Page 52 - Azerbaijan State University of Economics
P. 52

THE                      JOURNAL OF ECONOMIC SCIENCES: THEORY AND PRACTICE, V.82, # 2, 2025, pp. 32-60

                    Inflation  dynamics  are  regime-sensitive;  for  instance,  INFLATION(-2)  is  significantly
                    negative (z ≈ -2.72), pointing to corrective mechanisms such as monetary policy or adaptive
                    expectations. Unemployment exhibits delayed reactions, notably under regime 2, where
                    CHOMAGE(-1) significantly correlates with inflation (z ≈ 2.68), reflecting labor market
                    stress  amid  price  instability.  Residual  standard  deviations  (SIGMA)  confirm  higher
                    volatility in regime 2, especially  for unemployment  and inflation.  The error covariance
                    matrix reveals significant positive correlations between LPIB and LDEP, consistent with
                    fiscally-driven  growth  dynamics.  The  model’s  robustness  is  supported  by  a  residual
                    covariance  determinant  of 1.25e-07,  log-likelihood  of 82.24, and favorable  Akaike and
                    Schwarz information criteria.

                    Impulse Response Analysis
                    Using  Cholesky  orthogonalization  with  degrees-of-freedom  adjustment,  the  impulse
                    responses reveal substantial regime-dependent differences. In the sTable regime, a positive
                    public spending shock leads to a moderate and gradual increase in real GDP, temporary
                    inflationary pressure, and a slight reduction in unemployment—patterns consistent with
                    Keynesian multipliers. Conversely, in the crisis or volatile regime, the same shock results
                    in weaker GDP responses, sharper inflationary spikes, and an ambiguous or delayed impact
                    on  unemployment.  These  findings  align  with  Auerbach  and  Gorodnichenko  (2012),
                    highlighting that fiscal multipliers are state-dependent—more effective in sTable periods
                    and significantly constrained during recessions or turbulent conditions.
                                          Historical Decomposition using Cholesky (d.f. adjusted) Weights
                                    GDP from Exp                             Exp from Exp
                    29.9                                     24.6
                                                             24.5
                    29.8
                                                             24.4
                    29.7
                                                             24.3
                    29.6                                     24.2
                    29.5                                     24.1
                                                             24.0
                    29.4
                                                             23.9
                    29.3
                                                             23.8
                    29.2                                     23.7
                      02  04  06  08  10  12  14  16  18  20  22  02  04  06  08  10  12  14  16  18  20  22
                                     Actuals                                   Actuals
                                     Baseline                                  Baseline
                                     Baseline + Exp                            Baseline + Exp
                                  INFLATION from Exp                         Unemp from Exp
                    10                                        28
                     8                                        24
                     6                                        20
                     4                                        16
                     2                                        12
                     0                                        8
                      02  04  06  08  10  12  14  16  18  20  22  02  04  06  08  10  12  14  16  18  20  22
                                     Actuals                                   Actuals
                                     Baseline                                  Baseline
                                     Baseline + Exp                            Baseline + Exp
                         Figure 5: Historical decomposition of the switching Markov VAR model
                                                    Source: By author


                                                           52
   47   48   49   50   51   52   53   54   55   56   57