http://www.zmag.org/ZMag/articles/brill.htm
The Bureau of Labor Statistics (BLS) and the Census Bureau: Partners in Deceit
By Harry Brill
Many of those who track the official unemployment rate realize that the Department of Labor's Bureau of Labor Statistics (BLS) substantially understates it. Yet little or nothing is known about how the Census Bureau also contributes to underestimating the nation's unemployment rate and more generally, its social and economic problems. The important role of the Census Bureau and some lesser known shortcomings of the BLS deserve our attention.
We continually hear nowadays about the extraordinary accomplishments of the American economy, especially its ability to provide jobs for almost anyone who is willing and able to work. The main problem, according to the business community, is the serious shortage of qualified workers. We hear from employers that there must be something wrong with almost anyone who is still out of work. Echoing this view, a chief recruiter for the Census Bureau, concerned about hiring enough census takers, complained that since job opportunities are in abundance, those who are unemployed "are, for the most part, unemployable."
After all, doesn't our own common sense tell us that jobs are plentiful? At any shopping mall help wanted signs are ubiquitous. Although they are not usually the best of jobs, which can dispute that they are, at least, available jobs? But it is also possible that we are seeing a magic show. Barbara Ehrenreich, the writer and social critic, posing as a job seeker in the low wage service sector, found that many employers advertise vacancies even when openings do not exist. They build lists of applicants who they can readily access to replace those who will sooner or later be leaving. This practice is especially useful for filling vacancies in industries where jobs are very low paying and, accordingly, trigger high turnover. Is this what business leaders mean by their insistence that jobs are plentiful--that due to a combination of low pay and poor working conditions, too few or too many working hours, working people come and go as rapidly as a turnstile turns on a busy day?
Several years ago BLS changed the criteria for determining labor force status so that the higher the job turnover, the lower the official unemployment rate[Jannuzi: I believe this was in 1994 and the strangely Rooseveltian/WPA gov't homepage was none too forthcoming about 'certain' revisions.] . To illustrate, workers who expected to begin a new job but were not yet working were formerly counted as unemployed. Now most are defined as out of the work force and are no longer included in the unemployment statistics. Moreover, the worker who left with the intention of seeking another job is defined as out of the work force until that person applies for another job. To be counted as unemployed requires an "active" effort to find work.
But here's the catch. This restriction ignores the job seeking cycle. Many workers who have just been separated from their jobs first prepare for a job search, which the BLS does not consider an active effort. They may be writing their resume and soliciting references. They may be perusing newspaper ads, which, unlike in Canada and Japan, is not defined by the BLS as an active job search. So they would not be counted as unemployed. Neither the worker who expects to begin a job nor the individual who just left the same job are collecting a paycheck. Yet the official unemployment rate among them is zero. The greater the job turnover, the more this situation is replicated, and the larger the gap between the official and real unemployment rates.
That a serious job shortage exists is suggested by comparing our economic situation with Japan. It is widely recognized that Japan is in a recession and is experiencing substantial unemployment. Yet although our official unemployment rate has been generally higher than in Japan, our labor market is considered to be at almost full employment. Why, then, is a lower percentage of joblessness been considered bad times for working people in Japan while a higher rate in the United States is interpreted as good news?
The New York Times [Jannuzi: always willing to expose the truth about the alien abroad but never the universal US, right?] explained that the actual extent of joblessness in Japan is twice the official rate. As evidence, the newspaper pointed out that workers in Japan who worked as little as one hour during the survey week are counted as employed. But the Times fails to mention that the BLS also counts these workers as employed. The BLS studied how Japan counts its jobless and adjusted its figures to approximate the BLS own concepts. Although some of Japan's labor force definitions produce lower rates than BLS measures, they are higher in other ways. On balance, there is very little difference.
As in Japan, the official unemployment rate in the United States of over 4 percent is nothing to cheer about. As Robert Reich, the former Secretary of the Department of Labor called to our attention, not until the unemployment rate dips below 3 percent is the labor market really tight. Unlike the practice in Japan and elsewhere, information on job vacancies is not collected. But studies of job vacancies show that when the unemployment rate is in the vicinity of 4 percent, there are two or three jobless workers for each opening.
Among the widely known biases of the BLS is that jobless workers who haven't "actively" sought jobs for several weeks are not counted as unemployed. Also, part-time workers who want but are unable to find full-time jobs are counted as fully employed, although on average they are half-unemployed. There are other serious shortcomings in how the BLS determines the unemployment rate that have received little public attention.
Surprising for an agency of the Labor Department, is that the BLS does not take into account the job seeking cycle of unemployed workers. In this regard, workers who have applied for several positions and perhaps have for a while exhausted the possibilities typically wait a while to learn of the outcome. Or the want ads may not include jobs for which they are qualified. In both instances, their search is not considered active, and so they are not included in the unemployment count.
Also uncounted are the large numbers of older workers who are victimized by age discrimination but who are defined as retired. Perhaps they have accepted one of those "you better take it or else" so-called voluntary retirement packages. After unsuccessfully attempting to find another job, many give up the search. But at bottom, it is the federal and state governments that have voluntarily retired--that is, retired from their obligation to deter employers from excluding them from the work force.
Like older workers, many job seekers with disabilities have been repeatedly rejected for jobs that they qualify for, and as a result they too have given up looking for work. They appear in the BLS category of wanting a job but not in the work force because of "ill health or disability." This suggests, most often incorrectly, that the barrier to obtaining a job is their own physical limits rather than employer-based prejudice and disinterest in making reasonable accommodations.
Many women too are counted as out of the labor force who want to work, but their participation is discouraged. Federal and State governments have not accepted their responsibility to assist these families to obtain quality and affordable day care. From the perspective of working people rather than the viewpoint of BLS statisticians and the business community, these mothers, and older and disabled individuals who want to work should be counted as unemployed simply because that is what they are.
As the BLS's own data shows, counting workers as unemployed who want to work and part-timers who prefer full time jobs (each part-timer is counted as half-unemployed) would lift the official rate of unemployment by at least 4.7 percent to 9.2 percent for 1998. The number of those who have been statistically disenfranchised serves also as a measure of business and government callousness.
With regard to the role of the Census Bureau, it has an ongoing partnership with the BLS in flattering the American economy. The Census Bureau conducts a household survey every month on behalf of the BLS. A sample of 59,000 households is contacted each month. An adult in the household is asked about the labor force activities of its members. The questionnaire is written by the BLS, but who is approached for interviews and who is not is determined by the Census Bureau and the Department of Commerce.
Among the serious failings of the monthly household survey is that it is based on the error ridden decennial census. Even in 1999, on the eve of the next decennial census, the understated unemployment rate is due in part to the miscount of the population in 1990. The reason is that the biases of the census are incorporated into the monthly household survey, and also in other reports on social and economic issues that depend on census data. Calculating the distribution of the population depends on the information obtained by the decennial census. The decennial census serves as a benchmark for subsequent updates. Underestimating the percent of the population that is poor, and conversely exaggerating the percentage that is financially well off results in lower estimates of joblessness.
According to a study by the General Accounting Office, an investigative arm of congress, about ten million persons were missed in the 1990 decennial census, and well over four million higher income residents were counted twice. Not only were millions of people missed. As the Census Bureau realizes, the uncounted included a disproportionate percentage of Black and Hispanic residents, Native Americans living on reservations, illegal aliens, homeless persons, and the poor generally. Unwisely, the Census Bureau purchased addresses for the 1990 census from private vendors. As it now acknowledges, the lists were far less accurate in low income communities because the addresses were intended for market purposes that were mainly directed toward upper income residents.
Since the addresses selected for the monthly household labor force survey is mainly drawn from the decennial addresses, many who are poor and members of disadvantaged minority groups are excluded as possible respondents in the household survey. Those double counted were typically upper income residents, many of whom own two homes. These mistakes have produced an inaccurate profile of the racial and economic mix of the nation's population.
Among the serious stumbling blocks to an accurate count is that the return rate among those who do receive the census questionnaire is very low in minority communities. A study in Milwaukee, which had a higher than average 1990 census questionnaire return rate, revealed that in minority communities the non-response rate was 40 percent. "More Americans were becoming alienated from society in general," the Census Bureau explained, "and more distrustful of government in particular."
The Census miscount has received national attention for mainly partisan reasons. Many Republicans have been worried that using techniques that would produce a more accurate enumeration would justify the reapportioning of congressional seats in favor of the Democrats. But the history of the census has been a history of undercounting no matter which party is in power. The chronic problem of the census undercount prompted the following question in a black scholarly journal, "Scientists can count fish and wild animals with great success; why can't the Bureau count people correctly?"
For the first time in history, the decennial census was less accurate than the preceding one. The Census Bureau estimates that the undercount rate was estimated at 50 percent greater than the 1980 census. As a result, its impact on underestimating unemployment was larger than before. The problem was not because the Census Bureau was more lax than in 1980. In fact, more money was spent per household in 1990. The unusually large undercount reflected the increase during the decade of those who are most elusive to count--the homeless, illegal aliens, and the poor generally. As the Census Bureau's own data shows, both the magnitude and extent of poverty during the 1980s has increased. As a result of important shifts in the population, the mail-back rate of the census questionnaires dropped 12 percentage points from 1980 to 65 percent 10 years later. But in middle and upper class communities, compliance is high.
During the 1980s, the poverty population increased. For various reasons related to being poor, including distrust of government, the return rate among low income residents is low. Also, the Hispanic population has grown during the 1980s, and although many Hispanics can comfortably read only Spanish, many of them received English only census questionnaire.
As the Census Bureau recognizes, to improve the count by just spending more money will not work. Non-traditional approaches are necessary to take account of the social and economic shifts in the population. In a report to Congress, the Census Bureau explained, "Populations with high undercount rates under traditional methods of enumeration have grown more rapidly than the total population."
But despite the interest of the current Democratic administration in improving the census count, the Census Bureau and the BLS, lip service aside, is still disinterested in counting everybody. Its estimate of the 1990 undercount is no different than the numbers claimed by the Census Bureau under the Bush administration. which is about half the undercount claimed by the General Accounting Office (GAO). Accordingly, they have made no serious efforts to accurately update the count. Like the Republicans, neither the current BLS nor the current Census Bureau are interested in including an appropriate sample of the homeless population in the monthly household labor force survey in order to improve the count of the unemployed.
That the homeless population, the vast majority who are unemployed, is much larger than the Democrats and Republicans acknowledge is unquestionable. The 1990 census tallied a total of 233,000 individuals, although various independent estimates claimed as many as three million homeless. That the Census Bureau's estimate was much too low is indisputable. First, the agency concedes that it did not attempt to count all the homeless. Second, it acknowledged that its count was circumscribed, including limiting its efforts to large cities.
Third, several studies that the Census Bureau itself commissioned to study the issue all found that huge numbers of homeless persons were missed. One study revealed that it omitted 70 percent of the homeless in Los Angeles, and another 47 percent of New York City homeless populations. Significantly, despite the findings of the studies it commissioned and the Bureau's own admission of an undercount, the agency did not correct its original estimate. Nor has the current Democratic administration [Jannuzi: what makes think things haven't gotten anymore accurate under Bush?] made the appropriate adjustments in its updates.
Many of those who realize that the homeless population is large nevertheless believe that the vast majority is unemployable. But mythology aside, many homeless adults are not unemployable any more than their lack of housing means that they are "unhouseable." It is important to break the bad habit of viewing structurally generated problems as attributes of individuals.
Aside from the problems produced by the decennial census, another serious shortcoming is rooted in the sample.
Homeless people are generally not interviewed because the household sample includes only those individuals and families with addresses. Also, around 15 percent of the sample are uncounted because the dwellings they had lived in are presumably vacant or now used for nonresidential purposes. We are certainly a nation of movers, but those who vacate their dwellings are not evenly distributed along racial and economic lines.
Blacks, Hispanics, and the poor are more likely to move than whites and middle class dwellers, whether as a result of being evicted, needing a less expensive arrangement, job seeking purposes, or for other reasons related to their economic duress. Also, dilapidated units are often assumed to be empty although, as a census study revealed, many are actually occupied. The same applies to many buildings that are incorrectly assumed to be nonresidential. These are not dwellings that shelter the middle class. Hardship also prompts poor individuals and families to move in with others, even if the arrangement is illegal or unapproved. And some are even forced into the streets. According to one study, one out of ten extremely poor residents become homeless.
Among the 50,000 occupied households in the sample, over 3,200 (6.5 percent, are not interviewed. Half of those approached decline to be interviewed. Other occupants may not be answering the door. What we have learned about non-respondents from the decennial census applies also to the monthly household survey. In addition, a census study found that a major source of error is missing people in housing units where a respondent supplies some information. As we have learned from the decennial census, the economically distressed households are more likely to refuse to answer questions or to answer them completely and honestly, particularly by those representing government agencies.
Also troubling is the relative size of the sample in low income communities. The number of households approached is based on the population size of the community, not the economic status of its residents. But because of the disproportionate number of vacancies and non-respondents in poor neighborhoods, proportionately fewer residents are interviewed in poorer than better off communities. As a result, the risk of error is greater, and accordingly, so is the likelihood of missing and undercounting unemployed workers.
To gauge the exact extent that the Census Bureau's methodology contributes to falsifying the unemployment rate would require, ideally, an independent national survey free as possible of the built in census biases. Such a survey would use the same criteria for defining the unemployed in order to isolate just the Census Bureau's shortcomings and problems. To my knowledge, no such survey has ever been conducted. However, almost 25 years ago, in 1975, the City of St. Louis commissioned researchers at a local university to conduct such a study of joblessness but only in St. Louis. City officials were concerned that St. Louis obtained its fair share of federal monies, the amount of which depended on the official unemployment rate.
The study left untouched BLS concepts or definitions. It only changed how the survey was conducted. Its sample was more carefully chosen, considerably larger, and more dispersed. In addition, its addresses were more up to date than the household survey's list. Experienced interviewers were hired and trained. Residents were also shown a letter from the Mayor endorsing the study and encouraging them to participate.
Over a three month period the survey revealed an average unemployment rate 44 percent higher than the official estimate of 11.7 percent. In one particular month it averaged almost 63 percent higher. Also, the unemployment rate of blacks rather than being about twice as high as whites, as officially reported, was instead three to one. This is not surprising because blacks are disproportionately undercounted in the household survey, and apparently the undercount excludes many of the unemployed.
No major improvements have been made in the household survey since then that would produce more realistic findings. In fact, both the homeless and the illegal population, who are greatly undercounted, have soared in the last 25 years. Certainly the gap between the official and real unemployment rates in the inner cities is greater today.
It involves some conjecture to estimate the extent to which Census Bureau's methodology currently deflates the national unemployment rate. Because of the concentration of poor people and alienated residents in the inner cities, the miscount is undoubtedly greater there. Still, the majority of the poor and unemployed reside elsewhere--in rural areas, many lower income suburbs, and on the other side of the track in economically comfortable suburbs. Some are even interspersed between well off households, which unkempt lawns and the long peeling exterior paint glaring at inpatient neighbors occasionally reveal.
The official national unemployment rate of 8.9 percent during those months of 1975 although lower than in St. Louis, was still very high, about two-thirds of St. Louis rate. Although the gap between the official and real rates is not as high elsewhere as in the central cities, it is still substantial. Let us assume that instead of the real unemployment rate being 44 percent higher than the official figures, as in the St. Louis study, it was only half as much, 22 percent, which is a conservative assumption. This would have then lifted the official national unemployment rate then by two percent (22 percent of 8.9). In the St. Louis study, the unemployed homeless were uncounted, and also the illegal alien population was much smaller than now.
The official unemployment rate in 1998 was 4.5 percent. Adding on another 22 percent would increase the official rate by about one percent. To make an additional but conservative estimate that includes both the out of work illegal aliens and the unemployed homeless would certainly lift the official rate by at least another one percent. The Household Survey only interviews persons with addresses, which the overwhelming majority of homeless individuals lack. And unquestionably many jobless illegal aliens avoid interviews with government employees.
Just by taking account of the Census Bureau's omissions would yield an official unemployment rate of 6.5 percent (4.5 + 2.0 percent). Correcting also for the shortcomings of the BLS definitions discussed earlier yields a double digit unemployment rate of over 11 percent (9.2 + 2.0 percent). Instead of 6.2 million unemployed, as the BLS claims, the actual number of unemployed last year was almost 16 million, which is over 2 and a half as many as the official count. It must be emphasized that the assumptions used to correct the official estimate lean heavily on the conservative side.
There are plenty of great big beautiful theories that have been destroyed by a little ugly fact. The ugly fact is that there are substantial numbers of people who although unemployed are systematically ignored both by the BLS and the Census Bureau.
They are often referred to by other names and titles--among them, delinquents, drug addicts, the mentally ill, illiterates, the retired, mothers at home, the handicapped, school dropouts, and members of the underclass. All of these embody an explanation that exonerates the public and private establishment from responsibility for closing the doors to a substantial segment of working people.
Just as those individuals are marginalized in actuality, they are marginalized for statistical purposes. It is therefore foolish to rely on the manicured official sources, which almost always provide us with a more polished scenario than in reality. From the perspective of working people, the American economy leaves no room for complacency. Beyond the pastel trimmings, it is bleak out there.
--------------------------
Posted by Charles Jannuzi