A Dangerously Narrow View Of Risk

Fund houses are required by SEBI to classify the risk of each scheme as one of five levels: low, moderately low, moderate, moderately high, or high.  But what exactly should we make of a scheme whose risk level is defined by the fund house as, say, ‘moderately low’?  On the other hand, what should we make of Value Research or Morningstar telling us that the risk grade or risk rating of that scheme (relative to its peers) is, say, ‘average’?

It isn’t just their ambiguity: I would not rely on any of these labels as they largely stem from a narrow view of risk.   I believe that if we are not careful, these can lead us to make flawed assumptions about the riskiness of a scheme and, worse, act upon them. 

To illustrate the perils of relying upon these risk ratings, I’d like to take the case of a specific scheme whose risk ratings are currently poles apart from my assessment of its risk.  To be clear, this scheme is an extreme outlier: it would be hard to find a scheme quite like this.  However, the extremity of this example is what makes it useful to show the arbitrary nature of fund house risk ratings, and the limitations of the methodology followed by entities such as Value Research and Morningstar.

The scheme in question is a debt fund that has been around for over ten years.  From what I can see, for most of its existence, there has been a noticeable consistency in the way that its maturity/ duration and its credit profile have been managed.

As regards its performance, in each of the last 10 quarters, its return was above average (compared to its peers).  In 3 of the last 6 quarters, its return was exceptional.  In the month of June, this scheme gave a higher return than almost every non-gilt fund.  Its return in June was also higher than its return in any of the preceding 12 months.

The fund house has classified its risk as ‘moderately low’.  On the other hand, both Value Research and Morningstar have currently given it a risk grade/ rating of ‘average’ and an overall rating of 5 stars (based on the performance of its direct plan, growth option).

So, what’s the problem?

Just as with some other schemes, over the past several months, this scheme saw a significant fall in its AUM.  Consequently, there is a certain illiquid NCD in its portfolio, which it hasn’t been able to sell off, whose proportion to the portfolio has zoomed up as the AUM has fallen.  As on May-end, this NCD made up 71% of the scheme’s portfolio.  As on June-end, this NCD made up 87% of the portfolio.

Take a minute to digest that, if you will, because there’s more.

That single NCD is currently rated BBB (CE) and is under “credit watch with negative implications”.  In other words, that NCD is just about making the cut for being ‘investment grade’ and is precariously close to slipping below that.  If it does, well, there’s no saying how much an investor could be impacted.  If industry practices are anything to go by, for starters, the fund house would have to mark down the value of that investment by at least 25%.  And for those who have forgotten, here’s a bit of a flashback.  Last month, when DHFL was downgraded from BBB- to D, one scheme which had 67% of its portfolio in DHFL NCDs saw its NAV fall by 53% on that single day.

So how does a scheme with a portfolio like this get a risk rating of ‘average’ or a risk level of ‘moderately low’? 

From what I have gathered, in the Value Research/ Morningstar risk ratings, factors such as portfolio concentration, even credit quality are not considered.  In contrast, consider the approach that CRISIL takes for its fund ranking.  In the case of debt funds, for example, apart from return, the ranking gives weightage to elements such as asset quality, interest rate sensitivity, liquidity and company concentration, among other things.  As it happens, moneycontrol.com, which apparently uses CRISIL’s ranking, has given the abovementioned fund an overall rating of 2 stars.  While I am not suggesting that CRISIL’s process is perfect, it is certainly a lot better than anything else that I have seen.

On the other hand, in the case of the fund house classification, the issues may be more complicated.  For one, fund houses are currently bound by the way in which SEBI has defined the risk levels.  For another, product labelling is practically a one-time exercise.  Personally, I don’t see much utility to having such a risk classification, certainly not in its present form.  Regardless, I would prefer that fund houses gave investors a list of things to check before investing and also highlight issues that warrant caution. 

In any case, investors would do well to not blindly go by star ratings or risk ratings.  If we choose to use them, at the very least, we should understand their limitations. 

The Valuation of Junk Bonds

The recent, successive downgrades of securities issued by JSPL, have led me to ponder over the inconsistencies across fund houses in valuing junk bonds. Frankly, valuation of securities is not my forte.  But seeing the exposed fund houses take different approaches to valuing their holdings in JSPL, I felt the need to examine the choices that they made.  In that backdrop, and with some assistance, I have put my observations and thoughts here, as accurately as I could, in the hope that these encourage discussion around improving the fairness of valuations.

On Feb 15, CRISIL downgraded JSPL securities to junk status.  The next day, the schemes of Franklin Templeton MF (FT) and ICICI Prudential MF (I-Pru) holding JSPL securities reported a fall in their NAVs.  It appeared that both these fund houses had applied a discount of 25% to the face value of their holdings right after the downgrade, something that was subsequently confirmed by a reading of their month-end disclosures and FT’s additional disclosure.  However, the schemes of Reliance MF, which also held JSPL securities, did not report a decline in their NAVs after that downgrade. 

According to this news report, a spokesperson of Reliance MF said that the fund house had negligible exposure in the company and that the security that its schemes held had not been downgraded. He/she may have been referring to the exposure relative to the fund house’s total debt AUM because when looked at, relative to the AUM of the individual schemes which held JSPL securities, the exposure was, by no means, small.  In two of these schemes, the exposure to JSPL was in excess of 11% of the portfolio.  In a third, it was over 8%. As to the point about the security they held not having been downgraded, that was, in a way, accurate.  That security had been rated by CARE, and not CRISIL, and CARE had not, till then, downgraded the security.  Reliance MF’s stance, though, was at odds with that of I-Pru, which held the same security in a number of their schemes, and had chosen to mark it down right after CRISIL’s downgrade.  Eventually, on Feb 25, CARE downgraded that security.  The next day, the NAV of the schemes of Reliance MF fell marginally.  Going by the month-end portfolio disclosures, Reliance MF had applied a discount of 2.5% to the face value of its holdings, far less than what FT and I-Pru had applied. 

Then on March 9, CRISIL further downgraded JSPL securities to default status.  FT stated that it had sold off its entire holdings of JSPL, leaving nothing to be valued. This time around, Reliance MF reported a fall in the NAVs of all three debt schemes holding JSPL, ranging from –0.83% to –2.66%.  Unlike the last time, it seemed as if it had accepted CRISIL’s downgrade (CARE’s downgrade happened on March 11).  While the complete facts may take some time to be known, it seems that it marked down the value of the JSPL holdings to about 75% of their face value (i.e. to the level that FT and I-Pru had marked down their holdings in February).  In stark contrast, I-Pru did not report a fall in the NAVs of any of the FMPs that had significant exposure to JSPL.

So why did the NAVs of the schemes of I-Pru not dip?  After all, it would stand to reason that whenever a bond gets downgraded, its price would be expected to fall.  If anything, FT’s sale of JSPL securities (although different from the one held by the FMPs of I-Pru) on March 10, at a further discount to their February-end values, supported such reasoning.  And why did Reliance MF apply a discount in February that was only a fraction of the discount applied by FT and I-Pru?  Why did it refuse to acknowledge CRISIL’s downgrade in February, and then act upon it in March?  And why did it do an apparent volte face, and apply a more significant discount after the March downgrade?  What, indeed, were the reasons for all these inconsistencies in the valuation of this junk bond?

As far as I can make out, the valuation policies of most fund houses do not carry any specific instructions on valuing junk bonds.  These fund houses let the decision be taken by their respective valuation committees.  Even so, it seems that the norm is for junk bonds to be valued at 75% of their face value.  The decisions by FT and I-Pru appear to be in line with that.  Reliance MF, on the other hand, is one of the few fund houses to have explicit guidelines for valuing junk bonds.  According to its valuation policy that was in force in February, junk bonds (that were not NPAs) with more than 182 days to maturity were to be valued at 75% of their face value while those with less than 183 days to maturity were to be valued “after markdown by 2.5% to the Face Value every 2 weeks cumulatively starting from the day of the downgrade.”  In light of this, its valuation of JSPL holdings in February is very understandable: the security that it holds, is due for redemption next month.  However, there appears to be no explanation in their valuation policy for their subsequent decision to apply a cumulative discount of 25% at one go, right after the second downgrade.  There is also no explanation that I have been able to gather for I-Pru not marking down its holdings any further.

Regardless of fund houses sticking to their individual guidelines or the recommendations of their valuation committees, I believe that fund houses should disclose to their investors, in a transparent manner, the rationale behind decisions that may appear questionable.  At the same time, I also believe that SEBI needs to bring about greater uniformity and even conservatism in the valuation of junk bonds.  If you think about it, at the end of February, Reliance MF valued its JSPL holdings 30% higher than what I-Pru had valued its holdings.  By any yardstick, that is a vast difference.  Thankfully, if I may say so, Reliance MF’s holdings of JSPL were all in closed-end schemes.  If these were to have been in open-end schemes, the knowledge of the fund house’s relatively liberal valuation policy could well have triggered an exodus of its investors from those schemes.  Of course, it is entirely possible that someone may have actually exited those schemes based on that knowledge, irrespective of their closed-end structure.

Last but not the least, I would also make the case that SEBI should re-examine how NPAs are defined, and provisioned for.  Most fund houses classify NPAs based upon the definition in the SEBI guidelines.  But that definition ends up excluding securities such as the JSPL NCDs held by I-Pru and Reliance MF even though CRISIL has specifically stated that its downgrade reflected actual delays by JSPL in payment of interest on its term loans.  For a bond to qualify as an NPA under the current mutual fund valuation guidelines, a fund house has to have firsthand experience of the delay.  And it isn’t just the definition: even the provisioning guidelines, to my mind, are way too liberal.  As I see it, fund houses that strictly follow the guidelines, run the same risk that I referred to in the previous paragraph.

On a somewhat related note, an extract from the SID of one of the debt funds still holding JSPL NCDs was brought to my attention.  It reads thus:

“Credit evaluation is a continuous process.  It applies not only for issuers where investments are being evaluated for the first time but also for those where we already have credit exposures.”

I doubt if any fund manager would disagree with the truth in this statement.  Yet, a statement like this also begs these questions: Why would a scheme still be holding a security that has been downgraded to default status?  What does the fund manager of that scheme know that the rating agency does not know?  And as I think about it, there may be one other advantage to having more stricter and conservative guidelines for valuing junk bonds and for provisioning for NPAs: it may well act as an additional deterrent against fund managers buying and holding securities that are junk or borderline junk, without good reason.

Special thanks to Robin Jehangir for his inputs.

More Rating Mystery

After my earlier post, about two weeks ago, I was not expecting to have something to write about fund ratings this soon.  It seems I was wrong.

On Thursday, CRISIL announced that it had placed its rating of JPMorgan India Treasury Fund under “Notice of Withdrawal” based on a request by the fund house.  By itself, this would seem to be a straightforward development.  After all, no fund house, in its right mind, would want to use a BBBmfs rating.  But the release didn’t stop at that.  For reasons best known to CRISIL, in this release, it also chose to explain its current rating, updating its explanation in the last release with a reference to the segregation of the Amtek Auto investment.  This is what triggered this post.  I reproduce below a section of the release that I am unable to make sense of. 

On September 28, 2015, JPMAMIPL had allowed redemption in the scheme, following segregation of the specified security in the
scheme’s portfolio upon receipt of unit holders’ approval. This resulted in intensified redemption pressure in the scheme, and therefore, an increased weightage of the specified security. The downgrade in the scheme’s rating reflects the expected increase in the weighted average credit score for the portfolio.

The rating remains on ‘Watch with Negative Implications’ as CRISIL expects the redemption pressure on this fund may remain elevated. Should the redemption pressure exceed current expectation, the weightage of the specified security in the portfolio will increase significantly, further weakening the scheme’s credit quality.

As far as I have been able to make out, these are the facts of the segregation:

• As soon as the segregation was approved, the NAV of the scheme fell to reflect the segregation of the said security.
• Investors in the scheme on the date of the segregation took a hit on the value of their investments.  Investors who invest after that date will remain unaffected by the segregation. 
• If the value of the segregated security is recovered, only the investors who were there on the date of the segregation have a claim on that value. 

What this means is that there is no further downside on account of that security to existing as well as new investors.  To put it differently, an investor who is going to buy into the scheme, in effect, will not be buying into that security.

Thus, if a rating has to be given on the portfolio now, why would this security have any bearing on that rating, no matter how much the portfolio shrinks?

Rated AAAmfs

The recent downgrades of JPMorgan India Treasury Fund by CRISIL have left me wondering about the usefulness of its ratings for bond funds.  As CRISIL puts it, its rating of a scheme reflects “the likelihood of timely receipt of payments from the investments” made by it.  Schemes with a AAAmfs are considered to have “the highest degree of safety regarding timely receipt of payments from the investments that they have made”.  According to CRISIL, its ratings “serve as a tool to investors for selecting funds with a suitable risk-return criterion.”  These also provide “an independent opinion” on the credit risk associated with a fund’s portfolio.

All of this matches with my own expectations from a fund’s rating.  My issues begin when I look beyond the letter, and into the spirit of what a fund’s rating implies.  The core of my dilemma is this: when it comes to an individual bond, I recognize the possibility of a highly rated bond being significantly downgraded overnight.  But when it comes to a portfolio of bonds, I find it difficult to accept that there can be compelling reasons for a major downgrade overnight. In this post, I try to make my case.

Let me start by giving below the month-by-month share of lower rated securities (i.e. other than Sovereign/AAA/A1+) held by JPMI Treasury Fund:

	AA+	AA	AA-	C	Unrated
Jan	7%	16%	3%	0%	6%
Feb	6%	10%	3%	0%	0%
Mar	5%	19%	13%	0%	0%
Apr	0%	13%	6%	0%	0%
May	0%	7%	12%	0%	0%
Jun	0%	7%	12%	0%	0%
Jul	0%	6%	10%	0%	0%
Aug	0%	3%	7%	6%	0%

JPMI Treasury Fund enjoyed a AAAmfs rating from before the start of the year, and this was reaffirmed in May.  It enjoyed this rating despite having around 37% of its March-end portfolio in securities rated lower than AAA.

Then, on September 1, CRISIL downgraded the fund to A+mfs (it subsequently dropped this to BBBmfs).  Going by its rating rationale for the September 1 downgrade, CRISIL did not have the latest portfolio composition at the time of taking that decision.  If, therefore, one had to go by the July portfolio and speculate what the August portfolio would have looked like (after the Amtek Auto downgrade), this is what my guess would have been:

	AA+	AA	AA-	C	Unrated
Aug	0%	6%	5%	5%	0%

Frankly, I find it difficult to reconcile the rating profile of the securities in the fund with CRISIL’s rating of the fund.  CRISIL’s process may well be, as they claim, “derived scientifically”.  But to me, as an investor, there is a contradiction with my perception of what fund ratings imply, at least in spirit.  I expect a portfolio rating to warrant a major downgrade only if a large portion of the portfolio has been significantly downgraded or is likely to be so downgraded.

CRISIL explains the September 1 downgrade, saying that this was “on account of significant weakening in the credit risk profile of an underlying security.”   It goes on to say this: “Earlier, the scheme’s credit score had cushion to absorb some deterioration in the credit risk profile of the said underlying security. However, with the recent sharp weakening in the credit risk profile of the security, the credit score for the portfolio has been adversely impacted.”

I don’t dispute any of this.  But if the downgrade of a single security can cause such a significant downgrade in an entire portfolio’s rating, then it would seem to me that the risk of a concentrated holding in a single security was not factored in determining the rating.  In my view, it should have been a factor.  Put differently, if the downgrade of a single security can push the rating of a fund down by several notches, then I am not sure if a fund’s rating can be of much use to an investor.