Planning Strategies and Design Concepts
The Synergic Effect of Walkability and Metro Ridership on Housing Prices of Transit-Designed Neighborhoods in Delhi
2025 Volume 13 Issue 4 Pages 20-45
Details
2025 Volume 13 Issue 4 Pages 20-45
Urban regeneration, especially in dense cities of emerging economies, requires mass public transportation supported by improved walkability, which reduces traffic and promotes health. A pedestrian-friendly environment and public transit ridership change urban life, including housing markets. In markets, convenient public transit can affect housing prices and demand. Due to their convenience and affordability, neighbourhoods near public transportation nodes are more expensive and in demand. Transit-designed communities usually have better walkability, which raises housing prices. However, these parameters can overvalue housing assets, risking local housing markets. The study examines how walkability and public transportation ridership affect Delhi's housing markets. The housing market changes in the Delhi metro catchment areas from 2010 to 2019 were reviewed. A total of 1039 sample apartments were studied, and the catchments of metro stations were assessed based on average walking speed, fetched using Google's time-distance matrix API. It was observed that, in general, in station catchments, the housing was overvalued until 2013. The findings suggest that Delhi's housing market was regenerating, and housing investment returns were lower than expected, but demand was consistent. After this, the housing market appears to have readjusted to actual valuation. The synergic effects of walkability and metro ridership growth on housing prices were assessed using Binary Logistic Regression. The study found that walkability and annual ridership growth moderately synergistically affected housing prices. However, the synthesis of walkability with other complementary variables can accelerate the escalation of housing market dynamics more tangibly and rapidly.
Nowadays, urban planners employ public transportation to restructure and redefine urban development worldwide (Huang and Xia, 2011; Kheyroddin, Taghvaee et al., 2014), as mass public commuting plays a vital role in the city's economy and spatial structure as an essential component of transit-designed development. These Transit-Designed Neighbourhoods evolved as urban or suburban settlements, emphasising increased accessibility to public transportation facilities. The objective behind promoting transit-designed neighbourhoods is to create communities where dwellers have the convenience to move around sustainably. Such neighbourhoods mainly feature the proximity to transit hubs, mixed nature of land utility, pedestrian-friendly environs, transit-oriented zonings, and availability of affordable housing for lower- and middle-income category households (Agnihotri and Paul, 2023b). Restructuring and reinventing cities results in real estate market shifts, especially in transit-designed neighbourhoods. Typically, market movements favouring transit-oriented development are reflected in property values (Bartholomew and Ewing, 2011). According to the available literature, the effects of urban railroads on home values are significant. However, urban rails cannot merely impact housing values; they require a conducive environment in various dimensions. Indeed, it has been seen that public transit has direct but, in most cases, indirect effects on property markets.
Inversely, it is commonly conceived that new metro stations cause shifts in housing prices upward within the catchment zones. Given that the metro stations are merely the catalyst and not a fundamental market force that results in market readjustments, there is a similar probability that the conceived beliefs are false. Indeed, developers are often found to claim that the conceived belief regarding the chances of residential price surges due to increased connectivity with metro nodes isrue (Coskun and Jadevicius, 2017). Falling for such assumptions, the investors, in multiple instances, participate in overvalued transactions while investing in housing assets. In such cases, the actual returns really shortfall the expected returns. Multiple lossmaking transactions ultimately become a financial crisis in the local housing markets. Examining the causes of economic crises has become more crucial than merely handling the crises (Haughwout, Lee et al., 2011). Given the direct association between housing market downfall and local market crises, it is essential to investigate the interrelationship between public transportation and housing prices in transit-designed neighbourhoods in station area catchments. Multiple aspects determine the dynamism of housing markets; housing supply, integrated planning of catchment regions and spatial demand are a few of those critical parameters (Higgins, Ferguson et al., 2014).
In the case of metro rails, ridership and its subsequent expansion over time are analogous to the fluctuating spatial requirement and demand through time. Thus, this study aims to clarify the impact of the synergy between improved walking conditions and increased ridership on annual property price changes within the catchment areas of Delhi's metro stations. The study examines whether middle-income apartment housing prices and their annual fluctuations within the catchment areas of Delhi's metro stations from 2010 to 2019 were influenced by the synergistic effect of the walkability of the surroundings (complementary built environments) and ridership increase (spatial demand growth). Here, walkability is determined by measuring walking speed. The study's independent variables are the average walking speed within the catchment area of metro stations, the annual growth in the station's metro ridership, the serviceable (catchment) area, and the average unit costs (Rs per square foot) of apartments in station surroundings. In contrast, annual price changes in station neighbourhood apartments are kept as the dependent variable. Utilising binary logistic regression, we evaluate the synergistic influence of walkability and ridership on yearly price changes in Delhi's metro catchment districts.
The paper aims to investigate the synergic effect of walkability and ridership growth on annual price movements of residential properties within metro catchments. The paper deals with the literature background encompassing various studies on housing prices, mass transportation, and their interaction in the real market regarding walkability and ridership growth. The literature review is followed by sections on methodology, description of the study area, analyses, and results. Finally, the conclusions are made based on spatial analysis and logistic regression to identify the role of the synergic effect of complementary built environment and spatial demand growth on annual price movements in metro catchments.
The study's inception was inspired by the idea of housing overvaluations, followed by the formation of real estate bubbles. The causes of such overvaluation and market bubbles are different. The current study understood metro stations as a tool directing the housing price movements. Hence, the various domains were explored to understand the same. The literature domains include the association of the dynamism of the property market with mass transportation, the housing markets of transit-oriented neighbourhoods, and walkability causing housing value shifts. The takeaways are discussed hereunder in sub-sections.
Property market dynamics and metro railsReal estate market dynamics result from different local factors; accessibility to urban rails is one of them and usually does not influence the property markets independently. Therefore, the impact of metro rails varies from case to case. For example, the new metro line in Charlotte (USA) increased residential property; however, the commercial properties remained untouched (Billings, 2011). Past studies suggest that the influence of urban rails on surroundings is mixed in nature. On some occasions, it was positive, and on others, it was negative; however, the positive effects were traced more than the negative ones, as explained by (Yan, Delmelle et al., 2012). The housing markets of San Diago, Seoul, Atlanta, Shanghai, Santa Clara, Izmir, New Jersy, and Gold Coast experienced a surge in housing prices after the coming up of new metro lines (Agostini and Palmucci, 2008; Bae, Jun et al., 2003; Bowes and Ihlanfeldt, 2001; Pan and Zhang, 2008; Yankaya, 2004). On the other hand, some scholars observed minimal or no influence in different cities such as Charlotte, Seoul, and Norfolk (Billings, 2011; Wagner, Komarek et al., 2017). Bae, Jun et al. (2003)concluded that the influence of metro rails on property prices was limited within 1 km of the proximity to stations in Bangkok. Property prices in Santa Diago (USA), Holland, and Portland surged up to 23 per cent, 22 per cent and 6.5 per cent after the opening of new metro corridors (Debrezion, Pels et al., 2011; Duncan, 2011) . Similarly, property prices were influenced by urban rails in California, Pennsylvania, Toronto, Sydney, and Taiwan too (Bajic, 1983; Dueker and Bianco, 1999; Landis, Guhathakurta et al., n.d.; Li, S., Chen et al., 2019; Lin and Hwang, 2004). Evidence shows that falsely valuing properties is common in various hot markets worldwide, resulting in market volatility (Smith and Smith, 2006). Such volatilities are investigated in the real estate markets of Australia, Hong Kong, Malaysia, Philippines, Singapore, and Thailand (Glindro, Subhanij et al., 2008). Álvarez-Román and Garcia-Posada (2021) assessing the degree of overpricing of housing stocks is essential for marking investment risks and avoiding significant financial losses and overall economic growth.
In line with the above studies, the Delhi metro is also described as a gentrification tool through which the city is restructured for capital accumulation (Randhawa, 2012). Given that the metro is being used as a gentrification tool, the housing markets of Delhi are less investigated neighbourhoods well connected with metro rail.
Housing markets of transit-designed neighbourhoodsTypically, market shifts favouring transit-oriented development are reflected in property values (Bartholomew and Ewing, 2011). In an efficient transit-oriented development, good pedestrian access to work and non-workspaces without using automobiles increases the value of the real estate near metro stations (Duncan, 2011). Transit-designed neighbourhoods have superior built environments. When relocating residential areas, most households prioritise improved public transportation access and enhanced walkability (Olaru, Smith et al., 2011). It ultimately results in a greater demand for space due to increased riders in these areas and higher housing costs.
Visible impacts are more significant in the commercial area than the residential area near the station. Regardless of land use, the same holds if land prices increase. If a parcel of land is within walking distance of a metro station, its price will be higher, provided it is not adjacent to the station. As the distance from the station increases, the impact on property values decreases. Existing areas have been observed to improve. For example, single-family homes near station areas have been converted into apartments, mixed-use land parcels have been transformed into commercial plots, and commercial development has occurred on vacant land near stations. Literature suggests that higher-priced land parcels may have dense growth near station ranges. If the land is treated as a commodity, its value will increase when demand is high, but supply is low. The value of a property does not rise on its own but rather because of the inflation of the land upon which it is built. This circumstance demonstrates the undeniable origin of investors' efforts to derive the most significant profit from such terrains. These activities increase the likelihood of a higher rate of return and the revenues of urban local governments. Establishing a practical methodological framework to assess the potential effects of new metro rail corridors on the city's real estate is crucial.
Kang (2019) houses too close to the metro stations received lower premiums due to noise pollution, overcrowding, and traffic congestion negative features. The studies indicate that the residents prefer walkable neighbourhoods directly connected to metro stations with higher metro ridership; however, studies also emphasise further research to establish the causality of the relationship between walkability and housing price shifts in transit-designed neighbourhoods (Choi, Y., Seo et al., 2019; Washington, 2013) .
Walkability and housing pricesWalkability plays a significant role in housing location choices within given affordable ranges (Gilderbloom, Riggs et al., 2015). Prospective purchasers and developers desire a certain level of walkability between their properties and public transportation hubs. The buyer's willingness to pay is usually significant for properties within walking distance of a more substantial number of amenities (Yang, Wang et al., 2018). Transit-oriented development is incompatible if walkability is not considered. It is observed that walkability has a positive effect on home prices for households with no or one automobile (Chernobai and Ma, 2022). The case of Delhi is ideal for proving this, as 58 per cent of homes in the city do not own cars and mostly rely on public transportation. The walkability of a street is influenced by its width, green spaces, traffic volume, on-street parking, pollution exposure, and weather conditions (Vichiensan and Nakamura, 2021), and it is related to the direct-connectedness of two locations (Roozkhosh, Molavi et al., 2022). Different past studies investigated the effects of neighbourhood walkability on home values and whether walkability increased the value of homes, all else being equal. Due to differences between areas with high and low walkability, certain studies have reported erroneous results regarding walkability. Others, such as in Miami and Florida, have established the positive effect of walkability on housing prices (Boyle, A., Barrilleaux et al., 2014). Some studies have declared walkability to be the saviour of the markets, claiming that walkable communities tend to protect housing prices even during economic downturns. The studies also revealed that the financial benefits of walkability vary by neighbourhood type and dwelling type (Choi, K., Park et al., 2021). The survey of housing prices in Nanjing highlighted that the walkability of transportation hubs and the availability of educational resources contributed to the increase in housing prices in certain city regions (Xia, Li et al., 2018) .In contrast, Bereitschaft (2019) attributes the overvaluation of housing properties based on better neighbourhood walkability to the variable relationship between walkability and housing affordability in local markets.
Walkability is measured using different tools like walk-score or walking speed indexing. Walk-score is a technique for evaluating the walkability of any area, and the method is widely accepted (Carr, Dunsiger et al., 2010; Duncan, 2011). However, the walk score has been criticised partly due to its underlying assumptions. Other studies have suggested walking speed as an indicator of walkability (Finnis and Walton, 2008) ; consequently, this study analyses this variable. Based on existing literature, the study hypothesised that walking speed directly affects housing price fluctuations. In addition, the alternative hypothesis assumes that walking speed cannot influence the housing price alone. As discussed below, an empirical methodology was developed to evaluate the null and alternative hypotheses.
Historically, several theories have been proposed concerning the relationship between the built environment, land use, infrastructure, and property prices. The earlier theories in urban planning, such as Burges Model, Homer Hoytt's Model and Multiple Nuclei theory, were constructed on the fundamentals of accessibility of LIG and MIG workforce to the residential zones, which were well connected to the city centres with public transit facilities. The integration of land use and transportation later transformed into urban econometrics, resulting in a debate over property prices. The debate was strengthened by the classical theory of bid-rent, which Alonso proposed. The standard linkage between all these theories was the importance of accessibility to housing and its affordability by LIG and MIG communities (the largest caterers of housing demand). Due to the most considerable demand of the housing sector or these societal segments, the probability of overvaluation of existing stocks can never be negated.
The research raises concerns about the overvaluation of housing assets in Delhi's metro catchment areas. It determines whether the preconceived notion of housing price increases in transit-designed neighbourhoods due to improved walkability is accurate. In addition, the study attempts to confirm the synergistic effect of walkability and ridership growth on housing prices near Delhi metro stations. A novel research framework was developed to conduct the research. The framework incorporates contemporary data collection methods utilising web-based tools such as APIs, followed by logical analysis.
As previously mentioned, the critical variables in the analysis are annual ridership growth, catchment area, average walking speed, average unit prices, and annual price shifts. The monthly station-by-station ridership data was collected from Delhi Metro Rail Corporation (DMRC); thus, the average yearly ridership growth from 2010 to 2019 was calculated. Each metro station's serviceable area (catchment area) was outlined using Thiessen polygons in geographic information system software. Using the Google Place API, 1240 sample apartments were selected, and a unit price data set for 2010, 2013, 2016, and 2019 was compiled for the sample apartments through various real estate web portals. The study mainly focuses on the housing price shifts in neighbouring metro station communities throughout the decade. The previous decade can be marked as one of the most important in metro development, with increased service sector employment and spatial demand in Delhi-NCR establishing it as a significant reason for choosing the previous decade. The pricing data used for analysis was for the first quarters of the financial year. Post-2019 housing data was not considered to avoid anomalies due to the pandemic, which may lead to manipulated inferences. Finally, 1039 sample data were deemed usable, and annual price changes for these apartments between 2010 and 2019 were calculated. The data were incorporated into GIS software and interpolated utilising Kriging. It generated the average unit price and average annual price shift contours. The average unit price and average yearly price growth were extracted for each metro catchment area. This study quantified walkability by the average walking speed between metro stations and apartments within the catchment area. The average walking speed was determined using Google API's time-distance matrix.
The ridership data was taken from government sources, and walkability and price information were captured using APIs, which are based on crowdsourcing. Since the data used were from valid sources, the reliability of the data does not pose any scepticism. After station-wise mapping of walkability, annual price shifts, and annual ridership growth, the maps were assessed regarding these variables' high and low values and spatial interactions. A panel data analysis technique would have given better insights, provided the data is not cross-sectional. Since collecting the longitudinal data for walkability was challenging, panel data analyses were avoided, and statistical probabilistic modelling techniques were explored.
A total of 157 station catchments were spatially assessed using GIS. Lastly, 111 station catchments, deemed suitable for predictive analysis, were analysed further using logistic regression. A stepwise binary logistic regression was applied to test the synergic influence of walkability and ridership on annual price shifts of apartments in the metro catchments. The yearly price changes were labelled 'Below Average' and 'Above Average' and assigned values '0' and '1' respectively. Lastly, the different models were compared to conclude. The detailed methodology is described in Figure 1, whereas the data description, analyses, and research outcomes are elaborated on in further sections of the paper.
India's capital city, Delhi, is home to one of the largest city populations in the country and around the globe. The city is spread around 1500 square kilometres and accommodated approximately 16 million people until 2011 (Gopal and Shin, 2019). In the last three decades, the city has grown with a 4 per cent annual rise in population, making it one of the densest cities worldwide with 24 thousand inhabitants per square kilometre (Ahmad, Avtar et al., 2016; Ahmad, Balaban et al., 2013; Goel and Tiwari, 2016). Given such a scenario, the city's spatial demand and unit prices are higher. On the one hand, Delhi has the second oldest yet most extensive metro system in the country. The Delhi metro is the largest in terms of the number of stations, network lengths, and ridership. However, on the other side, the city 37 per cent of households without vehicle ownership and entirely depends upon public transportation (Directorate-of-Census-Operations, 2011a, 2011b).
Additionally, the city serves its inhabitants and dwellers from neighbouring cities like Noida, Ghaziabad, Gurugram, and Faridabad, which form the urban agglomeration with an approximately 25 million agglomeration population. Even after having such a substantial spatial demand in the city's housing market, the role of metro corridors in market movements has been explored less. The above facts make the city a suitable case for studying the housing markets related to mass transportation and its surroundings. Investigating the case of Delhi would provide insights into the housing dynamics around metro corridors in populated cities of emerging economies. Additionally, a comparative study can provide a better understanding of housing market influences of metro stations provided the equivalency in scale of metro operations. Since, in India, no other city has such an established and vast metro network in terms of ridership, network, and age of operations, it will be unjust to make a comparison, at least at present.
Delhi MetroDelhi's metro system is the seventh busiest urban rail in the world. The network has been developed in eleven routes in three phases till the present. The network has been expanded by 65 kilometres (58 stations), 125 kilometres (85 stations) and 161 kilometres (109 stations) during its first, second, and third phases of construction, respectively, since 1998. The city's metro has a network length of approximately 590 kilometres with 221 operational stations. The metro network is anticipated to expand up to 600 kilometres by the end of this year. Due to the large agglomeration size, Delhi's metro services cater to the demand of commuters in terms of inter-city and intra-city (Goel & Tiwari, 2016). The demand is still increasing, so the metro authority plans to extend the network length to 1000 kilometres in the coming years to make the metro systems more accessible (Iyer and Kumar, 2019).
The urban rail system of Delhi is often interpreted as a landmark for 'Modern Delhi' rather than just being a commuter service (Butcher, 2011). It is also depicted as a feature of a global city (Dupont, 2017) and a cosmopolitan attribute (Sadana, 2010). The current metro map of Delhi is shown in Figure 2.
Catchment areaThe catchment area of a public amenity is the geographic region from which most users access the service, such as riding the metro. This area's correlation with the station's land use pattern and transit ridership demonstrates the station's regional significance. It also aids in determining the density of public transportation riders and population concentration within a given urban zone. Considering the catchment area of mass transits is essential to delineate a logical study area.
Understanding catchment areas is vital to optimising land structure, enhancing walkability and bicycle accessibility, and implementing transit-oriented development. This study used the Thiessen polygon method to calculate the catchment areas of each metro station. The proximity method helps map the spatial significance of any given point. Thiessen polygons are generated by deploying Delaunay triangulation methods in GIS (Li, X. Y., Sinniah et al., 2022). Thiessen polygon methods are adopted in different domains of spatial studies, especially to delineate the service areas (Wang, Sun et al., 2014). These polygons are constructed based on the topological relationship between a set of points (x, y) in a two-dimensional plane. The station-wise defined catchment area is shown in Figure 2.
A metro station's ridership is a crucial indicator of population concentration, demand for amenities and services, and housing demand in the area. Previous research found a strong correlation between ridership and transit-oriented development along metro lines (Nyunt and Wongchavalidkul, 2020; Zhou, Liu et al., 2016), indicating that ridership represents the area's (Jun and Hur, 2015) attraction or repulsion. Periodic ridership growth manifests the performance of transit-oriented developments, including various parameters such as spatial demand growth, transit demand, and surrounding built environment. The increasing trend of ridership growth also depicts the increased job density, residential density, availability of affordable housing stocks, and improved access to transit-designed neighbourhoods (Kang, 2019). It is also a crucial parameter to consider while assessing the local natural market due to its quality to reflect the reliability of the users on metro commutes, their satisfaction towards the metro services, and the real market equilibrium in nearby communities (Gan, Feng et al., 2019).
Therefore, metro ridership was considered a variable in this study, so the month-wise station-wise ridership data was collected from Delhi Metro Rail Corporation from 2007 to 2019, and annual ridership growth was derived for further analysis. There are multiple instances where the annual ridership shifts were declining, especially in the dense localities of city cores, after opening up new metro stations on city boundaries. It indicates that push factors in settlements induce the population to choose other localities with metro connections. The push factors usually include higher crime rates, more congested traffic, less access to amenities, relocation of the employment base, changing affordability range, higher pollution levels, and overcrowding. The ridership growth pattern of metro stations in Delhi is represented in Figure 3.
Walkability is a measure of easiness of access for commuters. It affects the time required to travel by foot and an average commuter's walking speed. There are multiple ways to assess the walkability of any region. Finnis and Walton (2008) proposed walking speed as an indicator of walkability. For the current study, the average walking speed while commuting from the sample apartments to the nearby metro stations was considered an indicator of walkability in station area surroundings. The average walking speed, in kilometres per hour, was derived using the time-distance matrix of Google API based on crowdsourced data.
The time-distance-matrix application programming interface (API) provides the details of travel distance and travel time between the set of origins and destinations based on the crowdsourced data of Google Maps — a matrix containing all the metro stations as origins and sample apartments as goals were constructed. The distance API collects the distance between stations and apartments and the average walking time required to traverse the shortest route for pedestrian movement. Google Earth was used to obtain the exact addresses of sample apartments and metro stations, which were incorporated into this exercise.
Silva, da Cunha et al. (2014) deemed a walking speed of 1.5 m/s, or roughly five kilometres per hour, suitable for the average human. This study evaluated walkability or walk-friendly environments based on the average walking speed around metro stations. With an average walking speed of approximately five kilometres per hour, the station's surroundings were assumed to be more walkable. The city's walkability was deemed to be below average. Overall walking conditions in Delhi's metro catchments are depicted on Figure 4.
The number of rural villages in Delhi decreased from 300 in 1961 to 112 in 2011, according to the 2011 Census, as urbanisation reached nearly 97.5 per cent. In Delhi, housing prices have increased by approximately 14% annually, while unsold housing inventory has decreased by 11% compared to the previous fiscal year. However, the housing stock has a higher vacancy rate (Gandhi, Green et al., 2022), which could result from overvaluation. Particularly in outlying areas such as Ghaziabad and Noida, the demand for housing continues to rise due to immigration and employment opportunities. The tendency and quantum of the housing market imposed a concern about the possibilities of market risks, as the accuracy of actual housing price reflection in the market was a point of concern.
Since the development of the metro rail system, housing demand has increased significantly, particularly in outlying areas (Abhay and Sharma, 2023; Sharma and Abhay, 2022). Others criticise Delhi's continued reliance on automobiles despite its effective public transportation system (Goel and Tiwari, 2016; Randhawa, 2012). The population density in Delhi has increased significantly due to redevelopment, fuelled by a robust housing market. As described in the methodology, the final sample size for this study was 1,039 apartments in Delhi's metro station catchment areas belonging to the middle-income group. The average annual price changes (in percentage) were computed based on apartment unit prices in 2010 (the base year), 2013, 2016, and 2019.
The data was collected for the first quarters of the fiscal years 2010, 2013, 2016, and 2019 to prevent data anomalies caused by extreme events like demonetisation. As a dependent variable, the average annual price change was utilised. Since the scope of the present study does not include the characteristics of the housing property, it was assumed that unit housing prices were the same.
The unit housing prices and per-year shifts to the unit prices were mapped using GIS, and Kriging was employed to prepare the unit price and annual price shift contours throughout the city. Kriging is a spatial interpolation technique used by numerous researchers to evaluate the spatial distribution of multiple aspects such as slum development, housing prices, and network kriging (Chakraborty and Li, 2022; Derdouri and Murayama, 2020; Li, Y., He et al., 2013; Seya and Shiroi, 2019; Zhang and Wang, 2014) utilising geographic information systems. It is a method for approximating the unknown values of random fields or functions (Van Beers and Kleijnen, 2004). It is superior to many of the conventional GIS interpolation techniques. The mapped data allowed for station-by-station breakdowns of average unit price and average annual growth within catchments, as depicted in Figure 5 and Figure 6, respectively.
Unit housing prices and yearly shifts in Delhi:
The average per square foot price (INR) of apartment houses in the city was 11,200, 10,150, and 10,350 in 2013, 2016, and 2019, respectively. The data indicated that the housing market in inner parts of Delhi flourished in the early stages of metro operations till 2013; however, metro expansion in the city also embraced the city's expansion, where the communities shifted towards the peripheries of city cores. The average annual price shifts were 18.1 per cent, 3.9 per cent, and 3 per cent till 2013, 2016, and 2019, respectively. The declining trends of yearly price growth rates were especially noted in core areas. It can be interpreted as the readjusting of housing values to their actual worths, i.e., bursting the local market bubble (Agnihotri and Paul, 2023a, 2023b).
As illustrated in Figure 7 and Figure 8, the housing market in Delhi boomed until 2013, followed by a great sump during 2014. National and international media outlets such as NDTV, Economic Times, First Post, Business Today, and Business Standards have also reported the historical decline of Delhi's housing market. Looking at this decline alongside the metro expansion suggests that the increased accessibility to the peripheral developments was one of the primary reasons behind the great fall, as rider participation in outer metro stations increased during that period. This significant change in apartment values in station locations is analysed statistically to determine whether it was related to metro operations.
The unit price and annual price changes for apartments in metro catchment areas were initially analysed. Initial research indicated that real estate assets surrounding Delhi's metro stations were overvalued. The other variable was compiled to determine their relationships with apartment unit price and annual price growth. The highest and lowest observed values for each variable are shown in Table 1.
| Variable | Highest observed value | Lowest observed value | ||
|---|---|---|---|---|
| Station | Value | Station | Value | |
| Walking speed (kmph) |
AIIMS Malviya nagar Mundaka industrial area |
5.78 5.60 4.99 |
Badarpur border Chawri bazar Naraina vihar |
4.32 4.61 4.64 |
| Catchment area (acres) |
Patel chowk Rajiv chowk Chandni chowk |
48 86 117 |
NHPC chowk Rithala Dwarka sector 28 |
3625 3497 3197 |
| Ridership growth (per cent per year) |
Mayapuri Dabri mor Delhi cantt. |
98.51 98.20 96.08 |
NHPC chowk Yamuna bank Sarai |
-7.37 -7.28 -4.13 |
| Unit price of apartments (rupees per sqft) |
Punjabi bagh (west) Green park Panchsheel park |
15,631 15,470 15,405 |
NHPC chowk Tikri border Ghewra |
5825 5885 5906 |
| Annual price shift (per cent per year) |
Naraina vihar Paschim vihar west Paschim vihar (east) |
7.76 7.64 6.81 |
Arjangarh Patel chowk Janpath |
-0.10 0.25 0.29 |
The analysis comprised three stages: first, an examination of the association using the graphical method; second, a spatial analysis to underline the synergic effect of walkability and ridership on housing prices; and last, a logistic regression analysis. It is evident that the graphic plot matrix, as shown in Figure 4, did not reveal direct relationships between the chosen variables.
The synergic effect was to be investigated through spatial and statistical instruments. Looking into the fact that complex modelling instruments can pose a risk of overfitting data (as the dataset is comparatively less than it requires for complex algorithms), a modelling tool that works on the probabilistic approach and has the power of easy interpretability of complex problems was required to be applied. Therefore, logistic regression was used to examine the relationship between walkability, ridership, and their synergistic effects on the annual price fluctuations of apartments in Delhi's metro catchment areas.
The maps prepared for catchments, annual ridership growth, average walkability and average annual changes in apartment prices were processed using GIS. The values more than the 'mean value' of the variables were termed 'Higher', whereas the values lower than 'mean value' were termed 'Lower' for each variable. Processing of the maps using 'Higher' and 'Lower' scenarios, as shown in Figure 10, presented the interaction of annual price change with the combination of walkability and ridership growth.
The total instances were categorised with the combination of higher price growth and lower price growth with higher walkability, lower walkability, higher ridership growth, and lower ridership growth in eight different combinations, as shown in Figure 11. Although a clear relationship was not identified, the nature of price shifts was traced based on Figure 11. It was observed that positive shifts in annual price change of apartments in station catchments were less governed by the walkability of the station's surroundings and ridership growth of the station. However, comparatively, the declining rates of annual changes in apartment prices were better directed by the walkability and ridership conditions. Finally, the binary logistic regression was applied to assess the synergic effects of walkability and ridership growth on annual price shifts in housing markets of the metro station catchments.
Using the different combinations of variables with walking speed, binary logistic regression models were developed to determine the likelihood of the synergic effect influencing the housing price shift in the metro catchments of Delhi. This modelling resembles multiple regression modelling in some ways. Because multiple linear regression and discriminant analysis present challenges for case-specific comprehension, logistic regression was selected as the most appropriate method (Efthymiou and Antoniou, 2013). In these models, the dependent variable, i.e., the annual change in apartment prices, was considered binary (1 = above average, 0 = below average). Eight models were created and compared based on various parameters, including Cox and Snell R square, Nagelkerke R square, Hosmer and Lemeshow goodness-of-fit statistical tests, constant values, and non-standardised β-values of variables in various models. The models were developed by introducing one variable at a time in conjunction with walking speed. The preparation of eight blocks is detailed in Table 2. Block 1 consisted of a single variable, 'Walk,' while each subsequent block added a different combination of variables relating to walking speed. Consequently, the final block, Block 8, contained eight variables: Walk, Walk by RiderGrowth, Walk by Catchment, Walk by Price, Walk by RiderGrowth by Price, Walk by Catchment by Price, Walk by RiderGrowth by Catchment, and Walk by RiderGrowth by Catchment by Price.
| Block details | Entering variable(s)/combination(s) in addition to the preceding one |
|---|---|
| Block 1 | Walk |
| Block 2 | Walk by RiderGrowth |
| Block 3 | Walk by Catchment |
| Block 4 | Walk by Price |
| Block 5 | Walk by RiderGrowth by Price |
| Block 6 | Walk by Catchment by Price |
| Block 7 | Walk by RiderGrowth by Catchment |
| Block 8 | Walk by RiderGrowth by Catchment by Price |
The models' fit was evaluated using pseudo-R square values in the form of Cox-Snell and Nagelkerke R squares. Higher values of Cox and Snell R square and Nagelkerke R square indicate the goodness of fit of logistic regression-based models. The models did not have significant pseudo-R-square values regarding walkability and synergy with ridership. However, as new combinations with walking speed were added to the models, these pseudo-R square values increased, indicating improved model fit. Block 8 received the highest Pseudo R square, as depicted in Figure 12.
Consequently, it is initially evident that the synergistic effect of walkability and ridership on apartment price fluctuations persists. In addition, the statistical significance of the models was evaluated using the Hosmer and Lemeshow (HL) test. For an excellent logistic regression model fit, the HL test must have a value between 0.05 and 1.00 that is not statistically significant. However, the more excellent HL test value indicates a more accurate model. As depicted in Figure 13, the HL test statistics were insignificant for all logistic regression models; however, the lowest HL test value was for block 2, i.e., walking speed alone, and the highest HL value was for block 5, i.e., the combination of walking speed, ridership growth, and unit price. Consequently, walking speed, ridership growth, and unit price all influence the annual price change of apartments in Delhi's metro catchment areas.
The significance of chi-square and unstandardised β-values were then used to evaluate the variables and their combinations in different models. As shown in Table 3, the individual walking speed was insignificant in all prepared models. In addition, the interaction between walkability and other variables, such as annual ridership growth, catchment area, and unit price, was significant in various models. Thus, it is understood that walking speed alone cannot influence the annual price changes of apartments in the catchment areas of metro stations. In contrast, these factors drive the housing price dynamics in various combinations.
| Model block | A | B | C | D | E | F | G | H | I |
|---|---|---|---|---|---|---|---|---|---|
| Block 1 | 0.412 | 0.399 | |||||||
| Block 2 | 0.562 | 0.508 | 0.009 | ||||||
| Block 3 | 0.506 | 0.443 | 0.009 | 0.234 | |||||
| Block 4 | 0.309 | 0.145 | 0.019 | 0.019 | 0.011 | ||||
| Block 5 | 0.285 | 0.121 | 0.225 | 0.017 | 0.018 | 0.534 | |||
| Block 6 | 0.302 | 0.139 | 0.219 | 0.594 | 0.041 | 0.522 | 0.862 | ||
| Block 7 | 0.348 | 0.113 | 0.033 | 0.659 | 0.013 | 0.077 | 0.524 | 0.072 | |
| Block 8 | 0.450 | 0.199 | 0.325 | 0.245 | 0.163 | 0.938 | 0.059 | 0.188 | 0.035 |
| Model block | A | B | C | D | E | F | G | H | I | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Block 1 | 5.850 | -1.261 | ||||||||||||||||
| Block 2 | 4.233 | -1.014 | 0.004 | |||||||||||||||
| Block 3 | 5.279 | -1.279 | 0.004 | 0.000 | ||||||||||||||
| Block 4 | 7.920 | -2.465 | 0.004 | 0.000 | 0.000 | |||||||||||||
| Block 5 | 8.288 | -2.650 | 0.007 | 0.000 | 0.000 | 0.000 | ||||||||||||
| Block 6 | 8.076 | -2.585 | 0.007 | 0.000 | 0.000 | 0.000 | 0.000 | |||||||||||
| Block 7 | 7.041 | -2.664 | 0.023 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | ||||||||||
| Block 8 | 5.673 | -2.172 | 0.012 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | |||||||||
Note: A. Constant; B. Walk; C. Walk by Rider Growth; D. Walk by Catchment E. Walk by Price; F. Walk by RiderGrowth by Price; G. Walk by Catchment by Price; H. Walk by RiderGrowth by Catchment; I. Walk by RiderGrowth by Catchment by Price
The unstandardised β-values were non-zero only for constant walking speed, variable walking speed, and annual ridership growth, as shown in Table 4. Therefore, it is evident that only walking speed and annual ridership growth are responsible for housing price fluctuations in metro catchment areas. Thus, the study hypothesises that the synergy of walkability and ridership influences housing price movements in the catchment areas of mass transit nodes.
Lastly, the models prepared showed the dynamic performance based on the entry variable combinations, as shown in Figure 14. Walkability alongside rider growth (Block 2) was most apt while projecting the price movements, which were below average for the city. After that, the models' overall accuracy stabilised and reached the maximum during Block 7 (walkability, rider growth, and catchment), where the projection of 'above average' cases also reached maximum accuracy. The accuracy of all the formulated models was between 60 to 70 per cent. However, it is concluded that the model's performance can be increased if other variables are added.
Based on the notion that metro station availability inflates housing prices, the study assessed the annual housing price shifts in catchment areas following the synergic effect of walkability and ridership growth. Walkability was used as a criterion for complementary planning around metro stations, while annual ridership growth was used to indicate the yearly increase of spatial demands in station catchment areas. The unit price of houses was believed to represent the current value and housing characteristics. Apartment housing in station catchment areas was chosen for this study to comprehend the housing market dynamics. The analysis was conducted in three phases: first, through a graphical approach; second, through a spatial approach; and last, logistic regression was employed to examine the role of walking speed and its synergistic effect on housing prices near metro stations. The study initially confirmed the existence of overvaluations followed by value readjustment in Delhi's housing market in the previous decade. There may be multiple reasons for such overvaluations; however, the role of metro stations and the resulting misconceptions about real estate markets cannot be refuted. The initial findings represent the overvaluation of housing until 2013, followed by market value adjustments after 2013 and in subsequent years. Initial results strengthen the impetus to investigate the influence of metro stations on housing price changes in their catchment areas.
The spatial investigation of metro catchment regions concerning annual shifts in housing prices, walkability and ridership growth was done using GIS. The mapped data was categorised into eight combinations of yearly price growth, ridership growth, and walkability based on their higher and lower values. The spatial investigation concludes that lower walkability causes lowered annual price changes in apartments in metro catchments regarding individual influence. However, regarding the synergic effect, lower walkability and higher ridership growth significantly impact annual housing price shifts, especially when the city's price growth rates are below average. It demonstrates that dense transit-designed neighbourhoods with poor walkability and higher year-to-year ridership growth are more prone to capture lower annual returns on apartment investments. Furthermore, the binary logistic regression was employed.
The study considered four predictor variables for the dependent annual apartment price shifts: walking speed, catchment area, ridership growth, and unit price. The analysis establishes that walkability, annual ridership growth, and their synergy are statistically significant in predicting annual changes in home prices concerning public transportation. The logistic regression models demonstrate that walkability plays a minor but significant role in the dynamics of home prices. The findings of this study help refute the null hypothesis that walking speed does not influence housing prices in metropolitan areas. Therefore, it is concluded that the synergistic effects of walkability should be considered when analysing housing prices in mass transit station areas rather than walkability alone. The results of logistic regression modelling indicate that as new variables are added to the modelling process alongside the walking speed, the model's performance improves. In addition, the β-values (intercepts) of walking speed are enhanced by introducing new variable combinations. It indicates that walkability can be enhanced by incorporating new variables alongside walkability when predicting housing price growth. Example: In the current study, when only walking speed was included in the modelling process, the intercept revealed that a change of one unit in walking speed would result in -1.261 units in the annual price growth of apartments in metro station catchment areas.
In contrast, when the synergy of walkability, ridership growth, and catchment area was considered in modelling, a unit change in walking speed became capable of causing annual price changes of -2.664% units. In this manner, the findings reveal a direct influence of walking speed on housing price fluctuations if and only if it is supported by other complementary variables that cause a shift in station catchments for metro rail to affect home prices. In conclusion, it can be stated that the findings of this study support the assertions made in a few previous studies, such as those conducted by (Boyle, M. and Kiel, 2001; Choi, K., Park et al., 2021), that walking speed alone cannot influence housing price fluctuations. In contrast, synthesis with other complementary variables can accelerate the escalation of housing market dynamics more tangibly and rapidly.
Even though the synergic effect has been confirmed through spatial and modelling approaches, additional research is required to examine such effects on a larger scale by looking at other variables with price data collected more frequently or from one year to the next. The panel data prepared in such a way can provide better insights. On the other hand, a comparative study across cities with operational metro stations will also better understand housing market dynamics in station areas. The study can aid real estate researchers and policymakers in mitigating the dangers of housing market financial crises caused by irrational asset overvaluation in housing markets caused by increased accessibility to public transportation. Lastly, such studies can help urban planners and policymakers stipulate resource allocation, market risks, effects of transit-oriented development, affordability challenges for residents, ways of integrating infrastructure, and economic growth.
Conceptualization, Vivek Agnihotri and Saikat Kumar Paul; methodology, Vivek Agnihotri; software, Vivek Agnihotri; investigation, Vivek Agnihotri; resources, Vivek Agnihotri and Swechcha Roy; data curation, Vivek Agnihotri; writing—original draft preparation, Vivek Agnihotri and Saikat Kumar Paul; writing—review and editing, Vivek Agnihotri, Saikat Kumar Paul and Swechcha Roy; supervision, Saikat Kumar Paul. All authors have read and agreed to the published version of the manuscript.
The authors declare that they have no conflicts of interest regarding the publication of the paper.
