This paper examines Ghana's inflation dynamics using a combination of regime switching and self-exciting point process models, complemented with formal jump-dependence tests. A Gaussian Hidden Markov Model (HMM) is applied to monthly Consumer Price Index (CPI) year-on-year changes to identify latent inflation regimes. Structural jump events are detected using change-point methods, and their temporal dependence is analyzed through a univariate Hawkes process estimated with the Expectation-Maximization (EM) algorithm. Christoffersen and Ljung-Box tests are employed to formally assess jump clustering and autocorrelation.
Download