Friday, January 23, 2015

Hey Fitch, What’s Up With This?

Just a follow up from yesterday evening’s post. To refresh your memory, here’s the data I posted:

  GDP growth (2014e) Gross Debt to GDP Fiscal Balance Current Account to GDP
Australia 2.80% 30.60% -3.30% -3.60%
Canada 2.30% 88.10% -2.60% -2.60%
Malaysia 5.90% 56.60% -3.60% 4.30%

To recap:

  1. Australia has low debt to GDP but runs a twin deficit;
  2. Canada has high debt to GDP and also runs a twin deficit;
  3. Malaysia has a middling debt to GDP and has a fiscal deficit, but runs a current account surplus

Here’s the sovereign credit ratings for all three by agency:

  S&P Moody's Fitch
Australia AAA (Stable) Aaa (Stable) AAA (Stable)
Canada AAA (Stable) Aaa (Stable) AAA (Stable)
Malaysia A- (Stable) A3 (Stable) A- (Negative)

As far as I can make out, Australia has fairly low contingent liabilities (less than 7% of GDP), while Canada (not including pension liabilities) is about 25% of GDP. Malaysia has about 15% (also not including pension liabilities). All three of course are primary commodity producers with government revenue either implicitly or explicitly reliant on taxes and duties on that sector.

So Fitch, dudes, what’s up with the negative outlook on Malaysia?

How come Mexico – with an economy 4 times as large but oil production 5 times as large, places 53 spots below Malaysia in terms of corruption and 36 spots below in terms of doing business, has both a current account and budget deficit, and has a debt level only a little lower than Malaysia – gets away with BBB+ (Stable)?

We’re as risky as Mexico, really?


  1. This is like a vote of confidence, very subjective. If economics is nothing more than plus, minus, stats and data, it will be known as maths.

    This is a Game of Confidence. Given a choice, most people are likely to migrate, invest, travel to Canada and Australia over Malaysia.

    If I am not mistaken, prior to the Opium War in China, China was the richest country in the world.

    Simply put it, the reason the dollar is the safe heaven because the one with the most power to determine and chart the course of the future is the safest bet. When and if the dollar is about to collapse, America only needs to withdraw their troops from all over the world and WW3 will short commence, thus the dollar becomes valuable again.

    You can't fight those that controls the future, dance with them.

    1. @anon

      I used to work at a credit rating agency, and I've done sovereign ratings before. You'd be surprised at the number of variables that go into a rating.

      However, the three cardinal rules of credit ratings is consistency, consistency, consistency. Fitch is not being consistent here.

    2. For someone in the financial industry, I am appalled by your naivety, simplistic and misleading view on Malaysia’s sovereign rating.

      Besides, with the experience of working for a rating agency (I’m assuming one of the big 3s, if not, it’s not worth mentioning at all), I’m sure you are aware that Fitch uses the application of an Ordinary-Least-Square methodology on certain variables to derive a sovereign rating, and many of these variables were conveniently left-out in the post above.

      For starters, why did you not mention about the GDP and GNI per capita, the governance indicators (I’m sure u know what I’m talking about here). But more importantly, what about Malaysia’s debt AND interest payments as a percentage of revenues. Clearly, you have not grasped the concept of debt-affordability!

      Oh, also on the Mexico argument, can you reveal their Debt/GDP and fiscal balance data, please? I’m very sure that will shed some light as to why they deserve their BBB+ rating or better, at least vis-à-vis Malaysia.

    3. @misleading

      Sir, like almost everything in finance/economics, I believe the author is breaking problem down to its mechanics.

      Even if Hisham did include variables such as GDP/Capita and governance indicators, it would not have made a difference. Including all of the variables would need a whole different post altogether, which I suspect, would be different from the author's intentions in this post. Somehow, I suspect that the author didn't intend to write a complete rating methodology guide.

      I think to call this post naive, simple, misleading and to conclude that the author has not grasped the concept of debt dynamics is harsh. For one, I would argue in this post growth rate relative to interest payments is more relevant here than governance indicators.

      On another note, it is interesting that you wrote "Ordinary Least Square" instead of the usual convention for someone that opines "I'm assuming one of the big 3s, if not, it's not worth mentioning at all"

    4. @misleading

      I could have come up with any number of snarky answers, but I think a straight reply would be more educational.

      1. Since a rating scale is a bounded discrete range, while the explanatory variables are continuous, you should NOT be using OLS. Instead, the correct methodological approach would be an ordered logit regression. This is assuming you're trying to match variables against a rating scale.

      2. However, if you're attempting to discover the probability of default, the correct model would be a probit model.

      3. Some of the explanatory variables are also bounded (e.g. the debt ratio) while some are not (e.g. GDP per capita). This presents significant problems for OLS, as you should not mix variables of different orders of integration except under very special circumstances. All you'll get is an interesting but ultimately meaningless correlation, when what you're looking for is causality.

      4. In any event, I can't see how an OLS would actually be useful, as you're trying to match explanatory variables with either a rating scale or the probability of default (which can then be assigned to a rating scale), and an OLS can't do that. What's needed here is a panel (matrix) ordered logit or probit regression, with or without fixed effects. This will allow the creation of a "global" default/rating model, against which individual countries can be matched.

      5. Whichever model you use, adding more variables introduces as many problems as it solves. Most macro variables are correlated, which means any model specification you add them into causes multi-collinearity. For instance, using the debt to GDP ratio and the deficit ratio in the same regression results in estimates that almost cancel out, or look like nonsense. Then there's issues with degrees of freedom.

      If Fitch is really using OLS, no wonder the ratings are off base.

      How do I know all this?

      Because I've done it before. I've backtested the ratings of all three majors against the variables they say they use. What I (and others who've tried the same) found is evidence of significant bias in the ratings given.

      This blog post isn't based on a "simplistic" or "naive" approach to credit ratings. My rant is more based on the supposed rationale behind Fitch's negative outlook - oil prices, a declining current account surplus (of all things!) and contingent liabilities (see point 5 above). Sorry, that's nonsense.

    5. Bonjour Hisham, i see some sound responses you have! Getting straight to the point here: I’m not questioning the validity of the methodology but rather, I don’t think you should have not mention the many other indicators, that otherwise, would have suggested a weaker credit profile for Malaysia.

      And that, to me, is misleading, especially to those who are not aware of these credit indicators that a typical economist would.

      And neither am I defending the methodology that Fitch uses. Its not perfect, by any means and we can debate all day on this.

      Anyway, just for fun, my responses to the points raised, are as follow:

      1. An ordered logit regression, although great in theory, is difficult to implement and raises more debates on the assumptions used. For example, how does one assign a default probability on a particular variable? An OLS, as Fitch implements it, allows for relative comparisons against peers across similar variables (via notching differentials). I believe S&P/Moodys use a variant of the ordered logit model in their respective metrices, but that also raises the issue of where they are drawing that boundary lines for each sub-factor and rating class.

      2. The probit model would also be ideal but has similar issues as above. Of course, if I’m wrong (which I can be), why have we not seen any reputable sovereign ratings methodologies emerge using these models?

      3. This is true to a certain extent, but I doubt the overall difference is meaningful to the actual outcome.

      4. No, you are wrong here (I’m beginning to have questions if you know how the OLS is implemented by Fitch). The OLS methodology is not used to match to a rating. That’s all I can say.

      5. I partly agree with this statement but it leads the discussion to nowhere. For argument sake, which you are aware of, a high Debt/GDP but with rising fiscal Surplus gives an indication of stronger credits going forward. On the other hand, a high Debt/GDP with rising fiscal Deficits may indicate the opposite. I think u know where I’m going with this….

      Referring to your back-testing, many of these academic articles were written way back in the 1990s or early 2000s, a time period when rating agencies kept their methodology as a ‘black-box”. Only a few years ago, I’d assume due to market pressures, these methodologies were made public. Out of pure curiosity, how did you back-test these ratings?

      And going back to your “rant”, the rationale for the negative outlook is clear. Malaysia’s rating was already on negative outlook when oil was way above $100. The deterioration of oil prices only expedited the deterioration of these dynamics, and not a direct causation for the negative outlook.

      I don't deny that the other AAA countries have also been impacted by the declining commodity prices, but their overall credit metrices have not deteriorated beyond their credit strengths or buffers of their rating class...

      At the end of the day, we should be asking if investors are more willing to invest in Cad./Aus. Govt bonds or MYR. Govt. bonds given current market and economic conditions.

    6. Hisham got it spot on. Fitch's ratings are damn inconsistent internally and among its competitors.

    7. @Misleading

      Ah, an informed commentator! Thanks for the input, and the discussion.

      However, on point no 3, you've totally lost me. If this very basic statistical principle is being ignored, the results (no matter how plausible they may look) are invalid.

      Lastly, my rant was on the original decision for a negative outlook, not the recent affirmation.

    8. I am very very surprised that Fitch (and other rating agencies) uses the simplistic OLS model. No wonder their ratings are dodgy and so inconsistent.

      In any case, I cannot recall a time when the rating agencies produced a consistent or even accurate ratings at all whether sovereign or on companies. It always been driven by either vested interests or knee-jerk responses. In the case of Malaysia, it appears the latter.

      After the fiasco of 2008 GFC, not many people actually take these agencies and their ratings seriously. I wonder how they are still allowed to operate still as what they did was criminal and corrupt.

    9. Hisham, on point 3, it depends on how the OLS is used. there's a treatment to adjust for these inconsistencies.

      @calvinsankaran, Fitch does not based its ratings solely on a simplistic OLS model. There are further steps involve in the process.

      Consistency has always bugged these rating agencies but they have not been far off, at least for sovereign ratings.

      The fiasco of 2008 raises the issue of how these rating agencies rated the mortgage papers, not sovereign rating. It was a new type of instrument and it could have been just an honest mistake (by making the wrong assumptions)

      Also, In downgrading a sovereign, i doubt any rating agency would benefit (they'll incur the wrath of the government, i'm sure).

      But I can understand if they were to benefit by giving AAAs to dubious papers....

    10. @Misleading

      Mixing up orders of integration is not an "inconsistency", it's fundamental no-no. You do not adjust the OLS, the recommended treatment is to transform your data or use something else.

  2. Lol. Check out France's numbers (any of it's numbers!) and it's corresponding rating.

    Gives you an indication of the conflict of interest going on there.

  3. Well, Hisham...if we were to subscribe to your views, then we shouldn't mind if Malaysia's credit ratings are downgraded, should we?

    Because, as you aver, all these ratings stuff are subjective anyway.

    So, why worry if Malaysia is demoted from "investment grade" to "junk grade"?

    The government can then focus on more productive things.

    1. @anon 3.31

      No, I'm not particularly concerned about a downgrade. We're far away from junk status, and things will have to deteriorate a great deal more than now to get there. Even with a Fitch downgrade, that would still put Malaysia three notches above junk status.

    2. Not to mention S&P maintains a stable outlook and Moody's has a positive outlook on Malaysia : )

  4. Reuters reported that Cagamas will be marketing its first foreign currency sukuk this year and targeting new investors in Europe and the Middle East.

    You may recall that Cagamas set up a US$2.5 billion multi-currency sukuk in November 2014, after issuing conventional bonds in yuan, Hong Kong dollars and US dollars in 2014.

    It will be interesting to see what Cagamas will price this foreign-currency sukuk at.

    If your thesis is correct, they should price it the same as US Treasuries, Australian government bonds or Canadian government bonds.

    They shouldn't be adding in a "premium" to attract these "new investors" from Europe and the Middle East.

    So, will the "market" allow Cagamas to do so?

    1. @anon 3.50

      10Yr Govt Bond Yields, inflation and real rates of return (data from Trading Economics)

      Malaysia 3.95% 2.7% 1.25%
      Ireland 1.07% -0.3% 1.37%
      Latvia 1.19% 0.2% 0.99%
      Lithuania 1.29% -0.3% 1.59%
      Poland 2.09% -1.0% 3.09%

      The above are all rated A- by Fitch

      For reference, here's the BBB+ list:

      Thailand 2.76% 0.6% 2.16%
      Spain 1.36% 0.0% 1.36%
      Slovenia 1.44% 0.2% 1.24%
      Peru 5.24% 3.2% 2.02%
      Mexico 5.32% 4.1% 1.24%
      Italy 1.52% 0.0% 1.52%

      See a pattern? No? Neither do I

    2. C'mon there, chumps. This is comparing apples and oranges (or watermelons).

      i'm thinking Cagamas is issuing a USD paper (right?), so the comparable yields should be in USD! not some EUR yields at negative rates!

      With German Govt EUR 10yr-bonds at 0.4%, how is that comparable to USD Govt 10yr at 1.8%?

      We should be comparing, sovereign issuances within the same currency here!

      hisham, i'm sure you are better than this!

    3. @MIsleading

      That's why I quoted the real inflation-adjusted yield, not just the nominal yield. Technically, I should also have put in expected currency movements as well, though that might totally confuse people.

    4. Ah Hisham, let me break this down for you. Anon 3.50pm is asking if the prices will be similar (malaysia vs cad vs aussie).

      And the answer is "No way".

      I find it very misleading when you try to beat around the bush with such a statement, implying that there is no premium to malaysia's foreign currency bonds.

      I'm sure a person of your calibre can comprehend the "premium", or what we economist call, the "spread". Check out the latest/historical "spread" for Malaysia's USD benchmark bonds, and the conclusion is quite self-explanatory.

    5. @Misleading

      I'm not sure how useful that exercise would be. Malaysia's USD bonds are illiquid and form a very small portion of government debt.

      But what the hell, here's the numbers:

      Australia 2.57% 2.3% 0.27%
      Canada 1.36% 1.5% -0.14%
      Malaysia 3.95% 2.7% 1.25%

      Very different of course. But just for fun, here are the same numbers for Indonesia (Fitch rating: BBB- ) -

      7.13% 8.36% -1.23%

    6. @Misleading

      Oh BTW, thanks again. We're having a conference call with Fitch this Friday. You've given me quite a few pointers on what to grill them on.

    7. @Misleading

      I thought Hisham's posting to anon 3.50 was quite clear. No beating around the bush as to your posting though. economist who does not look at the real return. Wonder what they are teaching in school nowadays. I am sure you can do better than just throwing out jargons to win arguments...

    8. @ Hisham. Yea, good luck on that. Ask the right questions and give the right answers (u know what i mean here)... You and your bosses better ensure that you guys are not playing football in a rugby game, just like in the previous calls, and giving many ppl, including me, more headaches.

      @Anonymous @12.22. The reply wasn't meant for you. Hisham understands it.

      The real return is not what it appears as to how he explains it. There's a whole bunch of mechanics behind that assumption. For e.g. the inflation rate is a historical inflation rate while the interest rate is a current rate. who knows what the inflation rate or FX rate might be when you received your bond payment after 5 years.

      But if the risk premium is quoted under a similar currency (which is the global market convention), that eliminates the currency and inflation impact, making it comparable.

      Hisham knows this but still chose to post as such.

      @Hisham the liquidity argument does not hold...and you know that for a fact...

    9. @Misleading

      I will, though I make no guarantees I won't give you more headaches. Internally, we're shifting away from published ratings except for initial filtering. None of my doing (though I gave my input) - the distrust is institutional.

      As for the "historical" versus "current" - inflation trends, as opposed to inflation itself, are a lot more stable. What you're arguing for here is essentially that the market does not take expectations of inflation and currency movements into account at all. The realisation of those expectations are beside the point.

      As for the last point, please explain...I'd be interested in hearing your viewpoint.

  5. @Warrior

    I still have no clue where you think I made a Hayekian gaffe, but I think the laughter is getting the better of you.

    Surely you mean Moody's, not Morgan Stanley?

  6. One gets the impression that things can get heated if it comes to economics! Talk about a staid esoteric profession finally growing fangs and claws after all those years…….hahahahahaha. Thank goodness, I am just a professional bohemian, a biking hobo if you like who is blessed with oodles of time and noodles of dough (hahahaha…rhymes well doesn’t it) until the next consultancy rolls in……Alhamdullilah for life's lil mercies…..

    Pity the blogger who has to face the bluster of wrath if he does not jive with the empty blather and aimless prattle aimed at himself, when his models clearly divine otherwise or his professional principles don’t bend with to the gusts of angst and flood of vitriol flowing from axe grinders who have nothing better to do than grind personal/political/ racial axes and vent their spleen in others' personal writing spaces .

    And so it flows, stupidity like sewage cannot be legislated outta existence and curseth are those with delusional selves who cannot see their stupidity as anything else but bars of gold or nuggets of wisdom….hahahaha

    And so after that rather long appetizer let me get to the meat of the issue and cleave it to its bones.

    Apart from the US Council of Foreign Relations:

    The Europeans also have grave misgivings of CRAs

    And pray what CRAs based their quirky assessments on. These two will give us a pix into the mix:

    This is also informative

    Notice we are not talking about which stats best befits which variables or anything high maths or mindboggling Statistics here. Just simply, what are the criteria that move CRAs to issue their writs of doom and gloom or boom and vroom on whomsoever they hate or love….hahahaha.

    As I expected, I don’t see any consistency in standout criteria as is usual in the case of economists or their fellow journeymen/women working in CRAs….hahahaha.

    What I I find is Jaramillo’s robust modeling yields different outcomes to Afonso and Hill’s and vice versa .

    Warrior 231

  7. Part 2

    But I came away scratching my arse over this by Hill:

    1.S&P outlook data provide the strongest in-sample prediction performance of any agency-based rating forecast,

    2. However, since in-sample S&P outlook status has a particularly high prediction accuracy relative to Fitch and Moody’s watch and outlook data, those interested in predicting rating outcomes for Fitch and Moody’s might usefully augment watch and outlook data with a range of publicly available variables to improve prediction accuracy

    Why scratch my arse? Just amazed by the fact that the most reliable (S & P) of the unreliable three is just keeping mum….so why the itch besetting Fitch?….go figure…..

    So squabble on guys and gals, its fun to watch but go easy on the blogger and cut him a lot of slack will yah..After all you guys/gals stormed in to mislead uninvited and to vent your spleen in the din....hahahaha.

    For in the end, even those who have analysed CRAs in some detail agree tacitly on one point, CRAs are to put it mildly, inconsistent aka unreliable.

    Me?Alhamdullilah, I am in a profession that improves lives based on available data not one that massacres societies wholesale on the dint of a hunch with limited datasets at that!!

    That is why I caution everyone to read related information just to afford yourselves a better, objective and unemotional perspective when commenting:

    as well as access latest available public data:


    Morga……errr..Moody…..yeah too inebriated with laughing gas to think straight. Thanks, dude.

    Hayek gaffe, dude? Lets say it revolves around this phrase “efficient markets” you used and leave it at that for another skirmish on another day, ok… for some grass and a swig of rice wine in true hobo style......

    Warrior 231

    1. @Warrior

      Why does Blogger think you're a spammer? Hmmm...

      As for Hayek:

      1. The efficient markets hypothesis became popular some 20-30 years after Hayek's heyday.

      2. What I actually said was "The market exchange rate just adjusts to clear demand and supply." This is almost a definition of any market with a price mechanism. In no way does this imply that the price arrived at is "efficient" in Fama's sense of the term (rational expectations and full information). In fact, I would assert that the FX market is probably the least efficient of any of the capital markets.

  8. Australia & Canada have good anglo-saxon laws which are largely being observed. Malaysian inherited a good set of similar laws but over the years, due to lack of character and wise leadership, have given way to corruption, weak governance and tardiness.
    That's a world of a difference.
    Even crooks want to live in Oz and Canada.
    Malaysia? Sadly after years of independence, it was worst off than pre-independence days.