This study empirically analyzes the long-memory and regime-switching characteristics of apartment price indices in eight cities of South Korea (Seoul, Busan, Daegu, Gwangju, Daejeon, Ulsan, Gyeonggi, and Incheon). An R/S test and MS-ARFIMA model were applied for the period from June 2003 to September 2024.
Results of the R/S test revealed long-memory characteristics in apartment price indices across all the cities. Thus, we conducted additional analysis using the MS-ARFIMA model to capture long-memory and regime-switching characteristics simultaneously. The MS-ARFIMA model analysis confirmed both characteristics in all regions, with particularly strong long-memory characteristics observed in the contraction phase. Metropolitan areas demonstrate complex characteristics, with high volatility and strong long memory. Additionally, all cities show asymmetric characteristics where the contraction phase showed a higher persistence probability than the expansion phase. In contrast, nonmetropolitan areas demonstrated strong persistence during the contraction phases but relatively weak state persistence during the expansion phases. Therefore, preemptive policy responses when entering a housing market downturn, identification of appropriate intervention timing, and the implementation of sophisticated region-specific policy approaches are necessary.