Credit Quality: What a Debt Fund Return Cannot Tell You
Credit quality is how likely the borrowers behind a debt scheme's holdings are to pay what they promised. The scheme's return, taken alone, carries none of it: one figure can come from safe borrowers in a quiet year or from weak borrowers paying extra to be carried. So the reading goes to the holdings, and a rating is read as a dated opinion.
A debt scheme lends. Lending is the whole of what it does: it hands money to borrowers who promise to pay it back with something on top, and it collects. So there are only two ways the arrangement can go wrong. The money can come back late or short, or the price of the promise can move while the scheme is still holding it. The first of those has a name: credit riskThe chance that a borrower pays late, pays less than promised, or does not pay at all.. The second, how a bond's price moves when interest rates move, is a different exposure and is covered in this platform's fixed income material.
Here is the difficulty. A holder of a debt scheme sees a value per unit each day and a return figure for the year. Credit risk asks one question: who the money was lent to, and whether they can pay it back. Neither of those numbers contains any answer to it. Worse than that, and this is the part that catches careful people: in a year where every borrower paid on time, the return of a scheme that lent carefully and the return of a scheme that lent to borrowers on the edge look like the same kind of number, and the second is usually the larger of the two.
There is one more thing to settle before any mechanism starts, and it colours everything after it. The record this material works from covers Girnar Asset Management Limited and two of its schemes, the Girnar Large Cap Equity Fund and the Girnar Broad Market Index Fund. There is no debt scheme in it at all: no debt holdings, no ratings, no yields, no maturity profile, not one line. So no credit table, no rating symbol and no yield figure can be drawn from it, and where a figure would normally sit the space stands open and marked. The reason for that emptiness is itself part of the subject rather than an apology for it.
Four ideas arrive here already built, and the work behind them is not repeated. A portfolio disclosure is a snapshot with no history behind it, established where the fund portfolio document is taken apart. Where several causes act on one observed total, the honest output is that total with its causes named rather than a split invented to look complete, established where a scheme's cash position is taken apart. The return a scheme publishes has the charge inside it before anybody sees the figure. And what a bond is, and what a lender is exposed to on buying one, comes from the fixed income material.
What does credit quality actually mean inside a debt scheme?
Start with the promise rather than the paper. When a scheme buys a debt security it stands in the position of a lender to whoever issued it, and that borrower has promised two separate things: a stated amount, and a stated date. Credit qualityHow likely the borrowers behind a set of holdings are to pay the promised amount on the promised date. is a plain judgement about how likely that borrower is to do both. Not one of them. Both. A borrower who eventually pays every rupee eighteen months late has not kept the promise, and a scheme holding that promise has a problem no amount of eventual payment removes.
The judgement is already familiar. Most people have made it themselves without calling it that. A cousin who has run the same hardware shop for twenty years asks to borrow for two months against a delivery he has already been paid for. A nephew wants the same amount to start a food stall he has not opened yet. The lender is not thinking about interest rates. The lender is thinking about whether the money comes back, and if the second one is lent to at all, the lender will probably want more for it. The lender's instinct, formalised and written down, is the whole of credit quality.
Notice where the quality actually sits: in the borrowers, never in the scheme. The scheme is a container. The container has a value per unit, a set of holdings, a charge and a record. Not one of those things can be more or less likely to pay a holder. The scheme is not the party doing the paying. The borrowers are. So a scheme does not have credit quality of its own, it holds obligations of borrowers who do, and that single sentence is why every honest reading goes through the scheme to the holdings and then through the holdings to the borrowers behind them. Stopping at the scheme, or at the scheme's return, is stopping one full layer above the thing under examination.
Somebody asks what the credit quality of a particular debt scheme is. Which first move is the accurate one?
Two debt schemes report returns a full percentage point apart for the same one year period. Which of them took more credit risk?
Why does a debt scheme's return say nothing about the risk that produced it?
Because of the one fact that organises the whole of lending: a lender who accepts a greater chance of not being paid is compensated for accepting it. The compensation is not a favour and it is not a mistake. The compensation is the price of the money. A borrower whose payment is in some doubt has to offer a higher yieldThe annual rate a lender is promised on the money lent, measured against what the lender paid for the promise. than a borrower whose payment is not, or nobody would lend to them at all. The gap between the two is a spreadThe extra yield a weaker borrower must offer above what a stronger borrower pays for the same money., and it exists precisely because the extra risk is real.
Now run one year forward. A scheme holding the weaker promises collects a higher yield across the year. If none of those borrowers fails during that year, every rupee promised arrives, the extra yield lands in the scheme, and the scheme reports a higher return. The return in a quiet year looks exactly the same whether the risk was taken or not, and a quiet year is precisely when the figure gets quoted. Nothing was hidden and nobody lied. The number is simply not built to carry the information being asked of it.
The everyday version is sharper than the finance version. Two acquaintances both lent money last year and both were repaid in full. Both tell identical stories: I lent, I got it back, I am pleased. One of them lent to a salaried neighbour against a deposit already sitting in a bank. The other lent to a stall owner two months behind on rent at the time, who spent the year one bad week away from not paying at all. The outcome is the same for both. The outcome is all a return figure is, and so the two cannot be told apart. What separated them was never what happened; it was what could have happened, and a return figure records only the first.
What is a credit rating, and what kind of statement is it?
A credit ratingA published opinion about a borrower's ability to pay a specific obligation on time, expressed on the issuing agency's own scale. is an opinion. The word opinion is not a criticism and it is not a hedge; it is the accurate description of the thing. A rating agencyA firm in the business of forming and publishing opinions on borrowers, registered and supervised as such. studies a borrower, forms a view on how likely that borrower is to pay a particular obligation on time, and publishes that view as a symbol on a scale the agency itself defines and explains. Rating agencies in India are registered and supervised. The conditions attached to that registration belong to the Securities and Exchange Board of India (SEBI), and SEBI publishes them at sebi.gov.in.
The scale itself, its symbols, and what each symbol is meant to convey are set out in each agency's own published methodology. Scales differ between agencies, and the definitions attached to them are revised, so an account that prints a symbol from memory teaches a reader to recognise a label whose meaning it has not checked. The invented labels Grade One down to Grade Four stand in for steps on a scale here, and are drawn only to show the shape of an ordered set.
Two properties of a rating matter far more than the symbol, and both are easy to read straight off the document it is printed on. The first is that a rating is an opinion formed rather than a quantity measured, so it can be revised at any time by the party that formed it, and nobody has to agree with it. The second is that a rating carries a date: the day it was issued, or the day it was last reviewed and confirmed. A symbol without its date is half a statement. How often an opinion must be revisited, and what an agency and a scheme must publish about it, are matters SEBI fixes, and the current position on both stands at sebi.gov.in.
A borrower's rating has not changed in two years. Which of these does the unchanged rating establish?
How is a rating different from a measurement?
A measurement is a quantity somebody counted, and its distinguishing feature is that it can be counted again. Weighing the sack of rice a second time gives the same answer, or reveals that one of the two weighings was wrong. Nobody counted anything in the first place, and a rating cannot be recounted in that sense at all. The rating was formed. A measurement can be checked by repeating it; an opinion can only be replaced by a different opinion.
The consequence people trip over is arithmetic. The steps on a rating scale form an ordinal scaleA set of labels that are ranked in order but carry no measured distance between one rank and the next.: they are ordered, so one is higher than another, but they are not spaced, so there is no distance between them. Two ratings are ordered rather than spaced, so the gap between two steps is not a number and must never be treated as one. Averaging a scale like that, or subtracting one step from another, produces something that looks like a measurement, prints like a measurement, and is not one. The look is the danger. Once a number exists in print, a reader will compare it with another number, and by then the mistake is invisible.
The second consequence is about time, and it is quieter. An unmoved rating is evidence of one thing only: nobody has revised it. Nothing underneath it has been shown to hold still. A borrower can lose its largest customer, watch its costs run away and burn through its cash while the symbol stays exactly where it was, because the symbol changes only when somebody decides to change it. The date beside a rating is therefore read alongside the symbol rather than after it, and a long gap since the last confirmation is itself part of what is being read.
A colleague averages the rating steps across a set of holdings and reports the answer as a portfolio credit score. Which objection to the score lands?
What does a reader look at instead of the return?
Four things, none of which is a single number, and all four sit in the portfolio disclosure rather than in the performance figure. The portfolio disclosure, what it shows and what it leaves out, is taken apart where the fund portfolio is covered; here it is used as a source rather than explained again.
First, the holdings themselves and what kind of borrower stands behind each one. A government borrower, a large lender, a small manufacturer and a finance company that lends onward to households are four different propositions, and the disclosure names them one by one. Second, the spread of the holdings across quality rather than a single summary of it. Third, concentrationHow much of a portfolio depends on a single borrower or a single set of related borrowers., meaning how much of the whole depends on any one borrower. Fourth, how long the money is out for. The longer a borrower has to keep paying, the more time there is for that borrower's circumstances to change.
An average across quality is exactly the wrong summary, and the second of those four is where the reading most often goes wrong. Picture two portfolios. One is overwhelmingly lent to borrowers regarded as very likely to pay, with a small corner lent to borrowers who are not. The other is spread evenly across the middle, with nothing at the top and nothing at the bottom. Work out an average and the two can land in the identical place. Then let one borrower stop paying. In the second portfolio a middling holding goes bad and the portfolio takes a knock. In the first, the corner is where all the trouble was concentrated, and the loss lands on a small part of a portfolio that looked safer than the other one on the summary figure. Two portfolios can share an average and behave nothing alike, and only the shape tells them apart.
Concentration is the household example everyone already understands, and it deserves its own sentence. A house running on one salary and a house running on four smaller ones can bring in the same money each month. The two houses are not in the same position, and everyone in both of them knows it. One person losing work is an inconvenience in the second house and an emergency in the first. A portfolio with one very large borrower is the single salary house, whatever its average quality says.
Two debt portfolios report the same average credit standing. Do they carry the same credit risk?
Move the mix and watch the return refuse to move
One control below moves how much of a scheme's year came from the safest part of its lending and how much came from being paid extra to hold weaker borrowers. The strip underneath redraws every time. The bar above it does not move at all, at any setting, and that refusal is the entire point of the exercise.
Educational illustration. Move the mix and watch the return refuse to move.
Assumptions on screen, and all three matter. The record behind this material holds no debt scheme, so the bar and the strip belong to nothing and carry no figures. The split is drawn, not measured, and the twenty one settings are twenty one drawings rather than twenty one findings. And a year in which nobody failed to pay is assumed throughout, the condition that holds the bar still. The default sits at the middle setting, fifty parts of the drawn bar on each side, and the worked example above uses that same position. At every one of the twenty one settings the two segments are exactly thirty one pixels apart from the next setting and always sum to the full six hundred and twenty pixel bar, so the total is exact by construction rather than by rounding.
With the holdings, the ratings and the yields all in front of the reader, can the decision whether the extra yield is worth taking now be made?
Where does the judgement actually sit, and what cannot be computed?
Here is the honest position. The question at the end of credit quality is not a calculation. No answer can be computed for it. The reader is weighing an extra amount of yield, which is visible and can be written down, against the chance that a borrower does not pay, which is neither. The chance of not being paid is not published and cannot be measured from a disclosure. A rating does not state it either: a rating is an opinion about likelihood rather than a probability anybody has counted.
So the trade cannot be closed with the information a disclosure carries. A careful reader can do three things instead, and all three are worth more than a false number. The first is to name the risk being taken, in words. A named risk is stated rather than left in the background. The second is to state what would have to be true for the extra yield to be worth it, which forces the assumption into the open where somebody can disagree with it. The third is to say what will be watched. A failing assumption is then noticed before the payment fails.
The value of the extra yield to a particular holder sits in no record, and nobody can settle which side of the trade to take on that holder's behalf. This is the same refusal made elsewhere about the choice between a plan sold through a distributor and one bought directly, and it holds for the same reason: the arithmetic is stateable, the preference is not, and the two must not be quietly joined.
What happens when the record holds no debt portfolio at all?
Because the record behind this material does not contain one, and the absence is more useful than a plausible replacement would be. The record does hold an equity scheme and an index fund: the Girnar Large Cap Equity Fund, carrying Rs 4,200 crore of net assets spread over 120.00 crore units, and the Girnar Broad Market Index Fund. Neither is a debt scheme. There are no debt holdings anywhere in it, no ratings, no yields, no maturity profile and no second scheme of any kind that could stand in.
No credit table, no rating symbol and no yield figure can be drawn from a record like that. Every place one would sit is left visibly open and stamped as not carried. The temptation to fill those spaces is real, because an invented credit profile would be easy to write, would look completely ordinary, and nobody would query it. The ease of the invention is exactly the reason those spaces stay open: a table invented to look plausible would wear the same clothes as a disclosed one, and an account whose whole argument is that disclosed artefacts must be interrogated cannot teach a reader to trust an artefact it made up itself. The refusal is the lesson, not a gap in it.
The record is uneven, and the unevenness is what makes the temptation strong. On the equity side it is precise to a fault: net assets of Rs 4,200 crore, units outstanding of 120.00 crore, one stated year with a benchmark beside it. On the debt side there is nothing at all, not one borrower and not one promise. A record that is exact where it is populated and empty where it is not is the ordinary condition of working with documents, and the discipline is to work the mechanism on the populated part and name the empty part rather than blending the two. That discipline is the same one applied where a scheme's cash position is taken apart: report the total observed, name the causes that can be named, and refuse to invent the split that cannot be seen.
Why does no debt portfolio appear at all?
Does the same silence show up in the one scheme this record does carry?
The silence does show up there, and running the check on equity matters. Confined to debt, the argument would look like special pleading. For the stated year the Girnar Large Cap Equity Fund reported 13.4 per cent, net of the charge. The benchmark it names for itself came in at 12.1 per cent over the identical twelve months. One scheme and twelve months.
Subtraction gives a difference of 1.3 percentage points, exact on the two stated figures. The check runs backwards as cleanly: 12.1 plus 1.3 is 13.4, and 13.4 less 1.3 is 12.1. So the arithmetic is not in doubt. The 1.3 points say nothing whatever about how either figure was produced. A debt scheme's return keeps that identical silence about its borrowers. The scheme's figure could have come from steady holdings or from a concentrated position that happened to work, and the subtraction is the same either way.
Two further cautions belong in this same block rather than in a footnote. The first is about bases, and it is not a technicality. The scheme's 13.4 per cent is net: already after the charge that ran against the scheme's assets. The benchmark's 12.1 per cent has never had a cost taken out of it. Nobody holds an index directly and nobody is charged for its existence. Subtracting a costless figure from a net one is arithmetic between two different kinds of thing, and putting both onto one basis is worked through where a benchmark return is compared with a fund return. The second caution is about precision. Both figures are printed to one decimal place, so each is a rounded statement of something else: a printed 13.4 could be a 13.44 rounded down or a 13.35 rounded up. Run that through and the true difference lies somewhere strictly between 1.20 and 1.40 points, and only equals 1.30 if both printed figures happen to be exact.
| Step | The arithmetic | Result |
|---|---|---|
| Start | The scheme's stated year, net of the charge | 13.4 per cent |
| One | The stated benchmark's same year, carrying no cost | 12.1 per cent |
| Two | 13.4 less 12.1, on the two printed figures | 1.3 points |
| Check back | 12.1 plus 1.3, which must return the scheme's figure | 13.4 per cent |
| Check back | 13.4 less 1.3, which must return the benchmark figure | 12.1 per cent |
| Rounding | 13.4 printed to one decimal covers 13.35 up to 13.45 | rounded either way |
| Rounding | 12.1 printed to one decimal covers 12.05 up to 12.15 | rounded either way |
| So the gap | Lies strictly between the two bounds above, not on 1.3 exactly | 1.20 to 1.40 points |
Now the limit on all of it, stated beside the figure rather than saved for a closing caution. Twelve months of one invented scheme sit behind these figures, and nothing whatever beside them: no earlier year to compare, no month by month path underneath, no sibling scheme, and no vehicle that exists outside this teaching record. Twelve months like that will not stretch into an annual rate, will not run forwards or backwards into other years, will not line up beside anything traded in a real market, and settles no argument about picking holdings or following an index. Every sum done on it is exact, and almost nothing follows from it, and both of those hold at once.
Which leaves the rule worth carrying away. A return is an outcome and credit quality is an input. An outcome from one quiet year says nothing about the input. The Girnar Broad Market Index Fund's numbers are kept out of the discussion altogether. A tracking figure parked beside a credit discussion invites a comparison neither number can support.
Who actually opens a credit profile at work, and what are they looking for?
Three different readers open the same material, and not one of them is hunting for a score. Sohail Merchant heads operations at Girnar Asset Management, and he is not judging borrowers at all. When a rating on a holding moves, his concern is what has to happen to the record and to the valuation as a result. The question is a different one, and it belongs to the operations material. He needs the event captured, dated and reflected, and he needs it to be the same answer on every scheme it touches.
An analyst working through a scheme's disclosure keeps the task deliberately small. Pricing the credit risk off that document is not attempted at all. The inputs pricing would demand are simply not in it. The analyst looks for the shape rather than the average: how much sits in the weakest part, how much depends on any single borrower, how long the money is out for, and how many holdings carry no rating at all. Then the analyst writes down what cannot be seen, and that line is usually the most valuable in the note.
A household choosing where a year of savings sits reads it differently again, and more usefully. The question is not which scheme is better. The question is whether the extra yield on offer is being paid for something they can name, and whether they could tolerate the year in which that something turns up. A reader who can say out loud what risk they are being paid to carry has done the part of this work that actually protects them, and a reader who cannot has not been protected by the return figure either.
The mistake a reader makes here, and the bill that arrives later
A reader lines up two debt schemes, sees that one returned materially more over the same year, and moves towards it on the strength of that figure. The two numbers are comparable in every visible way: same period, same units, same kind of scheme. The reasoning feels sound. The trouble is that the extra return may be nothing more than the compensation for lending to borrowers who might not pay, and in a year where none of them failed, that compensation arrives looking exactly like skill.
The cost is specific and it is delayed. The reader has accepted a risk they never priced, and they will find out about it in the year the risk turns up rather than in the year they chose. Nothing warns them in between. The value per unit will keep printing every day and the return will keep looking good right up until it does not. A risk that was not priced is not a risk that was avoided; it is one whose bill has not been presented yet.
There is a second failure and it belongs to whoever quotes the figure rather than to whoever reads it. A return published without the credit profile beside it is not a lie. The number is accurate. The number is also not information, and it cannot answer the question the reader is actually asking. Presenting a figure that cannot carry the weight a reader will place on it is its own kind of error, and the fix belongs to the person doing the presenting.
The fix is not vigilance, and it is not easy. The holdings are read before the return rather than after it. The spread across quality and the largest single borrower matter more than any average. And where the extra yield cannot be explained by something visible in the portfolio, the unexplained part is the thing being paid for, and the reader decides whether that is a payment worth taking.
Who fixes the rules around this, and where should a reader look?
SEBI sets what a debt scheme must disclose about its portfolio and its credit profile, how often it must do so, how a holding is valued from one day to the next, what must happen when a holding is downgraded or a borrower fails to pay, and any condition a scheme category has to keep to on the quality or the concentration of what it holds. SEBI also registers and supervises the rating agencies themselves. Every one of those items exists, and each is stated by SEBI rather than summarised here: the periods, the thresholds, the percentages, the prescribed treatments and any rating symbol standing as a regulatory line. Requirements like these get amended, and a printed copy of one does not merely go stale on the day it moves, it starts misleading people. The position as it stands is at sebi.gov.in.
The rating scales themselves, the symbols on them and what each symbol is meant to convey are published by each rating agency in its own methodology documents, and those documents are where a symbol is read rather than any summary of them. Material gathered at an industry level sits with the Association of Mutual Funds in India (AMFI) at amfiindia.com, a body that assembles and publishes it without setting any of it. Where the tax treatment of a debt scheme matters to a decision, the tax authority at incometaxindia.gov.in is where the rate, the category and the length of holding are stated.
Nothing in the mechanism above leans on any of those requirements. Credit quality sitting with the borrowers, a return being silent about the risk that produced it, and a rating being a dated opinion rather than a measurement are true wherever the scheme is domiciled. A second market would want another paragraph here and would leave the rest of the mechanism alone.
A debt scheme has reported better figures than others of its kind for three quiet years. Which conclusion about its credit risk does the record support?
References
| Source | Document | Where |
|---|---|---|
| Securities and Exchange Board of India | What a debt scheme must publish about its portfolio and its credit standing, how its holdings are valued, what follows a downgrade or a failure to pay, any condition attached to a scheme category, and the registration and supervision of rating agencies | sebi.gov.in |
| The credit rating agencies, as a class of publisher | Each agency's own published rating scale and rating methodology documents, the place where a scale is defined and a symbol acquires its meaning | each agency's own site |
| Association of Mutual Funds in India | Collated material about mutual fund schemes at an industry level, one place such material is gathered and put out | amfiindia.com |
| The income tax authority | The tax treatment of a holding in a debt scheme, a matter a reader weighing yield against risk runs into sooner or later | incometaxindia.gov.in |
Girnar Asset Management Limited, the Girnar Large Cap Equity Fund, the Girnar Broad Market Index Fund, Kalyani Bhagat, Sohail Merchant, Scheme A and Scheme B, Portfolio P and Portfolio Q, and the Grade One to Grade Four labels are invented.
Educational material. Not advice on any investment, tax, budget or market position.