An Access Advertising EconBrief:
The 7.8% Unemployment-Rate Controversy
On October 5, 2012, the Bureau of Labor Statistics released estimates on employment and unemployment in the United States for the month of September. BLS does this every month, and these data are usually a source of interest but only rarely a source of controversy. This release was different.
The Bureau announced that its estimate of unemployment had fallen to 7.8% from its previous level of 8.1%. This came as a big surprise to economic forecasters and analysts, who had expected the rate to remain the same or even rise. The source of controversy was the magnitude of the decrease and its rationale.
The unemployment rate itself is estimated using a survey of roughly 60,000 U.S. households. The results of that survey have been quite volatile in recent years – last month, for example, they showed a seasonally adjusted decline of 119,000 in the number of those working. But the September survey estimated an increase of 870,000 employed. This was a staggering result – the largest total in this category since January, 1990 (1,251,000) and June, 1983 (991,000). (Two larger totals were attained earlier in the millennium, but BLS adjustments in the data make these totals non-comparable with others.)
This was the kind of increase in employment normally associated with rip-roaring growth in economic activity. In June, 1983, for example, annualized growth in GDP was 9.3%. In January, 1990, it was 4.2%. But here in 2012 it is a puny 1.3%. This seeming paradox raised suspicions in the minds of some people.
Much has been made during President Obama’s tenure that no U.S. president has ever been reelected with an unemployment rate above 8%. Conservative talk-radio host Rush Limbaugh went so far as to predict that the Obama administration would somehow contrive to bring reported unemployment down below 8% prior to the election – implying that deception might be involved.
In the face of the decline in the reported unemployment rate, former CEO of General Electric Jack Welch sent a text message to friends in which he directly accused the Obama administration (whom he characterized as “Chicago guys”) of somehow manipulating data to produce this result.
Tons of ink and reams of paper are consumed writing about markets and their misfortunes. Virtually nothing is said about the collection, preparation and presentation of economic data. This time is ripe for that discussion.
Political Theater vs. Political Economy
The brouhaha over the BLS’ handling of this data release is ironic. While clear wrongdoing occurred, it has been virtually ignored throughout the controversy. Public debate has instead focused on a hypothesized conspiracy to invent or distort data, to “cook the books.” As is so often the case, battle lines have been drawn along political lines. Meanwhile, the news media has been perfectly willing to dramatize the conflict as an exercise in political theater while ignoring the underlying issues of political economy.
The BLS, and particularly Director Hilda Solis, plays a key role in the drama, but that role has been miscast by both political factions. The right wing has cast the agency as accomplice and co-conspirator. Defenders of the administration have portrayed the BLS as staffed by politically independent professionals, completely devoid of political sentiment and as behaviorally pure as Ivory Snow.
In reality, the agency is a branch of the “permanent government,” the bureaucracy that keeps rolling along like Old Man River through Democrat and Republican administrations alike. Its only inherent goal is to maintain its existence, size and power. Ms. Solis is a political appointee, named by President Obama in 2009. As such, she has divided loyalties.
As political appointee, she owes her position to the President. The temptation to hew her actions and public pronouncements toward the positions of the administration is ever-present. This would be true regardless of her personal sympathies, but since presidents usually choose department heads whose views dovetail with their own, the sympathies of a director typically reinforce the incentive to side with the administration.
But as chief administrative officer of a federal bureaucracy, she is the only person capable of steering that agency away from its normal self-serving goals and toward the objective of serving the broad general interest. As far as the American public is concerned, that is her only valid function – to steer the agency between the Scylla of toadying to the administration and the Charybdis of bureaucratic inertia.
In this case, Hilda Solis failed miserably. That is the wrongdoing – indeed, the tragedy – of the 7.8% unemployment controversy.
Friday Morning, 8AM, October 5, 2012
On the morning of the announcement, Ms. Solis was presented with the statistical reports prepared by her staff. In order to contrast what she should have done with what she actually did, we must take a critical look at those reports. The BLS takes two surveys of employment that attract widespread public attention.
Its payroll survey uses payroll records of 60,000 businesses to estimate new hires during the target month. The results of this survey tend to be relatively stable. The September report not only presented results for that month but also upward revisions for the previous months of July and August. Payroll jobs for July were revised up to 181,000; the August estimate was revised up to 142,000. The September estimate was a job gain of 114,000.
The first thing to notice about this survey is the downward trend. This, combined with the fact that unemployment has long been considered a lagging indicator, influenced the expectations of many economists who expected the September unemployment rate to rise slightly. While there is no general agreement among economists, it would be fair to state that 142,000 jobs is close to a tipping point when it comes to lowering the unemployment rate – it is either barely adequate to nudge unemployment down or not quite enough, depending on how responsive one finds the labor force to be.
The 114,000 jobs chalked up in September, though, are not enough to make a dent. That is why the result of the other employment survey, the telephone survey of households conducted by BLS, created such a stir.
The household survey purported to locate a total of 873,000 new jobholders in September. Of these, some 582,000 were supposedly part-time jobs. The fact that this total had been exceeded only twice since 1983 – and both times when the economy was growing at elevated rates – made many anti-administration partisans doubt the veracity of the figures.
These job numbers were not only dubious on their face. They were also blatantly at odds with everything else we knew or conjectured about the state of the economy. Growth had begun the year promisingly but had stalled and slowed to an annualized pace of 1.3% in the second quarter. World trade slowed. Recession loomed in Europe.
Some good news tempered the general mood of gloom, but it was measured. Consumer confidence rose somewhat, perhaps buoyed by a stock market rally – but the rally was dampened. Labor force participation increased after steady decreases – but the increase was slight.
In order to believe in the veracity of the household survey’s jobs estimate, we would have to believe that the labor market had suddenly, inexplicably become the leading indicator for a roaring expansion that as yet had no other harbinger – that the household survey was telling us the truth while all other indices were lying, or at least keeping mum.
Historically, the household survey was known to be volatile. The previous month, August, it had recorded an estimated job loss of 119,000. Thus, the variance between the two surveys was still three times greater in September.
The only reasonable conclusion seemed to be that the household survey was wrong. “Wrong” doesn’t mean faked or fraudulent. It doesn’t mean that BLS employees didn’t make the survey calls, or didn’t record the answers correctly. It certainly doesn’t mean that somebody hid the results in the dead of night or bribed the BLS to suppress them.
All experienced economic forecasters and statisticians know that formulating estimates from sample data is far from an exact science. It is like dining out every night – sooner or later you’re going to get hold of something dreadful that needs to be purged. And that is exactly what statistics textbooks advise students to do with obviously aberrant values in a data set – omit them.
The argument for omission is fairly straightforward. The most basic type of statistical estimation technical, called linear regression, tries in effect to draw a straight line through a collection of data points for the purpose of estimating the course future data will follow. The line is an attempt to capture the central tendency of the data. Including a wildly aberrant value will pull that line off course and make the future estimation process less accurate.
What BLS Director Solis Should Have Done
For practical reasons, it may be difficult or impossible to simply cancel or postpone the release of the household survey and associated unemployment rate. This is an eagerly awaited statistic that is followed closely by analysts throughout the world. Regardless of any good reasons advanced for cancellation or postponement, such an unusual procedure would itself be suspect – people would wonder what the authorities were hiding.
Of course, that argument cuts both ways. The world isn’t waiting breathlessly in order to receive estimates that are worthless or downright misleading. Then there is the little matter of a Presidential election that probably won’t – but just might – turn on the result of these estimates.
What Hilda Solis should have done is: 1. order a double-check of all relevant figures and calculations in the household survey; 2. assuming the results check out, announce at the press conference that the data release contains survey data and a consequent estimate that defy common sense; 3. advise the general public that no weighty conclusions be drawn from the suspect estimates, since they are unsound; 4. invite all interested parties to inspect the Bureau’s data, methods, calculations and results.
She should have done this because the purpose of government is to aid and inform the American public, not to serve the political interests of any administration or the economic interests of bureaucrats. By presenting the data but warning the public, she would be telling the truth, the whole truth and nothing but the truth. She would be allowing anybody who still wanted to accept the figures to do so, but at their own risk. And she would be putting everybody else on notice. She would be behaving the same way as a fiduciary – a professional who has the legal duty to put the client’s welfare above all else. That duty covers both commissions and omissions; it is the obligation to place the full range of professional expertise at the service of the client. In this case, the client is the American people.
What Hilda Solis Actually Did
What Hilda Solis actually did was to release the household survey and unemployment-rate estimate without warning the public. Indeed, she not only refused to supplement the data release with a warning – she passed up opportunities in subsequent interviews. An interviewer from Bloomberg questioned her three times about the dubiety of the 7.8% unemployment rate and the 870,000 job gain in the household survey. She defended the household survey, citing job gains among 16-24 year-olds. At no time did she back away from or otherwise express reservations about the household survey.
Ms. Solis’s act had the effect of inviting the public to take the dubious household-survey results at face value. Some people did that. Others were shocked by the extremity of the 870,000 job-gain and 7.8% unemployment-rate figures. Still others were outraged by what seemed altogether too fortuitous a coincidence – that a bureau in the Department of Labor, long dominated by the Democratic left wing, would produce a wildly extreme employment report favoring a labor-union-supported Democratic incumbent on the eve of a presidential election.
But the people in the best position to evaluate the report were professional economists and forecasters. Here is a representative selection of their characterization of the two disputed estimates – the 870,000 September job gain and the 7.8% unemployment rate: “”Must be an anomaly;” “statistical anomaly;” “just a fluke;” “statistical quirk;” “implausible;” “almost certainly a statistical fluke;” “huge statistical outlier on the upside;” “not reality;” “an aberration.”
All of these comments came from respected economists, forecasters and consultants. One of them is a former director of the Congressional Budget Office. Some of them are known to be supporters of the Obama administration. None are rabid anti-administration partisans. Clearly, they all knew statistical salmonella when they saw it. Yet none of these people criticized Director Solis’ decision to release the estimates without warning or qualification.
The Harm Caused by the BLS Acts of Omission
The news media covered the issue as an exercise in political theater. They pitted right-wing claims of conspiracy against indignant denials and claims of pristine innocence on the left. When the conspiracy angle petered out for lack of evidence, the story died.
The real harm caused by BLS wrongdoing is much more mundane, but more hurtful than any partisan conspiracy. It concerns the day-to-day functioning of government, not the crimes of individuals. The unemployment rate is used by analysts throughout the world as a barometer and index of the U.S. economy. Investment company owners and fund managers use it to calibrate the timing of investments. Financial planners use it to manage their clients’ money. Large corporations use it to gauge the direction of consumer demand. Commercial and investment bankers use it; business and economic forecasters use it; employment agencies and corporate headhunters use it. Even small businesses use it.
All these people suffer when information disseminated by the federal government turns out to be disinformation. When people discover that they have been fooled, they will take the index less seriously in the future. As a result, their job performance will suffer. And their cynicism about government and the rule of law cannot help but harden – after all, they are already suffering their fourth year of being fed false information about interest rates by the Federal Reserve. The Fed’s QE series of government and private securities purchases is openly and deliberately designed to hold interest rates artificially low by increasing the supply of money. Interest rates are even more ubiquitously used and useful than government economic data.
The people best equipped to understand the abdication of professional responsibility by Hilda Solis and the BLS are the premier economists, forecasters and statisticians. They know that the household survey’s September estimates should have been released – if at all – with a stern caution to the general public. This is directly analogous to the warning labels that government regulators require private businesses to stick on products that present a potential hazard to consumers. The 7.8% unemployment-rate and 870,000 job-gain estimates were no less hazardous to the financial, intellectual and political health of the American public.
The quoted comments above demonstrate that these financial experts recognized this danger quite well. But while they noted it in casual asides and obiter dicta, they refused to take the obvious next step. They refused to call Director Solis and BLS to account. They refused to alert the American people to the true nature of the wrongdoing. They refused to limit the damage done. And they lost the opportunity to deter future episodes of misconduct.
The 7.8% Solution
The real wrongdoing in the 7.8% unemployment-rate controversy stems from negligent omission, not active conspiracy. It is patent in the reactions of professional economists and forecasters. The permanent government was derelict in its responsibility to aid and inform the American public. Instead, it catered to political and/or bureaucratic interests. That is not the kind of dramatic, theatrical conspiracy that attracts the attention of news media. But the failure of day-to-day government to do its job grinds down our living standards, morale and respect for law.