Looking at Regional Disparity through Employment Indicators
February 20 (Friday) 2009
Naoki Nagashima
Senior Research Fellow
Summary
- The Japanese economy maintained a roughly five and a half year period of growth from the January-March quarter in 2002 to fall 2007. Along with economic recovery, “widening disparity” also garnered attention during this time. “Disparity” is mainly referred to in economic terms and typically within the context of income disparity among regions, companies of differing sizes, and different types of employment.
- From a critical perspective of the Koizumi Administration’s economic policy, awareness of such disparity issues has been expressed in clichés by the mass media and opposition parties such as “preferential treatment of the wealthy” and “cities favored, rural areas ignored”. Discussion on disparity has not been solely economic, with books such as “Society with a Disparity of Hope” by Masahiro Yamada also attracting attention.
- However, even if limiting to “central and rural disparity” and “regional disparity”, there are various views on what to consider disparity and what to use as a measure. Per capita income is frequently used as a proxy variable to represent standard of living. Because each region has different price levels, however, income level cannot be equated to standard of living. There is also the problem of age demographics differing among regions, and the danger of underrating the standard of living in a region with a high ratio of elderly households with low income but large assets.
- In addition to income, disparity indicators such as the Indices of Industrial Production (IIP) are also used. Such indicators, however, are flawed in that they do not consider industries outside of manufacturing, such as the service industry. This research uses the effective ratio of job offers to applicants (annual average of the monthly index), a leading employment indicator, to compare from the perspective of living standards.
Fixed bottom rankings is a problem
Comparing the 2007 effective ratio of job offers to applicants (annual average) by prefecture, Aichi Prefecture with its strong manufacturing emerged on top at 1.95, while Okinawa Prefecture placed last at 0.22. Fukui (1997) and Tochigi (1987) Prefectures topped the list ten and twenty years ago, respectively, with periodic change in between. In terms of disparity between cities and rural areas, Tokyo has never finished in the top five, and similarly there is no evidence that prefectures included in the Tokyo metropolitan district such as Kanagawa, Saitama and Chiba are at high levels. The top five are in constant shuffle, and no prefecture has remained in the top five over the last 30 years.
Conversely, the other end of the rankings has been fixed with Kochi, Aomori and Okinawa prefectures rounding out the bottom both 10 and 20 years ago. More so than disparity, the industrial structure of these prefectures with stagnating employment can be viewed as the problem.
How has disparity transitioned over the years? Calculating the average of the top five prefectural effective ratios of job offers to applicants compared to the average of the bottom five reveals the following: 3.3 times higher in 2007; 3.0 times higher in 1997; 4.8 times higher in 1987; and 5.9 times higher in 1977. While the gap in 2007 was slightly larger than in 1997, compared to the previous decades prefectural disparity has actually shrunk.
Widening disparity, but only compared to mid-90s levels
For a more accurate long-term look at prefectural disparity, the coefficient of variation related to the effective ratio of job offers to applicants was measured on a monthly basis starting in 1977. The results are listed chronologically in the chart below (refer to PDF file for charts). Variation is typically shown using variance or standard deviation, but the coefficient of variation is derived by dividing standard deviation by the average, and thus dependence on units is avoided (1).
The chart confirms increasing variation (disparity) in prefectures since 2002. This disparity, however, is the same as 1996-1998 levels, and compared to the time before that regional disparity has narrowed. Regional disparity has been shrinking since the mid-1980s, with a reversal of this trend confirmed in 2002. There has been little fluctuation since 2005, with no movements towards widening disparity.
The discussion on disparity is a complicated one. Efforts by the media – which tends to assess the problem in terms of whether disparity has “widened or shrunk” in the short-term – to accurately interpret the current levels have been insufficient. It is necessary to understand the long-term trends. At the minimum, there should be discussion based on multifaceted verification results backed by concrete data. The news media, with its sensational examples and titles such as “Society of Widening Disparity”, should at times be viewed with a sense of skepticism.
(1) For example, when measuring household annual income for a group in Yen units and then in 10,000 Yen units, the former standard deviation is 10,000 times higher and variance is 100 million times higher than the latter. In contrast, the coefficient of variation does not vary according to the unit of measurement.