ESRI Discussion Paper Series No.341
Structural Change and Business Cycle Fluctuations in Japan: Revisiting the Stylized Facts

Satoshi Urasawa
The Economic and Social Research Institute, Cabinet Office


The Japanese economy has experienced massive structural changes since the end of the 1990s, including a decline in the working-age population, along with a decade of deflation, an increase in the number of non-regular workers, which has almost doubled since the early 1990s, contributing to a large reduction in wage costs, and a rapid advance in globalization. What are the implications of such changes for the understanding of Japan’s business cycle dynamics? This paper analyses the stylized facts of Japanese business cycle fluctuations under structural change.

The empirical evidence, based on traditional frequency domain analysis using more than 60 quarterly macroeconomic time series, provides robust facts. Among the most interesting findings is that the role of scheduled hours worked as a buffer for labor input has become increasingly important, suggesting that Japanese firms tend to adjust their labor input through hours worked, owing, in part, to the increasing number of non-regular workers, which allows firms to adjust labor input in a relatively flexible manner while keeping the number of employees unchanged. The increased role of hours worked is confirmed by an analysis based on a time-varying parameter structural vector autoregression (TVP-VAR) model taking the time-varying nature of the underlying structure of the economy into account. Meanwhile, in other areas such as private consumption and investment, wages, deflators and prices, and financial market indicators, the basic nature of business cycle fluctuations has remained broadly unchanged, implying that structural change does not necessarily affect the cyclical regularities in all macroeconomic time series.

Structure of the whole text(PDF-Format 1 File)

  1. Full TextI will open in a new window(PDF-Format 472 KB)
  2. page1
  3. page2
    1. Introduction
  4. page4
    2. Examination of economic structural change
  5. page5
    3. Frequency domain analysis framework
    1. page6
      3.1 BP Filter
    2. page7
      3.2 Data description and statistics
  6. page7
    4. Business cycle characteristics and changes therein
    1. page7
      4.1 Business cycle characteristics before 2000
    2. page9
      4.2 Changes in business cycle characteristics
  7. page12
    5. Analysis of changes in business cycle characteristics using a TVP-VAR model
    1. page13
      5.1 Structure of the model
    2. page15
      5.2 Bayesian estimation of the model
    3. page15
      5.3 Data and settings
    4. page16
      5.4 Estimation results
  8. page18
    6. Conclusion
  9. page19
  10. page21
    1. page21
      Figure 1: Cyclical component of real GDP
    2. page21
      Figure 2: Real GDP and its trend component
    3. page21
      Figure 3: Result of the CUSUM test
    4. page22
      Figure 4: Change in cross-correlation: Comparison of before 2000 and since 2000
      1. page22
        (a) GDP components
      2. page23
        (b) Employment
      3. page24
        (b) Employment (continued)
      4. page25
        (c) Wages
      5. page26
        (d) Deflators and prices
      6. page27
        (e) Interest rates, stock prices, money
    5. page28
      Figure 5: Developments in total labor supply (man hours): Contribution of hours worked and number of employees
    6. page28
      Figure 6: Impulse responses of GDP, scheduled hours worked, and employees to a demand shock: TVP-VAR model (left) and constant VAR model (right)
    7. page29
      Figure 7: Developments in simultaneous correlation coefficients obtained from TVP-VAR model
      1. page29
        (a) Simultaneous correlation of scheduled hours worked with demand shocks
      2. page29
        (b) Simultaneous correlation of number of employees with demand shocks
  11. page30
    1. page30
      Table 1: Statistics for cyclical component: Before 2000
    2. page31
      Table 2: Statistics for cyclical component: Since 2000
    3. page32
      Table 3: Selected parameter estimates of the TVP-VAR model
  • 1-6-1 Nagata-cho, Chiyoda-ku, Tokyo 100-8914, Japan.
    Tel: +81-3-5253-2111