends on the business model of the bank in question i.e. an investment bank will look different from a corporate bank, which will be different from a retail bank / nbfc. For most NBFCs though, this should largely be loans with small holdings of cash / liquid securities. Large holdings of cash or liquid securities may indicate inability to deploy funds, which is a growth concern. Analyzing maturity profile of assets, which is typically disclosed by lenders, is useful. For banks / NBFCs into corporate lending, concentration risk needs to be considered - they will usually disclose their exposures to key sectors and clients. Geographic concentration is also important, particularly for MFI.
Disbursals are a key metric to track, which is not captured on the balance sheet - shorter term lenders like working capital financiers and MFIs are able to cycle a lot more capital in a given period, and make that much more spread. This is usually disclosed in presentations. Disbursals are also where loan fees are generated, which in some categories like CV finance, used car finance comprise a large portion of income, since the loans are cycled fairly rapidly. For shorter tenor loans, this is the better metric to track. Many NBFCs also securitize and upsell loans to banks, so the book does not reveal the true extent of their origination capacity, and so disbursal / AUM (=loans on book/ loans securitized) are good to consider. Higher level of securitization would mean origination is higher than that shown on the book, and the NBFC is able to generate liquidity to fund new loans, creating a fly wheel effect, which is good.
NPA and stressed assets: This is the single most important thing to analyze for a lender. Loans and credit assets are classified as under / non-performing based on delays on payment or days past due (DPD). As per RBI, a NPA is a loan that has at least one payment overdue by 90 days or more. It also provides guidance on ‘special mention accounts’ (SMA) which may not yet be NPA, but are showing signs of stress. The appropriateness of 90 day DPD depends on the asset class - for home loans, it may be fine, but for microfinance or CV finance, anything more than 10-20 days is a big risk, and lenders would internally consider them non-performing even if they don’t report them as such. They may provision aggressively for such assets (for example, many MFI, CV finance guys take 50% provision within the first week of delays and fully provision and write off an asset with 90+ DPD). This will hurt profits but is actually a good thing from a prudential point of view, and helps maintain balance sheet strength. Likewise, other lenders may underprovision or underreport NPAs (such as recent revisions uncovered by RBI audit of Yes, ICICI). It is good to compare asset classification and provisioning norms across different lenders of the same type - i.e. Citibank vs Yes, Axis vs HDFC, SKS vs Equitas etc. Thumb rule being conservative positions are better for financial strength. Check the notes to accounts for the NPA / provision addition, recovery and net position schedule to get better insight into how many assets were stressed in the period, and how many were recovered. A good ratio to consider is Additions to NPA / (Loans disbursed in the period), which gives you a sense of incremental book quality (you can also add opening loan book in case loans are of longer tenor, for shorter tenor, disbursals are better). Re-structured assets is another important thing to note - this is usually done to a non-performing, SMA loan to improve viability of recovery, but is also a way to mask weakness. For analysis, it is best to consider the total of NPA + SMA + restructured assets - you will see many banks with double the restructured assets than the NPA. The large portion of restructured assets will fail eventually, and so this pool is basically future potential pool of NPA.
Security cover of loans is also important to consider - for example, a gold loan lender with 5% NPA is actually fairly safe since the loan is collateralized by a liquid asset under their control, as opposed to maybe a loan against property, which is much harder to foreclose and liquidate - the former can afford to provision less, while the latter cannot. The final thing to note is also the level of write offs and distressed asset sales, since these erase assets from the book for future periods. The level of stress is typically the hardest part to ascertain so careful analysis of management disclosures and peer data needs to be done to guess at true level of stress, and appropriateness of the same. Once a picture of this is formed, you can form a view of future provisioning and recovery / write off. V conservatively, you could opt to write off the entire estimated amount, and examine the impact on capital adequacy (useful for NPAs due to one offs like demonetization / farm loan waiver, but not for where lender’s origination is fundamentally poor - PSU banks).
Soft aspects: Banks and NBFCs are highly regulated, so the track record of management and investors and their regulatory track record is important to understand. Many NBFCs in real estate, gold loans and even MFIs face sharp regulatory changes due to evolving view of RBI on consumer protection and systemic risk, so one needs to be upto date on these changes, at least at a broad level. A company that has succesfully survived multiple cycles of regulatory change is that much stronger from an investment perspective. Bank / NBFC audits are complex, and while all lender auditors are empanelled by RBI, it is preferred that it is a Big 4 auditor - KPMG has I think a good reputation as the best in financial services audit in India.
Key metrics and valuation: RoA / RoTA, GNPA + SMA + restructured assets + write off + assets sold to total loan book, operating cost to income / contribution (=NIM+fees) ratio, NIM, contribution / AUM, growth in disbursals, growth in AUM, CASA, COF. I don’t use RoE, as it varies widely according to leverage levels and asset composition of the lender, and also regulatory requirements, and I prefer EV / Total Assets to Price to Book, since the asset base is what is essentially being valued. EV / Total Assets will typically be less than 1, significantly less, but the closer to 1 the more highly valued, and if its above 1 indicates high premium / growth prospects.
For most banks and NBFCs, the greatest component of income is interest, which is usually recognized on cash basis, so you can consider PAT-statutory reserves = FCFF for lenders as a rough measure – provisions, while non-cash, should be deducted as a penal measure. The entire pool of stressed assets may be deducted from PAT as a penal measure too. NBFCs and banks have statutory reserve requirements they need to provide for before distributions, and these should be deducted as well. Simple DCF then works fine imo, after PAT = ROA X AUM, and appropriate deductions. Cash flow statement is generally meaningless due to cash in cash out nature of operations. EV / Total Assets is also useful. Please note that one should not deduct net debt from EV determined by DCF as leverage is operating input for lenders, which also makes EBIDTA, EBIT meaningless.
Sorry, this got a bit long. Hope it helps, look forward to feedback, criticisms and comments.