Optimal building heights around Bogoti's first subway line By Esteban Castro Izquierdo Bachelor in Architecture Universidad de los Andes Bogotd, Colombia (2002) Submitted to the Department of Urban Studies and Planning in partial fulfillment of the requirements for the degree of Master in City Planning MASSACHUSETTS INSi MffE at the OF TECHNOLOGY MASSACHUSETTS INSTITUTE OF TECHNOLOGY June 2014 LIBRARIES C 2014 Esteban Castro Izquierdo. All Rights Reserved The author here by grants to MIT the permission to reproduce and to distribute publicly paper and electronic copies of the thesis document in whole or in part in any medium now known or hereafter created. Author Signature redacted Department of Urban Studies and Planning Certified by Signature redacted Departn Accepted by t of b Pyfessor Albert Saiz tudies and Planning z f Thesis Supervisor Signature red actedThssSprio sso Christopher Zegras air, MCP Committee Department of Urban Studies and Planning Associa JUN 19 2014 Master of City Planning Thesis Massachusetts Institute of Technology Optimal building heights around Bogoti's first subway line Esteban Castro Izquierdo Cra 4 No 70A - 82 / A615 Bogoti, Colombia estebancastroizquierdo@gmail.com May 2014 Abstract The necessity of a more efficient and clean mode of transportation, combined with the possibility to increase real estate supply in Bogotd, makes the first subway line a remarkable opportunity to redefine the city's future. In order to do so, it is fundamental to understand the potential market demand in the area. Historically, the market has not been taken into account when defining land use regulations, that is why incorporating market factors would definitely contribute to a more appropriate land use policy. The purpose of the present research is to develop a model to estimate optimal building heights that can help in calculating optimal market densities in the first subway line catchment area. Estimating potential densities would be useful not only to make better land use and regulation policy, but also to find real estate opportunities in the subway corridor. Acknowledgments I would like to thank the following institutions and people who contribute with information, advice, and their knowledge: Gustavo Marulanda and Ivan Dario Cubillos (Catastro Bogotd), Alejandro Forero (Galeria Inmobiliaria), Anuar Prez, Camilo Zea, Adriana Gutidrrez, Ben Golder, and professors Chris Zegras and Albert Saiz from MIT. 2 Table of Content 1. INTRO DUCTIO N .............................................................................................................. 5 1.1 Introduction..................................................................................................................................5 2. FRAM EW O RK .......................................................................................................................... 6 2.1 Bogotd's first subway line and its catchment area ................................................................... 6 2.2 Land values and redevelopment ............................................................................................ 8 2.3 Land supply................................................................................................................................10 2.4 Transportation im pact on real estate prices............................................................................ 12 25 M arginal prices and costs............................................................................................................14 3. DATA ........................................................................................................................................ 15 3.1 Subway station locations ..................................................................................................... 16 3.2 Locational explanatory variables .......................................................................................... 16 3.3 New housing supply...................................................................................................................17 4. 3.3 Cadastral data in catchment area ........................................................................................... 18 3.5 Developm ent costs ..................................................................................................................... 19 OPTIMAL HEIGHT DEVELOPMENT MODEL.......................................................... 22 4.1 Data processing ......................................................................................................................... 22 4.2 OLS Regression model .............................................................................................................. 23 4.5 M arginal revenues......................................................................................................................26 4.5 Optimal building heights ........................................................................................................... 5. 28 CO NCLUSIONS.......................................................................................................................32 5.1 M ain findings and further steps ............................................................................................. References ........................................................................................................................................... 32 33 3 Appendix ............................................................................................................................................. 35 A .1 Spatial distribution of dwelling density ................................................................................ 35 A .2 Spatial distribution of people density by UPZ...................................................................... 35 A .3 Spatial distribution of employment density by UPZ............................................................. 36 A .4 Spatial distribution of reported traffic accidents by UPZ ..................................................... 36 A .5 Spatial distribution of reported unsafe incidents by UPZ ................................................... 37 A .6 Spatial distribution of Transmilenio lanes and stops ............................................................ 37 A .7 Spatial distribution of bus stops ........................................................................................... 38 A .8 Spatial distribution of dedicated bicycle lanes...................................................................... 38 A .9 Spatial distribution of parks .................................................................................................. 39 A .10 Spatial distribution of public sport facilities ........................................................................ 39 A .11 Spatial distribution of museums......................................................................................... 40 A .12 Spatial distribution of churches ......................................................................................... 40 A .13 Spatial distribution of public libraries................................................................................. 41 A .14 Spatial distribution of public schools................................................................................. 41 A .15 Spatial distribution of non public schools.......................................................................... 42 A .16 Spatial distribution of higher education centers................................................................. 42 A .17 Spatial distribution of health institutions ............................................................................ 43 A .18 Spatial distribution of drugstores ....................................................................................... 43 A .19 Spatial distribution of malls ................................................................................................ 44 A.20 Spatial distribution of markets ................................................................................................ 44 A .21 Spatial distribution of hotels ............................................................................................... 45 A .22 Spatial distribution of tourist attractions ............................................................................ 45 A .23 Spatial distribution of police stations and military facilities ............................................ 46 A .24 Spatial distribution of major industrial centers ................................................................... 46 A .25 Reported police incidents in 2010/01 - 2011/08 ................................................................. 47 A .26 Python script to calculate euclidian distances .................................................................. 48 A .27 OLS mean observations by stratum ................................................................................... 49 4 1. INTRODUCTION 1.1 Introduction With nearly seven million people, and a mono-centric functioning model, Bogoti is experiencing an economic and physical growth process, caused mainly by large capital inflows, population migration, macroeconomic stability, decrease in violence, and until the beginning of this year, a stable political environment. While the demand for real estate is increasing at a fast pace, supply of land is severely constrained not only by the cities' geography and political boundaries, but also by historical land use patterns, the lack of available green fields, and current zoning and regulations dictated by its master plan. Over the past 16 years, Bogotd has undertaken several projects to enhance mobility. The city has built nearly 112 km of the BRT system Transmilenio, and has been trying to shift from an "informal" bus transit system to an organized transport system. It also has built 300 km of a dedicated bike lane network, as well as making unsuccessful effort to control private car usage. Still, citizens are experiencing unbearable congestion and intolerable levels of pollution. While some responsibility lies with the lack of infrastructure maintenance, and its capacity, some transportation experts and urban planners argue that a subway system is necessary to address transportation challenges in the city. Whether or not to build the first subway line -FSL- has been a very contentious debate. Opponents argue that the Transmilenio system should be the priority, and could be finished with the same resources the subway would cost. Another argument is that the city should not take the risk of investing in this high technology infrastructure which the country has no experience implementing. Conversely, promoters claim the subway will be more efficient, will have a lower impact on the urban fabric, and that property values are expected to increase more than in the case of the BRT. After many years of debate among the government agencies, academic institutions, politicians, and specialists, the city has recently embarked on an ambitious project to build Bogoti's FSL. The necessity of a more efficient and clean mode of transportation, combined with the possibility to increase real estate supply by redeveloping the subway corridor, makes this project a remarkable opportunity to redefine the city's future. The subway line will mean a complete urban renovation process, which implies improvement in transportation efficiency, and also positive impacts in social, economic, and politic domains. 5 This urban renovation process could only be accomplished by aligning land use policies to the FSL catchment area. In order to do so, it is fundamental to understand the potential market demand in the area. Historically, the market has not been taken into account when defining land use regulations. Incorporating market factors would definitely contribute to a more appropriate land use policy. The purpose of the present research is to include market factors for estimating optimal building heights that can help in calculating optimal market densities in the FSL catchment area. Estimating potential densities would be useful not only to make better land use and regulation policy, but also to find real estate opportunities in the subway corridor. Bogoti's master development plan, Plan de Ordenamiento Territorial -POT-, has not yet established a direct relationship between the the FSL corridor and land use policies and regulations. Even worse, there is no clear linkage between public transportation and real estate development. Consequently, redevelopment calculation scenarios are helpful to guide to policy and decision making to determine the kind of city that should be fostered. These scenarios can serve as a tool to estimate future property and betterment levy tax variations, future infrastructure needs, public space requirements, zoning and regulation strategies, social and income redistribution policies, or even to look for alternative financial strategies for the construction of the subway. This study is structured as follows: The second section explains the study framework, and the third section describes the data used in the analysis. The fourth section presents the methodology and results of the model, while the last section discusses conclusions and further steps. 2. FRAMEWORK 2.1 Bogoti's first subway line and its catchment area Even though more than a dozen initiatives to build a subway in BogotA have been made since 1942, it was not until 1997, when Antanas Mockus was appointed as the city mayor, that the FSL was included in the conceptual design of the integrated system of mass transit. Unfortunately, this initiative was frozen when Enrique Pefialosa was in office as mayor in 1998. Nevertheless, during his mayoralty, Transmilenio's first phase was designed and implemented. 6 In 2007, during his election campaign, Samuel Moreno promised that if he was elected, he would start the FSL construction. During Moreno's mayoralty, the city council defined the first subway line as a priority project. For that purpose, an inter-institutional commission was created. The commission was led by the city transportation department (Secretarfa Distrital de Movilidad) and had the support of other government institutions, the Inter-American Development Bank, and World Bank. The commission managed to elaborated the terms of reference for the "conceptual design of the subway mass transit network design and operational, legal, and financing modeling"'. Those terms of reference have served as the structure for the subsequent steps to develop the subway project. The terms of reference had five stages: The first stage was a comprehensive study of current and future transportation needs of Bogoti, the POT, and forecast projections about cultural, economic, social, demographic, environmental, and mobility and local characteristics. This stage also included a proposal of various alternatives for the subway line location. The second stage was the assessment of the alternatives presented and the priorities for implementing the selected one. The third stage consisted of the operational design for the FSL. The fourth stage involved the detailed designing, and the fifth stage was the financial and legal structure. All stages were open to public bidding. While the first three were developed by a consortium led by Sener Ingenieria and Transporte Metropolitano de Barcelona, the forth stage will be delivered in September 2014 by another Spanish consortium led by Euroestudios, Idom and Cano Jim6nez. An important milestone which reinforced the need the city for in building the subway, was in 2011, when Steer Davis Gleave did a revision, actualization, and calibration of the transportation model for Bogoti and its surroundings, including the subway project, Transmilenio expansion, the integrated public transport system (SITP), and La Sabana Light Rail. The FSL will connect downtown and the financial district with the Tintal neighborhood located on the southwest outskirts of the city, and the Usaqu6n neighborhood located on the northeast side (see Figure 1). The conceptual design of the FSL contemplates a total of 29 stations. Out of the total 29 km, 25 km will be underground, 3.5 km will run at ground level using existing rail corridors, and the remanning 0.5 km will be elevated. 1 Docurnento de Empalme, Proyecto Metro de Bogota. 2011 7 At first neighborhoods are rural land on the periphery of the city waiting to be developed. As the city expands, this arable land becomes occupied by single-family homes which are usually owned by similar socioeconomic or ethnic groups. After a long process of maturity and consolidation, the neighborhood could enter either a stage of more intensive redevelopment, or of decline. Redevelopment could occur because its location becomes a premium one, because its highest and best use shifts to a more profitable one due to proximity to certain uses, or because it benefits from public infrastructure improvements. This is the case of a new mass transit line. The possible downgrading could be caused by aging housing stock, insecurity, and migration. From the two possibilities explained above, redeveloped and depressed neighborhoods could either started again a process of renovation and densification, or enter a decline stage. An example in Bogoti of this process of double redevelopment, is the Chic6 neighborhood. In the early 1950s the area was a farm on the fringes of the city. Between 1951 and 1954 it was developed as a high class suburb with single-family homes. As the city grew, and the neighborhood was no longer on the border, its highest and best use changed from one floor houses to 4-5 story apartment buildings. Around seven years ago, due to a higher demand for high-class well-located dwellings, and surrounding concentration of offices, a second wave of redevelopment started. Nowadays, the former low-rise apartment buildings and the few remaining original family homes are being demolished and replaced by 8-10 or even 15-story apartment buildings. As happens with neighborhoods, single properties also evolve with time. They have their own life cycles which begin when sites are developed, followed by a natural decline process. Afterward, the buildings are demolished, and a new cycle starts over again with the redevelopment of the property (see Figure 2). Buildings could lose value because of physical, economic or functional obsolescence. While physical obsolescence has to do with natural aging processes, economic obsolescence refers to how a building cannot fulfill its highest and best use due to a higher density demand for the same use, or a demand for a new use. On the other hand, functional obsolescence is related to changes in habits and preferences such small-space housing trends, and high-flexibility office spaces. A property's value is equal to the sum of the land and the building values. Every time a site is redeveloped it reaches is highest and best use. Then, as time passes, the property value gradually decreases, until it equals the land value. As properties lose value, the option value of the land to be redeveloped increases 9 until it reaches the point R, where it is more profitable to redevelop. Redevelopment will occur when the land price of new developments exceeds the property value in its current use minus demolition costs. U E - ------- ------ L R R R Time P= Property value R= Construction or redevelopment points in time (typically between 30-100 years between) C=Land redevelopment call option value U= Usage value at highest and best use at time of reconstruction S= Structure value L=Land appraisal value (legal value) K=Construction (redevelopment) cost excluding acquisition cost Figure 2 - Property Value Components 3 2.3 Land supply Land as a factor of production plays a key role in real estate development; therefore, it has a major impact on housing prices. Land supply is principally determined by four factors: geography, historical land-use patterns, the financial asset market, and policies and regulations that rule a housing market. 4 Bogotd's physical growth has been defined by its geography. On the one hand, the steep mountains on the east have limited development, and due to various factors, such as better drainage conditions and views, higher-income neighborhoods have historically located there. Initially, high construction costs on these steep terrains were the reason that prevented development in this area, but as land prices rose and construction costs were no longer a constraint, environmental protection measures are now holding back expansion on this side of the city. On the other hand, the Bogotd River and its wetlands have restrained the city expansion to the west side. Being a high-risk flooding zone, the river area has drawn lower income communities, due to low land prices. "Historical land-use" patterns refers to the agglomeration of certain activities, such as employment concentration, government centers, and neighborhoods with historic value. When specific activities concentrate in an area, there is a point where land becomes scarce for the specific use. The FSL's 3 Geltner D, Miller N, Clayton J, Eichholltz P, 2014, Commercial Real Estate Finance and Investment, 3rd edition 4 Saiz A, 2013, International housing finance and economics course notes, MIT 10 As planning policies and regulations can constrain land availability, they can also increase land supply. By wisely matching transport infrastructure projects with land-use regulations, can create new land. With the FLS, Bogoti has a unique opportunity to renovate deteriorated large areas, which will greatly increase housing supply, improve accessibility and mobility, reduce air pollution, and create a more equitable social place to live in. Land is an asset, as well as a factor of production. A property owner can either develop a site, or wait for a higher expected appreciation of his asset, as immediate development might not be optimal. Willingness of some investors to hold sites that could be developed, also contributes to land shortage and therefore higher land values as developers have to pay this real option premium. 2.4 Transportation impact on real estate prices Proximity to mass transit not only impacts the ridership of the system, but also real estate values. This distance of influence, also known as catchment area, is determined by people's willingness to walk and their preferences. 5This distance can vary depending on people's age, purpose of the trip, climate, sidewalk conditions, topography, or crime rates, among others. Even though there is not a strict distance to determine these catchment areas, according to literature reviewed, range can vary between 250 m and 1,500 m. Figure 4 summarizes a series of studies on how new transit lines have impacted real estate prices on transit corridors. BogotA's FSL consultant group, which elaborated the conceptual designs, estimates that the catchment area should be 500 m (product No 42). It is not clear why the consultants chose that distance; however, as stations are proposed every kilometer, one could argue that overlapping areas were not desirable. The consultant group also forecast that prices would increase around 10% in places where prices are currently very high, and 15% where current land prices are low. After studying results for other cities and Rodrfguez and Targa's study on Transmilenio 2009, a catchment area of 1,000 m radius around the subway station is taken as a base for this research project. It is important to clarify that this research project did not take into account variations on the willingness to walk under hillside conditions at the east side of the corridor, where the steep roads might increase walking time and affect people's comfort. 5 Cervero R, Guerra E. 2013. Is a half mile circle the right standard for TODs? Access No 42 12 As there is no certainty about how much the subway line would affect real estate prices, it is desirable to simulate different alternatives. The present research study has evaluated four scenarios: a no-price-change scenario, and 5%, 10%, and 15% increment scenarios. T PLACE Amsterdam, Netherlands T AN(E PRItF CH MANrr- nATA P Ir-F 33,333 buildings (2003) Buildings <500m from the train station are 4m taller (-40%higher). There is an extra 10%effect if the building are located less than 500m to a highway Madrid, Spain 1,714 housing units asking price (2010) At a 1,000m distance from Metro line 12, prices decrease between 2.18% and 3.18%. At a 1,000m distance from Cercanfas commuter rail, prices decrease between 3.38% and 5.17% T Sunderland, Uk Asking prices before (1998) and after (2003) Subway line had no changes in property prices T Eastern MA, USA 1,896 single family sales price (2007) Housing prices increase between 9.6% and 10.1% in municipalities with commuter rail train T S. East England, UK 7,474 housing transaction between (1997 and 2001) Housing prices increase 9.3% more in areas where transportation infrastructure projects where developed T Chicago, USA 17,034 single family house transactions (1983 -1999) Housing prices near Midway rapid transit line increase 6.89% more T Seul, Korea 241 apartment units (1989 2000) Prices increase by 8.9% in a 1,000m radius around the new subway line T LA County, USA 3,803 sales price (2000) No changes in property prices T Washington DC, USA 250 rental price (1992) Rental prices increase between 2.4% and 2.6% for every T Chicago, USA 79 blocks (1980 - 1990) Residential land values increase by 17% in a 800m radius from Metra line to the airport T Chicago, USA 1,485,000 units (2004) Housing prices increase between 3% and 8% LR LR Manchester, UK San Diego, USA 795 housing units (1990s) 1,495 multifamily units (1990s) Housing prices increase between 2.1% and 8.1% LR New Jersey, USA 14,068 housing units (1991 2008) BogotA, Colombia 3,976 housing units (2001-2006) Housing prices increase between 13% and 14% around B Bogota, Colombia 130,692 housing units Housing decrease on average 4.5% near the BRT, than elsewhere in the city. No causality is attributed to the system. B BogotA, Colombia 494 housing units (2002) Housing prices increase between 6.8% and 9.3% for every 5 min closer to the station B Los Angeles County, USA 3,803 multifamily inots (2000) No changes in property prices B QTtmV 160m closer to the Metro station Housing prices increase between 2% and 6% - Housing prices increase between 8.3% and 9.8% around 400m 500m T is subway or commuter rail, LT is light rial, and B is bus rapid transit Adapted from Rodriguez and Vergel 2010 6 Figure 4 - Mass transit and its impact in real estate prices 6 Rodriguez D, Vergel E. 2010. Comentarios y Anotaciones en el Marco de la Primera Lfnea de Metro de Bogoti. 13 premium to enjoy better views, and better access to light, and to avoid negative externalities such as pollution or noise. The main objective in real estate development is to maximize profit. Assuming no zoning restrictions, a developer will build as many square meters as she could, as long as the cost of producing the last marginal square meter will be less or equal to the price she can sell it for. In other words, a developer will build until marginal costs are equal to marginal revenues. Even though this process of profit maximization could also be applied to a unit's features and finishes, such as type of kitchen or number of bathrooms, this research study is focus only on understanding how building heights are related with housing costs and market prices. Dwelling prices reflect both unit and location characteristics. To determine how people value certain product characteristics, hedonic modeling is used to decompose unit price into structural and neighborhood attributes. While "structural" characteristics refers to size, number of bedrooms, bathrooms, and building age, "neighborhood" characteristics could include variables such as socioeconomic attributes, access to public transport, crime rates, and proximity to amenities and points of interest. More complex models could also include land-use diversity indexes or school quality. This price breakdown is done through a linear regression model, where the structural characteristics of the unit and neighborhood attributes, known as explanatory variables, are calculated by an ordinary least square (OLS) method. The hedonic model explains the price of a dwelling with the following linear equation: Yi=a+/iXi +e where: Yi is the market value of property i, a is the constant, Pi are coefficients, Xi are the explanatory variables, and e is the error in the model. 3. DATA Data for the present study was gathered in Bogoti in January 2014, and comes from seven different sources: Bogoti's cadastral department (Departamento Administrativo de Catastro Distrital), Bogotd's city planning department (Departamento de Planeacidn Distrital), the national statistics agency (Departamento Administrativo Nacional de Estadistica - DANE), Bogot~d's subway final conceptual design document, Galeria Inmobiliaria, Construdata, and a series of interviews with local developers, architects, and planners. 15 3.1 Subway station locations Projected subway station locations have changed over the different design phases, not only because of technical recommendations, but also because of political pressures. When the data was collected, there were two versions of the FSL station locations, which differ in the location of stations 10 to 15. For the present research, the locations of the future subway stations were taken from the final reports of the consulting consortium in charge of the conceptual design. These final reports were downloaded from the official subway webpage (www.metrodebogota.gov.co). 3.2 Locational explanatory variables The city base shape files, such as the city's border, districts, neighborhoods, blocks, and lots, as well as points of interests and amenities, and stratum data, were downloaded from the spatial data infrastructure unit of Bogota's cadastral department -IDECA- (www.ideca.gov.co). This information was released to the public on January 24, 2014, and is projected in MAGNA SIRGAS coordinate system. Reported police incidents data was downloaded from the city planning department website (www.sdp.gov.co), and corresponds to all incidents occurring between January 2010 and August 2011. Reported incidents were classified into three categories: perceived unsafe incidents, traffic incidents, and other incidents. Only the first two categories are used in this research model (see Appendix 25). To be able to compare results along districts, this data was normalized by hectare (see Appendixes 1, 2) . Employment and population data was downloaded from the national department of statistics (DANE) website (www.dane.gov.co). The information contains number of people employed, housing units, and population. It was aggregated according to city planning zones (Unidad de Planeamiento Zonal -UPZ), and also normalized by hectare, so it could be compared across them (see Appendixes 3-5). The data mentioned above was processed to obtain the locational explanatory variables. Figure 6 summarizes points of interest, socioeconomic, and safety data, as well as the computed variables in the model. 16 COMPUTED VARIABLE SOURCE Socioeconomic IDECA (2014) DANE (2012) Safety Secretaria Distrital de Planeacion BogotA (2011) IDECA (2014) Social stratums Population density Household density Police reported incidents Traffic accidents Distance to police and military facilities Public Transportation IDECA (2014) Distance to BRT stations Distance to bus stops Distance to dedicated bike lanes Work places DANE (2012) IDECA (2014) Employment density Distance to major industrial clusters Education IDECA (2014) Distance to higher education institutions Distance to public schools Distance to private schools Shopping IDECA (2014) Distance to malls and department stores Distance to supermarkets Distance to drugstores Health IDECA (2014) Distance to health institutions Distance to doctor's office Recreation and culture IDECA (2014) Distance Distance Distance Distance Distance Tourism IDECA (2014) Distance to hotels Distance to tourist attractions to parks to public sport centers to museums to churches to public libraries Figure 6 - Locational explanatory variables computed in the model 3.3 New housing supply Galeria Inmobiliaria is the biggest private real estate data provider in BogotA, and it collects among other data, information about new housing supply. In January 2014, new housing supply consisted of 2,216 new projects that had 49,190 units. For every project Galeria Inmobiliaria collects information about the project such as name, address, stratum, neighborhood, total units, down-payment percentage and fiduciary trust, and its location coordinates. The information also records if the project is affordable housing, and if it has selling features such as a model unit and a showroom. Additionally, for every unit the data base has information about unit size, number of bedrooms and bathrooms, garage ratio, unit type (apartment or house), and selling price. For the spatial distribution of the new supply in January 2014 see Figure 7. 17 3.5 Development costs The development cost structure that is used in the present research was gathered from series of interviews with local developers, architects, and builders. Construction prices were consulted using a premium account in Construdata (www.construdata.com), a private company which delivers construction price information by location and social stratum in Colombia. Development and construction costs vary by location, use, size and height of buildings, as well as many other factors such as quality of finishes, construction technology, and other policy, economic, and political factors. After comparing several samples of cost structures of new developments across different socioeconomic stratums in Bogotd's housing market, it was found that cost structures are very similar (see Figure 8). While all design and technical studies, development and sale fees, financing, taxes and other costs remain constant as a percentage of total revenues, land and construction costs are the only factors that vary according to socioeconomic stratums. Although land and construction costs add up to a similar percentage of total revenues in all stratums, the difference is that land makes up a higher percentage in higher stratums. For example in stratum 2 land corresponds to 16% and construction costs to 57% of total development costs, while in stratum 6, land is 28% and construction cost is around 48%, of total development costs. The cost matrix in Figure 9 represents the average development cost by stratum and building height in the FSL's catchment area. As the real estate development market in Bogotd does not include stratum 1, that stratum was not considered in the cost model. Land price refers to its transaction price. Construction costs include construction materials, labor, equipment, contractor fees, urbanism costs, as well as a construction fee charged by the construction company. In most cases, construction fees are paid as a percentage of the total cost of construction. Design and technical study fees vary according to the complexity and size of the projects. All projects require architectural design, soil engineering, structural, mechanical, electrical, and pluming engineering, oversight, and co-property regulations. Some projects have special conditions that require specific studies, such as security and control, traffic and acoustic studies, landscape design, and energy efficiency consultants. The impact on the total cost of these additional technical studies is so low, that they were not taken in consideration for the present research. 19 1 2 3 Item Land (16% -28%) Construction Costs (47% - 60%) Direct construction costs Construction Fee Design and technical studies fees Architectural Oversight Soil engineering Structural design Water and sanitary designs Electrical designs Mechanical designs Security and control Co-property regulations Other % of Sales 28.00% 47.50% 43.00% 4.50% 2.64% 1.40% 0.75% 0.05% 0.12% 0.08% 0.06% 0.04% 0.02% 0.02% 0.10% Item Development fee 5 Sales fee and advertisement 6 Taxes, legal, insurance, & Other Gain Tax (Plusvalia) Permit fees Property tax Bank appraisal 0.004 tax Legal costs Fiduciary costs Administrative costs Public service fees 7 Financial costs 4 % of Sales 3.00% 4.75% 3.71% 1.00% 0.96% 0.06% 0.02% 0.25% 1.00% 0.15% 0.15% 0.12% 10.40% 1.40% Construction credit fees Return on investment Total Costs 9.00% 100.00% Figure 8 - Average development cost structure in the FSL catchment area Development and sale fees are around 3% each, while advertisement expenditure is close to 1.75%. There are four main taxes that apply to a housing development project in Bogotd. Gain tax applies to land that has benefited from an increase in its value due to improvements, derived from public infrastructure investments. Property taxes are paid to the municipality while the land is owned by the developer during sale and construction processes. The 4 x 1,000 is a national government tax, charted to all financial transactions, by an amount of 0.004%. Lastly, permitfees are paid to CuraduriasUrbanas,private agencies that revise and approve construction permits. After the 1998 real estate crisis in Colombia, almost all housing developments are financed with a presales scheme. To ensure a transparent money management process, fiduciary trusts are used as vehicles to administer funds in real estate projects. Fiduciary trusts receive buyers incoming funds, and make disbursements to the developers, subject to construction advancement milestones. Fiduciary costs are approximately 0.15% of total revenues. Other administrative costs, utilities connections fees, and legal consultancy sum up 1.27%, while construction credit fees are on average 1.4% of total development costs. Return on investment is the opportunity cost of capital, which in this kind of projects is around 9% of total development revenues. 20 BUILDING HEIGHT 5 10 15 20 25 30 STRATUM 2 16.00% 59.50% 2.64% Land Construction Costs Design Fees 4.75% 3.71% Sales fee and advertisement Taxes, legal, insurance, & other 9.40% 4.00% Financial costs Development fee Average development costs STRATUM 3 19.00% Land 56.50% 2.64% 4.75% Construction Costs Design Fees Sales fee and advertisement 3.71% Taxes, legal, insurance, & other 9.40% Financial costs 4.00% Development fee Average development costs STRATUM 4 22.00% Land 53.50% Construction Costs 2.64% 4.75% 3.71% 9.40% 4.00% Design Fees Sales fee and advertisement Taxes,legal, insurance, & other Financial costs Development fee Average development costs 160,000 595,000 26,400 184,000 684,250 30,360 202,400 752,675 33,396 232,760 865,576 38,405 279,312 1,038,692 46,086 349,140 1,298,364 57,608 47,500 37,100 54,625 42,665 60,088 46,932 69,101 53,971 82,921 64,765 103,651 80,957 94,000 40,000 1,000,00 108,100 46,000 1,150,000 118,910 50,600 1,265,000 136,747 58,190 1,454,750 164,096 69,828 1,745,700 205,120 87,285 2,182,125 323,000 345,610 362,891 381,035 407,707 448,478 960,500 44,880 80,750 1,027,735 48,022 86,403 1,079,122 50,423 90,723 1,133,078 52,944 95,259 1,212,393 56,650 101,927 1,333,633 62,315 112,120 63,070 67,485 70,859 74,402 79,610 87,571 159,800 170,986 179,535 188,512 201,708 221,879 94,416 72,760 76,398 80,218 85,833 1,700,00 1,81900 1,909,950 2,005448 2,145,829 2,360A12 616,000 1,498,00 665,280 1,617,840 698,544 1,698,732 733,471 1,783,669 777,479 1,890,689 831,903 2,023,037 73,920 133,000 103,880 263,200 112,000 2,800,00 79,834 143,640 112,190 284,256 120,960 3,024,000 83,825 150,822 117,800 298,469 127,008 3,175,200 88,017 158,363 123,690 313,392 133,358 3,333,960 93,298 167,865 131,111 332,196 141,360 3,533,998 99,828 179,615 140,289 355,449 151,255 3,781,377 875,000 1,767,500 92,400 166,250 129,850 329,000 140,000 3,500,00 945,000 1,908,900 99,792 179,550 140,238 355,320 151,200 3,780,000 992,250 2,004,345 104,782 188,528 147,250 373,086 158,760 3,969,00 1,041,863 2,104,562 110,021 197,954 154,612 391,740 166,698 4,167A50 1,093,956 2,209,790 115,522 207,852 162,343 411,327 175,033 4,375,823 1,159,593 2,342,378 122,453 220,323 172,084 436,007 185,535 4,638,372 68,000 STRATUM 5 25.00% 50.50% 2.64% 4.75% 3.71% 9.40% 4.00% Land Construction Costs Design Fees Sales fee and advertisement Taxes,legal, insurance,&other Financial costs Development fee Average development costs STRATUM 6 28.00% Land 1,260,000 1,348,200 1,415,610 1,472,234 1,531,124 1,607,680 47.50% 2.64% Construction Costs Design Fees 2,137,500 118,800 2,287,125 127,116 2,401,481 133,472 2,497,541 138,811 2,597,442 144,363 2,727,314 151,581 Sales fee and advertisement Taxes, legal, insurance, & other Financial costs Development fee Average development costs 228,713 213,750 178,637 166,950 452,610 423,000 192,600 180,000 4,500,000 4,81500 240,148 187,568 475,241 202,230 5,055,750 249,754 195,071 494,250 210,319 5,257,980 259,744 202,874 514,020 218,732 5468,299 272,731 213,018 539,721 229,669 5,741,714 4.75% 3.71% 9.40% 4.00% Figure 9 - Average development costs per sqm by stratums in the FSL catchment area 21 4. OPTIMAL HEIGHT DEVELOPMENT MODEL 4.1 Data processing The following four-step process was carried out to organize and prepare the data for the regression model in the present research. First, cadastral units were filtered by use to ensure only housing units were selected from the cadastral data base. Then, in order to get an approximation of transaction prices, all cadastral prices were divided by 0.7, because according to the cadastral department, assessed prices are on average 70% of transaction prices. Afterward, observations under 20 sqm and over 2,000 sqm, and with prices above $17,000,000 COP psm, as well as units below ground level, were discarded. From the initial 516,625 units in the FSL's catchment area, only 242,049 remained for the analysis. Second, as nearly half of the new housing supply projects lack adequate geo-referenced information, they were geocoded using Google Maps and ArgGIS. Third, cadastral and new housing supply databases were merged, only keeping the variables common to both databases. The resulting joint database had a total of 243,320 observations. Fourth, for every housing unit, two measurements were computed: the distance to the closest point of interest, and the gravity measure. While the former measurement only considers the closest point of interest, the latter takes into account how a household is located in relationship with all other points of interest of the same category. As seen in the gravity equation below, points of interest will be more valued if they are closer to the household, and less valued if they are farther away. G = gravity measure 1 = 2 ii (dij) i = household location 1 = point of interest location d = distance If the point of interest was a polygon or a line, such as parks and dedicated bike lanes, the distance was computed using the ArcGIS Near tool. When the points of interest were points, as in the case of BRT stops, public schools, or malls, the distance was computed in Python, using the script shown in Appendix 26. 22 4.2 OLS Regression model The following hedonic model was specified, where price per square meter of the housing units is the dependent variable, and structural and location variables described in Figure 10 are the explanatory variables. Yi=a+fiXi +e where: Yi is the market value of property i, a is the constant, Pi are coefficients, Xi are the explanatory variables, and e is the error in the model. To understand how the different housing markets in Bogoti can be explained by its structural and location attributes, dummy variables were created for every socioeconomic stratum. Consequently, building height and size variables were separated by stratum. Additionally, for every point of interest variable that involved a distance calculation, a new variable with its logarithmic transformation was created. For the preliminary models that were tested, these point of interest variables had 4 variations: the minimum distance, the gravity measure, and the logarithmic transformation of each of those measures. After testing the initial models for point of interest variables, only the most statistically variable out of the four variables was kept in the model; all others were discarded. Then, the model was run again, and the non statistically significant variables were discarded as well. The final regression model only contains the explanatory variables that were statistically significant. Afterward, squared variables for the unit size and the building height were created. The purpose of creating these squared variables was to account for nonlinear relationships the explanatory variables could have with the dependent variable. 23 Variable Description T Min Max Mean Dependent variable PRICE Cadastral and selling price (COP) Structural explanatory variables HEIGHT_1 Floor unit height in stratum 1 HEIGHT_2 Floor unit height in stratum 2 HEIGHT_3 HEIGHT_4 HEIGHT_5 HEIGHT_6 HEIGHTSQ_1 HEIGHTSQ_2 HEIGHTSQ_3 HEIGHTSQ-4 HEIGHTSQ_5 HEIGHTSQ_6 SIZE_1 SIZE_2 SIZE_3 SIZE_4 SIZE_5 Floor Floor Floor Floor unit height unit height unit height unit height in stratum 3 in stratum 4 in stratum 5 in stratum 6 Square of HEIGHT_1 Square of HEIGHT_2 Square of HEIGHT_3 Square of HEIGHT_4 Square of HEIGHT_5 Square of HEIGHT_6 Floor area in sqm in stratum 1 Floor area in sqm in stratum 2 Floor area in sqm in stratum 3 Floor area in sqm in stratum 4 Floor area in sqm in stratum 5 SIZE_6 Floor area in sqm in stratum 6 SIZESQ-1 SIZESQ_2 SIZESQ_3 SIZESQ_4 SIZESQ_5 SIZESQ_6 AGE Square of SIZE_1 Square of SIZE_2 Square of SIZE_3 Square of SIZE_4 Square of SIZE_5 Square of SIZE_6 Age of building Std. Dev. C 57,546.08 17,900,000.00 2,442,885.0 1,649,547.0 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 20.00 20.00 20.00 20.00 20.00 20.00 400.00 400.00 400.00 400.00 400.00 400.00 0.00 0.00 0.00 0.00 1.00 11.00 31.00 60.00 55.00 26.00 39.00 121.00 961.00 3,600.00 3,025.00 676.00 1,521.00 1 021.52 1,255.30 1,965.20 1,797.20 1,909.00 1,982.33 1,043,503.00 1,575,778.00 3,862,011.00 3,229,928.00 3,644,281.00 3,929,632.00 93.00 1.00 1.00 1.00 2,079.00 0.01 0.39 1.03 0.63 0.24 0.62 0.03 1.24 6.18 5.23 1.57 4.53 1.25 28.09 51.58 15.03 7.82 24.13 255.69 5,733.37 10,972.75 2,765.60 1,339.31 5,218.91 23.50 0.04 0.01 0.66 114.72 0.16 1.04 2.26 2.20 1.23 2.04 0.46 5.78 35.40 34.23 12.71 24.64 15.94 70.32 91.17 50.39 35.75 68.09 5,097.30 21,249.27 39,422.29 30,414.74 22,258.03 35,326.16 15.60 Building is a new supply project 0.20 Building is historical 0.11 JOINT Building is part of a joint property 047 UNITS Number of units in building 209.44 Location explanatory variables STRATUM_1 Building is stratum 1 0.00 1.00 0.01 0.09 STRATUM_2 Building is stratum 2 0.00 1.00 0.20 0.40 STRATUM_3 Building is stratum 3 0.00 1.00 0.40 0.49 STRATUM_4 Building is stratum 4 0.00 1.00 0.16 0.37 STRATUM_5 Building is stratum 5 0.00 1.00 0.07 0.25 STRATUM_6 Building is stratum 6 0.00 1.00 0.16 0.37 POPDENSITY People per Ha by district 1.19 498.81 192.34 121.79 DWELLDEN Housing units per Ha by district 0.28 110.45 57.90 26.44 SAFETY No of police incidents per Ha by district 0.00 2.87 1.16 0.71 TRAFFIC No of traffic accidents per Ha by district 0.00 12.55 4.22 3.04 POLICE Distance to police and military facilities 0.18 591.36 312.98 1,797.76 BRT Distance to a BRT station 0.50 17,982.57 1,436.82 1,959.44 CYCLING Distance to a dedicated bike lane 0.39 25,302.77 1,100.74 3,441.20 EMPDEN Employes per Ha by district 0.06 312.17 73.92 77.19 INDGRA-LOG Log of gravity to an industrial clusters -6.07 3.25 -5.00 1.67 H_EDU Distance to a higher education center 2.64 12,200.26 1,772.99 1,517.93 PUBSCHOOL Distance to a public school 2.39 505.43 2,152.45 449.22 PRISCHOOL GRA Gravity to private schools 6.43E-06 0.18 8.72E-05 4.49E-04 MALLLOG Log of distance to a mall -2.85 3.69 2.74 0.35 MARKETLOG Log of distance to a supermarket -0.85 3.67 2.70 0.37 DRUGSTORE Distance to a drugstore 0.00 146.49 1,041.02 105.99 PARKGRALOG Log of gravity to a park 2.02 -4.31 -3.34 0.31 SPORTGRALOG Log of gravity to a sport center -7.02 -5.87 -3.36 0.42 MUSEUM Distance to a museum 12,937.18 1.45 2,959.01 2,283.74 CHURCH Distance to a church 6.23 3,188.63 434.90 249.16 LIBGRALOG Log of gravity to a public library -6.88 -2.52 -6.13 0.42 HOTELLOG Log of distance to a hotel -4.91 2.71 4.11 0.58 TOURISTLOG Log of distance to a tourist attraction 0.21 3.63 2.83 0.40 N = 242,460 D = dummy variable, C = continuos variable, US 1 = 2,000 COP, and all distances are measured in meters. NEWSUPPLY HISTORICAL Figure 10 - OLS mean observations 24 Variable Coefficient Std. Error T statatistic Constant Structural explanatory variables 638,297.40 58831.370 HEIGHT_1 211,663.90 27,246.960 7.770 HEIGHT-2 HEIGHT_3 HEIGHT_4 HEIGHT_5 -1,205.00 -3,359.17 29,726.69 30,003.77 2,788.393 1,116.817 1,430.595 3,347.995 -0.430 -3.010 20.780 8.960 HEIGHT_6 87,963.20 1,587.287 55.420 HEIGHTSQ_1 HEIGHTSQ_2 HEIGHTSQ_3 -26,728.32 -218.90 155.38 5,514.609 363.556 59.628 -4.850 -0.600 2.610 HEIGHTSQA HEIGHTSQ5 -425.11 -1,593.11 73.949 255.116 -5.750 -6.240 HEIGHTSQ_6 SIZE-1 -3,477.13 -4,276.09 104.381 246.533 -33.310 -17340 SIZE_2 -4,128.41 64.600 -63.910 -3,101.27 -3,614.09 2,655.98 2,152.81 4.06 5.05 1.90 2.24 -2.53 -1.33 -15,497.80 2,100,316.00 -60,459.45 37.589 52.323 81.812 43.936 0.507 0.139 0.054 0.061 0.079 0.052 86.188 45,663.420 9,418.484 -82.500 -69.070 32.460 49.000 8.020 36220 35.100 36.690 -31.930 -25.340 -179.810 46.000 -6.420 JOINT 25,501.69 4,423.902 5.760 UNITS 3050 6.139 4.970 624,630.10 1,013,008.00 38,340.460 38,345.970 16.290 26.420 STRATUM_4 1,894,975.00 38,588.730 49.110 STRATUM_5 STRATUM_6 POPDENSITY 2,270,263.00 2,892,562.00 1,561.45 39,875.130 38,898.630 42.456 56.930 74.360 36.780 -7,45652 -105,083.60 133.605 4,794.332 -55.810 -21.920 12,96832 -109.22 1,159.683 3.894 11.180 -28.050 -6.500 SIZE_3 SIZE_4 SIZE_5 SIZE_6 SIZESQ_1 SIZESQ-2 SIZESQ-3 SIZESQ_4 SIZESQ_5 SIZESQ-6 AGE NEWSUPPLY HISTORICAL Significance 10.850 Location explanatory variables STRATUM_1 STRATUM_2 STRATUM_3 DWELLDEN SAFETY TRAFFIC POLICE -14.19 2.182 CYCLING 5.05 2.014 2.510 EMPDEN 581.90 26.481 21.970 -255,89750 59.82 231.74 5,043,466.00 64,482.85 -156,851.00 282.47 5,108.856 2.216 4.118 2,219,584.000 3,533.236 4,049.170 11.821 -50.090 27.000 56.270 2270 18.250 -38.740 23.900 BRT INDGRALOG H_EDU PUBSCHOOL PRISCHOOLGRA MALLLOG MARKETLOG DRUGSTORE ** ** -64,409.40 3,946.066 -16.320 SPORTGRALOG MUSEUM CHURCH LIBGRALOG HOTEL_LOG TOURIST LOG 123,658.30 -82.52 -167.39 -146,921.70 -154,622.60 12,453.61 3,882.276 1.499 5.391 4,091.014 3,274.800 5,137.566 31.850 -55.060 -31.050 -35.910 47.220 2.420 N= 242,460 -Adjusted R-square 0.91 - *, * *** denotes statistical significance at a 90%, 95%, and 99% level of confidence PARKGRALOG Figure ** 11 - OLS Regression results 25 OLS regression results are shown in Figure 11. Most of the explanatory variables' signs coincided with expected results. While building story height is valued in all stratums, unit size adds value in high stratums, and decreases value in low stratums. Building age, historical conservation, and the number of reported perceived unsafe police incidents have a negative impact on prices. With stratum, prices also increase. New supply projects, being part of a joint property building, and being closer to a police station or military facility also have a positive relationship with prices. Even though traffic accident results seem counterintuitive, they can probably be explained because higher income neighborhoods have more cars, and therefore more traffic accidents occur. Based on the reviews from previous studies, distance to the BRT was expected to have a greater impact on housing prices. Even though proximity to the BRT is statistically significant, and has a positive impact on housing prices, the regression model results show that a unit will decrease by less than 2% of its price when it is 1,000 meters away from a BRT station. 4.5 Marginal revenues The present study had the objective to estimate marginal revenues of dwelling buildings as a function of the following three variables: building height, unit size, and proximity to mass transit. As proximity to the BRT system results had such low impact on housing prices, this variable was discarded. Marginal revenues were then calculated only as a function of building height and unit size. For these two variables, marginal revenues were computed, taking into account socioeconomic stratums using the following quadratic equations: P* = a + #X + Pi Height_* + Pi Height2 _*+ e P* = a + /3X + /3i Size_* + Pi Size 2_*+ e where: P is the market property value, a is the constant or Y intercept, P are the variable coefficients, Height is the building height, Size is the area of the unit, Xi are all other explanatory variables, * is the stratum of the unit, and e is the error in the model. To compute the marginal revenues in both models, all explanatory variables (except building age), were added to find the Y intercept. Then, building height and unit size were computed using their corresponding stratum variables. In all cases, mean observations of building height and unit size for every stratum were used (see Appendix 28). 26 5. CONCLUSIONS 5.1 Main findings and further steps Two of the three studies reviewed about Transmilenio showed positive significant relationships between real estate prices and proximity to the system. In the present research, proximity to the BRT system had a lower effect than expected. These unexpected results were statistically significant, but did not have any major impact on housing prices. On average, property values decrease less than 2% when dwellings are at a distance of 1,000 m from the BRT stations. It is possible that better results could have been obtained, if the explanatory proximity variable to the BR[ system, had been measured by stratums, as lower socioeconomic groups rely more on mass transit. If the subway line does not impact real estate prices, 12 to 19-story buildings will maximize land values and development revenues, depending on the socioeconomic stratum. In the case where land values increase 15%, optimal building heights would be between 17 and 25 floors. For stratums 1 to 4, marginal prices decrease as the unit size increases, while for the upper two stratums, marginal prices increase as the size of the dwelling increases. Even though the construction price model was done carefully, in order to validate the precision of the model, it will be desirable to calibrate it with a construction cost analyst expert in Bogotd's development market. It will also be desirable to include in the model marginal costs as a function of unit size, in order to find out average optimal unit size per socioeconomic stratums. The model also relies on the accuracy of cadastral prices being on average 70% of transaction prices. Using a real transaction price data base will also improve the precision of the model. 32 References 1. Armstrong RJ, Rodriguez DA. 2006. An evaluation of the accessibility benefits of commuter rail in Eastern Massachusetts using spatial hedonic price functions 2. Bae CC, Jun M-J, Hyeon P. 2003. The impact of Seoul's subway Line 5 on residential property values. Transport Policy 10 3. Benjamin JD, Sirmans SG. 1996. Mass transportation, apartment rent and property values. Journal of Real Estate Research 12 4. Cervero R, Duncan M. 2002. Land Value Impacts of Rail Transit Services in Los Angeles County. Washington, DC. National Association of Realtors & Urban Land Institute 5. Cervero R, Duncan M. 2002. Land Value Impacts of Rail Transit Services in San Diego County. Washington, DC: National Association of Realtors & Urban Land Institute 6. Cervero R, Guerra E. 2013. Is a half mile circle the right standard for TODs? Access No 42 7. DiPasquale and Wheaton. 1996. "Economics and Real Estate Markets". Prentice Hall 8. Du H, Mulley C. 2007. The short-term land value impacts of urban rail transit: Quantitative evidence from Sunderland, UK. Land Use Policy 24 9. Forrest D, Glen J, Ward R. 1996. The Impact of a Light Rail System on the Structure of House Prices. Journal of Transportation Economics and Policy: 30 10. Geltner D, Miller N, Clayton J, Eichholltz P, 2014, Commercial Real Estate Finance and Investment, 3rd edition, OnCourse Learning 11. Gibbons S, Machin S. 2005. Valuing rail access using transport innovations. Journal of Urban Econ. 57 12. Kim K, Lahr M. The Impact of Hudson-Bergen Light Rail on Residential Property Appreciation. Urban Studies In press 27 13. Lehner M. 2011. Modeling housing prices in Singapore applying spatial hedonic regression 14. McDonald JF, Osuji CI. 1995. The effect of anticipated transportation improvement on residential land values. Regional Science and Urban Economics 25 15. McMillen DP, MacDonald JF. 2004. Reaction of house prices to a new rapid transit line: Chicago 's Midway line, 1983-1999. Real Estate Economics 32 16. Mejia-Dorantes L, Givoni M, Rietveld P, Koomen E. 2010. Assessing the impact of rail stations in Amsterdam: A third dimension approach. Madrid: ETSI de Caminos, Canales y Puertos 17. Mejia-Dorantes L, PaezA, Vasallo JM. 2010. Analyzing house prices to assess the economic impacts of new public transport infrastructure: Madrid Metro Line 12. Madrid: ETSI de Caminos, Canales y Puertos 33 18 Mendieta J, Perdomo J. 2007. Especificaci6n y estimaci6n de un modelo de precios hed6nico espacial para evaluar el impacto de Transmilenio sobre el valor de la propiedad en BogotA. Documentos CEDE 19. Munoz-Raskin R. 2009. Walking accessibility to bus rapid transit: Does it affect property values? The case of BogotA, Colombia. Transport policy 17 (2010)72-48 20. Rodriguez D, Mojica CH. 2009. Capitalization of BRT network expansions effects into prices of non-expansion areas. Transportation Research Part a-Policy and Practice 43 21. Rodriguez D, Targa F. 2004. Value of accessibility to BogotA's bus rapid transit system. Transport Reviews 22. Rodriguez D, Vergel E. 2010. Comentarios y Anotaciones en el Marco de la Primera Linea de Metro de Bogotd. 23. Rosenthal, Stuart, Helsley, Robert. 1993. "Redevelopment and the Urban Land Price Gradient". Journal of Urban Economics. Vol 35: pp 18 2 - 20 0 24. Ute grupo consultor primera linea metro Bogoti. 2010. Diseflo conceptual de la red de transporte masivo metro y diseiho operacional, dimensionamiento legal y financiero de la primera linea del metro en el marco del sistema integrado de transporte pdblico - SITP - para la ciudad de Bogota. Producto No 42, informe de resultados - Alcance inicial de la etapa 4 25. Ute grupo consultor primera linea metro Bogoti. 2009. Disefio conceptual de la red de transporte masivo metro y disefio operacional, dimensionamiento legal y financiero de la primera linea del metro en el marco del sistema integrado de transporte pdblico - SITP - para la ciudad de Bogoti. Producto No 20, planos de trazado de la primera linea de metro y su Area de influencia, Planos 26. Watts S, Neal K, and Davis L. 2007. "Tall buildings." Building (00073318) 272, no. 17: 74. 27. Zegras C, Grillo C, Jiang S. 2013. Sustaining mass transit through land value taxation? Prospects for Chicago 34 Appendix A.1 Spatial distribution of dwelling density Dwellings per Hectare 0 1.00 -20,00 20 01 - 40 00 4001 -6000 M60 01 - 80.00 80 01 - 110 00 A.2 Spatial distribution of people density by UPZ ~,1 People per Hectare 1i - 50 , ,fj51 -100 M101 - 200 201 - 300 301 - 500 35 A.3 Spatial distribution of employment density by UPZ Employment per Ha 0- 10 11 - 25 26 -50 51 - 100 M101 -200 201 - 350 ii 0 2.0 04 r 0 2E 5Km I I A.4 Spatial distribution of reported traffic accidents by UPZ Traffic accidents per Ha 0 00- 050 051 -1 00 101 - 2 50 2 51 -5-,00 5 01 - 10 00 1001 -1500 I 36 .............. -........................ ......... .. - -...... ..... - A.5 Spatial distribution of reported unsafe incidents by UPZ I F <A -J Reported incidents per Ha 00 - 0 50 051--10 11 - 1,5 1.6 - 2.0 21 -25 O 2.6 - 30 2,5 I I 0 2,5 5 I Km A.6 Spatial distribution of Transmilenio lanes and stops j / 7 p7$ ~ ./ /.~ II ,r ~ 1 ,'A> X \ I I I - Transrmlenio stops Transmtlenio lanes I I I 5 Km 37 A.7 Spatial distribution of bus stops 03.. . Sir .41 * ~ . * *v** ~;* I - ~ : * *:~* ~ *iK4 * *' '. .*. *~..* SITP Bus stops *. ~ -11A.d . : ILa *, ..IS .* 7 -*., 0 2,5 5 Km A.8 Spatial distribution of dedicated bicycle lanes *1i 38 A.9 Spatial distribution of parks Parks 0 25 5Km 0 25 5 Km A.10 Spatial distribution of public sport facilities - Public sport facilities 39 .... ..... .. .. .. .. A.11 Spatial distribution of museums - 0 I Museums 25 5 Km 25 5Km A I A.12 Spatial distribution of churches *1 - Churches 0 I I 40 A.13 Spatial distribution of public libraries Public - hbraries 0 I 5Km 25 A.14 Spatial distribution of public schools * * -. S* * -. * - * . - - *2 - - - . * - * * - - * .*. ''a 13 * .** * * * 1~. - ** * I * * * * *..**'.'.* *. ** 2.5.a 'I Public schools 41 A.15 Spatial distribution of non public schools .,.4 *.1 .. * - * . . .* .. : .. a.. .. s . . - -* ; ''.. J.. * * -o. * .-,*. -* .* ' . - '. 'y . * %. ".. . . : Non public schools , 0 25 5 Km 0 2.5 5 Km A.16 Spatial distribution of higher education centers 7 - Higher education institutions I. * *. .4 I I I 42 A.17 Spatial distribution of health institutions IM.X - . -...--.-. J, * . . -- . - . %....-4 *I .*. - JV.j 44 4. - Health institutions 0 2.5 I I 5 Km I A.18 Spatial distribution of drugstores 4r. j * *i~PAL ***..'.3..If **.*i..:. tq'a *Ask- ,s~. .;~ ~J~ 4 1k; . **. 4 lb' 0 - Drugstores I 25 I 5 Km K 43 ......... ..................... ............. - .. ............. ...... ... .. ... ..-............................ -.... .................. ..................................... -.............. ............... .......... ................ -..... .... . A.19 Spatial distribution of malls I * *4~ * 'I - *. - 0 Maps 25 5 Km I~~ 2.5 5 KM I A.20 Spatial distribution of markets f,. I,- S. i .. 0. - Supermarkets 0 I I 44 .. ........ .. .... ...... .. .. ...... .... .. .. ......... A.21 Spatial distribution of hotels -irk.* '. 4 - . Hotels 0 25 5Km 0 25 5 Km A.22 Spatial distribution of tourist attractions I ~ 1:.'* * E ** .1~~ * *s~W~. * 4 - Tourist attractions 45 A.23 Spatial distribution of police stations and military facilities - Police stations and miltary facilities 0 2.5 I 5 Km I 0 25 5 Km I I I A.24 Spatial distribution of major industrial centers .1 . *g - Major industrial sites i 46 A.25 Reported police incidents in 2010/01 - 2011/08 CODE DESCRIPTION REPORTS PERCEIVED UNSAFE INCIDENTS 904 Realized robbery 150 905 Robbery in progress 528 906 Sexual violence 910 Personal injuries 911 Shooting 210 6,699 580 922 Narcotics 76 923 Begging 46 932 Disturbance of the peace 934 Brawl 968 Juvenile gangs 976 Missing or lost person 120 5,074 11 22,787 TRAFFIC ACCIDENTS 942 Traffic accidents 87,605 OTHER INCIDENTS 602 Height fall 10,459 604 Respiratory 23,832 613 Unconscious / respiratory arrest 617 Gastrointestinal symptoms 901 Dead 924 Sick 926 Drunk 928 Floding 930 Landslide 931 Fire 941 Mental disorder 967 Minor in a inappropriate establishment TOTAL 19,073 8,480 55 29,206 1,114 911 1,323 3,047 10,621 3 232,010 47 A.26 Python script to calculate euclidian distances import csv import math householdsfile = "houses.csv" businessesfile = "businesses.csv" def get-distance(house, business): xO, yO = house['x'], house['y'] xl, yl = business['x'], business[y'] xO, yO = float(xO), float(yO) xl, yl = float(xl), float(yl) return math.sqrt( (xl - xO)**2 + (yl - yO)**2) def getgravity(distance): return 1 / distance**2 def get-distances(house, businesses): distances = [get_distance(house, business) for business in businesses] gravity = sum([get-gravity(d) for d in distances]) min distance = min(distances) house['gravity'] = gravity house['min-distance'] = min_distance def get-rows(path): rows = [] with open(path,'r') as f: reader = csv.DictReader(f) for row in reader: rows.append(row) return rows def writecsv( data, path): oneitem = data[O] with open(path,'w') as f: writer = csv.DictWriter(f, onejitem.keyso) writer.writeheadero writer.writerows(data) def mainO: houses = get rows(householdsfile) businesses = get-rows(businesses-file) for house in houses: get-distances( house, businesses) write csv( houses, 'calculatedhouses.csv') maino 48 A.27 OLS mean observations by stratum Variable Constant Coefficient 638,297.40 Mean observation by socioeconomic stratum Stratum 4 Stratum 5 Stratum 6 0 0 0 0 0 0 Structural explanatory variables 211,663.90 HEIGHT_1 -1,205.00 HEIGHT_2 Stratum 1 1.51 0 Stratum 2 0 1.972331 Stratum 3 0 0 -3,359.17 0 0 2.587793 0 29,726.69 30,003.77 87,963.20 -26,728.32 -218.90 155.38 -425.11 -1,593.11 -3,477.13 -4,276.09 -4,128.41 -3,101.27 -3,614.09 2,655.98 2,152.81 4.06 5.05 1.90 2.24 -2.53 -1.33 -15,497.80 2,100,316.00 -60,459.45 25,501.69 30.50 0 0 0 2.89 0 0 0 0 0 140.43 0 0 0 0 0 28,675.67 0 0 0 0 0 13.30 0 0 0.09 15.86 0 0 0 0 6.260245 0 0 0 0 0 142.3634 0 0 0 0 0 29055.51 0 0 0 0 18.2049 0.0103335 0.0013959 0.2910477 67.70027 0 0 0 0 0 15.54354 0 0 0 0 0 129.883 0 0 0 0 0 276294 0 0 0 30.54845 0.0163451 0.0111279 0.5713532 134.45 3.814761 0 0 0 0 0 31.8559 0 0 0 0 0 91.54089 0 0 0 0 0 16841.42 0 0 19.80563 0.0695957 0.0439604 0.9050194 163.0854 0 0 0 3.439338 0 3.796304 0 0 0 0 0 0 0 0 0 22.64186 0 27.8057 0 0 0 0 0 0 0 0 0 112.9759 0 147.8416 0 0 0 0 0 0 0 0 0 19351.76 0 31979.59 16.8253 19.83921 0.1534898 0.0707862 0.0063558 0.0048618 0.953965 0.9772527 128.3246 74.44369 1 0 0 0 0 0 129.72 34.18 0.54 1.78 654.48 1,288.24 388.23 12.73 -5.48 2,940.43 198.49 0.00 3.02 3.15 252.63 -3.36 -6.27 5,355.49 956.07 -6.14 2.98 3.23 0 1 0 0 0 0 329.4261 77.387 1.786583 1.95796 559.7379 967.302 350.3918 31.42513 -5.195909 3557.904 229.2039 0.0001592 2.855164 3.094486 116.846 -3.367395 -6.340527 6123.083 591.758 -6.240274 3.003726 3.237061 0 0 1 0 0 0 222.0139 68.1225 1.251064 3.948663 574.1577 1237.044 650.5065 66.32051 -4.955442 2102.947 311.2508 0.000103 2.761356 2.675124 122.3786 -3.231209 -5.864138 2709.047 362.476 -6.097827 2.981945 2.853385 0 0 0 1 0 0 111.2139 47.22534 1.056798 6.030701 480.884 1170.826 1589.833 105.3271 -4.813632 362.7401 446.6397 0.0000454 2.700647 2.541296 133.7298 -3.445361 -5.524798 1302.343 339.2842 -5.953472 2.19967 2.526559 0 0 0 0 1 0 80.73779 29.79102 0.5262321 4.193892 527.6389 2917.123 3451.596 99.98783 -4.533123 510.8341 854.5162 0.0000455 2.582008 2.475621 170.4007 -3.391017 -5.709198 1949.166 413.5464 -6.270346 2.43783 2.746809 HEIGHT_3 HEIGHT_4 HEIGHT_5 HEIGHT_6 HEIGHTSQ_1 HEIGHTSQ_2 HEIGHTSQ_3 HEIGHTSQA HEIGHTSQ_5 HEIGHTSQ-6 SIZE_1 SIZE _2 SIZE_3 SIZE_4 SIZE_5 SIZE_6 SIZESQ_1 SIZESQ_2 SIZESQ_3 SIZESQ_4 SIZESQ_5 SIZESQ-6 AGE NEWSUPPLY HISTORICAL JOINT UNITS 0 0 Location explanatory variables STRATUML1 STRATUM_2 STRATUM_3 STRATUM_4 STRATUM_5 STRATUM 6 POPDENSITY DWELLDEN SAFETY TRAFFIC POLICE BRT CYCLING EMPDEN INDGRA-LOG H_EDU PUBSCHOOL PRISCHOOLGRA MALLLOG MARKETLOG DRUGSTORE PARKGRALOG SPORTGRALOG MUSEUM CHURCH LIB GRA LOG HOTEL_LOG TOURIST LOG 624,630.10 1,013,008.00 1,894,975.00 2,270,263.00 2,892,562.00 1,561.45 -7,456.52 -105,083.60 12,968.32 -109.22 -14.19 5.05 581.90 -255,897.50 59.82 231.74 5,043,466.00 64,482.85 -156,851.00 282.47 -64,409.40 123,658.30 -82.52 -167.39 -146,921.70 -154,622.60 12,453.61 0 0 0 0 0 1 86.77878 33.44136 0.6131417 5.908802 806.1813 2138.708 1653.456 104.4576 -5.209075 702.3348 1239.875 0.0000197 2.622178 2.486818 237.9027 -3417978 -5.706513 1705.844 498.2813 -6.215311 2.323657 2.597847 49