Thursday, January 19, 2012

Kicking a Statistic Around

Politicians of various stripes and their backers delight in kicking each other around.  They’ve become so accomplished at mud slinging and twisting facts one would think they spend some time practicing.  Perhaps they do.

For the last several years, our pols appear to have been warming up for the big 2012 campaigns by kicking a statistic around.  A great many have been putting the boots to the National Unemployment Rate. 

A while back, Democrats claimed unemployment was much higher than the indicator showed, and thus painted an unrealistic picture of the horrible economic conditions created by George W. Bush.  A bit later, the blame game diminished in effectiveness as a tactic, but Democrats kept blasting away with the same theme to support the idea that we need more stimulus money and government job creation to pull out of our economic funk.

Republicans now are singing the same song—the monthly unemployment statistical reports are grossly understated—but for a different reason. The contention is that President Obama is falsely claiming economic achievements based on the unemployment indicator dropping slowly over the past few months. They are saying exactly what the Democrats have been claiming—discouraged job seekers are leaving the labor force in large numbers, and thus skewing the indicator. 

Usually, the Geezer would pass all this off as typical political posturing unworthy of serious attention.  However, a recent editorial in The Washington Times raised my ire. The headline read: “Obama Cooks the Unemployment Books.”

The last line in the opinion piece says a calculation posted on a web site, “ . . . exposes what the government statistics are intended to do: Get Barack re-elected.”

From start to finish, the editorial was an unwarranted insult to civil servants who work to provide useful statistics to policy makers and the American people, especially those employed by the Census Bureau and the Bureau of Labor Statistics.  Of course, it also insulted Obama, but there is nothing particularly noteworthy about the Times doing that.

I worked with civil service statisticians for many years.  It is hard to envision any group of people more dedicated to getting things right.  They are especially diligent about trying to develop controls to minimize or eliminate bias from any source. 

Even if President Obama was unethical enough to try to influence the work of the Census and Labor Statistics people, he certainly is not dumb enough to try.  Any high-level politician attempting to “cook the books” in this arena would touch off a storm of protest he or she could not survive. 

The National Unemployment Rate, a product of the Current Population Survey, is derived from statistics collected by the Census Bureau for the Bureau of Labor Statistics.  It is based on a sophisticated, carefully designed sampling of households throughout the U.S.  It does not count every unemployed or employed person.  Thus, it is inappropriate to claim the rate is either right or wrong.  The questions should be: “Does the rate reasonably portray the situation in the nation? Is it comparable to rates for previous months and previous years, and thus useful in showing trends?

No indicator based on sampling is perfect.  Government employees have been calculating unemployment rates since the 1890s.  From time-to-time they have sought to improve the accuracy of the rates by changing sampling methods or adjusting underlying assumptions when there is a consensus that newer assumptions are more valid. Thus, looking at rates for different time periods easily can result in “apples and oranges” comparisons.

For example, it would be folly to try to make a precise comparison between the current 8.5 percent unemployment rate and the 21.7 percent rate in 1934.  For one thing, before 1948 the data included people aged 14 and up; since then only those 16 and older are included.  However, the historical data have some value as long as the survey framework is generally the same. Any reasonable person would conclude that employment conditions were pretty awful in 1934. Exactly how awful compared to today’s conditions we cannot tell.

Reasonable people also could see that conditions were very good in 1952 when the rate was 3 percent.  Certainly those of us who lived back then know times were much better than they are now.  Again, however, the two rates are not strictly comparable.  The Bureau of Labor Statistics made a major reform in the calculation system in 1994.  There have been minor adjustments since, but no big change. Matching our present 8.5 rate with the 4.6 percent calculated for 2006 before the Great Recession started thus is very close to an “apples to apples” comparison.

The Times editorial writer casts aside the work of civil service statisticians and analysts over many years and cites statistics from an obscure web site claiming the “true” unemployment rate currently is 11.4 percent.  Oh?  There is no such thing as a statistically derived “true” unemployment rate for the United States.    

Saying any numerical value is better than the one the government provides is irrelevant when considering the important reasons an indicator was developed in the first place.   We should be interested in a rate that gives a decent portrayal of the unemployment situation and is the best available indicator of trends.  The Census Bureau and the Bureau of Labor Statistics give us useful numbers. The Washington Times does not.

The Geezer has deep sympathy for anyone struggling to find satisfactory employment and strongly supports efforts by both business and government to create new jobs. An 8.5 unemployment rate is not good any way you look at it. But directly or indirectly bashing those who work diligently to bring us the best rate data they can assemble does no one a service.


schmidleysscribblins, said...

Dick, funny you mention this. David reads the Times and asked me this morning if the stats could be manipulated. I was a Current Population Survey (CPS) expert when I worked for the Census Bureau. The CPS is the basis for the employment numbers. Although I could write a long boring answer about the strenghts and weaknesses of the CPS, suffice it to say, the statisticians who work on this survey are indeed professionals who would not distort the numbers for political gain.

The only pressure I saw brought to bear on the stats during the years I worked at the Bureau (Clinton and Bush Administrations) were indirect ones like suppression, where possible, of the numbers of that would prove embarrassing (both parties hate bad social numbers) and a delay in releasing some information (i.e. get it out on a Friday when the news is slow.) These tactics, encouraged by public relations offices (read branches of the White House), never worked because the Congress was all over the numbers and got them out in different ways (I spent much time on the phone with Kennedy's staff as well as Paul Samuelson).

Towards the end of the Bush administration, people like me (actual statisticians) were discouraged from talking to the press without a PR type on hand (paranoia?).

I saw no difference overall with the approach of the two White Houses although the liberal media would suggest otherwise where Bush was concerned. Dianne

JHawk23 said...

A very valid point, and well-made. We should let our government statistics speak for themselves, as benchmarks and indicators of trends. Political partisanship has intensified (but certainly not originated) making every statistic, no matter how conscientiously prepared, a political football.

And the sad fact is that, despite what politicians may tell us, neither party has reliable means of "creating jobs". Cause and effect are blurred by too many factors and too many independent decisions at myriads of different levels, for that to be possible.

Sightings said...

Thanks for the interesting & informative post. I'd like to know your view on the Valerie Plame affair. I don't know much about it, honestly, but I saw the movie which suggests that the Bush White House pressured the CIA to support the (presumably false) notion that the Iraqis were acquiring materials to use for nuclear weapons.

And then of course the question: If politicians can pressure the CIA, can't they pressure the Bureau of Labor Statistics, and everyone else?

I'm not saying they do -- I have no basis to suggest that. But surely you can understand why some people could be skeptical. So, are you saying the Valerie Plame incident was unusual, that politicians typically leave the technicians alone to do their job, and do not try to influence the numbers?

Dick Klade said...

Tom, you are quite correct that there is room for skepticism regarding government pronouncements.

Politicians do try to exert pressure when it suits their purposes. In a way, the Plame affair supports my contention that it often is not smart to do so. If I understand it correctly, Bush critics said Plame was exposed as a CIA agent because her husband questioned reasons given by the White House for invading Iraq. President Bush denied he knew anything about it. A series of investigations and court cases involving White House staff kept the question before the public for a long time.

It seems plausible that one reason the Republicans lost the White House in the next election was because of the Plame incident charges and other reports that the Bush administration tried to "cook the books" in justifying the Iraq war.

The general public doesn't find out about such pressures unless government workers either individually or collectively protest by contacting the media. Chances of this happening are very high when the responsible organization is staffed by professionals. Most are, in my experience.

Thus, my contention that President Obama would not be foolish enough to try to influence economic statistics. The risk would be too great. He would not want to take a chance on losing his job because he tried to cause anyone to produce false unemployment statistics.

Kay said...

Dick, I love coming to you for a balanced opinion on politics. Thank you!