This paper examines the impact of market conditions on the behavioural tendency of market participants in a housing market. The focus of this study is to investigate whether property sellers conform or resist the main stream of local housing market conditions during the boom and recession periods. At the heart of the empirical strategy, this study assumes that all the sellers’ behavioural tendency are embedded in the similarity among transaction prices in local housing market, so that their conformity can principally be inferred from the strength of spatial dependence in the prices of surrounding properties. The strategy is based on the presence of spatial spillover effects that can be capitalised into house prices. For example, a higher or lower transaction price of neighbouring house than a current market price might signal other sellers to ask for an increased price or to be asked for a decreased price in local market. Through the capitalisation of the spatial spillover effects into property prices, a house tend to exhibit a value similar to that of a house in a nearby location, meaning housing values depend on those of neighbouring houses.

The main hypothesis of this study is that property sellers tend to take advantage of an ascending trend of transaction prices whereas strongly resist a descending trend, thus the spatial dependence in house prices may be higher during the boom period than the recession period. The hypothesis is investigated by comparing the spatial dependence during the boom and recession period in the Seoul housing market, South Korea. This study applies 17,290 transaction data during the boom period (January 2007- January 2010) and 13,257 transaction data during the recession period (January 2010 - March 2014). The main findings suggest that the sellers are more likely to follow the ascending trend of transaction prices than the descending one, indicating a much higher spatial dependence during the boom period than the recession period as expected. 

For the empirical methodology, this study employs the spatio-temporal autoregressive model to control spurious spatial connections that future transaction prices influence past ones. Since transactions take place at different points in time, ignoring time dimension may lead to an inappropriate model specification. The results suggest that the spatial only model overestimates the spatial dependence in the house prices significantly and performs worse than the spatio-temporal model.