How misleading are Economic Stats?

[william the wie]

This is not the usual attack on UE or hedonics but a serious question. The stats put out by public and private bodies have a lot of problems such as not knowing for sure the error rate because of budget constraints. This tends to be a lesser problem with private sources because the customers will not continue to pay for non-actionable intelligence. And there is the secondary constraint of getting the computers needed to figure out what the collected data means when 535 members of congress have input on the purchase.

There are basically three problems with the use of private source data:

It has to be available on the market. Value Line and other sources for actionable data on stocks are not cheap and stocks are the fourth or fifth biggest capital market.

It has to be accurate. Case-Shiller was adapted after the meltdown and it has major problems such as not dealing with amenities, location and sq. footage. However it was such a huge improvement on the previous rental equivalent model (there may be other terms used in your sources) that it has become almost universally accepted. Real estate is the largest capital market.

It has to be actionable. It is relatively cheap and easy to get most types of insurance, particularly if you don't need it. Insurance/pensions may be one of the biggest capital markets out there. However it is only in the last 4-5 years that anything like comparable policies have been available for anything in insurance.

Government stats tend to be treated as a big step down from private sources. The Fed, treasury and congress all use different models to figure out what policies to implement. So here is my question, which I have also asked on other boards, how misleading are these economic stats?

 

[Toro]

I think there are a few problems, but for the most part, the stats are pretty good.

 

[william the wie]

I think there are a few problems, but for the most part, the stats are pretty good.

Good enough to cause severe injury.

The meltdown, cause bad housing stats masquerading as decent housing stats.

The Tech-wreck, cause not subtracting out one time Y2K revenues/expenditures when doing IT projections.

or as NN Taleb pointed out it usually profitable to bet on models failing.

 

[pinqy]

This is not the usual attack on UE or hedonics but a serious question. The stats put out by public and private bodies have a lot of problems such as not knowing for sure the error rate because of budget constraints.

Where did you hear that? While the error for Employment and Unemployment isn't explicitly published, the parameters are available to do the calculations. I have an Excel spreadsheet set up to calculate the standard errors and different levels of confidence for Employment and Unemployment. I'm pretty sure the errors and error calculations are available for all gov't stats.


ALL statistics can be misleading for people who have not studied statistics. Many terms, such as Employed, Unemployed are used in a technical sense, while people understand them in a colloquial sense. There are not many in the general public who have studied stats and economics and understand what's being measured, why, and the difficulties and problems with collection and analysis. Hell, some people talk as though BLS has a giant list of everyone in the country and classifies each individual every month. And it seems counterintuitive to many when both employment and unemployment go up or both go down..."How can the UE rate go up when we've gained jobs????"

And prices....the concept of average weighted price changes seems lost on many. A person will see that their grocery bill has gone up, guess at the percentage (I doubt many if any actually do the hard math), don't take into account any changes in purchasing, just noticing that their bill has gone up "a lot" and then see that the CPI only went up 1%. Somehow they think inflation should exactly reflect their indivdual experience.

So in short (too late), the stats ARE misleading to the general public, but not to most of the professionals who actually deal with them regularly.

 

[editec]

Even if the STATS are NOT flawed, the problem is that STATS seldom tell you the whole story.

A single swallow does not a summer make.

 

[william the wie]

Even if the STATS are NOT flawed, the problem is that STATS seldom tell you the whole story.

A single swallow does not a summer make.

Also label changing, a continuous DC process, violates all standards of recordkeeping in general and accounting in specifics.

Then there is compounding error as with Pinqy's battles over UE. The different surveys public and private imply an unacceptable error rate by not tracking closely enough to get anyone to believe that they reflect economic reality with the public surveys in particular highly suspect because they don't even agree with each other. I suspect that this is a budgetary problem but it provides continuous data for the CT crowd.

 

[8537]

Even if the STATS are NOT flawed, the problem is that STATS seldom tell you the whole story.

A single swallow does not a summer make.

Also label changing, a continuous DC process, violates all standards of recordkeeping in general and accounting in specifics.

Then there is compounding error as with Pinqy's battles over UE. The different surveys public and private imply an unacceptable error rate by not tracking closely enough to get anyone to believe that they reflect economic reality with the public surveys in particular highly suspect because they don't even agree with each other. I suspect that this is a budgetary problem but it provides continuous data for the CT crowd.

Which public surveys don't agree with each other?

 

[rikules]

This is not the usual attack on UE or hedonics but a serious question. The stats put out by public and private bodies have a lot of problems such as not knowing for sure the error rate because of budget constraints. This tends to be a lesser problem with private sources because the customers will not continue to pay for non-actionable intelligence. And there is the secondary constraint of getting the computers needed to figure out what the collected data means when 535 members of congress have input on the purchase.

There are basically three problems with the use of private source data:

It has to be available on the market. Value Line and other sources for actionable data on stocks are not cheap and stocks are the fourth or fifth biggest capital market.

It has to be accurate. Case-Shiller was adapted after the meltdown and it has major problems such as not dealing with amenities, location and sq. footage. However it was such a huge improvement on the previous rental equivalent model (there may be other terms used in your sources) that it has become almost universally accepted. Real estate is the largest capital market.

It has to be actionable. It is relatively cheap and easy to get most types of insurance, particularly if you don't need it. Insurance/pensions may be one of the biggest capital markets out there. However it is only in the last 4-5 years that anything like comparable policies have been available for anything in insurance.

Government stats tend to be treated as a big step down from private sources. The Fed, treasury and congress all use different models to figure out what policies to implement. So here is my question, which I have also asked on other boards, how misleading are these economic stats?

stats can be manipulated to present a lie as the truth

or just to present the image you want no matter how incorrect that image is

for example;

lets say 1% of the population has 99% of the wealth

the ruling class might not want you to know that

so

instead of telling you "1% has 99%"

they give you numbers that you are more comfortably with...

like...."the top 10% of the population has 50% of the wealth"

BOTH of these statements might be true but the first one might piss you off where-as the second one doesn't seem so bad....

 

[pinqy]

The different surveys public and private imply an unacceptable error rate by not tracking closely enough to get anyone to believe that they reflect economic reality with the public surveys in particular highly suspect because they don't even agree with each other. I suspect that this is a budgetary problem but it provides continuous data for the CT crowd.

But they shouldn't agree precisely...different surveys measure different things in different ways. General trends do tend to track, and the Gallup UE rate and the BLS not seasonally adjusted rate look to me like they're within each other's margins of error, even though the methodolgies are radically different.

The ADP survey is only of their own customers, so while large, is not really representative of all businesses, so BLS and ADP are off and on as far as correlation.

Which particular surveys were you thinking of?

 

[uscitizen]

Even if the STATS are NOT flawed, the problem is that STATS seldom tell you the whole story.

A single swallow does not a summer make.

Yep anyone who has worked in mid level management or above in a large corporation knows how statistics are "massaged" to show what you want them to.
another thing about stats is that they are ALWAYS past tense.

 

[boedicca]

Some stats can be misleading. They are spun and used to fool the uninformed. Reading the fuller reports is the best way to "get under the hood" and understand the actual meaning.

 

[william the wie]

Some stats can be misleading. They are spun and used to fool the uninformed. Reading the fuller reports is the best way to "get under the hood" and understand the actual meaning.

However that may not do much good at all. Lew Rockwell had a piece on hedonics and the impossibility of computing value and productivity of Youtube, Wikipedia and Google since they are all free to the user and the advertiser pays according to the click. But if I buy "Charmin" due to a video ad on a website I use, the website gets no click and I pay nothing for visiting the website. Trying to figure out the impact of that transaction is effectively impossible.

Also almost all of the tools for statistics are based on normal distribution only Pareto demonstrated in the 1920s that most economic data is not normally distributed. I cannot think of any operations formulae that use normal distribution although surely there are some. The Black and Sholes model of market pricing is based on normal distribution and while there is a great deal of debate about what distribution model best describes the market normal distribution is not a major candidate. What is being spewed out by the government often resembles poker stats at a bridge table, that is dangerous.

 

[iamwhatiseem]

It depends on who/where produces the stats.
As well as what other statistics and context are applied to the numbers.

In other words, a total flip of the coin.
Any analyst worth his weight in dust can produce virtually any outcome regardless of reality.

 

[william the wie]

It depends on who/where produces the stats.
As well as what other statistics and context are applied to the numbers.

In other words, a total flip of the coin.
Any analyst worth his weight in dust can produce virtually any outcome regardless of reality.

Well it depends on how transparent the process is and the US Government is generally good about this, but also budget and whether what is being measured is useful. Virtually all of the arguments about stats on this board boil down to the utility of what is being measured. The question of how to count a small businessman who goes out of business or is going out of business, discouraged workers and so on cause flame wars almost every time the subject of unemployment comes up. That BLS does not publicly track the self-employed or the availability of work in different job markets vs. the percentage of discouraged workers at a minimum smacks of spin and creates the image of deliberate deception.

 

[8537]

Some stats can be misleading. They are spun and used to fool the uninformed. Reading the fuller reports is the best way to "get under the hood" and understand the actual meaning.

However that may not do much good at all. Lew Rockwell had a piece on hedonics and the impossibility of computing value and productivity of Youtube, Wikipedia and Google since they are all free to the user and the advertiser pays according to the click. But if I buy "Charmin" due to a video ad on a website I use, the website gets no click and I pay nothing for visiting the website. Trying to figure out the impact of that transaction is effectively impossible.

I think Lew got the transaction backwards. If something - Google, Youtube, Facebook etc... - is free, then the person using it is the product, not the consumer. The consumer is the company purchasing the information about you, the product. It is quite simple to determine the value of that product.

 

[pinqy]

Some stats can be misleading. They are spun and used to fool the uninformed. Reading the fuller reports is the best way to "get under the hood" and understand the actual meaning.

However that may not do much good at all. Lew Rockwell had a piece on hedonics and the impossibility of computing value and productivity of Youtube, Wikipedia and Google since they are all free to the user and the advertiser pays according to the click. But if I buy "Charmin" due to a video ad on a website I use, the website gets no click and I pay nothing for visiting the website. Trying to figure out the impact of that transaction is effectively impossible.

Hedonics aren't used for all goods, and free goods aren't priced at all.

BLS does not publicly track the self-employed

Yes, they do. Tables A-8 and A-9 of the Employment Situation covers self-employed, and Table A-14 covers the unemployed self employed.
They're not counted in the non-farm payroll survey for a few reasons, but they are tracked in the household survey. Many many small business are included in the payroll survey, though.

the availability of work in different job markets vs. the percentage of discouraged workers

There's almost no way to do that. With only a million or so discouraged workers in the entire country, there's no way to break that down by job market. The margins of error would be outrageous.

 

[Toro]

I think there are a few problems, but for the most part, the stats are pretty good.

Good enough to cause severe injury.

The meltdown, cause bad housing stats masquerading as decent housing stats.

The Tech-wreck, cause not subtracting out one time Y2K revenues/expenditures when doing IT projections.

or as NN Taleb pointed out it usually profitable to bet on models failing.

It wasn't housing stats per se that contributed to the problem. It was how housing inflation is calculated in the CPI which arguably fueled the housing bubble, as inflation looked subdued, keeping the Fed funds rate lower than it should have been.

Statistics played no part in the Tech Bubble.

 

[uscitizen]

Ina then generally the inflations figures leave out energy and food costs. 2 of our most expensive items that everyone must have.

Ohh they provede them with food and energy included but that is not the stats that are promoted to the public.

 

[pinqy]

Ina then generally the inflations figures leave out energy and food costs. 2 of our most expensive items that everyone must have.

Ohh they provede them with food and energy included but that is not the stats that are promoted to the public.

Can you show an example of that? Every news report I've ever seen quotes the all items index as the headline number and only mentions "less food and energy" by referring to that as core CPI.

 

[psikeyhackr]

The vast majority of information spread about anything is bad or worthless. Be it economics, computers, stereo equipment or cars.

Most of the so called information is so unimportant it is not worth paying attention to.

Most people are supposed to be losers in the economic game and kept running on the treadmill. Why does no one suggest mandatory accounting in the schools. Why does almost everyone get 4 years of English literature in high school. What good does that do anybody except the people who become English Literature teachers?

How can there be growth in the automobile industry without the planned obsolescence of automobiles? Then the economists don't talk about the depreciation of all of those cars.

The statistics have to be distorted bullshit but you have to be smart enough to figure out how they are distorted and then they can be useful. The smart people are hiding the important information from the dumb people and spreading disinformation.

The Economic Wargame is a continuation of the Military Wargame by other means.

All warfare is based on DECEPTION. - Sun Tzu

psik