The Most Important Stat: Bloggers versus Experts

Market pundits always try to explain modest changes in the broad averages.  Today their stories are all wrong.

Today, the most important statistic came from Jack Hough at Smart Money

The Story

Hough did a piece on the credibility of government statistics.  It is a nice job compared to most prior discussions.  He provides a viewpoint on each of the key issues, citing arguments from the BLS and the principal conspiracy theorist.

It would be interesting for investors to read the entire article and also to read the supporting documents from the conspiracy guy and the BLS.

Almost no one will do this, since they are too lazy to do any real work.  We are now addressing to the handful of people still thinking and reading.

Of the few who try, most will not be able to evaluate the arguments, since it requires some understanding of data analysis to reach a real conclusion.

It is possible for someone to sell snake oil and the reader will not know the difference.

Basically, it comes down to how people feel about government.  Most students enter college -- we know, we taught them -- with a negative bias.  They see movies.  They make the mistake of thinking of Congress -- 535 people with different motives, coalitions, and partnerships, as if it were a rational unitary actor. Simple and facile interpretations can be persuasive.

The Data

As observers of the investment scene, we have documented a long-term process of dis-information.  To our amazement, the story was even worse. 

The data show that 85% of readers do not believe government data.  Wow!

This is an astounding result.  We knew that the big-time blogs were unduly influential.  We knew that the mainstream media sources accorded undue recognition to the extremists.  We also knew that every time we discussed government data in a positive light, we got hate mail -- both aggressive and plenty of it.

We are still astonished.  Normally the highest people in socio-economic-status (SES) also have a higher score on questions related to Confidence in Government.

But it is  an online poll -- tracking an audience known to be unrepresentative, wealthy and conservative. Does this matter?  What interests us here is that the unrepresentative sample  is so far out of touch with reality.

To Summarize

The data are complex.  We believe that any of the Internet pundits (name your candidate) or the conspiracy guy would be beaten like a drum if engaged in a fair fight with someone from the BLS or another suitable representative.

BEATEN LIKE A DRUM.  Any media source wanting to try this could do it.

We cannot cover this in one article, but it is an open challenge to government critics.  Any time, any place.

Investor Implications

Many individual investors are attempting to monitor economic developments.  There is a choice in data sources.

You can take information from one of two sources:

The first source consists of career professionals whose salary and benefits have no relationship to politics.  They develop methods and write articles for peer-reviewed journals. Their career chances depend upon professional performance as evaluated by others in the field.  You already pay millions of dollars for this information.

The second source has a profit motive.  None of the assertions are reviewed by peers or published.  Everything plays upon emotions and pre-conceptions.  There is plenty of support from bloggers who share these motives.

To us the choice seems obvious.  When we learn that nearly everyone disagrees, it makes the trade even more attractive.

Comments

[Bluce]

Experts can never be defeated.

[oldprof]

Dave - -Why don't you start by taking a look at this:

http://www.econbrowser.com/archives/2008/10/shadowstats_res.html

You need to follow the links in the article. James Hamilton is a top economist and not a market bull by any means. He is just trying to get it right. The key point is that the supposed data cannot be replicated.

This is just the CPI stuff, but the employment arguments are similar IMHO.

Jeff

[Dnarby]

Thanks, I'm slowly learning to give the benefit of the doubt to all but the most overtly hostile net-based communications.

More to the point though, I was wondering if you still discounted Shadowstats after reading their methodology/history/reasoning http://www.shadowstats.com/section/primers

The reason I bring this up is that IMO the scenario they put forth is very plausible.

[oldprof]

Dave - -This is not a problem for me at all. There have been a few "lost" comments and people send me an email. We fix ASAP.

I hope to write more on the issues you raise.

Thanks for your comments and the courteous follow up messages!

Jeff

[Dnarby]

Hi Jeff,



Much needed apologies if I erroneously and unfairly accused you of
improper comment deletion, and thanks for the cordial response.



I am curious though, if you have read the methodology/history behind
Shadowstats. http://www.shadowstats.com/section/primers  IMO, their
arguments are well reasoned and compelling.



Interestingly, they do not attribute any nefarious intent to the
government, just logical behavior by bureaucrats under pressure from
their superiors.  I suspect that something similar happened in the
former Soviet Union.



I could be wrong.



Best regards,



Dave

[Dnarby]

Hi Jeff,



This just in, and conveniently directly bears on the discussion (as
it's a blogger pointing out problems with the way the gov't creates
it's stats):



http://michaelscomments.wordpress.com/2009/06/15/joe-tries-to-spin-but-ends-up-twisting-in-the-wind/



It also includes the information I had originally posted re: 
Unemployment stats.



Best regards,



Dave

[oldprof]

Dave - we do not delete posts that disagree. We welcome any discussion that is germane and within the bounds of decency.

Occasionally TypePad thinks that something is spam and filters it, but I cannot find anything from you there.

As you can tell from the article, I disagree about shadowstats, but it is fine for people to disagree.

Thanks for calling this to our attention.

Jeff

[Dnarby]

OK...

You didn't like my post indicating the unemployment projections based on stimulus/no stimulus that were released by the government were at best, wildly optimistic. Because it's been deleted.

Ever poke around http://www.shadowstats.com/ ?

Believe none of what you hear and half of what you see.

[chartsandcoffee]

Weekly Market Analysis From Charts and Coffee Blog is up - http://chartsandcoffee.blogspot.com/2009/06/sunday-night-coffee-61409.html

[Tom Cole]

Hi Jeff,

I can agree with most of your post, but there is one area of the BLS data that at least in my mind is quite culpable in creating the Mess We Are In. It is "equivalent rents". If the BLS had used any reasonable measure of housing costs, the effect of the housing bubble would have impacted the CPI in a meaningful way and most likely Greenspan would have raised interest rates much earlier possibly deflating the bubble in time. A lot of guesswork and assumptions in the previous statement, for sure. And there are many causes to the Mess We Are In, I know. But consider that by using "equivalent rents" the housing bubble had the perverse effect of lowering the government's definition of inflation. This was due to the pricing pressures that landlords were under due to the easy money available to anyone breathing, or having recently breathed. "Equivalent rents" is a proxy. the relationship between rents and house prices is quite elastic. yet this elastic relationship was part of a supposedly tight feedback loop of Fed rate policy and its observed effect on inflation.

[oldprof]

Harry -- I agree that it would not be exciting. I suppose that means it would not work. As I wrote earlier this week, it is all about the ratings.

Barry Ritholtz would not be the ideal candidate for this debate, since he does not believe in the conspiracy theories. I think he has made this clear on several occasions.

He disagrees with many of the BLS methods and conclusions, but his points are based on actual issues.

Thanks for some good points on data presentation.

Jeff

[oldprof]

Steve - You have laid out nicely what most people think, and perhaps also an acceptable theory about why agencies grow.

But institutions do not have motives, the people do. There is a distinction between the relatively small number of political appointees and career professionals who watch Presidents come and go.

So let us think about how this would work at the BLS. The conspiracy types believe that these regular workers cook the books and later make revisions, etc. This would be a conspiracy involving scores of people. How do they get their orders? Why do they all keep secret? Did they all just switch from doing this for Bush to doing it for Obama?

These are realistic and practical questions. The conspiracy folks make it seem like every administration comes in and has complete control over data releases. The reality is that most of government functions do not change with elections, and the workers have differing party affiliations.

I invite you to elaborate a bit more on your idea. How is a tenured career economist at the BLS or the Census Bureau affected by these political influences and motives? They cannot be fired, and their pay is not determined by the political types.

Thanks for commenting and initiating a thread that I hope others will consider.

Jeff

[Mark Hines]

Jeff-
I like it.
Keep up the good work.
Mark Hines

[Steve]

"Peer review" usually refers to PHD holders from tax funded education institutions. To say they don't have an ax to grind after working most of their lives in these organizations is false.

"The first source consists of career professionals whose salary and benefits have no relationship to politics. They develop methods and write articles for peer-reviewed journals. Their career chances depend upon professional performance as evaluated by others in the field."

If they are from academia and the civil service, their success depends a lot on politics, who they know and who they have helped.

Politics dominates discourse and policy in tax funded institutions whether the military, academia, the courts, the civil service.

And in these institutions they all are motivated always by 1) how to acquire more power and 2) how to get a bigger budget.

[oldprof]

David - I am curious about what government statistics -- not forecasts, but statistics -- you think are the result of political pressure.

You frequently use your personal experience to inform your excellent articles. You are probably most confident when a topic hits the sweet spot of your experience.

Personally, I have worked in a government research operation and taught mid-career government professionals in a graduate program. I have former colleagues who have entered the public service. This experience helps me to identify issues and sources where politics is involved. From the perspective of my experience, the idea that routine data releases are manipulated for political effect is not only mistaken, it is silly.

Here is a small suggestion. The BLS is not so far from you. Have a visit for an hour or two. They would be happy to talk with you. Then see if your opinion is the same.

Thanks for your comment and support for the "debate" idea.

Jeff

[Harry]

While I agree that the creators of these monthly government reports are upstanding citizens and I also agree that it is inordinately hard to explain all the statistical caveats and assumptions that underlie the statistics in these reports. However, I think this also misses the point of the manner in which these reports are disseminated. That is, the reports are highlighted by business news pundits or octobox talking heads who talk their book.

So if you look at today's Retail Sales numbers the CNBC pundits will focus on the 0.5% growth but not review further why that number is much higher than the the previous month.

One quick look through the summary statistics that when aggregated generate the headline number will show that the majority of the growth is from Gasoline stations, meaning oil prices are the reason for the jump in retail spending. This probably mitigates any upside benefit that the headline number implies.

If you did bring in a Census bureaucrat, s/he would explain in a long-winded fashion an explanation that implies a lot of the uncertainties these that these numbers necessarily contain, including a warning similar to the paragraph above.

If you pitted that person versus a blogger like Barry Ritholtz I think you would have a fairly collegial interview that would not make for great television. Thus, most people are stuck with CNBC and its better television, while a few people actually read the figures and (hopefully) make money in the long run.

[jbr]

Before one can say that they don't believe in something they should know what it is they are questioning - the methodology, sampling error, assumptions, limitations of the approach, leading/lagging bias as pertaining to the economic cycle, etc.

Sure, there may be lots of reason for scepticism but more often than not it will be because of 1) methodology issues or 2) misunderstanding of what the data actually captures rather than the data itself.

For example, inflation data could be misleading if the basket being measured leaves out certain asset prices or is adjusted in some way that distorts the popular notion of inflation. The devil always lies in the details.

Lastly, I want to thank you for your posts on understanding some of the economic data and in general keeping things sober.

[www.google.com/accounts/o8/id?id=AItOawmVfqVqMVnx-8sLXEoRoiPyQOYtbSdGfXY]

PS -- aside from the weird Google Accounts jive -- I am David Merkel and I am not trying to hide it.

[www.google.com/accounts/o8/id?id=AItOawmVfqVqMVnx-8sLXEoRoiPyQOYtbSdGfXY]

Dr. Jeff, some bloggers are experts -- the categories are not mutually exclusive. Also, the American way is to not give the benefit of the doubt to the government. The bureaucrats have their own pressures that they face from politicians and business interests.

But I am a reasonable guy, and I say, let the debate begin. Let a highly trained defender of government statistics debate a detractor. I would find that fun, and my opinion is malleable in this matter, so maybe my opinion would change, which is middling; I trust some government statistics and not others. It depends on the level of political pressure involved.

[Paul in Kansas City]

jeff this article is an example of why i love your site. i like to believe i am better thinker because of it.