Sunday, March 17, 2013

Self-Censorship & The Brilliant Independence of Brokers' Research

I have yet to come across a top tier broker research that truly examines ALL the potential outcomes of the upcoming election and the implications. Yes, we had the research from Bank Islam and see what happened. Nomura Research only presented 3 possible scenarios. THREE, nothing more.

Even the dumbest person in the room would know that that is not the entire scenarios available or the available outcomes. Even in my "biased piece below", at least I have the decency to look at all the possibilities.


Nomura has painted three potential election scenarios and their impacts on the economy.
The first scenario is that the ruling coalition will win a smaller majority of 120-124 seats out of 222. BN currently holds 137 seats. The second scenario is that the number of seats won by the opposition and the ruling coalition are marginally less than 120 seats, which would likely be an accelerated ousting of the prime minister. Nomura said within this 20% probability, there is a small chance of the election leading to a hung parliament. The final scenario is that the ruling coalition winning around 125 to 130 seats. But Nomura said this outcome looks unlikely unless there is an unexpected fall in support for the opposition.

Three fucking scenarios ONLY??? Its a fucking coin toss, the last elections popular vote is like 49-51, and you fucking give only 3 scenarios??? Why don't you just say it out loud ... that you cannot make any comment on the other 2 possibilities, no matter HOW REMOTE they may be, only then can you say it is from a fucking research house!!! (the other two possibilities: PR wins by small majority, and PR wins by more than 10 seats majority). Why the self censorship, is there something you cannot say, ... its like predicting the recent US elections and you only give Obama winning by less than 2%, 5% or 10% ... asif the Republicans never existed. I know your hands are tied and you depend on the flows of IB deals but seriously, if you cannot comment or choose not to comment on the other side of the coin .... then don't fucking print the research idiots. It makes a mockery of the so call research piece.

Another one, just received from Morgan Stanley, the big US house, from the land of the purveyor and global sheriff of the maintenance and prevalence of democracy ... they also gave 3 fucking scenarios ONLY ... sigh, long live the USA ... OR come out with a note that says that you think there is a ZERO chance of Pakatan winning the elections and see how many legs you have to stand on.

I am sure we all understand the self censorship part, I would understand but not condone it, if it came from a local house ... but from a foreign house it smacks of something smelly.

MORGAN STANLEY

#3: Election Scenarios and macro implications 
 Scenario 1: 
BN Parliamentary seat share > 63% 
Positive surprise for investors
 Scenario 2: 
High 50% < BN Parliamentary seat share < 63%
 Scenario 3: 
BN Parliamentary seat share < mid 50% 
(To be fair to Morgan Stanley, they did a decent policy comparison between the two parties without really saying too much one way or the other. The snaps attached below are from MS).


The Edge/Sun Daily: Nomura Economics Research said the upcoming general election (GE), which is likely to be held on a weekend between April 6 and 20, 2013, could make or break its relatively positive economic outlook on Malaysia.
"Our baseline scenario is for the ruling Barisan Nasional (BN) to win, but by a smaller majority of 120-124 seats, which is lower than what we had previously penciled in, partly because recent surveys show an emboldened and well-organised opposition," said Nomura in its "Asia Special Report: Southeast Asia" dated March 6, 2013.
"This raises uncertainty over whether there would be an orderly transition of power, whether Prime Minister Datuk Seri Najib Abdul Razak remains in power and whether the much-needed economic reforms can continue," it added.
Nomura has painted three potential election scenarios and their impacts on the economy.
The first scenario is that the ruling coalition will win a smaller majority of 120-124 seats out of 222. BN currently holds 137 seats.
"Although there is scope for some relief on this baseline result because a small minority of the market sees the opposition actually winning the election, we believe this will be short-lived owing to the immediate implications of a small majority.
"A simple majority in parliament calling a no-confidence vote could emerge when parliament convenes for its first session 60 days after the election. This would raise political and economic uncertainty," said Nomura.
The second scenario is that the number of seats won by the opposition and the ruling coalition are marginally less than 120 seats, which would likely be an accelerated ousting of the prime minister.
Nomura said within this 20% probability, there is a small chance of the election leading to a hung parliament.
"The risk of such an outcome includes policy paralysis, potential protests, calls for a recount or even an annulment of the election results, which can be called within three weeks of the election announcement," said Nomura.
Risk of a no-confidence vote from the opposition on the prime minister will also be even greater, while internal party pressure could also lead to a change in Umno leadership.
"Currently, Deputy Prime Minister Tan Sri Muhyiddin Yassin is seen as a potential replacement for Najib and viewed by the market as somewhat hard-line. This change could emerge earlier than the planned October/November 2013 party elections.
"With the negative political backdrop and economic risks, as well as possible delays in investment projects, these could lead to a sovereign rating downgrade," the research firm said.

The final scenario is that the ruling coalition winning around 125 to 130 seats.
But Nomura said this outcome looks unlikely unless there is an unexpected fall in support for the opposition.
"(However, if it does happen,) the pressure on Najib to step down on this outcome would be significantly reduced and economic policy continuity will likely be maintained."
Nomura believes that increased political and economic concerns if the opposition gains power could be overstated as there may not actually be any major policy changes.
"The opposition has a common objective and is likely to initially focus on governance issues (which could lead to slower government investment), keeping key government staff in place such as the central bank governor and heads of key state companies, consolidate the civil service, before focusing on the direction of economic policy," it said.




hishamh said...
Dali,

I've looked at the problem myself, and for the life of me, I can't figure out how anybody can make an objective forecast of GE13, much less tease out the probability distribution of outcomes.

We are nowhere near being able to replicate basic electoral prediction methodologies, much less aggregate them as e.g. Nate Silver did recently for the US presidential election.

Problem 1: The overall Malaysian electoral sample size (across time) is too small, both for votes and seats.

Problem 2: The sample size for the predictors normally used (opinion polls, quarterly economic data) is even smaller - small enough that a regression estimate can't be generated, which is a precondition for estimating the probability distribution of outcomes.

Problem 3: GE12 may represent a structural break from the past, but that can't be determined statistically until confirmed by results of GE13.

A time series analytical approach can handle the first two statistical issues, and predicts a BN victory with well above a two thirds majority (point estimate) but with a sample error so large as to make any forecast worthless (I suspect this is due to problem 3).

I don't understand how Nomura or MS or BIMB can predict such tight probability outcomes given these constraints. There's nothing in the scenario analysis methodology that allows you to estimate the probability distribution. Scenario analysis is more of an ...if...then... decision tool, not a forecasting methodology per se.

In short, i don't think self censorship has much to do with this, rather everyone's just pissing in the dark. You're absolutely right - nobody should be publishing research on this, because its too damaging to their credibility.


No comments:

Post a Comment