Investment Analysis by Using Tax Assessment Database on Land Lots in Gwangjin District, Seoul, Korea

Abstract

This paper addresses the issues regarding the performance evaluation of generic investment strategies in land lots in a retail market trade area in a district in Seoul, Korea. To this end, this paper investigates 1993–2016 data on 6,478 parcels of commercially used land parcels in Gwangjin District, one of 25 districts in Seoul, the capital city of South Korea. Gwangjin District extends over 17 km2, with33,307 land lots and a population of 366,939 as of 2015. Because tax assessments in this district are determined by both certified property appraisers (appraisal) and government staff (mass appraisal) in a pegged order, it is generally accepted that the resulting assessed numbers are biased to certain level. Thus, we use the gradient measure to eliminate the most likely biases. Our innovative term of gradient is calculated by subtracting the assessed value of one land lot from that of another in the neighborhood; simply put, the difference of assessed values of two land parcels within certain distance bands. We analyze whole land lots in this district currently used for commercial purposes (retail shops and small offices such as FIRE and clinics), along two dimensions: assessed values and their gradients. What we call the HVHG group consists of land lots with higher assessed values and higher gradients. The HVLG group consists of land lots with higher assessed values but lower gradients. We also consider LVHG (lower assessed values and higher gradients) and LVLG (lower assessed values and lower gradients) groups in our analysis. This research finds that the differences in mean returns between HVLG and other groups except HVHG group are statistically significant based on a paired t-test.We test a couple of investment analyses. Our investment analyses control for the impact of both the value of land lots and distance from subway station on the realized rate of returns. We find that both the smaller valued and remote from the station groups beat the other groups in a statistically significant manner. In addition, a simple one-year momentum land investment strategy is again tested and the result is statistically not significant. Finally, we also visualize the retail market area and intertemporal changes in the market area.