Diverging Employment Indices and Politics

When President Jimmy Carter was running against President Gerald Ford in 1976, the economy was by most standards doing very poorly, and Carter wanted to focus on this condition to make his case for the presidency. In the process, he coined the term “misery index.” Typically, unemployment and inflation tend to run in counter cycles with one running higher, while the other runs lower. In the 1970’s, we suffered under both high inflation and high unemployment and the sum of the two is what Carter defined as the misery index. Carter’s new index had a saliency because it was easy to understand and it reflected the sad concurrent economic experience of most people.

Carter inherited an historically high misery index in the low teens from Ford, but managed to steer the economy into a misery index over twenty before handing over to President Reagan an economy at a misery index in the high teens. The misery index plummeted thoughout Reagan’s two terms. Reagan’s second term ended at the post-war average of ten for the misery index. We have either been just a little over ten or substantially below that figure since then. Indeed, the first four years of George W. Bush’s Administration had a lower misery index than Clinton’s first four years.

The current misery index is about where it was when Bush took office despite an inherited recession and the attacks of September 11. The current inflation rate of about one percentage point less than is less than the post-war mean of 4.4% and the current unemployment rate of 5.5% is less than the post-war average of 6.4%

Under these conditions, the traditional misery index was useless as a political bludgeon to go after Bush. Hence, Democratic presidential candidate John Kerry yielded to the temptation to conjure up a new misery index. Kerry’s index was so contorted and convoluted that it made Jimmy Carter’s record of double digit inflation and double digit unemployment (and should we add double-digit interest rates) appear to be better than our current, comparatively benign conditions. Not even Democratic partisans bought into the index because it was more likely to highlight Kerry’s intellectual dishonesty than it was to persuade people that Carter’s economic experience of the 1970’s was to be preferred. Voters were not convinced in 1980 that the economy was doing well when they dumped Carter in a landslide and they were not likely to be convinced that conditions are worse now.

While the employment rate, the traditional measure of unemployment, has been steadily declining, Democrats reverted to citing to everyone who would listen, the payroll survey numbers. This employment index shows a net decrease of 1.1 million jobs since Bush assumed office. Now, in fairness the peak in employment in the payroll survey data came in late 2000 and the downward trend began before Bush took office. Indeed, during the first year of the Bush term, which included not only an inherited recession but the September 11 attacks the total employment as measured by the payroll survey dropped by 1.7 million. The employment bottomed out in August 2003. With fits and starts, the payroll employment survey indicates that 1.5 million jobs have been added in the last year.

However, the payroll survey is not the only measure of employment published by the Bureau of Labor Statistics. The Bureau also computes an employment index based on household surveys. They literally call 60,000 households and ask them if they are employed. This last month, the payroll survey showed just a 32,000 increase in employment, while the household survey showed a 600,000 person employment increase. Indeed, this latter measure has show a significant increase of 1.8 million in total employment over the last four years. This represents an over 3 million person disparity between the two employment indices. This clearly represents more than statistical fluctuations between the two surveys. Thise recent divergence between the surveys has puzzled economists.

Economists have generally preferred to use the payroll survey because the sample size (400,000) is larger, reducing the month-to-month sampling variability and the two surveys have tracked reasonably well in the past. However, there is more to accuracy than statistical sampling errors. Part of the problem may be associated with the fact that the press has been focusing primarily on the preliminary rather than revised monthly numbers. The payroll survey is often revised months even years later. These revisions have often been dramatic. The payroll employment survey for 1992 was adjusted so many times in the following two years that 1992 (the last recovery) went from showing a net job loss to a net gain. Hence, the payroll survey is a much better retrospective tool than when considered in “real-time.”

It is also well known that certain corrections have to be made to the payroll survey data. If during a single month a person moves from one job to another, that employee is counted twice. This will tend to inflate the payroll survey data. Attempts are made to adjust for this. However, if during different parts of the business cycle, employed people are more or less apt to switch jobs than average, it can introduce biases in the survey. In addition, self employment and employment in new firms is often missed in the preliminary payroll survey measurements, but caught in the household survey. However, unemployed people may report themselves a “self-employed” perhaps out of denial.

All these are rather technical issues and there is not doubt that the Bureau of Labor Statistics does a professional and credible job attempting to capture snapshots of the state of the dynamic and diverse economy. However, the question must be asked why an esoteric and heretofore little known statistic has gained such prominence. One cannot blame Democrats for trying to use it because, of the three employment measures: the employment rate, the payroll survey, and the household survey, the payroll survery was the single most politically exploitable. Partisans often pick and choose indices to suit their purposes. However, one can blame the press for grasping on to this particular index, downplaying the more traditional unemployment rate without a clear reason why.

At the very least, attention should have been given the different measures of employment and their respective advantages and disadvantages. These indices must be considered in the light of other measures like the number of unemployment claims, withholding tax receipts, and indices of real earnings. One wishes that the national media would devote the same level of professionalism to covering economic statistics as the Bureau of Labor Statistics exhibits in their creation and maintenance. In the end of course, reporting on the economy can only effect perceptions at the margins. Though the margins can be important in close elections, by-and-large, people vote based on their personal economic experiences not on indices.


  1. US Bureau of Labor Statistics.
  2. Kane, T., Diverging Employment Data, The Heritage Foundation, March 4, 2004.

One Response to “Diverging Employment Indices and Politics”

  1. […] no longer look at the unemployment rate, the traditional measure, because it was too good. Instead, they used other Bureau of Labor statistics that notoriously lag the economic growth that was beginning even in 2003 and […]

Leave a Reply

You must be logged in to post a comment.