Testing for Long Memory in Housing Prices in Korea

This paper investigates the persistence property of Korean housing prices, using a fractional integration framework. Based on the ARFIMA model, even though non-stationarity of the house prices of Seoul and its metro area cannot be ruled out statistically, the estimated persistence coefficient is not one rather less than one. For the other cities, the house price series is stationary and mean reversing. Our empirical results suggest that the property of Korean house prices is in favor of fractional degree of integration during entire sample period from January 1986 to October 2025. It is also found that the long-memory parameters are much larger in Seoul and its metro area than in other cities. For all house price series, three subsample periods are also found to have different parameters lying between zero and one, showing that the price series have the property of long memory in each period. Based on our empirical finding, housing policies need to be dependent on the regions and on time periods. House authorities require different policies, considering the fact that the differences both in regional and in time factors affect the degree of responsiveness of each housing market to shocks.