Fantastic post about "why macro statistics are COMPLETELY meaningless"

[bushi]

"Figures Lie, Liars Figure, the Math of Mainstream Economists"

...discuss, please - that was my long time feeling, that statistics are just that, and they do NOT picture the Reality in any meaningful way. This above post is a concise, brilliant, thoughtful reasoning, that punches very precise holes in that whole house of cards, and depicts the absurdity of worshiping some macro-numbers - and the absurdity-squared thing, TARGETING these very same abstract, meaningless numbers. It simply DESTROYS it, that whole nonsense "economy by statistics" thing.

Excellent write up, and something that everyone should read, to immune himself from BS of macro "hard economic data". No BS taken or given, no conspiracy theory, just pure, iron-clad reasoning.

I love it.

 

[ancona]

In my work, I frequently compile statistics as regards various species of molds, yeasts and rusts. We perform a geometric mean for each season of the year to determine what the average concentration of spores might be expected to be at any given time and under which set of circumstances. The Spearman profile is the best for this as it throws out the one or the two as statistically irrelevant, leaving only those species that are truly represented in meaningful numbers.

Since mold spore concentrations are constantly in flux due to the vagaries of weather, we create this documentation to understand the trends rather than the norms. A single sample, or even a set of five or six samples are nothing ore than a snapshot of spore populations at that particular moment in time, meaning that the various factors surrounding their presence can change from hour to hour, radically altering the concentrations of spores of various species, since each taxa reacts somewhat differently to the diurnal cycle.

When I read statistics that are given for various financial and economic cycles or timelines, I always have a little chuckle because I know from direct experience that I can make a set of data say whatever I wish it to.

Depending on how data sets are weighted, and depending upon which data sets you use, it is quite possible to cancel out one set with another set.

I take all statistics with the grain of salt that came with them. ; - )

 

[benjamen]

Statistics, finance, and economics can provide very useful results. The problem can usually be found if you follow the money. If you are paying me to forecast something, but you really really really want a lower number... guess what, I will give you my lowest estimate if I want your business in the future.

A good modeler/forecaster will use at least 3-5 different methods to estimate what ever they are looking for and make a final pick somewhere within the range of those answers based on experience, judgement, ect. However, if the person with the money really want that low number... chances are the final answer will amazingly be equal to the lowest model output.

 

[bushi]

A good modeler/forecaster (...) and make a final pick somewhere within the range of those answers based on experience, judgement, ect.

EXACTLY, so there it is, where the math is put aside, and common sense, experience & one's wisdom takes the lead.

Same goes for the assumptions made, as part of the inputs to our govt macro statistics - they are completely arbitrary, and they are strongly correlated to a broad economic picture of the country - like "workforce" definition.

So these simply CANNOT serve as inputs to measure the Economy in any meaningful way, if they are dependent on the Economy's performance, benefit's system in place and gfadzillions of other things (all related and dependant to the thing being measured, ie the Economy) themselves.

And yes, I do agree that statistics in general can and is very useful - just not for portraying any kind of macro economic picture. Let alone "targeting" some of these "objective, hard" numbers. Because in fact, they are both very subjective, and also soft as a soft poo.

 

[pmbug]

There is no spoon. /Matrix

You can alter reality (ie. macro stats) by changing your measuring stick (how stats are computed).

 

[benjamen]

EXACTLY, so there it is, where the math is put aside, and common sense, experience & one's wisdom takes the lead.

Same goes for the assumptions made, as part of the inputs to our govt macro statistics - they are completely arbitrary, and they are strongly correlated to a broad economic picture of the country - like "workforce" definition.

So these simply CANNOT serve as inputs to measure the Economy in any meaningful way, if they are dependent on the Economy's performance, benefit's system in place and gfadzillions of other things (all related and dependant to the thing being measured, ie the Economy) themselves.

And yes, I do agree that statistics in general can and is very useful - just not for portraying any kind of macro economic picture. Let alone "targeting" some of these "objective, hard" numbers. Because in fact, they are both very subjective, and also soft as a soft poo.

The two biggest things that really determine how well a model works are your assumptions and your data. If either are incorrect then even the best model is completely useless. The more you learn about statistics the more the polls, official numbers, ect make you laugh. Unfortunately, most people do not have such a background and have no idea how misleading a lot of these numbers really are.

 

[DCFusor]

Yes - for a couple examples.
"Banks are trading under book value." Really? Since the FASB ruling they can mark to unicorn, how do we know book value is even a positive number? If if was, at least one major bank would be hollering with every buck they have for advertising and lobbying to remove that rule and bring back mark to market, yet...crickets.

Another "interest rates on risk free (UST) predict - whatever they used to about the market.
Not when there's a very deep pocket manipulating interest rates downward, explicitly and publicly. Those relationships might have been true in a normal market, but...

This is why I don't trust long term models. There are some things different this time. Human nature never is, but how about the examples above?

An awful lot of terrible decisions are made based on those two alone, and if you're not paying attention, you can get sucked into them. If you are, it's a great time to be trading - the other guy is often wrong due to these. For once, the other guy at the table is the mark.

 

[Jay]

How to Lie with Statistics:
[ame="http://www.amazon.com/How-Lie-Statistics-Darrell-Huff/dp/0393310728"]How to Lie with Statistics: Darrell Huff,Irving Geis: 9780393310726: Amazon.com: Books@@AMEPARAM@@http://ecx.images-amazon.com/images/I/51UT7gYNn6L.@@AMEPARAM@@51UT7gYNn6L[/ame]

 

[DCFusor]

What's worse is that after their "goal seeking", eg cooking the books, they then believe their own, known-stupid numbers, and act as if they were really true. Classic circular cluster-fsck.

 

[bushi]

YES exactly, it is like in that example from the article - I move my neighbour's lawn, he pay me a hundred bucks. Then he moves my lawn, I pay him a hundred bucks. We just have contributed $200 to the GDP growth, but is ANYONE better off, regardless increased GDP? Is anyone wealthier, because of that GDP growth? Or is this increase just a number, that doesn't represent anything meaningful, after all. The answer is pretty clear. We both had to do the same work as before, and we both have the same amount of money to spend on other things after money changes hands.

Another example, even more drastic would be, if government decided to hire one million people to digg holes, and another million to fill them back in. Assuming they were paid 50k a head, hey, it is 100 billion instant GDP increase ;-)


Sent from my GT-I9100 using Tapatalk 2

 

[DSAbug]

Numbers are bullshit..