The following post is by a hedge fund manager friend of DB who shall remain nameless. He runs the emerging markets desk at his firm.
The man who first hired me to work in emerging markets investing used to claim that understanding the idiosyncrasies of these markets was entirely a matter of understanding corruption. Predicting policy choices, he argued, was a matter of figuring out which domestic interest would pay the highest bribes. Capital flight? Just a question of how corrupt actors were feeling about the security of their ill-gotten gains. At the time, my youthful enthusiasm resisted accepting wholeheartedly such a cynical conclusion. It wouldn’t be unfair to say that a good deal of my many years of practical education since then has consisted of coming to a full understanding of my old boss’s dictum. It turns out that countries run by crooks tend to blow up. Go figure.
Today marked the release of Transparency International’s annual Corruption Perceptions Index. The index isn’t a perfect measure, as corruption is a labile concept, but TI’s methodology seems well thought-out, incorporating data derived from resident and non-resident country experts as well as business leaders. Readers can get further details on the index here.
The development of the sovereign CDS market means that more countries than ever before have observable credit spreads. As a result, it is possible to look at the relationship between CPI score and credit spreads for a large sample. Of the 180 countries evaluated by TI, there is CDS spread data available for 64. Take a look at the results: corruption matters.*
That one can very clearly illustrate the fiscal cost (via higher borrowing costs) of corruption would suggest one motive – though hardly the only or most obvious one – for a country to clean up its act. Why hasn’t it happened already? Consider my old boss’s dictum. Corruption makes sense for the corrupt. The burden of higher borrowing costs falls on the great mass of citizens. The fruits of corruption accrue to… the corrupt!
Scrutiny of the outliers seems to support the basic story. The biggest underperformers of their CPI score (i.e. higher credit spreads than expected) fall into two groups: the very highest-spread countries (foremost Argentina and Ukraine) and a couple of low-spread countries (Iceland and Latvia). The former group’s outlier status appears to arise from the fact that even using log of spreads and a quadratic term doesn’t fully capture the non-linearity of the relationship. The latter were prominent victims of the credit bubble, whose very tight sovereign credit spreads have blown out due to the actual or contingent sovereign liability for the banking system’s losses. The biggest outperformers of their CPI score are actually not outperformers by much – residuals here are much smaller – with Philippines and China at the top of the list. Perhaps DB’s commenters would care to tender some possible explanations for these two…
married to da mob.
-cg
great read
r sq of 51 – why bother?
This rocked my world, so you’re saying the more corrupt a country is the higher its cost to borrow? My quant models had this all wrong…summa bitch.
–AIG quant
@2
Get off his nuts, dickrider
I agree with #3. You’re trying to fit too hard.
Maybe the 5 yr spreads are not the way to go. Maybe corruption has a better correlation with short term CDS quotes?
@5 I’ll ride whatever dick I want, thank you very much.
- barney f
Similar concept to how a company with good corp. governance practices is generally more valuable/outperformns…except applied to countries.
This post has motivated me to not only short corruption in the financial markets but in all aspects of life. As such, I’ve traded Chad Johnson off my team.
AQR quant here, what does “segfault” mean?
-@6
actually, bet its even lower. both variables are too noisy and have too much ‘other’ stuff in them.
this is lazy thinking stuff done by stats illiterates.
@3 and @6
0.5 R^2 is good: it’s just telling you that there are a bunch of omitted variables that would presumably be included if this was a paper, not a quick post on a blog.
As for the part of the CDS curve to look at, it’s the 10y I would like to see used for the analysis (or better yet, the 10y5y slope), not the short end. Corruption is something that screws you over in the long term – it does not make you jump to default next year.
-@11
ah, indeed, unfrtunately “..it’s just telling you that there are a bunch of omitted variables ..” implies we cant really do much with what we have here.
the long end of the cds curve is typically made up, and not updated frequently enough – not really useful for real stat analysis.
@11 makes an excellent point. Merely loan money to corrupt folk and make sure they “pay you back” with their commodities. Oh, Pinky and I used to launch a thousand companies when we’d do that. After he had his heart attack in 1992 I was left to my own devices and the help of my “Lehrlings”, bless their hearts. In a house full of Parisien hookers I am “El Matador”…we called Pinky “The Admiral” but for different reasons than that of the young oil trader on Fox News from time to time.
~Zaftig in Zug
@7 – Do not shit in my mouth and tell me its ovaltine, asshole.
5
@3, 6 & 12, seems pretty highly explanatory in this context especially considering CDS only insures actual default. Plus profit is the area under the curve, not the frequency of positive returns.
@AHFM Chinese growth and assets? Philippine capital controls?
Post more tits.
CG
@8
After last week’s miserable performance, 85 is on the verge of getting 86′d on my squad as well, however, I’m long douchebaggery so we’ll see how that contrarian trade plays…
-@11
0.5 R^2 is good? To even get 0.5 he had to use a log of the values and a polynomial fit. I think this is more of a wishful thinking exercise in Excel than an observable trend.
I don’t have the CDS values on hand, but my guess is that short term would be a better fit simply because someone buying CDS to cover short term debt that is due by a corrupt government would be more worried. This is just total speculation.
@18 it’s called math.
@18
I take your points. On using log of spreads, that’s standard for statistical analyses of credit spreads. Think about how credit spreads map onto probabilities of default and its pretty clear why one would do this. As for the polynomial fit, it doesn’t really improve the R^2 much relative to a linear fit. .519 vs .508, so not really that meaningful.
The bigger potential problem I see is that since the CPI is survey-based, it’s possible that the contributing “country experts” have their corruption perceptions colored by credit spreads. I assume many of the respondents are market-aware.
Just start a list: Russia, all the arab/persian/packi/indian/east euro/ former USSR colonies/ All African countries/south america, mostly/Zug/ and the bezzle is around 12% everyplace and this goes on and on. I should know.
@12 & @19
*Look* at the scatterplot. It’s not an observable trend? The points are distributed randomly? Take any bivariate model of CDS spreads of 64 craptastic EM countries – vs. debt, vs. current account deficit, whatever – and you’re not going to get R^2 >> 0.5. You’d know that if you had ever tried it.
Perhaps Chinese outperformance reflects a crowded trade by external investors and Philippine outperformance reflects a thin market with a strong local bid.
Given that corruption is rarely transient, has the outperformance been persistent?
@22 ?
-19
23 came to work today.
@24/19
Think he meant 18
fuck that. russia’s a berry. can’t wait to get funding.
-JM Advisors
@mc
I’m just saying that a fit of 0.5 isn’t something that I’d build a model around.
@AHFM
The problem with corruption is, obviously, the lack of transparency in the system. Things could work smoothly, or the president for life could decide that next week he’s not paying back the IMF. I think it’s hard to say that CDS for EM has the same information baked into it that you’d see for non-EM countries, and makes it a suspect measure. Not that I’m saying it’s not useful, but that it’s not the most informative.
Was I the only one who read, “full understanding of my old boss’s dictum” as “full understanding of my boss’s rectum?”
not sure it makes sense to imply some cause/effect relationship there…
please stop saying he “runs the desk” and say what the person does for a living. the office manager runs desks.
@31 you’re a douche. it’s ben heller, and he is the pm on the emerging markets desk. so yeah, he runs it.
Is the guy who hired him Simon Treacher? I know he used to come up with these things and he definitely was very close to the local policy makers to know how corrupt they were. Wouldn’t be surprised if our anonymous source hence works at BlueBay. Anybody else got any views?
Amazingly, the upper left quadrant is dominated by 4 datapoints that are not countries at all, rather DC, Chicagoland, 85 Broad, and Charlotte respectively.
@34 FTW!
this article states the obvious. more corruption means more market inefficiencies meaning more costs to the economy. if the writer is Ben H. at HBK, he should get back to trading and cut back on all these postings.
@36 I disagree on both points.
A “corrupt” yet stable regime is more efficient and cheaper to work with, especially on long-term, large-scale projects that need financing. Political instability adds cost, just as an open and transparent process (with its attendant bureaucracy) adds cost.
AHFM benefits from floating his ideas here without having to own them.
to InfiniteGuest@37
we are not disagreeing. “corrupt” and “stable” are not mutually exclusive. i agree that a corrupt but stable regime will usually beat out a less corrupt but volatile regime. however,in general, a corrupt regime will be less efficient.