Modeling emission reduction benefits of the premium point-to-point bus service in Metropolitan Manila, Philippines – a scenario analysis

Abstract Road transport has been one of the largest contributors to global carbon emissions. In the Philippines, one such measure is the introduction of a relatively new public transport mode called the “Premium Point-to-Point (P2P) Bus Service”, which features shorter travel time and lower emission intensity than conventional buses. In light of the significant deficiency in transport emission studies for Philippine cities, this paper presents a scenario-based method for quantifying the potential emission reduction benefits of the P2P bus in Metropolitan Manila. Based on a pro rata mode shift assumption, it is estimated that a 1% mode shift of passengers from existing motor modes to P2P buses would lead to 0.64% reduction of total emissions over selected travel routes in Metro Manila. The analytical method addresses the acute emission data scarcity in the Philippines, and provides early evidence on the aggregate environmental benefits of P2P buses. Extensive sensitivity tests are conducted to verify the simulation results and identify key determinants of emission reduction. The tests confirm the robustness of research findings and further uncover the great potential of increasing vehicle occupancy levels to mitigate vehicle emissions in Metro Manila. Policy implications for enhancing the environmental benefits of P2P buses are provided.


Introduction
Road transport alone constitutes 10% of global greenhouse gas emissions (GHGs), posting the highest rate of increase among all sectors in most developing countries, and it has been the dominant source of carbon emissions within the overall transport sector (Sims et al., 2014;United Nations Climate Change, 2021).In the Philippines, the transport sector and the energy supply sector are the two most significant sources of GHG emissions (Asian Development Bank, 2017).In 2015, road passenger transport in Metro Manila emitted 13.78 million tonnes of CO 2 , which could balloon to 27.90 million tonnes in 2040 without interventions (Ahanchian & Biona, 2014).Among major cities in East and Southeast Asia, Manila is one of the most polluted, with the level of particulate matter three times higher and nitrogen dioxide twice higher than what is deemed acceptable by the World Health Organization (Japan International Cooperation Agency (JICA) and National Economic Development (NEDA) Authority, 2014a).Despite the heavy emissions observed in major cities, the national average GHG emissions in the Philippines (1.6 tons CO 2 e per capita in 2012) remained much lower than the global average of 6.5 tons (Asian Development Bank, 2017).
The Philippines, cognizant of its vulnerability to climate change (World Bank, 2013) and as a party to the United Nations Framework Convention on Climate Change (UNFCCC), submitted its Intended Nationally Determined Contributions in 2015, which declared its ambition to reduce greenhouse gas emissions by 70% by 2030 (United Nations Framework Convention on Climate Change (UNFCCC), 2015).The Philippine national government has been promoting cleaner and more energy-efficient modes of transport, especially public transit modes.One of these initiatives is the "Premium Point-to-Point (P2P) Bus Service", initially introduced in Metropolitan Manila (Metro Manila) and supported by the national Land Transport Franchising and Regulatory Board (LTFRB).P2P bus is an airconditioned express bus service generally with no stops in between the origin and destination (Department of Transportation (DOTr), 2015).In contrast to the existing and heavily polluting buses, P2P buses are required to comply to the Euro-4 (diesel) emission standard or better (DOTr, 2015;Land Transport Franchising and Regulatory Board (LTFRB), 2015).The P2P bus service aims to provide quick and clean access to major employment, commercial, or residential centers in Metro Manila, and has gained much traction since its launch in 2015.
This study represents one of the earliest studies aiming to estimate the aggregate environmental benefits of the P2P bus service in Metro Manila.The emission analysis covers a wide range of air pollutants: carbon monoxide (CO), nitrogen oxide (NO x ), sulfur oxide (SO x ), particulate matter (PM), hydrocarbon (HC), and carbon dioxide (CO 2 ).It should be noted that there is a significant research deficiency in motor vehicle emissions in the Philippines and other low-to-middle income countries in Southeast Asia, where timely policy interventions are urgently needed to curb car dependence as income rises.There is also a lack of detailed, up-to-date vehicle emission data for Philippine cities in the literature, and emerging modes of transport such as the P2P bus tend to be overlooked.
The paper is expected to contribute to the literature on several aspects.First, to address the vehicle emission data deficiency for Philippine cities, we collate vehicle emission data from various sources and provide a benchmark vehicle emission data set which could be further applied, verified, and updated in future research.Second, this study presents one of the first empirical evidence on the environmental benefits of the emerging P2P bus service in Metro Manila.The study also identifies key determinants of the magnitude of emission reduction through extensive sensitivity tests.Policy implications are drawn which are expected to inform the upscaling and management of cleaner modes of public transport in the Philippines.The paper is structured as follows.Literature review is provided in Section 2, which is followed by introducing the study area and the method for emission estimations including assumptions in Section 3. Section 4 discusses the scenario analysis results, sensitivity tests, and policy implications.Conclusion is presented in Section 5.

Motor vehicle emissions in Asian cities
Policy research on vehicle-related emissions in Asia has focused on emission control and improving energy efficiency, comparisons between traditional and emerging modes of travel, and the feasibility and impact analysis of novel energy sources as a replacement for fossil fuels.Sun et al. (2021) evaluated Beijing's policies on transport energy conservation and emissions reductions, pointing out that Beijing achieved a reduction of nitrogen oxide (NO x ) by 43% for petrol vehicles through a phased uplifting of vehicle emissions standards for both new and existing vehicles.In Shenyang, China, CNG-fueled buses and taxis were compared with other buses and taxis, respectively, and the empirical analysis suggested that new energy modes have distinct cost-benefit advantage due to their outstanding performance in reducing environmental costs (Geng et al., 2013).Studying the Trans-Jogja bus system in Greater Yogyakarta, Indonesia, Dirgahayani (2013) estimated that, over a period of 14 years, over 65,000 tons of CO 2 and 546 tons of NO x could be reduced if the existing buses were upgraded to Euro-4 diesel buses, and a cumulative reduction of 1.3 million tons of CO 2 was estimated when all passengers shift from private cars, motorcycles, and conventional buses to Euro-4 diesel buses.
Nakamura and Hayashi (2013) investigated low-carbon transport policies of several cities across different continents and stressed that in spite of the idea that "various levels of CO 2 mitigation can be generated by types of measures, there is no single solution to achieve the challenging target of mitigation for each country and city."Moreover, there is empirical evidence that compact development prevalent in Asian cities may have a considerable effect on carbon emissions, which is further complicated by their particular context such as fast urbanization and the tendency of urban sprawl due to soaring demand for mobility and space.Promoting public mass transit could be a key component for managing the anticipated growth of GHG emissions (Nakamura & Hayashi, 2013).
However, Jin et al. (2017) stress that there is urgency for enacting policies that stem rising car ownership for fast growing areas such as Greater Beijing, as rising household income means decreasing price elasticity for travel, and ultimately implying insensitivity to public transport.It was also found that densification may slow down the rise of CO 2 emissions, albeit limited, and that the simultaneous implementation of road charging and land use-transport coordination may attract passengers to shift to public transport to a greater degree than when only one of either policy is implemented (Jin et al., 2017).
In terms of emission studies on the Philippines, Lucero et al. (2019) made an initial assessment of P2P bus routes in Metro Manila through rider surveys and found that among the surveyed P2P bus commuters on selected routes, 13% used to drive their own cars, 19% by taxi or ride-hailing services, 8% by jeepney 1 , 31% by regular bus, 1% by motorcycle, and 13% by utility vehicle.Fabian and Gota (2009) forecasted that annual CO 2 emissions in Metro Manila would range from 10 to 12 million tons during 2011-2015.In a hypothetical scenario where vehicle kilometers traveled (VKT) of two-and three-wheelers, cars, and jeepneys were reduced by 30%, the number total emissions would go down to less than 9 million tons by 2015.Moreover, the study pointed out that the rate of upgrading the vehicle fleet in the Philippines tends to be sluggish, indicating lagged effects of raising emissions standards (Fabian & Gota, 2009).Vergel and Tiglao (2013) also conducted a simulation of emissions in Metro Manila and estimated that the baseline transport emissions (HC, CO, NO x , SO x , and PM) in Metro Manila amounted to 1,710 tons daily by 2015.Through scenario analysis, the study found that an additional 164.1 kilometers of railway lines and 19.7 kilometers of busways in Metro Manila could reduce total daily transport-related emissions by 15% due to a mode shift to mass transit and the consequent reduction of vehicle-kilometre traveled (VKM) of private cars; it was also found that the 1 There are some unique travel modes in the Philippines.First, a jeepney is an affordable mode of public transport that is similar to a minibus.A typical jeepney normally has a fixed service route and features crowded seating.A utility vehicle (UV), also known as the Asian utility vehicle, is a small-size sport utility vehicle that is usually airconditioned and has a fixed service route.
implementation of the motor vehicle inspection system, which would enable authorities to enforce emission standards on public vehicles, could reduce daily emissions by 34%, and a complete switch to CNG fueled buses by 2015 could reduce emissions by 3.2% (Vergel & Tiglao, 2013).In terms of the daily reduction of fuel consumption, the aforesaid mass transit network expansion as a single intervention could reduce diesel and gasoline consumption in Metro Manila by 13% each, while a complete switch to CNG fueled buses could reduce daily diesel consumption by 28% (Vergel & Tiglao, 2013).
In terms of cross-country studies in Asia, Lin and Omoju (2017) estimated the transport emission elasticity with respect to income growth based on Southeast Asian countries, and found that a 1% increase in income level leads to a 0.853% increase in CO 2 emissions from transport, confirming that Asian countries "are still in the early stages of the environmental Kuznets curve (EKC) when economic development is still strongly associated with environmental degradation".Lee and Van de Meene ( 2013) studied 22 cities in Asia and found that an expansion of public transport infrastructure was associated with a reduction in CO 2 emissions.

Managing motor vehicle emissionsinsights from developed countries
In their meta-analysis on the relationship between travel and the built environment, Ewing and Cervero (2010) found that destination accessibility and distance to downtown are highly associated with vehicle miles traveled (VMT), although inversely, but variables such as population and density did not associate with VMT as much as expected.Access to a transit facility, road networks, and mixed land use were the most highly associated with transit use (Ewing & Cervero, 2010).Stevens (2017) found that distance to downtown, along with household or population density, are inversely related and most highly associated with driving, proposing that planners promote more housing and density in urbanized areas and their environs in order to curb car use.The experience of Singapore has shown how to manage private car use and promotion of public transport through a package of policies.The vehicle quota scheme (VQS) and certificate of entitlement (COE), as well as congestion pricing, were the main instruments that Singapore used to regulate the ownership and use of cars (Lam & Toan, 2006).The reason behind such measures is to encourage the use of public transport and deter high ownership of cars and their frequency of use, but the key to the overall transport policy of Singapore is integrating land use and transport design, using smart technologies to monitor and regulate private cars, and the enhancement of public transit services and infrastructure (Diao, 2019;Lam & Toan, 2006).

Study area
To identify the study area, first, major employment, commercial or residential centers in Metro Manila are identified based on the authors' local knowledge, supported by aggregate statistics and verified with local experts.Second, the selected centers are then superimposed with the endpoints of existing P2P bus services.This allows for the selection of centers served by P2P buses.Third, based on the observed origin-destination (OD) commuting matrix for Metro Manila and its environs, the labor catchment area of each selected center (i.e.all possible origins including the center itself) is identified and added to the research area.We should note that the OD data are not based on traffic analysis zones, but on aggregated districts or local areas.For each selected locality, we set point geometry in a GIS software based on authors' local knowledge, representing the travel origin and destination.Figure 1 shows the spatial extent of the study area, where red dots represent P2P bus stops.Appendix A contains a list of the areas and their representative land use functions.

Research design
The research aims to quantify the potential environmental benefit of P2P buses in terms of reducing air pollution and carbon emissions by attracting commuters from conventional modes of transport to P2P bus.The possible net increase of commuting demand caused by P2P buses, i.e. induced travel demand, is not considered.This analysis assumes a constant commuting demand across scenarios at the origin-destination pair level, and focuses primarily on how varying take-up rates of P2P buses would affect the total emissions through modal shift.Specifically, we focus on two research questions: (1) Can the P2P bus service mitigate air pollution and carbon emissions in road passenger transport in Metro Manila? and (2) What are the key policy variables that affect the magnitude of the mitigating effect?To answer the first research question, scenarios of varying levels of P2P bus uptake (1%, 3%, 5%, and 10%) are tested, and the emission estimates are compared against the baseline scenario where no P2P bus is in operation.A 1% P2P bus uptake assumes 1% of total commuters on the selected routes would shift from existing modes to P2P bus on a pro rata basis.To address the second research question, a series of sensitivity tests are conducted, focusing on the impact of the following parameters on emission estimation: pattern of mode shift from other modes to P2P bus, vehicle emission factors, and occupancy levels.

Modeling vehicle emissions
Given the nature of the study as a strategic assessment, the authors follow existing literature, adopting a distance-based method for estimating vehicle emissions in consideration of average vehicle speed.For vehicles under the same travel mode category as per the empirical data, only fuel-type difference is considered while other variations such as the engine size and the age of the vehicle are excluded due to data unavailability.The total emission (E f ijk ) of type f pollutant for traveling from location i to location j by travel mode k is defined by: Eq.1 where D ij is the distance (km) between location i and location j and U f k is the unit emission factor of type f pollutant per passenger-kilometre traveled (PKM), which is sourced from existing literature and is used as exogenous input in the analysis.We do not use the unit emission per vehiclekilometre traveled (VKM) because different travel modes have distinct passenger occupancy levels.Using a VKMbased measure for calculating emission changes induced by mode shift is subject to significant errors, as mass transit modes tend to have high VKM emission intensity but low PKM intensity due to significantly higher occupancy levels than private modes such as the taxi.For calculating D ij , the authors first calculate the Euclidean distance between all possible origin-destination pairs, and then multiply it by a detour index of 1.4 to approximate the network distance.This approximation method has been applied in a number of empirical studies for converting the crow-fly distance to approximated network distance, for example, Cole and King (1968), Boscoe et al. (2012) and Wan et al. (2021).A recent review is provided by Barth elemy (2011).

Benchmark vehicle emission data
Benchmark vehicle emission refers to the U f k variable in Eq.1.Transport and climate change data in the Philippines are difficult to obtain, especially granular data pertaining to vehicle types and speeds, due to institutional constraints in the generation, management, and sharing of such data (Mejia et al., 2017).Official transport data in the Philippines are not longitudinally consistent and the few existing policy analyses based on transport models heavily relied on foreign assistance (Regidor, 2019).Through reviewing data from various sources including both academic papers and published industrial/government reports (Appendix B), a comprehensive vehicle occupancy and emission data set for Philippine cities is developed.) can be calculated by dividing the VKM-based emission intensity by the average passenger number.For car, taxi, and UV, weighted average emission intensity with respect to the total PKM per fuel type is used, while emission factors for all other modes are based on single fuel type.Table 1 shows that passenger car and taxi have significantly higher emission intensity than public transport modes.In particular, the taxi has the highest emission intensity due to relatively low average occupancy level.

Baseline origin-destination commuting flow data
The baseline ( 2015) origin-destination commuting data set used in this study is the most recent of its kind in Metro Manila, which include a total of 14 travel modes and 432 designated zones, mostly constituting Metro Manila and surrounding provinces.The subset used in this study consists of six vehicle modes and almost 3,000 unique origin-destination pairs.In terms of aggregate mode share over the selected travel routes, 339,900 trips (46.3%) were by motorcycle, 252,300 (34.3%) by private car, 44,200 (6%) by taxi, 84,000 (11.4%) by jeepney, 9,500 (1.3%) by bus, and 4,800 (0.7%) by UV.Appendix C shows the mode share and the total vehicle volume by distance band in the baseline flow data.

Scenario Analysis -Quantifying emission reduction effect of P2P bus
A total of four P2P bus uptake scenarios are considered, with the following rates: 1%, 3%, 5%, and 10%.A pro rata mode shift from existing modes to P2P bus (based on the baseline mode share) is assumed for all four scenarios.The discussion in this article is based on the 5% P2P bus uptake scenario, whereas the results for all other scenarios are provided in Appendix D. The 5% uptake scenario is selected for two reasons: (1) results across all scenarios show a broad linear trend, and (2) the 5% uptake rate of P2P bus appears a realistic level of market penetration in Metro Manila.Total emissions from the 5% P2P bus uptake scenario are compared against the baseline where P2P bus service was absent.The baseline scenario generates a daily total of 1,836 tons of gas emissions (all pollutants plus CO 2 ) from the selected commuting routes, while the 5% scenario would produce 1,778 tons, with a net decrease of 58 tons (around 3.2%).It implies an emission reduction elasticity of À0.64 (3.2% emissions reduction divided by 5% P2P bus uptake), i.e. 1% P2P bus uptake would lead to 0.64% total emissions reduction, assuming a pro rata model shift from all other modes.Furthermore, for the P2P bus to achieve its emission-reduction benefit, the minimum occupancy level should be no less than 16.3 persons.Lastly, to achieve a lofty 30% reduction in CO 2 emissions, it requires a 57.5% pro rata mode shift to the P2P bus.

Sensitivity tests
To verify the model results and identify key policy variables that affect emissions reduction, a series of sensitivity tests is conducted.Specifically, the authors considered the following two tests: (1) different mode shift patterns from other modes to P2P bus, as opposed to the prorated approach in the baseline scenario and (2) different levels of vehicle emission factors and/or different levels of occupancy by travel mode.The first test explores complementary policies regulating the use of other travel modes, and the second test aims to identify minimum occupancy requirement for the P2P bus to realize its potential environmental benefits.Both tests are based on the 5% P2P bus uptake assumption for consistency.

Sensitivity Test 1: Different mode shift patterns to
P2P bus This test considers a range of more nuanced mode shift assumptions (see Table 2).The test aims to assess the effect on emissions reduction if complementary policies for promoting the P2P bus could be implemented, e.g.targeting travel modes with higher emission intensities.
Results on total emissions are presented in Figure 2. First, all schemes would produce a lower level of total emissions than the baseline scenario, implying an enhanced mitigating effect.It further corroborates the environmental benefits of P2P bus.Second, an interesting tradeoff is uncovered, where targeting the most polluting modes of travel (as in Schemes 5 & 6) does not necessarily lead to the lowest total emissions.Among the tested schemes, an equal split among all modes (Scheme 1) produces the lowest total emissions, with Scheme 3 producing the second lowest.This is because different commuting routes have a distinct mode split due to differences in the distance and mode optionality.Consequently, the most polluting modes do not necessarily generate the highest emissions because their passenger-kilometre traveled may not be the highest.
It should be noted that, although an equal split (Scheme 1) produces the lowest total emissions, it does not indicate that Scheme 1 is optimal for Metro Manila.In fact, targeting the most polluting travel destinations (see results in Appendix D) may generate the largest emissions reduction.For example, given Fort Bonifacio as the commuting destination, identifying the most polluting modes for each travel origin and tackling them accordingly would, in theory, maximize the emission mitigation effect.Nonetheless, the authors are prevented from determining such a maximum for two reasons: (1) the data quality of the observed mode split data at origin-destination level varies significantly and is subject to possible sampling and measurement errors and (2) designing and implementing origin-destination-specific and highly sophisticated policy interventions in Metro Manila seems unrealistic given its current development stage of the local governance system.Deriving such a mathematical optimum, though tempting, would have minimal practicality in terms of informing policy making.In contrast, the results of Scheme 3 are encouraging, as the scenario is based on preliminary empirical data on mode substitution by the P2P bus.Wang et al. (2011); and (e-g) authors' own estimation due to data deficiency.For (e), estimated based on figures for the gasoline cars and the ratio of the average emissions of a gasoline car and diesel car from the UK Department for Transport (2021); the same method was applied to (f).For (g), the authors assumed that SO x emissions of a P2P bus are 80% of a regular bus.
Lastly, while targeting the top three most polluting modes (i.e.car, taxi, and UV), as per Schemes 5 and 6, would generate a similar emission mitigation effect as Scheme 1, implementing the former would, however, be much more controversial than the latter.It suggests that a blanket approach of implementing prohibitive measures on certain modes in the name of emissions reduction may not be costeffective.Instead, improving the competitive advantage of the P2P bus renders a low-stakes and more feasible strategy that may achieve comparable outcomes.

Sensitivity Test 2: Perturbation of emission factors
and occupancy levels The vehicle emission factor and level of vehicle occupancy are crucial factors affecting the estimation of total emissions in this study.As a response to the emission data deficiency issue identified in the literature review, the authors developed a set of provisional vehicle emission factors by collating relevant data from various sources and the authors' own estimations (see Table 1).While the authors call for future research to verify and refine the provisional emission factors, it is acknowledged that some of the provisional emission factors are outdated at the original source and may be inaccurate.This sensitivity test thus aims to explicitly address this possible issue by investigating how different emission factors and average occupancy levels may affect the analysis results and alter the previous findings.
Table 3 provides the list of all perturbations to be tested in this sensitivity test, where the individual effects of the two variables are first tested (E1 to E9 and O1 to O8), then the combined effects (E1O1 to E3O8).Results are presented in Figures 3 and 4. In this test, the authors only consider the independent impact of each parameter and disregard the interdependence between the two parameters, i.e. changing occupancy level would alter the estimation of emission factors.
Figure 3 shows the individual effects on total gas emissions (E1 to E9 and O1 to O8).The blue marker indicates the baseline scenario, which has no perturbation of vehicle emission factors and occupancy levels at the 5% uptake scenario.Green markers indicate perturbations of vehicle emission factors while the red ones indicate perturbations of occupancy levels.Overall the changes in total emissions are   marginal for most of the perturbations, except for O8 to E9 which produced much lower total emissions.A 25% reduction of emission factors for the car would reduce total emissions by 13.7%, implying an elasticity of À0.55 for the car mode, i.e. a 1% reduction of the vehicle emission factors for the car would lead to a 0.55% reduction of total emissions for the selected commuting routes.By contrast, despite taxi being the most polluting mode in terms of emission factors, the emission elasticity of the taxi is much lower (-0.07).This is because the base year mode share of taxi and total passenger-kilometre traveled is small.
In terms of occupancy level changes, the elasticity of total emissions with respect to unit occupancy increase on all public modes (O7) and the car (O8) is around À0.27 and À0.51, respectively.This demonstrates great potential for mitigating vehicle emissions in Metro Manila through encouraging higher vehicle occupancy, particularly among passenger cars.Figure 4 further shows the results of the combined effects (E1O1-E3O8), where the three lines represent E1 series, E2 series and E3 series, and the blue line passes through the current and baseline scenario, the former having no shift to P2P buses.It shows that changes of emission estimates across E1 to E3 are broadly linear, which further indicates the stability of the baseline results.Overall, the results of the sensitivity tests suggest that the authors' primary findings are relatively robust under minor changes of vehicle emission factors and occupancy levels.

Policy implications
Given the scenario analysis and the subsequent sensitivity test results, policy implications for reducing transport emissions in Metro Manila are summarized as follows.First, P2P bus, as an emerging and cleaner mode of public transit, is expected to generate desirable environmental benefits for Metro Manila.The emission mitigation effect does not require a complete substitution of any existing motor mode, albeit a targeted approach (e.g. by mode, location, and travel route) could boost the mitigation effect.Our scenario analysis based on a 5% pro rata mode shift to P2P buses would reduce overall emissions by 3.2% from the baseline.A much higher uptake of P2P buses (57.5%) could reduce CO 2 emissions by 30%.
Second, policies aiming to reduce transport-related emissions in Metro Manila need to address the complex tradeoff between targeting travel modes.For example, tackling passenger car mode may lead to increasing use of taxi due to the substitution effect, which would cancel the emission reduction effect (see Scheme 4 in Figure 2 increasing emissions from taxis leads to higher emissions than the baseline).Blanket measures without considering such tradeoffs may see backlash from commuters and even undermine the environmental benefits of P2P buses.
Third, exploring the complementarity between P2P bus and other modes of travel, particularly public modes, may be considered for boosting the uptake of P2P buses and hence amplifying the emission reduction effect.For example, the aggregate mode split in the base year data (see Appendix C) suggests the potential of promoting multi-modal transit involving P2P bus, regular bus, and UV as viable alternatives to private car use.It is also suggested to further differentiate the services between P2P bus and the regular bus, in terms of service range, in order to manage the possible competition between the two public modes.
Fourth, while improving energy/emission efficiency and increasing vehicle occupancy level could both reduce total emissions, the former would require considerable capital investments and fiscal incentives to enable the transition, and thus put a strain on public finance.By contrast, increasing occupancy levels represents a consumer-side behavioral change, which, in theory, would be less capital-intensive.The emission mitigation effect from travel behavior changes such as increasing vehicle occupancy levels could be substantial, which is in line with the findings in Ng (2018) and Sch€ afer and Yeh (2020).Note that increasing vehicle occupancy level does not necessarily curb travel demand; instead, its emission mitigating effect is achieved through lowered emission per passenger-kilometre traveled.This implication is particularly relevant for Metro Manila and other fastgrowing city-regions of similar context, where demand-side behavioral changes are critical and imperative for promoting a low-carbon lifestyle, in light of the increasing mobility demand associated with income growth.
Lastly, the explicit spatial dimension in the selection of major business centers and major residential areas, as well as P2P bus routes in this study enables a high-level investigation into the spatial distribution of jobs and housing in Metro Manila.Similar to other established metropolitan areas such as Beijing and London, employment in Metro Manila is highly concentrated in central areas while residents are relatively dispersed.For suburban locations, existing public modes are hardly viable in terms of access, comfort, and travel speed, while P2P bus appears to be a competitive public transit alternative for long-distance car commuters.
To mitigate the transport emissions over the long term, spatial planning in Metro Manila may experiment with novel strategies that integrate land use and transport, e.g.providing subsidized P2P bus services to new and dense housing development in suburbs as an integral package for new residents.The P2P bus could be further complemented by micro-mobility solutions for first/last mile access for large neighborhoods.For suburban residents, P2P bus can either provide door-todoor access to major business centers or serve as a feeder mode to other mass transit.To make such an aspiration a reality, it requires extraordinary coordination among P2P bus operators and between P2P bus operators and local transport authorities.The government is expected to play a vital role in terms of facilitating, mediating, and financing.

Conclusion
This study aims to quantify the potential emission mitigating effect of P2P bus in Metro Manila and identify key policy variables that determine the magnitude of emission reduction.It should be noted that there is a significant gap in literature examining transport emission for Philippine cities, while the need for timely policy interventions is acute.To address the significant research gap, first, the authors establish a provisional vehicle emission data set for the Philippines through extensive literature review.Second, a scenario-based method is proposed for simulating potential reduction of vehicle emissions over selected commuting routes in Metro Manila.Based on a pro rata mode shift assumption, it is estimated that a 1% mode shift from existing modes to P2P bus based on observed mode share would lead to a 0.64% reduction of total emissions over selected commuting routes.Furthermore, a minimum occupancy of 16.3 persons for the P2P bus would need to achieve such that the emission-reduction benefit could be realized.The study provides one of the first evidence on the environmental benefits of P2P bus in Metro Manila.
For verifying the simulation results, extensive sensitivity tests are conducted.The sensitivity test results confirm the robustness of research findings from the baseline scenario and further identify key policy variables that affect the emission mitigating effect: (1) patterns of mode shift from existing modes to P2P bus, (2) vehicle emission factors, and (3) average vehicle occupancy levels.In particular, it demonstrates the significance of reducing vehicle emissions through increasing vehicle occupancy levels and considering mode-or location-specific policy interventions.The study found that BF Homes, Ortigas Center, Phil-Am, Antipolo (South), and Dasmariñas would see the greatest reduction in the scenario with a 5% uptake of P2P buses, due to high share of car and jeepney use for commuting.The detailed ranking of locational emissions reduction potential is provided in Appendix E. Our findings support a place-based approach for promoting and prioritizing P2P bus provision.
In terms of limitations, first, the benchmark vehicle emission data are not empirically sourced and validated.Despite extensive sensitivity tests, the deficiency of up-to-date local vehicle emission statistics remains as an outstanding gap, which calls for more research contributions.Second, while place-based policy implications have been drawn, it should be noted that air pollution and carbon emission are not physically confined within the selected localities and along the routes.Lastly, it must be noted that this paper disregards other factors affecting emission intensity which include, but are not limited to, vehicle age, engine size, vehicle weight, and terrain traveled due to data deficiency.Expanding the emission estimation by considering detailed road network and variable travel speed is suggested for future research.

Figure 1 .
Figure 1.Spatial extent of the study area (Red dots are P2P bus stops).

Ã
Scheme 5:Equal split among top 3 most polluting modes

Figure 2 .
Figure 2. Results of Sensitivity Test 1: Different mode shift schemes to P2P bus.
Table 1 reports emission intensity by pollutant type and vehicle type, as well as the average occupancy and the determined U f k per vehicle.It should be noted that the emission intensities in Table 1 are based on mode-specific average speed obtained from the literature.The effect of variable speeds on emissions was not considered, which is a limitation of this research.Based on the VKM-based emission data and the average occupancy level in Table 1, the PKM-based unit emission (U f k

Table 1 .
Emission intensity by pollutant type and by vehicle and fuel type.Taxis are treated like cars by the authors.2Regularbusesareallassumed to diesel only.Unverified data suggested that less than 5% of buses run on gasoline in Manila.3Dataweresourced from the following: (a) Vergel and Tiglao (2013); (b) Nilrit and Sampanpanish (2012); (c) Japan International Cooperation Agency (JICA) and National Economic Development (NEDA) Authority (2014b); (d)

Table 2 .
Sensitivity Test 1: Different mode shift schemes to P2P bus.