ESRI Discussion Paper Series No.292
Measuring the Effects of Monetary Policy: A DSGE-DFM Approach

Hirokuni Iiboshi
Graduate School of Social Science, Tokyo Metropolitan University


I propose new method of measuring the effect of monetary policy on a large number of macroeconomic series by combining dynamic stochastic general equilibrium (DSGE) with a dynamic factor model (DFM) in a data-rich environment including a broad range of useful information for both central banks and the private sector. This method can resolve two problems inherent in the FAVAR approach proposed by Bernanke, Boivin and Eliasz (2005), yet can enjoy its benefits as well. Other positive aspects of this method are to split off components of model concepts (or common factors) and measurement errors from all observable variables, and to identify structural shocks, including monetary policy shocks, from the point of view of DSGE based on a microeconomic foundation with rational expectations. This new method calculates historical decomposition, which provides useful information for policy analysis for researchers and policymakers, as well as theoretically reasonable impulse response functions and variance decompositions for all observable variables of large panel data. Through this method I will provide empirical illustrations of 55 Japanese macroeconomic series during the late 1980s and the 1990s using a Smets-Wouter (2003, 2007) type medium size DSGE model.

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

  1. Full Text別ウィンドウで開きます。(PDF-Format 2.6 MB)
  2. Abstract
  3. page1
    1 Intoduction
  4. page5
    2 DSGE-DFM approach
    1. page5
      2.1 Framework of DSGE-DFM model
      1. page5
        2.1.1 State Space Representation of Dynamic Factor Model (DFM)
      2. page6
        2.1.2 FAVAR Approach for Measuring the Effects of Monetary Policy
      3. page7
        2.1.3 State Space Representation of DSGE-DFM
      4. page8
        2.1.4 Two Types of Data Indicator Xt
      5. page11
        2.1.5 DSGE-DFM Approach for Measuring the Effects of Monetary Policy
    2. page11
      2.2 The DSGE model
      1. page12
        2.2.1 Equilibrium Conditions from Housing/Investor Sector
      2. page13
        2.2.2 Equilibrium Conditions from Firm Sector
      3. page14
        2.2.3 Miscellaneous Equilibrium Conditions
      4. page14
        2.2.4 Structural Shocks and Forecast Errors
      5. page15
        2.2.5 System of the Log-Linearized Model
    3. page15
      2.3 Estimation Method
    4. page18
      2.4 Policy Simulation
      1. page18
        2.4.1 Impulse Response Function
      2. page19
        2.4.2 Historical Decomposition
      3. page19
        2.4.3 Variance Decomposition
  5. page20
    3 Application
    1. page20
      3.1 Empirical Implementation
      1. page20
        3.1.1 Observable Varaibles and Measurement Errors
      2. page21
        3.1.2 Calibration and Prior of Parameters
      3. page22
        3.1.3 Data
    2. page23
      3.2 Empirical Results
  6. page25
    4 Conclusion
  7. page26
    A Appendix
    1. page26
      A.1 Sampling of State Variables St from Simulation Smoother
    2. page28
      A.2 Sampling of Parameters Set Γ of Measurement Equation (2.7)
  8. page30
  9. page45
    Appendix Table 1
  10. page46
    Appendix Table 1(continued)
  11. page47
    Appendix Table 2. Prior Distributions of the Parameters
  12. page48
    Appendix Table 3. Posterior Distributions of the Parameters
  13. page49
    Figure 1. Smoothed Variables of 21Sensor Series
  14. page50
    Figure 2. Smoothed Variables of 34Information Series
  15. page51
    Figure 2. (continued)
  16. page52
    Figure 3. Contribution Ratio of 7Model Concepts for 34Information Series
  17. page53
    Figure 3. (continued)
  18. page54
    Figure 4. Structural Shocks
  19. page55
    Figure 5. Impulse Response Function of 21Sensor Seriesto Monetary Policy Shock
  20. page56
    Figure 6. Impulse Response Function of 34Information Seriesto Monetary Policy Shock
  21. page57
    Figure 6. (continued)
  22. page58
    Figure 7. Historical Decomposition of Sensor Series.
  23. page66
    Figure 8. Historical Decomposition of Information Series.
  • 1-6-1 Nagata-cho, Chiyoda-ku, Tokyo 100-8914, Japan.
    Tel: +81-3-5253-2111