The Rapid Rise of Middle-Class Vehicle Ownership in Mumbai By Manish Shirgaokar A dissertation submitted in partial satisfaction of the requirements for the degree of Doctor of Philosophy in City and Regional Planning in the Graduate Division of the University of California, Berkeley Committee in charge: Professor Elizabeth Deakin, Chair Professor Robert Cervero Professor Joan Walker Fall 2012 The Rapid Rise of Middle-Class Vehicle Ownership in Mumbai © Copyright 2012 by Manish Shirgaokar All Rights Reserved Abstract The Rapid Rise of Middle-Class Vehicle Ownership in Mumbai By Manish Shirgaokar Doctor of Philosophy in City and Regional Planning University of California, Berkeley Professor Elizabeth Deakin, Chair In India, demand for urban mobility is increasing rapidly because of growth in urban populations, establishment of multiple employment sub-centers, suburbanization of households, better education, higher workforce participation rates, and rising incomes. An increase in discretionary spending is leading to higher household transportation budgets. Middle-income households in particular are investing in private vehicles such as motorized two-wheelers (TWs) and cars. At the same time, policies to reduce vehicle ownership through regulations and user costs remain underdeveloped and weakly enforced. This further increases households’ willingness to use vehicles, especially for non-discretionary work trips. Higher private vehicle use is affecting other quality of life issues such as time spent commuting, accident rates, noise pollution, and particulate and greenhouse gas emissions. In part, this higher vehicle ownership and use is driven by land use dynamics in Indian cities, where growth within city municipal boundaries is constrained by regulations limiting floor-area ratios. As a result, much of the new growth has taken place in urban peripheries where land is cheap and building costs are low. In these peripheral areas, existing small and medium towns have become anchors for agglomeration, transforming into bedroom communities for emergent middle-class groups. Urban peripheral areas are usually undersupplied with transportation infrastructure such as roads or bus transit. This dissertation unpacks the question of why the middle-class in India is driven to owning and using TWs and cars by asking the following: (1) How does work location influence travel by public and private modes? (2) What factors encourage vehicle ownership in middle- class households? (3) What factors drive up vehicle use in middle-class households? The research was conducted using a travel survey dataset from the Greater Mumbai Region (GMR) that represents 1.5% of the households there. The GMR is among the most populated megacity regions in the world, housing over 22 million people. Its growth illustrates the transformation from a monocentric to a polycentric city which is seen in many rapidly growing Indian cities. In seeking to develop an understanding of how work location affected travel, this research identified employment sub-centers using work destination data. Of all middle-class 1 home-based work trips, 67 percent ended in a sub-center, while 33 percent did not. Mean travel times and mean travel distances by train, TW and intermediate public transportation (IPT) modes such as rickshaws were longer for work destinations in sub-centers than for work destinations in the urban periphery, but trips made by buses were shorter in sub-centers. Car users traveled longer and farther compared to TW users for home-based work trips in the GMR. Trains were the speediest mode of travel in the GMR, but traveling by a TW or car was speedier than bus or IPT travel—confirming that having a private vehicle has advantages. This research used a multinomial logit model to analyze households’ choice of having no vehicles, only TWs, or at least one car. Results indicated that household utility from both TWs and cars increased with household characteristics such as per capita annual income, living in an independent house or an apartment, number of rooms in the housing unit, housing location farther from a railway station, the presence of children under 5 years, and larger household size. Moreover, vehicle utility for households increased with the primary wage earner’s characteristics including college education, employment, being married, making more trips across all modes, traveling during the morning peak, and working in the urban periphery. Household utility from both TWs and cars decreased when the primary wage earner had longer work trips and higher employment density at the work location. Regression models for vehicle kilometers traveled (VKT) and person kilometers traveled (PKT) for cars and TWs showed that vehicle use increased with number of employed persons in the household, and if the primary wage earner worked in the urban core. Vehicle use decreased if density of housing and jobs went up at either the home or work location. TW use went down with per capita annual household income. Overall findings indicate that demand for private vehicles is rising due to the following factors: better education, employment, higher incomes, suburbanization, peripheral employment node formation, and lack of public travel options. However, higher density decreases vehicle use. Without changes in policies encouraging higher well-managed densities, jobs-housing balance, and supply of adequate transit and IPT travel options, vehicle ownership and use will likely continue to grow rapidly in India. 2 To that which is To that which may yet be To hope i TABLE OF CONTENTS CHAPTERS Page 1 Factors Contributing to the Rapid Rise of Private Vehicles in India: Framing the 1 Issue of Vehicle Ownership and Use 2 Case Study: Greater Mumbai Region (GMR) – Growth Dynamics and Household 23 Travel Survey Dataset 3 Employment Centers and Middle-Class Travel Behavior in Mumbai 50 4 Vehicle Ownership in Middle-Class Indian Households: Mobility in Mumbai 66 5 The Impact of Suburbanization on Car and Two-wheeler Use in Mumbai 81 6 Conclusions 94 REFERENCES 98 APPENDICES 1 Statistical Test Outputs for Employment Center Indicators 104 2 Why Not A Nested Logit Model for Vehicle Ownership in the GMR? 116 3 Output for the Multinomial Logit Model for Vehicle Ownership 122 4 Regression Model Outputs for Vehicle Use 126 ii ACKNOWLEDGEMENTS Many forces shape what we are, what we do, and how we do things. Soyabbhai, Pravinbhai, Manharbhai. Friends, more often guideposts to my moral compass, always men with dignity who gracefully accepted the hand life had dealt them. Sometimes chauffeurs. You helped me understand how the other lives. You shared the hopes and dreams you had for your children with me. I may have never embarked on this journey if you had not been around to drive me from airports to corporate offices, often very early in the morning, sometimes too late at night. A disaggregate analysis of consumption behaviors is rare in the context of public policy in India. I was lucky to have a series of contacts that led me to MMRDA (Mumbai Metropolitan Region Development Authority) who held an extensive household travel survey dataset – making such an analysis possible. I thank Mr. PRK Murthy (Chief, Transport and Communications Division) for granting me access to this non-public dataset, and Mr. Manoj Dandare for his help at MMRDA. Mr. B. M. Setty of LEA Associates South Asia Private Limited was helpful with other relevant pieces of information. Elizabeth Deakin (Betty) has helped me evolve into a better policy thinker. She supported me not just by being a mentor during the many years of working together, but also financially through research positions. She constantly challenges me to think critically, and I hope she will keep doing so in the future. Robert Cervero, though busy with travel and other engagements, was supportive with careful critique and encouragement. Joan Walker was helpful with the choice models. I thank her for bringing this elegant technique to those of us interested on the policy applications side. Lee Schipper was an optimist on policy questions in India, and always had a contact person to go to when I was stuck with an issue. His passing on during my dissertation writing phase was saddening. The University of California Transportation Center’s Dissertation Grant and the Dean’s Normative Time Fellowship at UC Berkeley provided funding for the analysis and writing phase of my Ph.D. My professors and colleagues have helped me tremendously over the years. However, all opinions, errors or omissions in this work are mine. Jin Murakami was my sounding board on specifics of research design and strategy. Liming Wang and Sheng Li helped me with questions on database management, statistical and GIS analysis. Ye Kang Ko, Sara Jensen, and Allison Lassiter helped me with the nitty-gritty of GIS analysis. Akshay Vij was my go-to choice modeling guru. Colleagues in the Ph.D. Program at Berkeley – Bruce Appleyard, Gautam Bhan, Andrea Broaddus, Ian Carlton, Asavari Devadiga, Reim El-Zoghbi, Lisa Feldstein, Julie Gamble, Erick Guerra, Sara Hinkley, Carrie Makarewicz, Sergio Montero, Greg Newmark, Wei-Shiuen Ng, Billy Riggs, Renee Roy, James Rubin, Rebecca Sanders, Bob Schneider, Allie Thomas, Jake Wegmann, and Miriam Zuk – were there to talk and share during these years of production. I have learned a lot from them. In Mumbai, Amita and Amitabh RoyChowdhury opened their home and hearts to me during the months of field work. This family adopted me and made me feel welcome in my case city. They were inquisitive about my research and helpful when I was frazzled. In Berkeley, Robert Lew was not just my landlord, but also a friend and mentor. He saw me jubilant and stressed during these years. He was always there; for this I thank him. Last but iii not the least, my parents Anand Shirgaokar and Jyotsna Shirgaokar, my sister Rashmi Vaidya and brother-in-law Sachin Vaidya understood my need to indulge in original research.