Does Urban Rail Increase Land Value in Emerging Cities? Value Uplift from Bangalore Metro

This paper seeks to understand how urban rail can influence land value uplift, especially in emerging cities which are largely unstudied. It examines the Bangalore Metro and shows that the uplift from the metro rail was substantial in the ‘catchment area’ and ‘across the city’. The analysis was based on the panel data hedonic price model for around 160,000 apartments over the period 2012-16 and a crosssectional data hedonic price model for 314,000 apartments in 2016. The panel data resulted in a stronger model and show significant land value increases, even beyond the traditional 500m catchment (Figure 1). A ‘before’ and ‘after’ from the commencement of the metro rail operations shows a price uplift of 4.5% across the whole city and indicates a major agglomeration economic event resulting in substantial willingness to pay of USD 306 million from the metro rail accessibility. Emerging cities can expect metro rail to substantially improve their economies and other co-benefits as long as finance can be obtained by capturing this value. © . This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/ licenses/by-nc-nd/4.0/


Introduction
In the latter half of the 20 th century, governments favoured urban road systems and failed to allocate substantial public funds for urban rail projects. This approach contributed to removal of urban rail across most of the cities around the globe in the 1950's and 1960's. Those that remained like London and New York's subway were significantly underfunded (Black 2007; Green, 2016;Sharma & Newman, 2017).
Urban rail is back on the urban development agenda. It is thriving in densely populated cities of Asia, Europe and the Middle East and in the American and Australian cities which are heavily reliant on cars.
Over the last two decades, China and India introduced over 25 high capacity urban rail systems (metro rail) with another 25 currently under construction 1 . This surge is driven largely by rapidly growing demand for rail in cities due to increasing travel time differentials between urban rail and urban traffic as well as a growing need for dense urban centers that are facilitated by urban rail. However, financing remains a constant struggle with the conventional model of government grants and fare-box revenue proving to be inadequate to meet the increasing rail demand. An alternative is the need to realize the 1 In addition to urban rail, China is constructing 18,000 km network of high-speed rail to support its growing economy (Chen, 2012 Land value capture (LVC) mechanisms have shown significant potential as a sustainable source of finance for urban rail projects. This financing alternative emerges from the potential of urban rail to increase the land value in transit catchment. Most of the LVC studies have been done on cities in developed nations. This paper seeks to enable a better understanding of LVC in emerging cities as the need for alternative funding is even more significant in emerging cities where there is high deficit in social infrastructure as well as lean budgets.
It is speculated that urban rail (metro rail) has increased real estate value in Indian cities ( This paper attempts to answer these questions in stages. It begins by examining the relevance of location theory, land rent and demand, and WTP for transit infrastructure in any city including emerging cities to see how metro rail influences land value. The subsequent section discusses prominent studies on residential land value uplift due to metro rail in developing and developed countries based on the hedonic price model (HPM) method. Following this, a methodology is proposed to evaluate the impact of metro rail (as a property attribute) on the residential property market using HPM to estimate user's WTP. The methodology is applied to the Indian case of Bangalore 2 . The case study uses cross sectional and panel data to prepare HPM's for calculating WTP for different property attributes, particularly metro rail accessibility. The estimated HPM's are used to capture the increase along the metro rail catchment land market and at city-level. The latter is rarely done in LVC studies.
In the next sections, theory is used to show a) why particular HPM variables were chosen to evaluate land value in Bangalore and b) how the land value uplift can be explained. Value capture requires that the extent of impact in a city be quantified, followed by analyzing if urban rail generates sufficient value (demand -WTP) to be captured. Econometric models have been extensively used to assess the impact of urban rail on real estate. The most popular among these models is the hedonic price model (HPM).

Hedonic Price Model (HPM)
The HPM is a regression model with its basis in economic thinking of the early 20 th century. It involves the application of least squares regression analysis which requires a linear relationship between the dependent variable (eg. property value) and independent (explanatory) variables (eg. characteristics of property). It estimates separately the contribution of each independent variable price to the total estimated (hedonic) price. The HPM functional forms include linear, linear-log, log-linear and log-log.

Equation 1
Parametric Land Price Equation Where Di is the estimated land price of the i th observation, Xj is a vector of quantitative and qualitative property attributes, β j is the unknown hedonic price of the property for attribute j, and ε i is the stochastic error term.

Location Theory
Von Thunen's (1826) classic location theory analyzed the spatial division of different production activities to minimize transportation expenses between production area and marketplace (Fischer & Nijkamp, 2014). His concepts were applied to urban activities by Hurd (1903), Haig (1926) and Ratcliff (1949) to suggest that urban activities reflect rent competition for locations that minimize movement.
On the other hand, Robert Park (1929) theorized that improvement in transportation and population growth augments benefits of the city center. Alonso (1964) built his location theory based on earlier pioneering studies (Isard, 1956 andWingo, 1961) to suggest that minimization of transportation cost (spatial friction) between residence and work increases land rent in urban settings, as high accessibility to central areas activates competition for locations closer to the central business district (CBD) (Capello, 2011).
The above location theories when applied to an urban context, essentially explain the economic rationale of choosing to situate a firm or household at a specific location in an urban space to minimize transportation costs in the context of agglomeration economies (Capello, 2011). The primary consideration in selecting the location for an urban household is to ensure efficient access to the benefits of agglomeration viz. urban resources, services and workplaces. Thus, location theories highlight the significance of activity centers and travel time in a city.
Saving on travel time is economically significant in a city as it contributes to decreases in transportation and opportunity cost. Over the last decade travel time by car has exceeded that by urban rail in cities across the globe (Newman & Kenworthy, 2015). The importance of saving on travel time is driving the demand for urban rail that is further catering to the urban knowledge economy and the culture of people-centered urban form to support this process (Matan & Newman, 2016;Glaeser, 2011).
Based on the above discussion, this study examined HPM variables on activity centers, urban infrastructure, locational attributes, density and mixed land use as factors affecting the residential land market in Bangalore.

Urban Land Rent
An important feature common to urban location choice theories is the cost of land, or land rent (Capello, 2011). Urban land rent represents the manifestation-in price terms-of the economic value of a scarce resource, e.g. urbanized land endowed with general accessibility characteristics (to the center and to specific facilities like railway stations, airports, parks and green areas) or agglomeration benefits discussed above. Camagni (2016) suggests that rent emerges from two preconditions, first being a limited supply that leads to a 'scarcity absolute rent' (Scott, 1976;Sraffa, 1960) and the second, a 'demand for city' i.e. a household's willingness to pay more than the supply cost for a desirable good or production factor such as access to transit. This demand is generated by the need to benefit from an urban environment which is a product of agglomeration economies. Demand may increase due to time/space specificities when a city becomes crucial for economic activities, for instance the knowledge economy emerges or a city provides an innovative environment or introduction of a public urban infrastructure like metro rail (Camagni, 1992, as cited in Camagni, 2016).

Urban Land Demand
As a scarce resource, urban land displays certain peculiar characteristics to qualify as a marketable commodity (Johnston, 1977, as citied in Kivell & Shaw, 1988). It plays a role in optimization processes in the locational choice of actors, in allocation decisions of land-owners, and in minimizing of mobility and interaction costs. These processes and characteristics contribute to the demand of urban land. Alonso (1964) suggested that the demand of urban land is a utility function of the characteristics of land, geographical location and income constraint. He suggested that an individual household buyer trades off between accessibility, land characteristics and money to reach a decision. The trade-off forms a three-dimensional relationship to represent householder's equilibrium demand. He expressed this in the form of bid-rent curves: the householder's indifference surfaces yield a set of alternative combinations of price and quantities of land at a location for required income and transport costs (Kirwan, 1966). Muth (1969) accorded Alonso (1964) on the positive relationship of land value with its proximity to a CBD. Muth (1969) expanded Alonso's model and showed that population density and proximity to CBD are relative to household income and age of dwellings. Alonso (1964) added that the only way to channel high income demand for land back to central areas 3 is by up-zoning it (as cited in Kirwan, 1966).
Based on these urban rent and demand theories, a city-level assessment of urban rail impact on residential land markets was conducted on Bangalore in order to assess if urban rail results in agglomeration benefits at city level in addition to the generally accepted impacts on catchment areas. These case studies indicate that LVC has significant potential but it needs further development in emerging cities through better data and more parameters to explore the impact of urban rail accessibility on WTP and hence LVC. The paper thus moves to a more advanced study of Bangalore using better data and more parameters than have been used in other emerging cities.

HPM Case Studies -Influence of urban rail on residential real estate
independent variables and the model are statistically significant (Gujarati & Porter, 2004). Achen (1982) and Granger & Newbold (1976) share the same view. Cross-sectional data of 6,700 residential properties located within Helsinki city for years 1980, 1985 and 1989 -The first year represents premetro rail times, and the last two years post-metro rail times.
11% property price increase due to metro rail on the most desirable locations, and price decrease by 8% in the most remote feeder transport areas. Positive impact is highest at the distance of 500 m -750 m from the metro station, lower at 250 m -500m and lowest at less than 250m. In Bielany, properties located within 1 km of a metro station show 6.7% higher selling price than those located beyond 1 km.

Location -Transit System
In Targówek

Bangalore Metro Context
The Bangalore Metro project is being executed by the Bangalore Metro Rail Corporation Limited  (Satyanarayana, 2016). This signifies that the market is responding to a location-based speculative demand of real estate due by Bangalore Metro project. BMRCL could well be in a position to package a rail network with land development to finance the metro rail project. For this, their financial model must include the land value appreciation at different stages of the project. This paper provides the basis for such analysis.
Subsequent sections discuss the methodology and analysis of the impact of Bangalore Metro (Phase 1) on Bangalore real estate.

Real Estate Data Collection
Land/property valuation and registry is essential to efficiently manage this important economic factor of production. India is among the worst ranked countries in land/property registry (The World Bank, 2016). Indian cities lack a comprehensive system to maintain and update urban land records and construction profile (Bheenaveni, 2011). Municipal bodies maintain records of properties for collection of property tax but do not update them annually. If a city like Bangalore is growing rapidly then much of the properties will not be assessed for property tax. Economic Survey of India 2016-17 (Ministry of Finance, 2017) notes that Bangalore has over 80% of built-up area not assessed.
Government authority real estate prices were not used in this study for the following reasons:

Hedonic Price Model's for Bangalore
This study considered cross sectional data HPM and panel data HPM to evaluate the impact of Bangalore Metro Phase 1. The two different HPM's were included in the study to see if crosssectional data is adequate for achieving LVC results. This is because many emerging cities do not have panel data. Data used for the panel and cross sectional HPM's are at the city-level to estimate the impact of a metro rail project at both city level and the metro rail catchment area. Independent variables and dependent variable used for HPM's are discussed in the next two sections.

Dependent Variable
This study uses average sale price of property in Bangalore as a dependent variable for both the

Independent Variables
Independent variables mentioned in the literature review were expanded based on the availability of data and due to the absence of existing empirical studies on factors influencing real estate price in Indian cities to check their impact.
The literature review (Table 1) suggested that independent variables should include city specific, structural, neighborhood and locational variables. Independent variables influencing property price (dependent variable) were considered based on property variables, neighborhood/ socio-economic variables, accessibility variables and metro rail specific variables, as listed in Table 2. City specific independent variables like lake and airport were also included. A total of 22 independent variables were considered in this study but only statistically significant variables were included in both the cross sectional data HPM and panel data HPM.   Four functional forms (linear, log-linear, linear-log and log-log) were tested for both the HPM's. This investigation into the different functional forms of the HPM was necessary as the studies presented in Table 1 used differing functional forms of HPM each and a guidance for best suited form could not be established.

Comparing Panel and Cross Sectional HPM's
Statistical software (SPSS 22) was used for estimating both HPM's. 'Enter OLS' method was used in SPPS to delineate statistically significant independent variables by multiple iterations and estimate the best fitted model with up to 95% confidence level. The analysis results in Table 3 shows that both cross sectional and panel data are statistically significant. As other emerging cities rarely have property price panel data, Bangalore results suggest that cross sectional data may well be good enough for the city to assess its value capture potential.  Table 3 shows that the Ordinary Least Squares (OLS) for cross sectional HPM for the linear, loglinear, linear-log and log-log functional forms displayed varying levels of success in modelling property price and all four functions are not random up to 99.99%. For panel data HPM, only the log-linear functional form was statistically significant for the desired independent variables and not random up to 99.99%.
The log-log functional form of the cross sectional HPM explain the highest variation 7 (54%) in the dependent variable and the log-linear functional form of the panel data HPM explain 64% of the variation in the dependent variable, thus these were selected for further analysis. 6 Analysis of variance 7 An R-square comparison is meaningful as the dependent variable is the same for the models. Table 4 shows the cross sectional HPM results for the impact of statistically significant independent variables on Bangalore's property price with descriptive statistics. All metro rail related variables were statistically significant in the model and suggest an upward trend of property prices due to metro rail accessibility. It shows 35.8% value uplift in properties located within 500 m catchment of a metro station and 19.3% value uplift in properties located within 500 m to 1 km catchment of a metro station.

Results from Cross Sectional HPM
Value uplift in properties located within 1 km to 2 km catchment of a metro station is 13.9%. These value uplift trends in properties based on proximity to metro station are similar to that of global cities cases represented in Table 1. 1. LN -Log

* -Dummy variable
The results of cross sectional HPM suggests that change in metro rail's operational status from underconstruction to operational raises the property price by 10.9% across the city. This increase reflects a significant citywide land market response to the availability of new rail transit and the substantial capital investment it brings in the city. Also, the policy to increase FAR along the catchment area after the operation of metro rail could be playing a major role to this increase. This signifies that operation of metro rail is an agglomeration (urban) event that can increase economic productivity of the whole city. Panel data model also shows similar impact at across the city.
Other metro specific variable suggests property values decreases by 1.7% across the city with each passing year after the commencement metro rail the value uplift due to metro rail in property values.
On the contrary this variable has positive impact in the panel data HPM which is a stronger model than the cross sectional one. Table 5

*Dummy variable
Model shows an upward trend of property prices due to metro rail related variables. It suggests 10.7% value uplift in properties located within 500 m catchment of a metro station and 8.1% value uplift in properties located within 1 km to 2 km. The value uplift in properties located within 500 m to 1 km catchment of a metro station is 25.3% value uplift in propertieshigher than the properties with greater accessibility to metro stations. The difference is attributable to noise levels, vibrations due to high speed rail and prolonged construction related inconvenience due to delay in construction and other reasons discussed in the next section.
At city level, change in metro rail's operational status from under-construction to operational raises the property price by 4.5%. This increase reflects a significant citywide land market response to the availability of new rail transit and the substantial capital investment it brings in the city. Also, the policy to increase FAR along the catchment area after the operation of metro rail could be playing a major role to this increase. This signifies that operation of metro rail is an agglomeration (urban) event that can increase economic productivity of the whole city. Other metro specific variable suggests 1.8% value uplift in properties with each passing year after the metro rail became operational.
The analysis shows that developer grade is a significant variable as it yields 53% appreciation in property prices with improvement in grade. This underscores the importance of quality of development, facilities in the property, neighbourhood and other property specific parameters. Whilst a year's delay in possession of a property reduces its price by 1.6% --in practical terms, the property owner loses rental value with delay in possession.

Discussion
Although the cross sectional data and the panel data HPM's cannot be directly compared due to different independent variables used in the models, both models displayed metro rail specific variables as statistically significant and generally reflected similar trends. As the panel data HPM is a stronger model and shows the impact of metro on property prices over the years, it has been used to assess the significance of WTP and draw conclusions in the study. In the next two sub-sections, we attempt to explain the panel data HPM results that are summarized in Figure 2 for the rail catchment area and across the whole city. The generic reasons of the decreasing land value in the catchment area can be due to the negative externalities for residential land market adjacent to the high intensity stationshigher noise levels, vibrations due to high-speed rail, prolonged construction and intense traffic flow as most of the rail stations are located on the arterial roads. These negative externalities can be applicable to most of the cities. Specific reasons for the case of Bangalore seems to bethe construction prolonged for over 2 years than planned; due to absence of policies to guide any land use change before or during implementation of metro rail, the metro rail triggers gradual commercialization prior to its operation commencement (Singh & Sharma, 2012 residential property buyer willingness to reside in 500 m to 1,000 m catchment area over residing within 500 m catchment. This also help explain the 8% increase out to the 1-2 km catchment. This decreasing land value in the catchment area of Bangalore Metro also extend Luca Bertolini model on 'node' and 'place' by finding a decreasing land valuation without proper planning within the urban rail catchment (Bertolini & Spit, 2005). He suggested that real estate value is likely to be generated more from the place than the node, although both are important but people choose to live in places, not nodes.

Property Price Impact -City Level
The metro rail specific variables 'operational year of metro rail' and 'metro rail operational status' in the city level HPM have revealed substantial agglomeration benefits of the metro rail even out to 29 km radius from metro stations 8 . The 4.5% increase in property price across the whole city 9 due to the opening of the metro is a very strong economic impact for an infrastructurethis city level increase was suggested in the urban rent and demand theories discussed in this paper. This is rarely measured in any HPM study on urban rail impacts and has significant policy implications.
The extent of the impact, across the whole city, is not generally understood by some economists and agencies who do not see property uplift from rail as a general economic benefit but only as a local catchment area benefit shifting economic value from one area to another. This research shows that urban rail value uplift covers the whole city. It may suggest that other value uplift studies could examine the extent to which the whole city benefits however it may be that it is too small to measure in a developed city.
It is perhaps easy to understand why there would be such an economic impact in an emerging city, such as Bangalore for the following reasons: a) Investment: The substantial capital investment metro rail projects bring in the city helps to accelerate economic activity of a city and such investment is highly significant in Bangalore.
The cost of Phase 1 of Bangalore Metro is USD 2,068 million (BMRCL, 2016), about two times (USD 1,005 million) the size of the municipal budget of Bangalore (Nag, 2015). Such investment can be hypothesized to have a larger economic impact than has been seen in developed cities due to its proportional investment impact. Transportation being the engine of the urban economy tends to have much more accumulated impact as a sector since it has relevance to all the existing industries and inhabitants of the city. Recalling the location theories, metro rail projects (transportation) can be seen to be playing an elemental role in shaping a city's demand. b) Travel Time: Significant economic growth potential is being blocked by large scale traffic congestion in most dense, emerging cities like Bangalore; hence a metro rail can unlock significantly greater economic opportunity through this improvement in accessibility. The difference in accessibility between emerging cities and developed cities means that far greater agglomeration benefits can be obtained due to the proportionally bigger accessibility gains. The combined effect of low traffic speed and increasing trip length affects travel time reliability (Taylor, 2013). A significant factor metro rail adds over other modes is the reliability to reach destination on timea major factor in workforce travel behaviour (Carrion & Levinson, 2012) and hence in generating accessibility benefits.
A study done by a cab aggregator, OLA (2015, as cited in Rao, 2015), showed that average traffic speed in all major Indian cities is about 19 kph and the average travel time is 34.8 minutes to reach workplace. This would explain why Indian cities are demanding and implementing metro rail projects that have a reliable average speed of 35 kph. Delhi, where transport has the largest share of road land use among all major Indian cities, witnessed traffic speeds of 20 kph in 2013 and is constructing the biggest metro rail system in India (MoUD, 2013). A study on mode choice modelling of private and public modes in Delhi showed that travel time is more significant than travel cost and other factors according to both modes users (Sharma, 2011).

Willingness to Pay (WTP)
WTP for metro rail transit is calculated by multiplying the catchment hedonic price and the average property value. WTP with respect to proximity to metro stations in Bangalore is shown in Table 6.

Conclusion
This study shows that urban rail has substantially increased property value in Bangalore. The impact of metro rail is beyond the traditional 500m and it to have reached right across the city. The increase in the whole city indicates a major agglomeration economic event resulting in substantial willingness to pay of USD 306 million. This increased willingness to pay in Bangalore now will demand changes in the policy and density zoning that will benefit land markets by pushing them to their highest value and best use. These benefits qualify to be recognized by the policy makers and be used to build urban rail as a maximizer for economic development. Although it is too late for this phase of the Bangalore Metro, other phases could plan to tap such an increase for funding. Other emerging cities can be given some confidence about using the value capture mechanisms as well to build or expand urban rail.
The findings of this study disrupts the traditional theory of increasing of land value with the proximity to urban rail. We have explained these findings based on theories and peculiar factors in Bangalore, and implications of these findings on urban policy have been discussed. The analysis of this study can help other emerging cities to quantify the impact of urban rail and help explain its findings as there are limited academic studies on emerging cities. Thus the analyses and the detailed literature review of this paper can benefit policymakers to make informed decisions on urban rail projects.

Disclosure:
1. The authors of this paper would like to thank: a. Dr. Giovanni Perucca for his important comments on methodology of HPM's.
b. M/s LJ Hookers for providing the real estate data.

Funding for research was provided by Curtin University Sustainability Policy Institute, Curtin
University, through the provision of a PhD scholarship to Rohit Sharma.
3. This paper is one of a series of papers, as part of the PhD on Financing Urban Rail Projects through Land Value Capture -The Indian Case.