Annotatsiya:

Abstract: The stochastic structure of Gross Public Debt to GDP ratio of Japan is studied for the first time with the help of a 150-year-long longitudinal dataset running from 1875 to 2024. The ARIMA Box-Jenkins framework was employed to help isolate internal time dependence and provide a statistically valid baseline for forecasting. The analysis revealed that the ARIMA(2,1,1) setting represented the preferred time series model, supported by the lowest AIC (6753.298) and BIC (6765.260) values. The results confirmed a highly autoregressive, persistent structure with both L1 and L2 coefficients highly statistically significant (p=0.000). Diagnostic tests including the Ljung-Box Q-statistic further supported the model's ability to filter fiscal signals from historical noise.