Official unemployment statistics answer specific questions. A broader indicator may answer a different question, but it must still use transparent definitions, reproducible data and plausible assumptions.
This archive article follows an earlier critique of an alternative inflation measure. It turns to an alternative unemployment series that stood substantially above figures published by the US Bureau of Labor Statistics in 2015.
The inquiry has two parts: what is the alternative indicator intended to describe, and is its method a reasonable way to describe it? The analysis treats this as a question about measurement rather than a dispute over whether labour-market hardship exists.
From unemployment to the employment gap
The standard US unemployment rate, U-3, covers people who are not working, are available for work and have actively sought work during the preceding four weeks. It expresses that count as a share of the civilian labour force. Because active job-seeking changes with the business cycle, U-3 is useful for tracking near-term labour-market conditions.
That narrow definition does not capture every person who wants more paid work. A broader concept—the employment gap—can include people who want a job but have recently stopped searching, as well as people working part time because they cannot obtain the hours they want.
The BLS broad measure, U-6, starts with U-3 and adds two groups: people marginally attached to the labour force and people working part time for economic reasons. Its denominator also adds the marginally attached.
| Measure | Who it includes | Primary use |
|---|---|---|
| U-3 | People without work who actively searched in the prior four weeks. | A focused indicator of cyclical unemployment. |
| U-6 | U-3, marginal attachment and involuntary part-time work. | A broader view of underused labour. |
| Proposed U-7 | U-6 plus other people who report wanting a job. | An experimental, still broader employment-gap measure. |
In BLS terminology, marginally attached people wanted and were available for work and had looked during the previous 12 months, but not during the previous four weeks. Discouraged workers were a subset who cited labour-market reasons for not searching. Involuntary part-time workers were employed for fewer hours than they wanted because of business conditions or an inability to find full-time work.
What changed in 1994?
The disputed alternative added an estimated category of long-term discouraged workers to U-6. It argued that a 1994 redesign had removed these people from official measures. The history is more nuanced.
Before 1994, survey respondents outside the labour force were asked whether they wanted a job and why they were not seeking one. Classification depended partly on the reasons supplied. The revised survey added a question about job search during the preceding 12 months and asked directly whether a respondent was available to work. It also introduced the broader “marginally attached” category.
The redesign narrowed the discouraged-worker category by requiring a search during the prior year, while broadening aspects of how reasons for non-search were handled. Contemporary research found that the net change reduced the count classified as discouraged. People who wanted work but had not searched in the prior year did not disappear from the survey; they remained identifiable in broader tables outside the headline measures.
Reconstructing the alternative estimate
The public presentation did not provide a complete model for the long-term category. The implied count can nevertheless be reconstructed from the reported rate.
For May 2015, the non-seasonally adjusted U-6 numerator was reported as 16.595 million and its denominator as 159.581 million. Substituting the alternative rate of 23 per cent and representing the added group as x gives:
Implied x: approximately 26.1 million people
The question is not merely whether a broader category could exist. It is whether available surveys and reasonable transition assumptions support a long-term group of that size.
Three checks on plausibility
1. Evidence surrounding the survey redesign
Estimates of the 1994 methodological effect were far smaller than the reconstructed 26.1 million. Applying the reported scale of the redesign to 2015 categories would suggest a count measured in low single-digit millions, not tens of millions.
2. Broader BLS tables
In May 2015, the BLS reported 92.025 million people outside the labour force. Most said they did not currently want a job. After separating those who wanted work, the marginally attached and those not then available, roughly 3.7 million reported wanting a job without having searched during the previous year. That was about one-seventh of the implied private estimate.
3. A nongovernmental survey
A late-2014 survey of prime-working-age adults provided another perspective. Its responses suggested several million people wanted and were able to work but had not searched during the previous year. Extending that evidence cautiously to other age groups might produce a larger total, but still one well below 26 million. Availability also remained uncertain because respondents cited overlapping reasons including care responsibilities, health limitations and not needing income.
Testing an attrition model
One possible reconstruction is an attrition model. Each year, a new cohort passes beyond the 12-month search limit and joins a long-term category. Older cohorts remain only until their members find work, resume searching, no longer want work, retire or otherwise leave the category.
For a steady-state illustration, let m be the average marginally attached population, r the annual retention rate and n the number of cohorts. The accumulated category can be represented as a geometric series:
Long-term component = total accumulated group − m
Using an average marginally attached population of about 2.035 million over the decade to April 2015, the model needs an annual attrition rate of only about 7 per cent to approach the reconstructed estimate. In other words, approximately 93 per cent of each cohort would have to remain in the category from one year to the next.
Observed transitions suggested much faster movement. A longitudinal comparison around the 1994 redesign found that most people who had been marginally attached one year were employed, actively unemployed, again recently searching or no longer wanted a job by the next year. Only a small share moved into the implied long-term category.
- 30.8 per cent were employed in the following survey year.
- 16.7 per cent were actively unemployed.
- 9.3 per cent remained recently attached through job search.
- 34.6 per cent no longer wanted a job.
Taken together, those transitions imply roughly 91.4 per cent annual attrition from the specific long-term pathway. Applied mechanically to the later average, that would yield a long-term group of about 191,000 rather than 26 million. The exercise is illustrative, but it demonstrates how strongly the result depends on retention assumptions.
Bottom line
The 23 per cent alternative rate reported for May 2015 was not supported by the public methodology or by the comparison evidence examined here. Reaching it required counting an exceptionally large long-term group and assuming that very few members left that status from year to year.
A broader measure than U-6 can still be useful. One transparent option would add all survey respondents who said they wanted a job, rather than only those classified as marginally attached. For May 2015, that experimental “U-7” calculation was approximately 12.9 per cent—above U-6, but far below the disputed private estimate.
Definitions matter. U-3 remains useful for short-run cyclical analysis; U-6 captures additional underemployment; and broader experimental measures can illuminate unmet demand for work. None should be presented without its date, denominator, inclusion rules and limitations.