A Study on the Impact of Home Occupant Characteristics on Housing Renewal and Improvement in Management-type Residential Environment Improvement Project Areas in Seoul: Focusing on the Moderating Effect of Home Occupancy Status

1. Research background and purposes

The rapid and significant economic development in Korea during the late 20th century coincided with radical transitions in the country’s housing supply strategies and policies. Specifically, in the 1960s, a series of land readjustment schemes reshaped the country’s housing landscape, transitioning from naturally developed housing to a more planned and organized urban approach. This resulted in the formation of residential areas primarily comprised of detached houses (Ha, 2010; La Grange and Jung, 2004). A rising population within cities, coupled with increased density in detached residential areas, led to the widespread construction of illegal buildings and structures, such as unauthorized rooftop houses (Kim and Kwon, 2020). In response to the increasing population influx into the Seoul Metropolitan Area, housing development shifted its focus to apartment complexes in the 1970s (Green et al., 1994; Thomas and Hwang, 2003).

Furthermore, the execution of housing redevelopment projects as part of national policies since the 1980s, along with the emergence of various new housing types, such as multifamily houses and multi-unit houses (Jang and Kwon, 2017; Kim et al., 2005), resulted in a continuous decrease in detached dwelling residential areas. As a result, as of 2019, detached dwellings accounted for 23.6% of Seoul’s entire residential areas (Lee and Nam, 2020).

Meanwhile, low-rise residential areas are often identified as land suitable for potential development, aimed at supplying mass housing for urban residents, due to their low population density and affordable costs. Consequently, these areas are prone to high levels of development pressure (Kim, 2021). However, low-rise houses hold significant importance as affordable housing options while also contributing to diversity in housing types (Hong and Ahn, 2013). Therefore, it is essential to devise effective plans and policies to proactively organize and manage them. Many local governments have thus far made efforts, such as management-type residential environment improvement projects, reshaping the paradigm of urban planning for aged low-rise residential areas from complete redevelopment to regeneration and small-scale development. The gradual execution of residential environment improvements, following this paradigm shift in urban planning, requires a primary focus on improving individual housing units rather than pursuing redevelopment schemes at the complex level.

Here, housing improvement refers to any specific actions undertaken to repair and enhance dwellings, aimed at addressing their physical deterioration over time.

While homeowners can control the pace of physical deterioration for their houses by investing in housing repair and maintenance individually, the costs required to maintain housing quality and services continue to increase as the houses age (O’Sullivan, 2004). Thus, homeowners choose housing improvement methods in a way that maximizes the benefit at the moment when the marginal cost becomes equivalent to the marginal benefit of housing improvement. In Korea, housing improvement encompasses any services required to be reported under the Building Act for detached houses, including new constructions, extensions, and major repairs, as well as those that do not require any legal procedures.

On the domestic front, management-type residential environment improvement projects¹⁾, which were first introduced in 2012, have been supplemented by subsequent R&D initiatives and policies. These efforts aim to address the country’s deteriorating housing environment resulting from the increasing number of aged houses.

The purpose of the management-type residential environment improvement projects is to supplement the infrastructure for low-rise residential areas in need while encouraging residents to voluntarily engage in housing renewal and improvement. However, these projects have not been highly effective in promoting voluntary housing renewal and improvement efforts due to their failure to consider regional characteristics and variations (Maeng et al., 2015). According to Maeng et al. (2017), although these projects are implemented under the Act on the Maintenance and Improvement of Urban Areas and Dwelling Conditions for Residents, the provisions of the act primarily prescribe methods for residential improvement when apartment complexes are built through complete demolition and redevelopment. The discrepancy between the purpose of the project and the specific provisions of the act has resulted in the initiative’s failure to devise detailed action plans or strategies for housing environment improvement. Moreover, despite the availability of loan programs for housing improvement, the lack of institutional systems to support and implement them limits the contribution of these projects to promoting voluntary housing improvement efforts.

Against this backdrop, it is essential to make R&D efforts and develop necessary policies that encourage residents to voluntarily engage in the improvement of their aged houses within the project locations.

Meanwhile, previous studies have reported that the decision for housing improvement is influenced by various factors, including the distinctive characteristics of homeowners, houses, and surrounding environments, as well as the nature of the housing market (Littlewood and Munro, 1996; Plaut and Plaut, 2010; Sari, 2014). However, there has been little research conducted on this issue on the domestic front. Unlike redevelopment projects for apartment housing, decisions regarding the improvement of individual houses are solely made by homeowners, regardless of whether they reside in them or rent them to tenants. For house owners, the purpose of housing improvement is either to enhance the quality of their own residence or to increase the value of their houses as part of economic activities. Under these perspectives, it is also necessary to consider whether homeowners themselves reside in the houses as a factor that may affect decisions for individual housing improvement, rather than defining all homeowners as a single category.

Within this framework, this study identified factors influencing the decisions of homeowners regarding housing renewal and improvement within the target areas of the management-type residential environment improvement projects, with special attention given to the nature and characteristics of homeowners. In addition, the factors proposed in previous studies were reinterpreted from the perspectives of homeowners.

Based on the results, this study aimed to identify factors that could further promote housing renewal and improvement efforts, while also providing relevant insights for future urban planning.

2. Research scope and methodology

The spatial scope of this study includes eight sites among the target areas of the management-type residential environment improvement projects in Seoul. These sites are of similar size and have already completed the public sector portions of the projects. During the selection of target sites, the project size was considered because factors beyond individual characteristics may influence decisions for housing renewal and improvement, particularly if each project varies significantly in size. The temporal scope of this study was set from 2011²⁾, when the management-type residential environment improvement project areas were first designated, to 2020. In this study, a broad definition of housing renewal and improvement encompasses even non-official housing repairs without validated records. However, among them, only those repairs that can be verified in building registers were counted due to difficulties in data acquisition. More specifically, housing renewal was limited to new constructions, while housing improvement included extensions, renovations, major repairs, and changes of intended usage³⁾.

Variables affecting decisions for housing renewal and improvement were identified and selected through a literature review. For the 874 land parcels within the selected eight management-type residential environment improvement project areas, a series of variables regarding the nature and characteristics of homeowners, parcels, houses, regions, and urban planning were derived. First, t-test and chi-square test methods were employed to gain insights into the fundamental characteristics of homeowners and their preferences for housing renewal and improvement, with a particular focus on whether they reside in the houses themselves. Subsequently, multinomial logistic regression models were used to analyze factors influencing their decisions regarding housing renewal and improvement.

A cluster analysis was conducted using SPSS Statistics 26, IBM’s statistics software package, to identify target analysis sites. A logistic regression analysis, which is particularly useful when dependent variables constitute nominal scales, was then performed using Stata 17 SE, another statistics tool. A multinomial logistic regression model, used when there are three or more dependent variables, was employed as the regression model. For the housing renewal and improvement type as a dependent variable, “No change” was included as a reference category to derive individual probabilities for each provided option, including “Renewal (new construction)” and “Improvement (extension and renovation).” The analysis in the present study utilized a series of data from the Ministry of Land, Infrastructure and Transport, including building registers, land ledgers, continuous cadastral maps, and officially assessed land prices.

1. Factors influencing housing improvement decisions in low-rise residential areas

Since studies on housing renewal and improvement have predominantly been conducted by overseas researchers rather than domestic researchers, our literature review focused on overseas research cases. Oxley and Smith (1995) confirmed that decisions for individual housing improvement were affected by various factors, including housing investments and other financial reasons, demographic factors, and factors associated with economic growth. The researchers, however, noted the need for further specification of these factors by introducing microscopic parameters. Mendelsohn (1977), Pollakowski (1988), Ziegert (1988), and Bogdon (1996) examined microeconomic factors influencing decisions regarding housing improvement, particularly for homeowners residing in the houses themselves. Shear (1983) identified the characteristics of households, houses, neighborhood environments, and regions as factors that could affect the decisions of residents to either improve their detached houses or relocate. Montgomery (1992) utilized various datasets from the 1985 US Census of Housing to devise a probit model that includes the cost spent for housing improvement as a variable. With this approach, the researcher assessed how this cost was affected by different factors, including the characteristics of homeowners (such as income, period of ownership, marital status, and age), houses (including the level of deterioration, parcel size, and floors) and neighborhood environment (quality and available services).

Boehm and Ihlanfeldt (1986) revealed the noticeable effects of various factors, including crime rates, the presence of public schools, the quality of media reports, and rates of littering, on residents’ decisions to improve their houses. Culp (2011) found that environmental factors influenced the decision regarding housing renewal and improvement, particularly among homeowners who moved in within the past five years. Clark and Kearns (2012) focused on the perspectives of individuals who actually reside in the houses, whether homeowners or tenants, revealing that a close homeowner-tenant relationship positively influenced decisions regarding housing renewal and improvement. Studies by Sternlieb (1966) and Porell (1985) reported that in the US housing market, the quality of housing was more effectively managed when homeowners actually resided in their detached houses or in apartments, serving as managers, rather than living in other regions as lessors. Davila-Ash (2002) also pointed out that the external appearance of houses appeared more poorly maintained when the homeowners did not actually reside in them, especially as the rental period grew longer. In addition, Lang et al. (2021) highlighted that financial factors were not the sole determinants of decisions regarding housing renewal and improvement from the perspectives of rental housing investors in Europe, Australia, and the US. These studies are significant in that they considered the type of ownership for individual homeowners in the analysis, despite that their interpretations cannot be directly applied to the Korean housing market due to significant differences between the markets.

In Korea, empirical research on housing renewal and improvement has only recently begun, resulting in a limited number of relevant studies.

Lee and Bae (2007) employed a housing value maximization model under the assumption that homeowners want to maximize the value of their houses. They used this model to determine the optimal timing for housing development with respect to economic growth rates through simulations for different development types, including maintenance, remodeling, and reconstruction. Specifically, in areas with low economic growth rates, housing development did not start on time, leading to rapid deterioration. In contrast, when economic growth rates were high, development occurred promptly within a short period of time, maximizing the value of houses. Hong and Ahn (2013) employed game theory to investigate the effect of rent regulations, density regulations, and homeowners’ preferences on decisions regarding housing renewal and improvement. The researchers confirmed that suitable combinations of polices, such as rent regulations and gross floor area incentives, could be effective in promoting decisions for housing renewal and improvement. As an empirical study on housing renewal and improvement, Yim and Choi (2016) used data from the Korea Housing Survey to assess individuals’ intent for housing improvement. They also employed logistic regression and tobit models to examine the effects of the characteristics of detached house residents and their houses on both decisions regarding housing improvement and the cost that they would be willing to pay for such improvements.

As seen above, empirical studies using data on the execution or cost of housing renewal and improvement began among overseas researchers in 1980. In Korea, this approach has only recently started to gain momentum gradually. This shift aligns with the gradual establishment of methods for residential environment improvement, highlighting the necessity for more specific empirical research on this topic. Meanwhile, previous studies regarding factors affecting decisions for housing renewal and improvement (Mendelsohn, 1977; Pollakowski, 1988; Ziegert, 1988; Bogdon, 1996; Boehm and Ihlanfeldt, 1986; Yim and Choi, 2016) did not differentiate between homeowners who resided in the houses themselves and those who rented their houses to tenants. Most studies analyzed specific houses in which their owners resided themselves, regarding them as the primary decision-makers for housing renewal and improvement. Given that housing rent transactions represent a significant portion of the housing market, deriving determinants for housing renewal and improvement without considering ownership type is considered a limited approach. Notably, management-type residential environment improvement projects, which will be discussed later, often target low-rise residential areas, encouraging residents to actively engage in renewing and improving their individual houses. Consequently, it is crucial to accurately understand the nature and characteristics of residents, particularly concerning ownership type.

2. Management-type residential environment improvement projects

First introduced under the Act on the Maintenance and Improvement of Urban Areas and Dwelling Conditions for Residents in 2012, management-type residential environment improvement projects aim to promote public sector investments in improving the overall quality of infrastructure, including streets and roads. This, in turn, encourages residents to voluntarily make efforts for housing renewal and improvement. Creating this synergistic cycle is the ultimate goal of these projects. Although the public sector does not directly participate in housing renewal and improvement tasks, its efforts, such as infrastructure improvement, can effectively control the pace of deterioration while also encouraging residents’ voluntary participation in housing improvement initiatives.

However, since the implementation of these projects, public-sector initiatives have been completed in some target areas, raising questions about the effectiveness of the projects in promoting decisions for housing renewal and improvement (Maeng et al., 2015). While private-sector investments directly lead to the renewal or improvement of individual houses, public-sector efforts are indispensable for attracting such investments (Smets and Weesep, 1995). Therefore, it is necessary to specify and accurately understand the characteristics of residents residing within the target areas of management-type residential environment improvement projects, which aim to promote voluntary participation in housing renewal and improvement initiatives.

Previous studies on management-type residential environment improvement projects can be largely divided into those focusing on understanding, assessing, and improving the projects and others conducting empirical research.

Yeo and Yang (2015) analyzed the Doil Traditional Market in Siheung to develop methods for addressing the limitations of the current form of the projects, particularly with respect to project execution, residential participation, and the activation of residential communities. Maeng and Baik (2017) also emphasized the activation of residential communities. They analyzed the current status of public policies, community organizations, and residential activities to identify the limitations of policies currently in place, as well as methods for improving them.

Kim and Shin (2018) focused on public facilities for residents within the target areas of management-type residential environment improvement projects. The researchers conducted a literature review and in-depth interviews to assess the impact of the presence and operation of these facilities on the competence of residential communities. They also proposed methods for improving efficiency in the operation of these facilities, thereby enhancing the capacity of residential communities. A study by Bae and Park (2019) also examined residential communities and their public facilities, identifying sustainable methods for more efficiently operating these facilities. Kim (2019) evaluated residential satisfaction with management-type residential environment improvement projects. Despite residents recognizing the need for the renewal and improvement of aged houses, their actual participation was not as extensive. Thus, the researcher emphasized the necessity for more effective methods to promote voluntary participation in housing renewal and improvement initiatives.

Meanwhile, several empirical studies have been conducted within the target areas of management-type residential environment improvement projects.

Jeong (2015) argued that public-sector investments during management-type residential environment improvement projects would lead to an increase in land prices and rents, resulting in gentrification and potentially displacing existing residents. To support this notion, the researcher compared land prices and rents before and after the projects. Kim and Kim (2017) also analyzed Yeonnam-dong, one of the target areas of management-type residential environment improvement projects, to identify patterns in residents’ decisions for housing improvement across each housing type, especially with respect to specific categories, including the age of owners, period of ownership, and ownership type. Their research holds significance in that it categorizes the characteristics of housing improvement decisions based on the age of land or building owners and the period of ownership, utilizing data available in building registers. However, the researchers only described the nature of land or building owners by dividing parcels within the target areas into those with changes and others left unchanged, failing to specifically identify the factors influencing owners’ decisions regarding housing improvement.

Overall, a literature review of management-type residential environment improvement projects revealed that most previous studies have focused on residents, with earlier studies emphasizing the need to activate residential communities while also providing assessments and improvement methods.

In addition to the activation of residential communities, some studies have examined the operation of public facilities for residents, evaluating residential satisfaction with them through surveys conducted after the completion of public sector portions of the projects. However, all these studies were conducted from the perspective of public sector projects. There have not been many empirical studies conducted from the perspective of the private sector aimed at promoting housing renewal and improvement efforts.

3. Unique aspects of this research

This research differs from previous studies in the following aspects.

First, this study examines cases of housing renewal and improvement, which have not been very common domestically. Notably, factors affecting decisions regarding housing renewal and improvement are analyzed with a primary focus on the characteristics of homeowners, which are known to be important determinants for housing renewal and improvement decisions. According to previous studies, homeowners tend to become more passive in engaging in housing improvement as they age, and as the length of their ownership period increases (Shear, 1983; Montgomery, 1992; Baum and Hassan, 1999; Yim and Choi, 2016; Kim and Kim, 2017). However, these results did not consider whether the houses were owner-occupied or rental properties, and nor did they account for the size of housing renewal and improvement projects. In general, for owner-occupied houses, housing renewal and improvement are aimed at enhancing housing services. In contrast, renewal and improvement efforts for rental properties, where owners rent the properties to tenants, are often made for investment purposes (Meijer, 1993). Thus, this study specifies the occupancy status of homeowners, determining how this distinction will affect the results compared to those of previous studies.

Second, most previous studies on factors influencing decisions for housing renewal and improvement were conducted through scenario analyses using theoretical models (Lee and Bae, 2007; Hong and Ahn, 2013). Despite employing empirical analyses, some studies failed to take a microscopic approach at the parcel level because they used national-level data, such as those from the Korea Housing Survey (Yim and Choi, 2016).

This study utilizes empirical data at the parcel level to conduct quantitative research, distinguishing its approach from previous studies.

Finally, similar to the empirical research on domestic housing renewal and improvement initiatives discussed above, the present study also focuses on the target areas of management-type residential environment improvement projects. However, it extends the scope by considering additional areas where public sector portions of the projects have already been completed. These are the unique aspects of this research that differentiate it from previous studies.

1. Research problem and hypothesis

Does the occupancy status of homeowners affect the characteristics that influence their decisions for housing improvement? This question aligns with the intent to specifically identify the factors that drive homeowners’ decisions in improving their houses from their own perspectives. This study proceeds under the common-sense assumption that the purpose of housing improvement may vary depending on the occupancy status of homeowners. With this notion in mind, our aim is to verify the moderating effect of occupancy status on the individual characteristics of homeowners.

Research problem: Does the degree of the effect of homeowners’ characteristics on decisions for housing improvement vary depending on their occupancy status?
→ Hypo thesis: The occupancy status of homeowners has a moderating effect, influencing the impact of homeowners' characteristics on decisions regarding housing improvement.

2. Subject of analysis

As of 2021, public-sector portions of projects had been completed in 28 sites among the target areas of the residential environmental management projects implemented by the City of Seoul. According to Niu and Han (2018), the construction of a building typically requires approximately three years from the completion of the land transaction. Thus, 20 sites where public sector portions were completed at least three years prior, specifically before 2018, were initially selected for analysis. Subsequently, IBM’s SPSS Statistics 26 was employed to conduct a cluster analysis on the selected sites with respect to their area. As a result, a total of three clusters were identified. Among the sites within the three clusters, eight sites with an area close to 50,000 m² were finally short-listed. The area of 50,000 m² is a threshold considered appropriate for the projects by the City of Seoul (Maeng et al., 2015). The eight selected sites are as follows: Dolsam Haengbok Village, Hanminjok Village, Bakmi Sarang Village B, Onsu-gol, Woori Village, Sansae Village, Saedongnae, and Hwigyeong Village. The results of the cluster analysis and selected sites are summarized in <Table 1> and <Figure 1>.

Table 1.

Area of analysis selected by clustering analysis

Figure 1.

Case study area

Next, the age of homeowners, period of ownership, and their occupancy status were examined by cross-referencing building registers and land ledgers.

During this verification process, parcels for which owners could not be properly identified due to the establishment of land rights for multi-owned buildings were excluded. As a result, a total of 874 parcels were selected for analysis.

The Dolsam Haengbok Village is located at 306, Samseong-dong, Gwanak-gu. Among the selected sites, this village exhibits the highest average homeowner age and the longest average period of ownership. It has a grid road system, with most of its square and rectangular parcels situated on sloping land. The Hanminjok Village is at 1027-1, Daelim-dong, Yeongdeungpo-gu. This site includes a relatively large number of parcels, totaling 198 units. These parcels are mostly situated on flat land, with many of them being square, rectangular, and trapezoidal, while some are irregularly shaped. Bakmi Sarang Village B is situated at 957, Siheung-dong, Geumcheon-gu. This site has the lowest average homeowner age and a relatively short average period of ownership. It includes the smallest number of parcels subject to analysis. This is attributed to its distinct housing type, which is primarily composed of apartment housing, including multifamily houses. Onsu-gol is located at 67, Onsu-dong, Guro-gu. This site is characterized by the highest proportion of irregular-shaped parcels among all parcels subject to analysis.

In addition to these parcels for analysis, there are a number of apartment houses, such as multifamily houses and row houses within this site. The Woori Village is located at 330-6, Bukgajwa-dong, Seodaemun-gu. The village is situated on completely flat land, with some of its parcels being trapezoidal and flag-shaped. It is the site where the management-type residential environment improvement project was first implemented, and its public sector portions were completed first among all selected sites. The Sansae Village is situated at 237, Sinsa-dong, Eunpyeong-gu, with over half of its area designated as Class I General Residential Area. This site has the largest number of parcels for analysis, totaling 242 units. This suggests that its housing type was primarily composed of detached houses and multi-unit houses in the past. Saedongnae is at 280, Dobong-dong, Dobong-gu, with all its parcels analyzed being situated on flat land. This site is characterized by its spatial structure composed of two separate zones. Finally, the Hwigyeong Village is located at 286, Hwigyeong-dong, Dongdaemun-gu. Most of its area is situated on sloped land, and similar to Hanminjok Village and Sansae Village, the site is predominantly composed of detached houses and multi-unit dwellings, both in the past and present.

3. Variable derivation and data sources

In this study, variables were derived through a literature review, and data were prepared to match each selected variable at the parcel level. The selected variables are described in <Table 2>.

Table 2.

Variables utilized in the analyses

Data for the housing improvement type, as a dependent variable, were sourced from the building permit data available from the Ministry of Land, Infrastructure and Transport. According to the data, the housing improvement type was initially categorized into New construction, Extension, Large-scale improvement, and Change of intended usage⁴⁾ and further refined into New construction and Small-scale improvement⁵⁾ for further analysis. Although previous studies (Yim and Choi, 2016; Kim and Kim 2017) have varied in their categorization of housing renewal and improvement types, this study adopted the categories of New construction and Small-scale improvement for variation derivation. This decision was made considering that the City of Seoul provides its housing improvement subsidies differently based on this categorization. Parcels, in which no construction works had been undertaken since the commencement of the management-type residential environment improvement project, were categorized as No change.

The homeowner characteristics, considered a key variable in the analysis, were sourced from the building registers and land ledgers available from the Ministry of Land, Infrastructure and Transport, specifically regarding their age and period of ownership. Generally, older homeowners are less likely to undertake renewal or improvement projects on their houses, and they tend to spend less on such improvements (Shear, 1983; Montgomery, 1992). Additionally, the longer they reside in their home, the less likely homeowners are to undertake housing renewal and improvement projects (Montgomery, 1992; Baum and Hassan, 1999). Accordingly, in the present study, the age of homeowners and the period of ownership were selected as variables to represent their characteristics. Furthermore, homeowner records specified in the building registers and land ledgers were cross-referenced to enhance accuracy.

Montgomery (1992) employed a wider range of variables, including household incomes and marital status. However, adopting this approach directly in our study was challenging because the researcher utilized data available from the Census of Housing. Moreover, other previous studies have utilized the age of homeowners and the period of ownership as variables representing homeowners’ characteristics. Consequently, in this study, both the age of homeowners and the period of ownership were defined as independent variables.

Meanwhile, the occupancy status of homeowners was selected as a moderating variable. Homeowners who reside at the same address as their houses were categorized as “Owner-occupied,” while those who reside elsewhere were categorized as “Non-owner-occupied.” While the occupancy status may be considered part of homeowner characteristics, previous studies have reported almost no cases where the occupancy status serves as an independent variable.

This can be attributed to the unsuitability of this variable for use in the modeling process or difficulties in obtaining the required data. Based on our hypothesis that preferences for housing renewal and improvement may vary between owner-occupied and non-owner-occupied cases, the occupancy status of homeowners was categorized as a moderating variable in this study.

Parcel characteristics, house characteristics, neighborhood characteristics, and urban planning characteristics were also selected as control variables. Parcel characteristics have frequently been employed as variables in previous studies on housing renewal and improvement, often represented by specific individual variables, such as parcel area, form, slope, and land price (Montgomery, 1992; Iwata and Yamaga, 2007; Kim and Kim, 2017; Lee and Nam, 2020). Similar types of variables have been used in other studies employing statistical models at the parcel level (Cho and Lee, 2018; Kim et al., 2019; Park and Kwon, 2023), although the focus of these studies extends beyond housing. Data for physical variables, including parcel area, form, and slope, were obtained from the continuous cadastral maps available on the Korea National Spatial Data Infrastructure Portal, with the exception of land prices. Data for the land price variable was sourced from officially assessed land prices provided by the Korea Real Estate Administration Intelligence System of the Ministry of Land, Infrastructure and Transport. For parcels with no changes, year-on-year land price increases from 2021 were measured, while for parcels that had undergone any construction work, year-on-year land price increases from the year of the corresponding work were estimated. Meanwhile, average land price increases vary from year to year and from region to region. Thus, the obtained data did not accurately represent net price increases in each target area. To address these limitations, this study adopted relative land price increase rates⁶⁾, which were estimated by dividing each parcel’s price increase by the land price change at the gu (district) level in Seoul. Seoul’s land price changes at the gu level were sourced from the Korea Real Estate Board’s data on year-specific, region-specific land price changes.

House characteristics were often represented by the age of houses in previous studies (Montgomery, 1992; Baum and Hassan, 1999; Baker and Kaul, 2002; Yim and Choi, 2016). While overseas studies primarily focus on detached houses, domestic studies have examined a broader range of housing types due to the distinct features of the housing market in the country (Yim and Choi, 2016; Kim and Kim, 2017). Therefore, the age of houses and house type were ultimately selected as variables to represent house characteristics.

Data for both variables were obtained from the building registers provided by the Ministry of Land, Infrastructure and Transport. More specifically, the house type was categorized into detached houses, multi-unit houses, and houses for commercial use. For parcels where multifamily houses, row houses, and apartment buildings already existed or were newly constructed before the commencement of housing renewal and improvement, it was impossible to extract homeowner data at the parcel level because land rights had already been established.

Meanwhile, previous studies have utilized a wide range of specific variables to analyze neighborhood characteristics.

For example, some researchers employed variables, such as crime rates, construction cost indexes, and tax rates (Boehm and Ihlanfeldt, 1986; Montgomery, 1992). However, this study only adopted a variable representing neighborhood deterioration (Shear, 1983; Boehm and Ihlanfeldt, 1986; Sari, 2014) due to variations in microscopic characteristics at each site and difficulties in obtaining the required data.

The variable of neighborhood deterioration was estimated based on the average age of all houses within the target area of the management-type residential environment improvement project. The required data were sourced from the building registers, similar to the estimation of the age of individual houses.

Finally, zoning was selected as a dummy variable to represent urban planning characteristics, which specified the type of zone for each parcel.

The required data were obtained from EUM (formerly known as the Land Use Regulations Information Service). Since building-to-land ratios and floor area ratios varied from zone to zone, the effect of these variations on the housing improvement type was examined.

1. Analysis of occupancy status

An independent sample t-test was conducted to determine the effect of homeowners’ occupancy status on homeowner characteristics, specifically their age and the period of ownership. The analysis results are summarized in <Table 3>. There were variations in both the average age of homeowners (t=-10.492 and p=0.030) and the average period of ownership (t=-11.950 and p=0.000) between the two groups. The average age of homeowners in the owner-occupied group was 68.66 years, 9.01 years older than that of the non-owner-occupied group, whose average age was 59.65 years. The period of ownership was 19.21 years for the owner-occupied group and 10.01 years for the non-owner-occupied group. The difference between the two groups was 9.2 years.

Table 3.

T-test of homeowner characteristics regarding occupancy status

Next, the effect of homeowners’ occupancy status on the type of housing improvement was examined. The results are summarized in <Table 4>. In the non-owner-occupied group, 329 homeowners (44.3%) were categorized as No change, 65 (15.5%) as New construction, and 26 (6.2%) as Extension. In the owner-occupied group, 414 homeowners (91.2%) were categorized as No change, 27 (5.9%) as New construction, and 13 (2.9%) as Extension. A chi-square test was conducted to compare the distribution patterns of the two groups. There was a significant difference in their distribution trends, with a Pearson chi-square value of 28.474 and a significance probability (p-value) of 0.05, indicating that the observed differences are statistically significant at the 0.05 significance level. These results suggest that the occupancy status of homeowners is associated with the type of housing improvement. Specifically, among the non-owner-occupied groups, higher proportions of New construction or Extension cases were observed compared to the owner-occupied group.

Table 4.

Chi-square test of house improvement type regarding occupancy status

To further determine the effect of homeowners’ occupancy status, t-tests were conducted on two continuous variables: the age of homeowners and the period of ownership. Additionally, chi-square tests were performed on the housing improvement type, a discontinuous variable.

The results confirmed that homeowners in the owner-occupied group are older and have owned their homes for a longer period compared to those in the non-owner-occupied group. The non-owner-occupied group exhibited more than twice the proportion of both New construction and Extension compared to the owner-occupied group. This trend indicates that homeowners in the non-owner-occupied group may have a greater inclination toward housing improvement. This behavior can be attributed to the fact that these homeowners are in a more advantageous position to undertake new construction or housing improvements because they have the flexibility to reside elsewhere during the construction period. Alternatively, their motivation for housing improvement may be to facilitate the renting of their houses rather than improving housing quality for themselves while they reside in them.

2. Analysis of determinants influencing housing improvement

In this section, the effect of homeowner characteristics on housing improvement is examined, while controlling for other variables, to assess the moderating effect of homeowners’ occupancy status. Stata 17 SE, a statistics software package, was utilized to conduct multinomial logistic regression analysis. This method is suitable for cases where the dependent variables are categorical, with three or more categories. In this analysis, parcels with no changes were defined as a reference category. Generally, a multinomial logit model is interpreted primarily based on the relative risk ratio (RRR). The RRR quantifies the relative change in the probability of a certain event occurring, given the presence of specific risk factors, when holding all other conditions constant. Specifically, an RRR greater than 1 indicates a positive (+) effect, while an RRR less than 1 indicates a negative (-) effect. A one-unit increase in a variable results in an increase in the probability of choosing the dependent variable by a factor equal to the RRR.

In regression analysis of moderating effects, the three-step hierarchical regression analysis is widely used. This method involves three hierarchical models, the independent variable model (X); the model with independent and moderating variables (X, M); and the model with independent, moderating, and interaction variables (X, M, X×M).

That said, Kim (2019) noted that the method proposed by Baron and Kenny (1986) had been mistakenly adopted by many researchers. This erroneous adoption, based on the notion that moderating effects are observed only when independent and moderating variables significantly influence dependent variables, has led to applications of the method lacking a solid theoretical basis. The researcher also argued that the literature specifying the relevant methodology (Frazier et al., 2004; Hair et al., 2005; Hayes, 2013; Whisman and McClelland, 2005) suggested the use of a two-step hierarchical regression analysis, demonstrating weak theoretical grounds for the aforementioned method.

Therefore, in this study, Model 1, which involves only independent and moderating variables, was employed to examine the effect of homeowner age and ownership period on the type of housing improvement. Furthermore, Model 2 was also utilized to verify the moderating effect of moderating variables on the relationship between independent and dependent variables. This model incorporates not only independent and moderating variables but also interaction variables.

The natural logarithm of the land price and parcel area variables’ values were used in the analysis to assess their normal distribution. The obtained analysis results are presented in <Table 5>.

Table 5.

Multinomial logit model for housing improvement influence factors

3. Sub-conclusions

Overall, the results above confirm that both the age of homeowners and the period of ownership have a negative (-) effect on the probability of choosing New construction and Small-scale improvement. Additionally, the period of ownership has a greater negative (-) effect on the probability of choosing New construction compared to Small-scale improvement. These results align with previous studies indicating that homeowners tend to become more passive in undertaking housing renewal and improvement as they age (Shear, 1983; Montgomery, 1992; Yim and Choi, 2016; Kim and Kim, 2017). They are also consistent with previous reports demonstrating that an increase in the period of ownership leads to a decrease in the probability of choosing housing renewal and improvement (Montgomery, 1992; Baum and Hassan, 1999; Kim and Kim, 2017).

This study further classified housing renewal and improvement into New construction and Small-scale improvement. It also confirmed the moderating effect of homeowners’ occupancy status on both homeowner age and ownership period, offering more specific and practical insights than previous studies.

In Model 1, both homeowner age and ownership period were statistically significant factors in determining choices for New construction and Small-scale improvement. However, in Model 2, which incorporates interaction variables, homeowner age and ownership period were statistically significant for New construction, whereas only the ownership period was significant for Small-scale improvement.

In the non-owner-occupied group, the moderating effect of homeowners’ occupancy status was confirmed: this variable negatively moderates the negative (-) effect of ownership period on the probability of choosing New construction over No change. Simply put, an increase in ownership period reduces the probability of choosing New construction to a greater extent in the non-owner-occupied group than in the owner-occupied group. This suggests that in the non-owner-occupied group, the period of ownership has a more significant effect, particularly when homeowners opt for New construction. As homeowners own their houses for a longer period, they are less likely to choose New construction, and the interpretation above suggests that this decrease in possibility will be greater in the non-owner-occupied group than in the owner-occupied group.