Ⅰ. Introduction
According to the Korean government, 387 buildings that were left incomplete and abandoned were reported nationwide in 2016 (MOLIT, 2016).1) Because uncompleted and abandoned buildings (hereafter, “abandoned” buildings) harm the landscape, the demand for housing in the area decreases. This leads to a decrease in the prices of nearby houses and a deterioration in construction investment (Han, 2014; Mikelbank, 2008). This decrease in nearby house prices and the decline in construction investment bring about a recession in the local real estate market and ultimately worsen the real estate-related financial revenues of local governments. This may cause a vicious cycle that weakens the base of public financing for residential improvements (Scafidi et al., 1998). Additionally, abandoned buildings are sometimes engaged in criminal activities (Cui and Walsh, 2015; Immergluck and Smith, 2006).
White (1986) shows that the ratio of property tax to rental income has a statistically significant and positive effect on the abandonment rate of buildings in an area and that the elasticity of property tax on abandonment is high enough to range between 1.65 and 2.10. Arsen (1992) argues that it is problematic for White (1986) to define abandoned buildings as those with delayed property tax for 18 months to three years. He restricts the definition of abandonment to cases in which the building is undergoing the foreclosure process due to delinquent property tax. He suggests that the ratio of assessed value to market value has a significant positive effect on abandonment of the building, with elasticity ranging from 2.0 to 3.7. According to the results of Bender (1979), who conducted a case study in Chicago, while the local abandonment rate is negatively related to the average price per unit area of land and the median income of the area, and positively related to the median age of the buildings, the ratio of assessed value to market value does not have a significant effect.2)
Meanwhile, to promote maintenance work on an abandoned building by public intervention with substantial public financing, it is necessary to ensure the feasibility of the project by objectively and scientifically demonstrating the benefits, such as a reduction in the negative impact of the abandoned building. This paper empirically analyzes the negative impacts of an abandoned building on the prices of nearby houses in Cheonan, Korea. The analytical method used in this study is a difference-in-differences estimation, where unique geographic characteristics are utilized as an identification strategy. According to an empirical analysis, a building abandoned due to the suspension of construction appears to lower the prices of the neighboring houses by 432 thousand Korean Won per m2. This finding justifies public intervention with financial investment for the proper management of abandoned buildings.
The rest of this paper is structured as follows. The next section discusses the theory of the negative economic effects of an abandoned building, specifically neighborhood effects or neighborhood choice. The data and variables used in the empirical analysis are described in Section 3. Section 4 presents the results of the empirical analysis on the economic impacts of a building that has been incomplete and abandoned for at least two years. The last section summarizes the results of this paper, presents the implications of the analysis results, and suggests future tasks based on the limitations of the study.
Ⅱ. Literature Review and Analytic Models
According to Kim and Kim (2013), buildings left unattended for a long period of time after an interruption in the construction process undermine the image of the city, which is a decisive factor affecting the city’s competitiveness as a residential area. It is necessary to address this through active public intervention rather than simply regarding it as a matter of private ownership.
Abandoned buildings and homes blight the landscape and can thus reduce the demand for housing in the affected areas. This leads to a decline in housing prices and a deterioration in investments in housing construction, causing a downturn in the local real estate market. Ultimately, it may also result in a vicious circle that weakens the public finance base for improving residential environments by aggravating the real-estate-related financial revenues of local governments. In addition, relatively lower profitability would shrink demand for real estate development by the private sector.
Mikelbank (2008) examines the impact of both foreclosures and vacant/abandoned properties in Columbus, Ohio, in 2006. Its key finding is that for a property located near foreclosed and vacant/abandoned properties, the price value is reduced by an average of $8,600 ($4,256 by foreclosed properties and $4,411 by vacant/abandoned properties).
Han (2014), based on the Baltimore, Maryland case, examines the impact of abandoned properties on nearby property values. It controls for both nearby foreclosures and local housing market trends by using longitudinal data sets.
According to Kobie and Lee (2010) using U.S. data, the value of houses near foreclosed properties falls, and negative effects occur not only in the year immediately following foreclosure but also correlate with the length of the foreclosure period and the value of nearby real estate. In other words, while an immediate effect is seen within a relatively short period of time, one year after foreclosure, the magnitude of these negative effects gradually decreases over time.
The presence of an abandoned building implies a neighborhood effect that negatively impacts the region. This effect is not limited to the drop in prices of nearby houses. Rather, as the neighborhood selection model predicts, due to the overall downturn in the local economy and the deterioration in residential environments, the population would decrease and only relatively poor people would remain. Since a small population and the residents’ low income do not guarantee sufficient demand for the development of local real estate projects, the local residential environments would worsen.
Cui and Walsh (2015) explores the impact of residential foreclosures and vacancies on violent and property crime by utilizing a difference-in-differences framework. The results indicate that violent crime rates increase by approximately 19% once the foreclosed home becomes vacant whereas foreclosure alone has no effect on crime.
Using more than 9,600 single-family property transactions in Chicago, Immergluck and Smith (2006) finds that foreclosures of conventional single-family loans have a significant impact on nearby property values. One of the estimation results show that each conventional foreclosure within an eighth of a mile of a single-family home results in a decline of 0.9% in value.
Hwang and Choi (2024) shows that spatial and transportation facilities have significant effect on the prices of nearby multifamily attached housing units. However, their effects are opposite each other, that is, a positive effect of spatial facilities and a negative effect of transportation facilities are observed.
Choi (2022) empirically investigates the welfare consequences of gambling facilities by using local housing markets. The main results show that the facilities have a negative effect on the price of nearby houses, meaning that local residents have negative willingness-to-pay for gambling facilities.
In this paper, a quantitative analysis of housing prices is conducted for a specific abandoned building using a difference-in-differences framework. If the decision to abandon the incomplete building is exogenous, that is, not caused by a recession of the local real estate market and if the two groups, i.e., the treatment and comparison groups, would have shown a similar trend in variables related to real estate markets without the abandonment, the difference-in-differences approach would be appropriate to control for counterfactuals.
A typical regression equation for the difference-in-differences framework can be written as follow:
Here, treatj is a dummy variable indicating the treatment group; it is 1 if the housing unit, j is assumed to be negatively affected by the abandoned building, and 0 otherwise. postt is a dummy variable representing the time points after the building was abandoned. The effect of the abandoned building on the prices of houses in the neighborhood is estimated through the coefficient of the interaction term between treatj and postt, β3. There is a possibility that the price of neighboring houses may drop due to the abandoned building, but if house prices fall due to other factors, it is likely that the building under construction may be left incomplete. In this paper, the building was abandoned not due to the decline in the prices of nearby homes but due to a private financial problem that affected the construction firm.3) Thus, there are no major concerns about endogeneity issues.
The empirical analysis in this paper focused on one abandoned building, specifically an apartment complex located in Sungjung-dong, Seobuk-gu, Cheonan, Korea. Because it has been left unfinished for a considerably long period of time (approximately, 74 months), has a large area of 46,034m2, and is up to 70% completed, its negative effect on nearby houses is significant enough to be acknowledged. The Ministry of Land, Infrastructure and Transport evaluated its safety as grade “C” which indicates how it is serious in the safety aspects.
<Figure 1> shows the location of the abandoned building to be analyzed, indicated by the arrow, and the geographical distribution of the houses included in the regression analysis. For the application of the difference-in-differences model, the selection of nearby houses is important. The comparison group, which is assumed not to be affected by the abandoned building but by other factors, such as cyclical factors or improvements in residential environments, should be defined by houses that would have experienced the same trend in sales prices as those in the treatment group if the building had not been abandoned. First, as shown in <Figure 1>, it is assumed that houses within 500m from the abandoned building experience a negative influence from it. The houses in the red rectangle in <Figure 1> are defined as the treatment group.
On the other hand, the comparison group is defined as houses located in the blue rectangle in <Figure 1>. As shown in the figure, these houses are very close to the treatment group as well as the abandoned building, but a railroad lies between the abandoned building and the comparison group. This suggests that changes in the prices of houses belonging to the comparison group due to economic factors other than the abandonment of the building are similar to those of the treatment group, whereas the prices of houses belonging to the comparison group are not influenced by the abandoned building because the two groups are geographically separated by the railroad and thus socially disconnected. The width of the railroad is 128m, which is sufficient to prevent people from coming and going between the two areas. Therefore, the existence of the railroad would preclude the negative effects of the abandoned building from spilling over from the treatment group to the comparison group.
Ⅲ. Data and Variables
The data used in the empirical analysis of this paper consist of actual housing transactions provided by the Ministry of Land, Infrastructure and Transport through public systems. Actual sales transactions have been disclosed since January 2006 for houses (for example, apartments, townhouses, single-family/multi-family homes), officetels, land, and houses for which real estate transaction reports and housing transactions have been reported. This applies to real estate such as commercial/office use, factories, warehouses, etc., and apartment sales/occupancy rights signed after June 29, 2007. The date on which the contract was made is provided, and the housing types are categorized into apartments, multi-family housing, single detached housing, and officetels, a category unique to Korea. In addition to the address of houses, including the lot number or road name, information about the name of the complex, the size of exclusive area, the transaction price, the number of floor, and the year of construction is provided.
<Table 1> presents the basic statistics of the actual transaction prices per m2 of the exclusive area included in the analysis, along with the number of years since construction, the size of the exclusive area, the number of floor, and the type of housing (i.e., whether it is an apartment or not). On average, approximately 19 years have passed since the housing units were constructed, the size of exclusive area is 70.57m2, and the units range from the basement level to the 23rd floor.
With regard to single-family homes, the heterogeneity between individual houses is significant, and variables that can be used to control for this heterogeneity are not sufficiently available. Thus, they are excluded from the analyses. In this paper, apartments and multi-family homes are differentiated, and 93.4% of the actual transactions used for empirical analyses correspond to apartments.
<Table 2> shows the distribution of the actual transactions by year. The proportion of actual transactions by year is generally similar throughout the analysis period of 2008–2018. However, it is relatively high in 2008, 2013, and 2014, while it is somewhat lower in the more recent years, from 2016 to 2018. The total actual transactions are divided into the treatment group and the comparison group. The distribution of actual transactions of the treatment group by year ranges from 3.7% to 15.7%, while for the comparison group, it ranges from 5.6% to 12.3%. A relatively large difference in the frequency of actual transactions by year can be observed in the treatment group, but this is due to the fact that the total actual transaction frequency for the treatment group is very low compared to that of the comparison group. However, the regression analysis of the difference-in-differences model assumes that the negative impact of the abandoned building would appear significantly starting in 2013, and the proportion of the actual frequency between the two groups before and after 2013 is not significantly different. Therefore, certain types of bias in the estimation, such as sample selection bias, are unlikely to be of concern.4)
Ⅳ. Empirical Results
<Table 3> shows the regression results of the effect of an abandoned building on the prices of neighboring houses based on the difference-in-differences model. Column (1) does not include the characteristics of individual housing units (i.e., the exclusive area, floor, years elapsed since construction, whether it is an apartment or not); whereas, Column (2) includes these characteristics.
First, considering the differences in actual transaction values according to the characteristics of the houses in Column (2), it can be observed that the actual transaction values are higher on average when the exclusive area is smaller. This occurs because the dependent variable is defined as the actual transaction value per m2 of exclusive area. The smaller the house size, the higher the price per unit area. This reflects the recent preference for small and medium-sized housing rather than larger units. In addition, the prices of high-rise units are higher than those of low-floor units. Furthermore, the older the house, the lower the price.5) These results are compatible with the usual expectations found in previous studies. In terms of value as an asset, ordinary apartments are preferred over multi-family houses, which means that real-estate transaction prices are higher for apartments. These results are in line with Henderson and Ioannides’ (1983) theoretical arguments that housing is both a consumer good and an investment good.
According to the analysis in Column (1), the prices of the houses near the abandoned building are as low as 924 thousand Korean Won (equivalently, 770 dollars) per m2 of exclusive area, regardless of the time. However, much of this difference is due to differences in the characteristics of the houses (i.e., the area, the floor, years since construction, and whether it is an apartment or not) between the two groups. After controlling for the difference of these characteristics between the treatment and the comparison group, it can be seen in Column (2) that the difference in the average actual transaction price per m2 between the two groups is drastically reduced to 185 thousand Korean Won (equivalently, 154 dollars).
The difference in actual transaction prices before 2013 and since 2013, within the period of 2008–2018, is statistically significant, with prices since 2013 being relatively higher. Given that year-fixed effects are controlled for, the rise in prices since 2013 is not due to the increase in the price index but rather to some good news about the real estate market common to both groups.
The change in the price of nearby homes due to the abandoned building is captured by the interaction term, ‘Homes near the abandoned building* 2013 or later.’ Here, ‘Homes near the abandoned building’ is the dummy variable representing the treatment group, and the dummy variable ‘2013 or later’ represents the year 2013 or later. In Column (1), where the characteristics of individual houses are not controlled for, the prices of homes near the abandoned building are estimated to have fallen more than those of homes across the railroad by 386 thousand Korean won (equivalently, 322 dollars) per m2. In Column (2), where the characteristics of individual houses are controlled for, the degree to which those homes near the abandoned building have experienced a relatively larger fall in their price is greater, i.e., a drop by 432 thousand Korean Won (equivalently, 360 dollars) per m2.6)
<Table 4> shows the results for different analysis periods. The years 2009–2012 are excluded under the assumption that people continued to recognize the negative impact of the abandoned building during this period, while 2007 is added to the analyses in order to make up for the loss of samples. In <Table 5>, the estimates of the coefficients for ‘Homes near the abandoned building* 2013 or later’ are –84.5 in Column (1) and –88.2 in Column (2), indicating that the negative impacts of the abandoned building on the local housing market were considerably underestimated in <Table 3>.
<Appendix Table 1> in the appendix shows the results when analyzing all housing units in Sungjung-dong as the treatment group and those in Shinbu-dong as the comparison group when also expanding the geographical range of the analysis. In Column (1), where the characteristics of the individual houses are not controlled for and the negative effect of the abandoned building, that is, the relative price decline of the houses in Sungjung–dong, does not appear. However, in Column (2), where the characteristics of houses are controlled for, the estimated value of ‘Sungjung-dong* 2013 or later’ is –31.8 and statistically significant. This implies that the geographic range of the negative impacts is wider than in <Tables 3 and 4>. However, it can be concluded that the geographic range of the negative impacts is limited because the absolute value of the estimates is much smaller when larger areas are included in the analysis.
<Table 5> presents the results of the analysis of the model with the fixed effects of the individual complexes controlled for. Column (1) does not control for the characteristics of each house, whereas Column (2) does. The variable ‘Years since built’ is excluded from the model because it is correlated with the fixed effects of the year and apartment complex. Estimates of the coefficient for the negative impacts on neighboring houses near the abandoned building, that is, ‘Homes near the abandoned building* 2013 or later,’ are –70.2 in Column (1) and 71.7 in Column (2). This implies that not controlling for individual complex-fixed effects would not cause any bias in the estimation.
Ⅴ. Conclusions
Abandoned buildings that have been left incomplete and unattended for a long time deteriorate the image of the city, which is a determinant of urban competitiveness. They also act as a deterrent to the revitalization of commercial areas and are considered a matter for public intervention rather than private ownership, given the locational individuality and uniqueness of land resources. These issues, however, must be resolved through public intervention as a common good (Kim and Kim, 2013). Nonetheless, if the maintenance project for an abandoned building is driven by the public, a substantial amount of public funds will be involved. Therefore, public intervention decisions must be based on an objective and scientific estimate of the project’s benefits or the negative effects of the abandoned building.
This study empirically analyzes the negative effects of a building that has been left partially constructed on the prices of neighboring houses. The analysis framework is a difference-in-differences model, and the data used in the analysis are actual transaction prices provided through the open trading system of the Ministry of Land, Infrastructure, and Transport. According to the empirical analyses, the abandoned building with incomplete construction reduced the value of neighboring houses by 432 thousand Korean Won per m2, indicating that the negative effect is quite significant.7) This result may justify public intervention to address the negative externalities from an abandoned building.
Setagaya-ku in Japan has promoted life-based application projects for vacant homes through the “Building Support Program for Local Communities” (Nam, 2014). This strategy is also being implemented at the local government level (Lee, 2013). In Korea, the Ministry of Land, Infrastructure, and Transport is promoting policies such as the Long-term Unused Structure Construction Projects and LH’s Illegal Building Maintenance Support Organization, but local governments need to take the lead in promoting projects that align with the characteristics and conditions of the areas near abandoned buildings. Most importantly, the central government may extend the investigation period from two to three years, which may not reflect the reality due to the lack of timeliness in the only basic data available for establishing relevant policies. It is necessary to not only conduct a survey but also establish an annual survey cycle. Currently, a building is defined as abandoned starting two years after it has been left unattended, but this is only an administrative definition. It would be more reasonable to define a building as abandoned from the point at which residents near the building recognize its negative effect. As observed in the empirical results of this study and in the case of the US (Kobie and Lee, 2010), the negative effects of abandoned buildings begin to appear relatively quickly.
The results of this study empirically suggest a decline in the prices of neighboring houses due to an abandoned building. However, because there are limitations in clarifying the geographical and temporal ranges of these effects, it is difficult to estimate the total extent of the negative impacts of abandoned buildings. If the changes in the prices of neighboring houses can be examined at various points in time when an abandoned building is to be demolished or if construction is resumed through the original plan or an alternative project, it will be possible to analyze the geographical and temporal scope of the negative effects caused by abandoned buildings.
