Printer Friendly

Japanese unemployment: BLS updates its analysis.

In a 1984 article in the Review, we presented an analysis of Japan's Labor force data and concluded that the official Japanese unemployment rates are only slightly understated in relation to U.S. concepts.(1) The data analyzed in the article were from the "Special Survey of the Labour Force Survey" (referred to as the special survey hereafter), conducted in Japan in March 1977 through 1980.

This report updates the article by analyzing data from the 1984 through 1986 special surveys which were conducted in February. Unlike the March surveys, the February surveys indicate that official Japanese unemployment rates are slightly overstated relative to U.S. concepts. In any event, the February results confirm the broad conclusion drawn from the earlier study: Japanese unemployment rates are virtually unchanged when U.S. concepts are applied.

Our article noted that it was difficult to draw firm conclusions from the March data because March is a very unusual month for the Japanese labor market. It is both the end of the fiscal year, when Japanese firms traditionally take on new workers to start April 1, and the end of the school year, when new graduates enter the labor market. Although February is also a month of higher than average unemployment for Japan, there is less seasonality associated with this month than with March, and the February results for 1984-86 provide new information about what may be expected in a more typical month.

The original BLS article was partially a response to a 1983 Review article by Koji Taira which also analyzed the March 1977-80 surveys.(2) In contrast to the BLS view of these surveys, Taira concluded that the Japanese jobless rate would be "nearly double the official unemployment rate" if U.S. concepts were used. Although both BLS and Taira found it necessary to make several adjustments to Japanese unemployment to bring it more in accord with U.S. concepts, BLS, by contrast, found Japanese unemployment to be only slightly understated.

A 1984 article by Sadonari Nagayama, former director of the Japanese Statistics Bureau, also reached conclusions different from Taira's.(3) Nagayama argued that Taira's adjustments were too large, particularly the adjustment in which he classifies as unemployed more than 500,000 students who graduated in March and would start work in April. Information from the February 1984-86 surveys throws further light on this issue.

The special surveys of February 1984-86 were not available to Taira or BLS when the earlier articles were written. After reviewing the surveys, BLS believes they support the contention that the Japanese unemployment rate is only slightly changed when U.S. concepts are applied. This report presents an analysis of the February surveys, including a breakdown of the results by sex. In addition, unemployment rates using an expanded concept of unemployment are calculated and compared.

Japan's special survey

To supplement its monthly labor force survey, the Japanese Statistics Bureau conducts special surveys once or twice each year to investigate, in more detail, the labor force status of the population and provide data needed for making employment policies. The themes of the special surveys change according to the social and economic circumstances and data needs at the time of each survey.

The underlying purpose of the special surveys from 1977 through 1980 was to investigate, in detail, the rise in the unemployment rate which began after the first "oil crisis." Later surveys had other emphases. For example, the March 1981 survey highlighted the situation of part-time workers and the 1983 survey presented a current labor force status versus usual status comparison. The differing underlying themes necessarily influenced the whole structure of the survey questionnaires. Modifications in questions and wording were made, not without a sacrifice to the continuity of the time series. As a result, the special surveys of 1977-80 were useful in quantifying the differences between Japanese and U.S. unemployment concepts, while the 1981-83 surveys were unsuitable for that purpose.

The 1984-86 special surveys returned to a questionnaire format similar to that used in the 1977-80 surveys, again producing the kind of data needed for adjustment to U.S. concepts. Moreover, the 1984-86 surveys were taken in February instead of March, thus eliminating at least some of the seasonality associated with the end of the fiscal and school years.

Adjustment to U.S. concepts

Several adjustments are made to the special survey data to bring them closer to U.S. concepts. Some persons counted as unemployed in the surveys should be excluded from the labor force, and some reported as not in the labor force should be included among the unemployed. The magnitude of each of the adjustments is significant, but, on balance, they tend to cancel each other out, leaving the Japanese unemployment rate virtually unchanged. Table 1 shows the 1984-86 adjustments for February along with the March 1980 figures, generally in the same format as table 4 in the 1984 article.

Unemployment. Most of the adjustments relate to the unemployed. The Japanese surveys report as unemployed a number of persons who did not actively seek work during the past month. The reasons for this relate to the wording of the survey questions (this is explained in our March 1984 article). "Inactive jobseekers" are subtracted from the reported unemployed for comparability with U.S. concepts. They amounted to 20 to 25 percent of the reported unemployed in the February surveys, compared with over 40 percent in the March surveys.

However, there are two groups of persons reported as not in the labor force who, upon further questioning, reveal that they should be counted as unemployed under U.S. concepts. Both groups are classified as not in the labor force in the Japanese survey because they initially respond that their status is housewife, student, or retired, rather than jobseeker. One group, responding to more probing questions later in the survey, stated that they had sought work in the past month and could have started work immediately if a job had been found. These persons amounted to 7 percent of adjusted unemployment in the February surveys and about 30 percent in the March surveys.

The other group from outside the labor force comprises persons who were waiting to begin a new job within 1 month and available for work. In March 1980, the following data were reported in the survey results:
 Total waiting to begin a new job .......... 860nds)
 Within 1 month ............................ 740nds)
 After graduation in March ................. 550nds)
 Other ..................................... 190nds)
 After I month ............................. 120nds)

Taira's adjustment on this point was to add 740,000 persons-that is, all persons waiting to begin a new job within 1 month-to the Japanese unemployed. In contrast, the BLS adjustment added only the 190,000 persons who were not students to the unemployed figure, excluding those who were waiting to start jobs after graduation in March. BLS omitted school graduates rather than including them in the upward adjustment to the unemployed for three reasons: (1) although most had already attended graduation ceremonies, it was questionable whether they were available for work prior to April 1; (2) they would not be included in the count in any month but March; and (3) there is hardly any chance that the jobs they were waiting to start would disappear (the surveys were taken during the last week of March and new jobs traditionally begin the first week of April).

The availability of the graduates was open to question because graduation ceremonies usually take place in early March. The students were not asked whether they wanted to begin work sooner than April 1, but this was a possibility. However, we maintain that the compelling reason to exclude them is to put the March surveys on a more typical basis.

Taira's method has the effect of using the March surveys as representative of the Japanese labor market over the course of the year. He compares the March results for Japan with annual average data for the United States and other countries.

When weturn to an analysis of the February surveys, the situation becomes clearer. There is no longer any valid argument to include the students waiting to begin new jobs in the Japanese unemployed count. Because these surveys are taken in February, students are still in school and, therefore, not available to take up their new jobs until after graduation in March. U.S. concepts require that persons waiting to begin new jobs within 30 days must be available to start work during the survey's reference week in order to be classified as unemployed. We suggest that none of the Japanese students should be included in the unemployed in February even under Taira's conceptual framework.

The following results were reported in the February 1984-86 surveys:
 1984 1985 1986
 (thousands) 1986
 Total waiting to begin a new job.... 1,640 1,460 1,640
 Within 1 month...................... 1,340 1,130 1,300
 After graduation in March........... 1,170 960 1,100
 Other............................... 170 170 200
 After 1 month....................... 310 340 330

The 170,000 to 200,000 persons who were not students waiting to begin a new job within 1 month are added to the Japanese unemployed for comparability with U.S. concepts. This adjustment accounted for 10 to 12 percent of the adjusted unemployed. The BLS adjustment to the March surveys on this point represented 15 to 20 percent of the adjusted unemployed.

Japanese unemployment was higher in 1984-86 than in the 1977-80 period. Nevertheless, all of the adjustments to unemployment were smaller in the February surveys than they were in the March 1977-80 surveys. There were fewer inactive job seekers to subtract and fewer unemployed from outside the labor force to add. These results are a reflection of the higher seasonality of March compared with February. Moreover, there are a greater number of inactive jobseekers to subtract in the February surveys than there are jobseekers not in the labor force to add. This is the reverse of the situation in March 1977-80 when jobseekers outside the labor force surpassed (1977-79) or balanced (1980) inactive jobseekers.

The earlier BLS study presented the adjustments based on the March surveys as "upper limits" because of the high seasonality of the March period. The February surveys support this view.

Labor force. The adjustments to the labor force for comparability with U.S. concepts are relatively small. Japan includes and the United States excludes unpaid family workers who worked less than 15 hours in the survey week. As indicated earlier, a number of unemployed persons officially classified as not in the labor force should be added to the Japanese labor force for comparability with U.S. concepts. However, some of the officially unemployed should be subtracted-the "inactive jobseekers." On balance, these adjustments reduce the reported labor force by 1 percent in both the February and the March surveys. (See table 1.)

The outcome. Whereas the March 1980 and earlier surveys indicated that the reported Japanese unemployment rates needed to be increased slightly to be in accord with U.S. concepts, the later surveys of February 1984-86 indicate that the Japanese rates should be decreased slightly. The reported rate of 3 percent in February 1984 is reduced to 2.8 percent after adjustment; the data for 1985 and 1986 are reduced from 2.8 to 2.7 percent. (See table 1.)

Comparisons by sex

Although the overall Japanese unemployment rate is changed only slightly when the special survey data are adjusted to BLS concepts, there is a more noticeable difference in the adjusted rates for men and women. The conventional Japanese data by sex show virtually no difference between the unemployment rates for men and women. However, according to the BLS adjustments, there is a significant differential between the male and female rates which may have converged somewhat between the two periods under consideration. For instance, in March 1977-80, the female rates were about double the male rates, but in February 1984-86, the rates for women are about one-third higher than the rates for men. This convergence may be real, but it may also be attributed to higher sex differentials in March than in February. Without February and March data for the same years, it is impossible to tell. The following tabulation shows unemployment rates for men and women from the March and February surveys (based on civilian labor force):
 Period As published U.S. concepts
 Men Women Men Women
 March 1977 ......... 2.4 2.3 2.0 4.3n
 March 1978 ......... 2.7 2.4 2.2 4.3n
 March 1979 ......... 2.5 2.4 1.9 4.1n
 March 1980 ......... 2.2 2.3 1.7 3.3n
 February 1984 ...... 3.0 3.0 2.5 3.3n
 February 1985 ...... 2.9 2.8 2.4 3.1n
 February 1986 ...... 2.8 2.8 2.4 3.3n

Thus, after adjustment, the Japanese situation appears more like Western countries where women usually have higher unemployment rates than men.

The reason for the wide male-female differential for Japan after the adjustments are made is that women account for the great majority of jobseekers classified as not in the labor force, while men account for most of the reported unemployed who did not actively seek work in the month of the survey. (See table 2.)

An expanded unemployment concept

Japan's unemployment rates, both on their official basis and adjusted to U.S. concepts, are well below U.S. rates. Annual U.S. jobless rates of 7.5 percent in 1984, 7.2 percent in 1985, and 7.0 percent in 1986 contrast with adjusted Japanese rates of 2.7-2.8 percent in February. These February rates for Japan were probably slightly higher than the annual averages because published February rates were 0.2 percentage points above the annual average. Other Western nations (Canada, France, the Netherlands, the United Kingdom) had rates in the 10- to 14-percent range during this period.(4) Is the efficiency of the Japanese labor market really 2 to 5 times better than that of the Western nations? A strict comparison of unemployment rates would arrive at that misleading conclusion. However, a substantial part of Japan's labor underutilization falls into the realm of underemployment (workers on reduced hours) and discouragement, or labor force withdrawal. These forms of labor slack do not show up in the conventional unemployment rate.

The March 1984 article provided comparisons based upon expanded concepts of unemployment which exist in the United States within the unemployment measures designated as U-1 to U-7.(5) These monthly measures include the official unemployment rate U-5. While U-1 to U-4 represent narrower measures of unemployment, U-6 and U-7 represent expanded concepts. Persons on part-time schedules for economic reasons are incorporated in U-6, and U-7 brings in discouraged workers, that is, persons who want a job but are not looking for work because they believe their search would be fruitless.

Table 3 updates the expanded concepts comparisons to 1984-86, and revises the U-7 calculation for 1980. Data from the February special surveys for Japan are compared with annual average data for the United States. The Japanese figures should be considered as only approximate indicators of U-6 and U-7.(6)

Since publication of the 1984 article, BLS has reassessed the Japanese data on discouraged workers and has concluded that they should more properly be expressed as a range. The Japanese survey questioning procedure differs substantially from the U.S. procedure, and it is difficult to make an exact fit to the U.S. concept. Discouraged workers are, by nature, a subjective phenomenon, and precise measurement in any country is an elusive proposition. An appendix to this articleprovides further information on the discouraged worker comparison between the United States and Japan.

In Table 3, the lower rate of the U-7 range includes persons who seem to fall strictly within the U.S. concept of discouraged workers; the upper rate of the range includes some who may not be counted under the U.S. definition, but they would fall under a broader concept of labor underutilization,

Comparisons of the U-6 and U-7 rates in relation to the conventionally defined rate (U-5) show that the Japanese "expanded" rates are increased to a greater degree than the U.S. U-6 and U-7 rates. In other words, there is a convergence in the "unemployment rates" for the two countries when the definition is broadened. The convergence was somewhat greater in 1984-86 than in 1980.

Under the conventional definition of unemployment (U-5), table 3 shows that the U.S. rate is 2.5 to 2.7 times the Japanese rate in 1984-86. Expanding the concept to include persons working part time for economic reasons (U-6), the U.S. rate is about twice the Japanese rate. When defining unemployment even more broadly to encompass discouraged workers (U-7), we find that the U.S. rate falls to only 1.2 to 1.4 times the Japanese rate at the low end of the U-7 range. At the higher end, the rates converge even more, to the point that the Japanese rate surpasses the U.S. rate in 1985 and 1986. But it should be emphasized that the upper Japanese U-7 includes some persons who might not be classified as discouraged under U.S. definitions.

Expanding the unemployment concept to include other elements of labor slack-economic part time and discouraged workers-draws the Japanese rate closer to U.S. levels. Explanations for any remaining differential lie in such factors as the composition of the labor force, levels of frictional unemployment, and economic growth rates.


(1) Constance Sorrentino, "Japan's low unemployment: an-indepth analysis," Monthly Labor Review, March 1984, pp. 18-27.

(2) Koji Taira, "Japan's low unemployment: economic miracle or statistical artifact?" Monthly Labor Review, July 1983, pp. 3-10.

(3) The Nagayama article was originally published in Japanese in Nihon Rodo Kyokai Zasshi, March 1984. An English translation of the article appears in "Are Japan's Unemployment Statistics Too Low?" Economic Eye (Economic Affairs, Keizai Koho Center), June 1984, pp. 14-18. Copies are available from BLS upon request.

(4) For international unemployment rates approximating U.S. concepts, see tables 45 and 46 in the "Current Labor Statistics" section of the Monthly Labor Review.

(5) The U-1 to U-7 framework was introduced in Julius Shiskin, "Employment and unemployment; the doughnut or the hole?" Monthly Labor Review, February 1976, pp. 3-10.

(6) See Sorrentino, p. 26, for further discussion of measurement problems and estimating methods.
COPYRIGHT 1987 U.S. Bureau of Labor Statistics
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1987 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:Bureau of Labor Statistics
Author:Sorrentino, Constance
Publication:Monthly Labor Review
Date:Jun 1, 1987
Previous Article:Weekly earnings in 1986: a look at more than 200 occupations.
Next Article:OECD meeting calls for job growth, flexibility, and readjustment.

Related Articles
Revisions of state and local area labor force statistics.
Adjusted Japanese unemployment rate remains below 3 percent in 1987-88.
Labor force data in the next century.
Japan's low unemployment: a BLS update and revision.
U.S. labor market performance in international perspective: from 1960 to 2000, U.S. unemployment rates improved from relatively high to the lowest...
The World According to TED.
Obtaining information from the Bureau of Labor Statistics.
Hurricanes and labor markets.
Focus on unemployment.

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters