Monday, February 4, 2013

"Don't believe any statistics which you haven't forged yourself!"

People who have management experience in large corporations know what beautiful things one can do with statistics. In the extreme case, it can lead to a situation where all divisions report good profts while the company as a whole reports a loss.

Much excitement is currently under way in connection with the statistical reporting of the budget deficit back in 2009. That is undoubtedly an interesting case for experts.

The more ordinary people would probably prefer to just have a simple set of statistics; meaningful statistics which tell the story correctly. That seems to be impossible because different institutions have different ways of putting numbers together. None of them are probably wrong but the ordinary observer doesn't know what to make of them.

Below are the export/import statistics of Greece for 2011, as reported by the Bank of Greece, the Troika, ELSTAT and Eurostat. The comparison speaks for itself.



2011 (in BEUR)






BofG Troika ELSTAT Eurostat





Exports 20,3 26,3 52,2 56,6
Imports 47,5 55,5 69,1 77,2

------ ------ ------ ------
Balance -27,2 -29,2 -16,9 -20,6


This is obviously not a matter of 'cheating' but, instead, a matter of different reporting methods. It seems that the Bank of Greece and the Troika follow relatively similar methods, and so do ELSTAT and Eurostat. Personally, I rely on the statistics of the Bank of Greece because they have always been very consistent and credible answers to questions could always be easily obtained.

The question, however, is: which of the above statistics represent reality the best and why is there such a huge difference between them? Below is what my contact at the Bank of Greece had to say about that: 


This is everybody’s usual complaint. The reasons for the discrepancies between the two sources are:

1. Differences in timing. The ELSTAT figures are custom (or VAT) based whereas the Bank of Greece (BOG) ones are settlements based. This means that the former have a 3 to 6 months lead over the latter.
2. Differences in valuation. There are cases e. g. in which the BOG data translate exports or imports denominated in dollar terms as reflecting transactions to and from the US which is wrong especially when it comes to oil imports.
3. Differences in coverage. There are e. g. different views between the two sources regarding thresholds beyond which transactions are reported.

These differences, however, are more pronounced on a monthly basis and smoothen out when it comes to annual data. You must also bear in mind that the ELSTAT figures rely on the philosophy of “garbage in garbage out” without inquiring as to the reliability or reasoning of the reported transaction. Inevitably, therefore, they revise their data on a monthly basis and the picture often changes considerably from one month to another.

In short, rely on the BOG data. You know there is a Greek saying that goes, ”If you do not praise your own house the roof is bound to collapse on your head”!!!!!! 


All clear???

6 comments:

  1. "Don't believe any statistics which you haven't forged yourself!"
    In fact you make a statement here even to us not to believe you as well, even when you are here with the most wonderful statistics.
    A not wanted contradictio in terminis?

    And because I knów what you say, I do not believe them, I do not believe anything else than what is reality, what is proved to be true. Words are just words, and numbers are just numbers. Brains are just brains and all brains that can help to write or count are human. With limitations.

    I prefer the believe a lioness (lions do not chase) when she has calculated her speed, the distance and the moment to catch the deer. Without any failure. Human beings cannot anymore. They cause accidents in traffic because they cannot calculate, not think. Though they think they can calculate, like O'Leary (Ryanair). Statistics are everywhere, and those who use them know how to manipulate, with numbers, creating confusion, blending the one with the other thing, so that finally people are so completely brainwashed that they cannot think clear anymore, and quit.
    If interested:
    Mayday Mayday (English subtitled)

    And:
    ”If you do not praise your own house the roof is bound to collapse on your head”!!!!!! "
    My comment on that: even when you praise your own house the roof will fall down anyway if you do not do anything else than praising, because words do not keep your house in a good condition, do not maintain it. You have to use your head, eyes, hands, arms, legs and feet, and work, to keep it well maintained. The mouth (to hail and praise) is not important here. The roof is honest. Sometimes you can hear a house screaming, for paint, for instance. But people do not listen, often. When it is rotten, finally, we cannot blame the house.

    I found a nice sentence when I tried to find the right translation for "in een goede staat van onderhoud verkeren" (this is Dutch, my English is not so good and search for the right words on internet)
    "Ships may be very old, yet if they are well maintained and have a good crew, they are highly seaworthy."

    My own contribution here:
    "Greece is like a ship, and may be very old, yet if it is well maintained and has a good crew, it is seaworthy."




    ReplyDelete
  2. There is always a problem with statistics. It's data.

    It gets worse - what do people think data is, and what qualifies it to be data in the first place. That might sound philosophical, the reasoning behind it is very practical.

    Getting proper data is an art in itself, and people like me are expert at cutting through cr**p. That's because all data is made by humans - or is witnessed by a human.

    Remember this because it is important: all data has a human element.

    So don't look at the data. Look at the humans behind it, that are creating it or using it. That is when you see very different shapes forming. Things that uncover the reality of the figures. It tells you if they are worth anything at all!

    Which brings me to another thought. The Tax-offices use this too which must mean it has some meaning. A Benford analysis of the first digits. It is a powerful reverse-engineering of decade growth. Because there are always more 1s than 2s than 3s wherever you find numbers that express growth.

    If however there are more 4s than 3s - it means that something hasn't occurred in the manner of decade growth. That is to say, one number has not occurred in sequence. The only way this can happen is if somebody has done something to the numbers. What's more, changing one digit on a spreadsheet can set of consequential errors that fall like a track of dominoes. All over the place.

    No honest person tampers with figures. Do I make my point?

    That's why the Belastingdienst use Benford's Analysis.

    ReplyDelete
  3. Sorry I clicked Publish too soon. I created that graph at the TradingEconomics site. I prefer graphs to tables, particularly for looking at trends.

    For Greece TradeEcon gets Balance of Trade from the BoG which is atypical, it uses "Statistics Office" sources for US, Ger, UK, Aus, Sng and Chn. TE does use ELSTAT for other numbers.

    Seem odd that the BoG should 'book' all imports priced in US$ to USA, you'd think it would know about such things.

    CK

    ReplyDelete
  4. I did my own sleuthing in BoG. My contacts pretty much confirm the above. However they also claimed that the ELSTAT data are better from 2010 onwards, that BoG internally is using ELSTAT and that they are trying to reconcile the data methodologies to ensure consistency (even try to convert the before the 2010 data to the new format) but it turns out to be pretty difficult. I guess we must wait and see how it develops.

    ReplyDelete
    Replies
    1. ...meanwhile, it would be good to know what the trade deficit really is (at least 1 BEUR +/-) because that amount, net of the services balance, is the amount which the Greek banking sector needs to borrow abroad...

      Delete