中東の紛争件数に見られる短期的な時間的持続性

Conflict events in Middle Eastern countries exhibit complex temporal fluctuations. As a simple baseline, we examine whether their conflict-frequency time series contain measurable linear temporal structure using the autoregressive integrated moving average (ARIMA) framework, which assumes linear dependence on past observations and errors. Despite this simple linear assumption, the temporal dependence is largely captured for most countries by ARIMA. Predictions based on the selected ARIMA models reproduce the overall temporal behavior to some extent, although abrupt changes remain difficult to predict. We further examine the short-term persistence of changes in conflict activity using the first zero-crossing point $k^*$, which characterizes the duration of the initial positive temporal correlation, and the integrated correlation time $\tau_{\mathrm{int}}$, which additionally reflects its strength and decay. The resulting $k^*$ values are approximately 2--6 days, while $\tau_{\mathrm{int}}$ ranges from approximately 0.6 to 1.7 days. The significant negative association of $\tau_{\mathrm{int}}$ with mean GDP per capita suggests a possible relation between short-term conflict persistence and broader socioeconomic conditions. These results reveal measurable linear temporal structure and country-dependent short-term persistence in the temporal patterns of conflict activity across the Middle East.

11 pages, 8 figures