copy the linklink copied!Chapter 6. Beyond the case study: interpreting the findings and assessing their wider relevance

This chapter explores the degree to which the key policy recommendations of the study are relevant to contexts beyond Auckland and New Zealand. To that end, it revisits the projected reductions in greenhouse gas emissions and other key outcomes under different assumptions about the carbon intensity of the electricity generation sector, population growth and income evolution. Moreover, the chapter offers an extensive sensitivity analysis with respect to various model parameters including the evolution of preferences for open space, fuel efficiency, electricity and fuel prices, as well as of the pace at which advantages of conventional vehicles vis-à-vis electric cars fade out. That analysis identifies the factors that have an important impact on the key conclusions. Finally, the chapter enumerates a series of methodological limitations of the analysis and the impact they have on the key conclusions of the study.

copy the linklink copied!6.1. Sensitivity analysis and external validity

This section assesses the extent to which the results presented in Chapter 5 are driven by the assumptions made about the evolution of exogenous variables. Population, income, the pre-tax purchase cost of vehicles and the prices of electricity and fuel are some of these exogenous factors. They play a key role in determining vehicle use, emissions, housing prices and other key outcomes of the model simulations. The reasons for which these variables are considered exogenous in MOLES are elaborated in Chapter 2. Conducting sensitivity analysis with respect to the exogenous variables is essential for two reasons.

Sensitivity analysis examines the stability of the main findings under alternative assumptions about the evolution of exogenous factors. That is, it enables to test whether the findings are qualitatively stable. In the context of this study, this means that the positive impact of various policies on emissions, welfare and other key outcomes remains positive, and vice versa. Qualitative stability also implies that the ranking of policies remain intact, i.e. the most socially desirable policies remain at the top of the list. Moreover, sensitivity analysis allows gauging the quantitative stability of findings, i.e. the extent to which their magnitude changes when the basic premises of the simulation exercise are altered. In the context of this study, policies induce welfare gains or losses, emission reductions or raises and budget surpluses or deficits. Sensitivity analysis helps to identify the background factors whose change causes the aforementioned effects to vary considerably.

Most importantly, sensitivity analysis is essential to gauge the degree to which the key findings and policy recommendations of this study are externally valid. That is, the analysis has a high degree of external validity if socially desirable policies identified for Auckland can be successful in achieving the same objectives in other urban areas. The sensitivity of the study’s findings to its background premises is therefore key to advocate in favour or against the adoption of the policy recommendations included in this report by other cities that differ regarding these background factors.

6.1.1. Evolution of the attributes of electric vehicles

The assumptions regarding the pace of technological change in the EV industry are important for aggregate emissions. The model explicitly incorporates factors such as the purchase cost and the driving range of EVs. However, there is also a number of EV attributes, whose effect is modelled only implicitly. These attributes include, but are not limited to: charging time, the variety of EV models available and the cost of their spare parts. The evolution of such attributes matters for the take-up of EVs.1

The reference and counterfactual scenarios are simulated under a relatively pessimistic assumption about the evolution of these factors. More specifically, it is assumed that the advantage that ICE vehicles possess over EVs shrinks over time, but remains substantial until 2050. This assumption turns out to be important in determining the share of EVs in the vehicle fleet. In the reference scenario, that share remains modest through 2050. The share of households that own an EV in the model increases from 0.6% in the benchmark year to 10.9% in 2050. The “promote EVs” policy package increases this share to 13.1% by raising the pecuniary costs of owning and operating an ICE vehicle and by subsidising EVs. That difference is represented by the distance between the solid and the dotted line in Figure 6.1.

Adopting more optimistic view on the evolution of factors driving the advantage of ICE over EVs changes the picture considerably. Figure 6.1 shows that, under such view, the share of households that own an EV turns out to be 41.6% in 2050. This “alternative reference” scenario is represented by the long dashed line in Figure 6.1. It assumes that the various advantages of ICE vehicles vis-à-vis EVs disappear completely by 2050. The EV share in that alternative case can be increased further, from 41.6% to 46.8%, with the “promote EVs” policy package.

Figure 6.1. The share of EV ownership is highly influenced by assumptions about pace of technological change

Figure 8.2 shows projected emissions to 2050 under various assumptions about EV attributes. The higher rate of EV penetration in the alternative reference scenario translates to substantial aggregate emission reductions compared to the original reference scenario. Emissions in the latter case increase by 6% between 2018 and 2050, as detailed in Chapter 5. In contrast, the alternative reference scenario that adopts a more optimistic view on the evolution of EVs yields a 30% decline in emissions during the same time interval (2018-2050). When the “promote EVs” package is implemented in combination with an optimistic view on the evolution of EVs, aggregate emissions fall by 64% in 2050 relative to 2018.

Figure 6.2. Optimistic expectations regarding the evolution of EV attributes and aggregate emissions

6.1.2. Evolution of conventional vehicle fuel economy and grid carbon intensity

The study makes specific assumptions about the evolution of the GHG emissions per vehicle kilometre. For ICE vehicles, greenhouse gases are emitted directly from the tailpipe. Therefore, the emission intensity of ICE vehicle use depends on the fuel economy of the vehicle. For EVs, emission intensity is indirect: it depends on the electricity consumption per kilometre and on the carbon intensity of the electricity grid.

The evolution of ICE vehicle fuel economy is subject to uncertainty. On the one hand, the increasing share of hybrid cars, which have both an electric and an internal combustion engine, contributes to an improvement of their fuel economy. On the other hand, a decreasing investment in R&D that aims to improve the fuel economy of ICE vehicles contributes to keeping their fuel economy stagnant.

Improvements in the fuel economy of ICE vehicles lowers the fuel cost and therefore the per-kilometre cost of their use. The study assumes a 55.4% increase in the fuel economy (kilometres per litre) of ICE vehicles between 2018 and 2050. This makes ICE vehicles relatively cheaper to operate over time and hampers the switch to other transport modes.

Assuming no improvement in the fuel economy of ICE vehicles, aggregate GHG emissions from road transport are projected to increase by 60% between 2018 and 2050. That change, which is depicted by the solid black curve in Figure 6.3, has to be juxtaposed against the solid grey curve in the same figure. The latter displays the change in aggregate GHG emissions predicted by the original reference scenario over the same time period (6%). The dashed curves in the same figure indicate that the difference between the two scenarios is almost entirely due to the additional greenhouse gases ICE vehicles emit in the alternative reference scenario.

The sensitivity analysis displayed in Figure 6.3 provides useful insights on how urban transport GHG emissions could involve in contexts with lower fuel economy standards than those imposed in New Zealand.

Figure 6.3. Keeping conventional fuel economy fixed at its 2018 level significantly increases emissions from ICE vehicles

The carbon intensity of New Zealand’s electricity grid, which is significantly lower than in other countries, is the main driver of the low level of GHG emissions from EVs. A more carbon-intensive electricity grid results in higher GHG emissions per kilometre driven by EVs. Therefore, it hampers policies that promote EVs to reduce GHG emissions.2

The sensitivity analysis uses an alternative reference scenario, in which the carbon-intensity of the electricity generation sector is given by the global average value. This analysis allows revisiting the effects of policies examined in Chapter 5, in a world where electricity generation is much more carbon-intensive. In order to conduct this analysis, historical data from the IEA (2019[1]) are extrapolated to the period 2018-2050. This extrapolation generates an upper bound estimate of the average carbon intensity of the global electricity supply. This is presented in Table 6.1, alongside with the projected evolution of the carbon intensity of New Zealand’s electricity supply. The latter is used in the original reference scenario, whose results are presented in Chapter 5. Using the upper bound values of carbon intensity results in higher GHG emissions per EV kilometre, as Figure 6.5 indicates.

copy the linklink copied!

Table 6.1. Projections of the carbon intensity of electricity
	Unit	2018	2030	2050
New Zealand	kgCO2e/km	0.119	0.057	0.022
World	kgCO2e/km	0.502	0.454	0.375
Source: IEA (2019[1]); authors calculations.

Figure 6.4. More carbon intensive electricity significantly increases the emission factors of EV

Aggregate emissions increase from both EVs and public transport, particularly in the long run, when buses are projected to be fully electric. In the reference case, aggregate GHG emissions increase by 6% relative to the 2018 benchmark. In contrast, these emissions are 13% higher in 2050 than in 2018 with a more carbon intensive grid. This is displayed in the upper panel of Figure 6.5. The contribution of EVs and public transport in total GHG emissions increases as well. In the original reference scenario (clean New Zealand electricity grid), EVs and public transport are responsible for 5.4% of emissions in 2018. That number falls to less than 1% in 2050. In contrast, with a more carbon intensive electricity grid, EVs and public transport make up 8% of total emissions in 2018. That number falls only marginally in 2050, to 7%. This is displayed in the lower panel of Figure 6.5.

Figure 6.5. Higher carbon intensity of electricity increases aggregate emissions and the share of emissions from EVs and public transport

6.1.3. Evolution of electricity and fuel prices

Electricity and fuel prices are two determinants of the operational costs of EVs and ICE vehicles. The original reference scenario Chapter 5 assumes that both electricity and fuel prices increase moderately between 2018 and 2050. The electricity price (NZD/kWh) increases by approximately 20% over the course of the study, while the fuel price (NZD/litre) increases by approximately 14%. These projections are based on extrapolations of historical data. For instance, the projection of electricity prices exploits the observation that, since 1980, these prices have increased on a yearly basis, with the exception of the period 2013-2017 where prices remained relatively stable.

A limitation of the study is that electricity prices do not depend on the penetration of EVs. However, higher EV adoption rates imply higher demand for electricity. That would require an increase of the total capacity of the electricity grid, or it could otherwise translate into higher prices. On the other hand, if electricity prices develop along the trend of the most recent years, prices will not increase and may even fall over time.

The analysis that follows explores the implications of deviating from the electricity price path assumed in the original reference scenario of Chapter 5. The alternative electricity prices, which can be considered as upper and lower bounds to the reference scenario, are presented in Table 6.2.

copy the linklink copied!

Table 6.2. Range of electricity prices simulated in sensitivity analysis
Variable	Unit	2018	2030	2050
Electricity prices used in the original reference and counterfactual scenarios	2017 NZD per kWh	2.31	2.68	3.07
Lower bound electricity prices	2017 NZD per kWh	2.31	1.34	1.53
Upper bound electricity prices	2017 NZD per kWh	2.31	4.02	4.60
Source: IEA (2018[2]); authors calculations.

Using the upper and lower bound of electricity price paths gives rise to alternative reference scenarios in which aggregate emissions do not differ substantially. Assuming that electricity prices evolve according to the lower bound path is associated with a marginal decrease (0.5%) in aggregate emissions. Similarly, assuming that electricity prices evolve according to the upper bound path is associated with an increase in aggregate emissions of a similar magnitude. This implies that electricity prices play a secondary role in predicting EV penetration rates.

Since price of gasoline affects the purchase and use of ICE vehicles, it is among the factors that determine aggregate GHG emissions. In New Zealand, that price has evolved in line with global oil prices in recent years. The projection used in the original reference scenario of Chapter 5 assumes a moderate increase of 14% in gasoline price between 2018 and 2050. The sensitivity analysis explores alternative scenarios for the evolution of gasoline prices. Upper bound and lower bound paths for the price are constructed as 50% higher and lower than the price in the original reference scenario, respectively. These are used in alternative reference scenarios, which are compared to the original reference scenario presented in Chapter 5. The resulting alternative price paths are shown in Table 6.3.

copy the linklink copied!

Table 6.3. Range of fuel prices simulated in sensitivity analysis
Variable	Unit	2018	2030	2050
Fuel prices used in the reference	2017 NZD per litre	0.97	1.11	1.21
Lower bound fuel price in alternative reference	2017 NZD per litre	0.97	0.55	0.61
Upper bound fuel price in alternative reference	2017 NZD per litre	0.97	1.66	1.82
Source: IEA (2018[3])

Changes in fuel price alter the attractiveness of ICE vehicles relative to EVs and public transport. Increasing the reference fuel price in 2050 by 50% reduces aggregate emissions by 2% relative to the reference case. That effect is roughly symmetric: decreasing the reference fuel price in 2050 by 50% increases aggregate emissions by 2% relative to the reference case. This is displayed in Figure 6.6.

Figure 6.6. Higher fuel prices lowers aggregate emissions, while lower fuel prices increase emissions

6.1.4. Evolution of income and population growth

Income and population are key determinants of GHG emissions in the study. Income bounds the expenditure capacity of individuals and shapes aggregate demand for housing, travel and various commodities. Population affects aggregate housing demand and therefore housing prices.

The original reference scenario of Chapter 5 uses the historical evolution of the per capita disposable income in New Zealand to extrapolate income growth (Statistics New Zealand, 2017[4]). The outcome of the projection is a real disposable income that grows steadily over time. Compared to 2018, it is 20% higher in 2030 and 54% higher in 2050.

To test the evolution of GHG emissions under alternative assumptions of income growth, income is kept fixed throughout the period 2018-2050. This slows the growth in aggregate GHG emissions, as emissions grow by 2% between 2018 and 2050 instead of 6%. This is displayed in Figure 6.7.

Figure 6.7. Keeping income fixed to its 2017 level reduces emissions relative to the reference case

Keeping income fixed alters the evolution of housing prices in Auckland considerably. Housing prices increase by more than 200% in the period between 2018 and 2050 in the original reference scenario examined in Chapter 5. In contrast, the alternative reference scenario, in which income is fixed, yields a much more moderate increase in housing prices. That is approximately 50%, as displayed in Figure 6.8. The juxtaposition of the original and alternative reference scenarios in that figure implies that a substantial part of the housing price increase is driven by real income growth.

Figure 6.8. House prices under different scenarios of income growth

Population growth has important implications for aggregate GHG emissions, as already discussed in Chapter 5. The projection of population used in the original reference scenario assumes that Auckland’s population will grow by 75% between 2018 and 2050. However, there is substantial uncertainty regarding the rate of population growth in that long time interval. Therefore, two alternative reference scenarios are constructed using upper and lower bounds of population growth. The associated paths of population evolution are presented in Table 6.4.

copy the linklink copied!

Table 6.4. Population growth scenarios simulated in sensitivity analysis
	2018	2030	2050
Population used in study	1,300,000	1,690,000	2,600,000
Lower bound population	1,300,000	1,518,400	1,604,200
Upper bound population	1,300,000	1,604,200	2,282,800

Figure 6.9 displays the evolution of GHG emissions in the alternative reference scenarios that assume a lower and an upper bound for the evolution of population. If population growth is more rapid than assumed in the original reference scenario, aggregate GHG emissions in 2050 may be 20% higher than 2018. The population in this case will continue growing, in the entire period from 2018 to 2050, with the same pace observed during the last years. On the other hand, if population growth is slower than that assumed in the original reference scenario, GHG emissions in 2050 will be 15% lower than in 2018.

Figure 6.9. Emissions under alternative scenarios for population evolution.

Like income growth, population growth has a significant impact on housing prices. Figure 6.10 displays the evolution of housing prices from 2018 to 2050 under the two alternative reference scenarios, which assume different population growth rates. If population growth is more rapid than what was assumed in the original reference scenario, real house prices in 2050 will be almost four times their level in 2018. On the other hand, if population growth is slower than expected, the associated increase will be approximately 110%.

Figure 6.10. House prices under different scenarios of population growth

6.1.5. Sensitivity analysis of preferences for open space

The welfare impact of densification policies depends on whether private open spaces, such as backyards, are substitutes or complements to residential floor space. The original reference and counterfactual scenarios, which were examined in detail in Chapter 5, assume that the preferences for private open spaces are particularly strong in New Zealand. Under these scenarios, it is also assumed that residential floor space and private open spaces are strong complements. This means that household welfare increases with a larger home, but that increase is relatively small if it is not accompanied by an increase in private open space.

Assuming strong preferences for open space therefore implies that densification policies may lower welfare. The reason is that densification programmes increase building height and decrease the space between buildings. By doing so, such policies result in reductions in the amount of open space available for each unit of residential floor space. This may have a negative welfare effect, which could offset the positive welfare effect from the decrease in housing prices these programmes cause. This possibility is illustrated clearly in Table 6.5 and Table 6.6. Both tables show that certain densification programmes explored in the report cause welfare losses, when preferences for private open space are strong, as assumed in the original reference and counterfactual scenarios. The negative welfare effect of some densification packages is reversed in the long run, as the positive effects of reduced housing price growth offsets the detrimental effect densification has on individual preferences. It can also be seen that part of negative welfare effects of densification are mitigated when such policies are combined with the “promote public transport” package, which has a positive welfare effect per se.

This section conducts sensitivity analysis to identify the impact of the assumptions made on preferences for open space. More specifically, the analysis is based on an alternative reference scenario, in which the preference parameters that govern the demand for private open spaces are relaxed. The alternative reference scenario also relaxes the degree to which residential floor space and private open spaces are complements. The findings from applying the “promote public transport” and the various densification packages under weak preferences for open space are displayed in Table 6.5 and Table 6.6.

copy the linklink copied!

Table 6.5. The welfare impact of densification policies
The welfare effect of stand-alone densification packages, when preferences for open space are strong or weak
	2030		2050
Preference for private open space	Strong	Weak	Strong	Weak
Widespread densification	-0.62%	3.84%	7.36%	14.31%
Transit-oriented densification	-4.10%	-2.06%	-0.83%	2.13%
CBD-surrounding densification	0.23%	1.46%	1.50%	3.16%
Isthmus densification	-1.01%	1.51%	2.11%	5.63%
Job hub-surrounding densification	-9.50%	-5.87%	-1.87%	3.84%
Note: Welfare impacts are expressed as percentages of net income.
Source: Authors’ calculations based on outcomes from MOLES.

copy the linklink copied!

Table 6.6. The welfare impact of combining densification policies with the promotion of public transport
The welfare effect of various densification packages combined with the “promote public transport” package, when preferences for open space are strong or weak
	2030		2050
Preference for private open space	Strong	Weak	Strong	Weak
Widespread densification	0.78%	5.11%	7.37%	13.52%
Transit-oriented densification	-2.84%	-0.92%	0.09%	2.65%
CBD-surrounding densification	1.28%	2.40%	2.21%	3.64%
Isthmus densification	0.26%	2.70%	2.80%	5.91%
Job hub-surrounding densification	-7.86%	-4.30%	-0.94%	4.15%
Note: Welfare impacts are expressed as percentages of net income.
Source: Authors’ calculations based on outcomes from MOLES.

Both tables suggest that a substantial shift in preferences causes the welfare effects of various densification packages to improve. This is intuitive as densification brings about a smaller welfare loss if preferences for open space are weaker. The most profound change is that widespread densification, combined with a promotion of public transport, turn out to be the optimal policy response even in the mid-term. The reason for this is that under weak preferences for open space, well-being is primarily determined by housing affordability. This can be verified by comparing welfare impact of the “CBD-surrounding densification”, the “Isthmus densification” and the “Widespread densification”. The sequence of these packages gradually generalises densification from the inner core of the city to the entire urban area. When preferences for private open space are weak, the welfare-enhancing effect of densification increases with the area of the city in which the densification program applies. This is because densification attenuates house price growth over a larger area of the city.

Furthermore, when preferences for open space are weak, all densification packages are welfare-improving in the long run. This is in sharp contrast with the case of strong preferences for open space, in which some of the densification packages are detrimental even in the long run.

The analysis is relevant for contexts outside New Zealand, where preferences for open space may be weaker. It can also provide insights about alternative outcomes in the long run, as it can be argued that a shift away from strong preferences for low density could be possible, even in New Zealand.

copy the linklink copied!6.2. Limitations of the study

6.2.1. Limitations stemming from assumptions about exogenous variables

The conclusions reached in this report depend on concrete hypotheses about the evolution of factors that the model has no predictive power over. These factors include population, income, the energy efficiency of vehicles, the carbon intensity of the electricity generation sector and the prices of fuel and electricity. The assumptions about the intertemporal evolution of these factors were elaborated in Chapter 2. The impact of these assumptions on the key outcomes of the study are examined in the previous section, which provides extensive sensitivity analysis. That analysis expands the validity of several findings to other settings. These may resemble Auckland in some respects, for example in urban morphology, but differ in others. For instance, the first part of this chapter has shed light to the case of a sprawled city of comparable characteristics, such as population and income, where electricity is produced in a more carbon-intensive way. That analysis is therefore relevant for several urban areas in Canada, the United States and Australia.

6.2.2. Limitations stemming from model specification

The external validity of the analysis, i.e. the extent to which the findings of the study can be widely generalizable to contexts outside Auckland, is to some extent limited by the spatial configuration of the model. The latter is elaborated in Chapter 3.

The study used a medium resolution, stylised representation of the spatial layout and the actual transportation networks. This resolution is higher than the usual degree of detail in scientific work, where multiple abstractions are necessary to derive insights about the properties of policies in generic contexts. For instance, the road networks in this study are represented by hundreds of nodes and links. Similarly, urban development is represented by hundreds of residential zones. Therefore, the spatial resolution of the model closely mirrors the actual spatial layout of the city, reflecting in detail its unique geographic characteristics.

Consequently, the recommendations of the study are to some extent specific to Auckland. The exact degree to which the spatial specificities of Auckland drive the reported results is unknown. That degree could be identified in an analysis that would remove all idiosyncratic characteristics of the urban area from the spatial configuration of the model. These include, but are not limited to, the large water bodies and the isthmus area that characterize Auckland. Thus, that analysis requires generic hypotheses about spatial structure. Insofar such an exercise is feasible, it goes beyond the scope of this study. It is therefore left as a topic of future research.

Nevertheless, it the main policy conclusions of the study are generally robust in the context of the specific spatial configuration used in the model. Repeating the modelling exercise with a more generic network and spatial configuration would alter the results of the study from a quantitative viewpoint. For instance, it can be argued that without the water bodies, overall private transport costs in the city would be lower. In turn, that would affect the share of income allocated to transport. It would also affect housing prices, which are affected by accessibility. However, it is unlikely that the findings of the study would change qualitatively, as the analysis keeps the spatial and network configuration fixed across the reference and the various counterfactual scenarios. Therefore, that configuration is not affected by policy changes.

On the other hand, the model abstracts from explicitly modelling a series of local, particularly idiosyncratic characteristics. These features may bear some relevance in the context of Auckland, but that relevance is very limited to other contexts. One of the most important idiosyncratic characteristics is the existing set of regulations regarding volcanic view shafts. These are spatially refined land-use regulations that aim to protect the visibility of volcanoes from various locations. Explicitly accounting for such characteristics would shift the focus away from deriving overall recommendations for optimal land use and transport policies. Instead, such an approach would direct the focus to how these optimal policies should be modified in order to fit the various local idiosyncrasies. The latter adjustment is left as a task to policymakers in New Zealand. The overarching messages of this report should be read in the context of the various local constraints.

Another important abstraction adopted in the modelling exercise is that the location of employment hubs is fixed throughout the entire horizon of the study. This implies that job locations do not respond directly to the policies examined in the study. However, the model still mimics the real-world dual response of households and jobs to changes in accessibility. That is, in MOLES an increase in the pecuniary cost of private vehicle use creates incentives for individuals to relocate closer to their workplace or to pursue a job in an employment hub that lies closer to their residential location. This implies that the density of employment, i.e. the number of workers employed at different locations, is also endogenous.

It can still be argued, however, that the indirect ways in which employment responds to policy changes in MOLES do not capture the full relocation effects these policies may trigger. To the extent that this is valid, the report underestimates the actual welfare benefits of the examined policies. This holds true because firms tend to move closer to location of labour. The policies examined in the study, especially densification policies, create compact urban forms. Therefore, they generate incentives for some of the firms located in peripheral job hubs to move to the inner part of the city, where a larger share of the labour pool resides. Insofar these incentives are not offset by an increase in land rents in central areas, which these firms also have to face, job relocation reinforces the positive effects of household relocation.

Another limitation stems from the choice to avoid modelling the provision of public transport services in an explicit way. The latter choice would substantially increase the computational burden as the behaviour of the various transport providers and authorities would have to be specified and fit into the model. Instead, MOLES uses fixed coefficients to convert passenger kilometres in public transport to rail and bus vehicle kilometres. Some of the examined policies in the report drastically increase passenger kilometres, causing a proportional increase to the kilometres traversed by buses and trains. Therefore, the findings do not account for the substantial part of these additional passenger kilometres in public transport that can be accommodated only through an increase in occupancy rates in buses and trains. A similar argument holds for the way passenger kilometres in public transport are converted to greenhouse gas emissions from public transport vehicles.

An immediate outcome of that is that the model underestimates the potential of the various policies examined in the report to reduce congestion and emissions. The most profound case is the policy package entitled “promote public transport”, which causes the largest change in the private-to-public transport passenger kilometre ratio. The impact of the policy on emissions and congestion is large, but the estimates get even larger by revisiting some of the model’s assumptions. For instance, someone can argue that the frequency of public transport service does not have to increase proportionally to bus and rail passenger kilometres, as the study assumes. Instead, most of the additional demand for public transport services will be absorbed by the current activity of public transport modes. This argument is particularly strong for Auckland, where existing occupancy rates are relatively low, but can apply to other contexts where public transport is far from being the dominant mobility option.

The above underestimation of emission reductions is much smaller in 2050. This is because the emission intensity of public transport vehicles declines steadily with time and is projected to be very low in 2050. Therefore, underestimating the potential of policies to reduce traffic in the long run implies a smaller bias in the estimation of their potential to reduce emissions.

The operational costs of public transport are affected by the policy packages examined in the report. That holds particularly true for the “promote public transport” policy package, which induces a massive change in the modal split in favour of public transport modes. The potential impact of such policies on the existing deficits of public transport operators should also be taken into account by local policymakers.

The counterfactual scenarios in the study did not consider the impact of a large investment in public transport infrastructure. Such an investment could considerably increase the capacity and frequency of service and might prove more cost-effective than the large fare subsidy included in the “promote public transport” package. The study does not explore this possibility due to lack of stated- or revealed-preference data on the way that individuals respond to public transport attributes such as frequency, comfort and reliability.

Finally, the model adopts the standard approach of general equilibrium models to individual preferences, assuming that these are fixed over time. This approach is necessary in order to identify the true effect of policy interventions and price shocks on individual behaviour and economic outcomes. Keeping preferences constant may be perceived as a general limitation of equilibrium models. However, that assumption is essential in order for all of the outcomes of the simulation outcomes to be attributed entirely on the policy shocks that generate them.

copy the linklink copied!6.3. Beyond the study: additional considerations and future extensions

The study exclusively considers the impact of land use and transport policies on exhaust emissions and emissions from the production of the electricity used to charge electric vehicles. Well-to-tank emissions (i.e. emissions from extracting, producing and transporting fuel) are not within the scope of emissions considered. It should be stressed that the policy packages explored in this study do not affect that latter category of emissions.

Unlike the estimates for tank-to-wheel emissions, the expansion of the study to a full-blown lifecycle analysis is a more challenging task. This is because the policies in the report affect the ratio between EVs and ICE vehicles and because the carbon footprint in the production stage differs between these two vehicle types. In particular, several studies find that EVs arrive on the market with a higher carbon footprint than ICE vehicles, as the production of their battery is energy-intensive (Wu et al., 2018[5]; Ellingsen, Singh and Strømman, 2016[6]).

The report focuses exclusively on emissions from mobile sources. These emissions arise from fossil fuel combustion and the electricity used to power EVs, electric buses and rail. However, the policy reforms examined in the report, especially those involving densification policies, alter the structural density of urban development. There is preliminary evidence in the literature that structural density is related to the energy consumption of buildings, as their energy needs for cooling and heating depend, among others, on the way urban development is organised across space. Future extensions of this study may examine the degree to which the policies examined in this report increase or decrease the amount of greenhouse gas emissions that can be attributed to static sources.

The policies examined in study could generate broader benefits that are not explicitly considered in the analysis. For example, a shift towards low-carbon transport improves air quality and reduces noise. These benefits are not included in the welfare calculations. Additionally, the health benefits of a shift towards active modes of travel such as biking or walking are not considered.

The different policies tested eliminate only a part of aggregate greenhouse gas emissions from urban transport. Further reductions require policies that promote research and development (R&D), in order to increase the pace of technological progress in key areas relevant to the transport sector. Computing the social cost and benefit of such R&D policies goes beyond the scope of this report, as such policies are mainly relevant in EV-manufacturing countries.

Other technological developments, such as shared mobility and autonomous vehicles, can have a direct effect on the trajectory of transport sector emissions. In a series of case studies, the International Transport Forum (ITF) assessed the likely impacts of shared mobility and found that a shift to shared autonomous vehicles has the potential to dramatically lower emissions and congestion (ITF, 2015[7]; ITF, 2017[8]). These possible developments are not addressed in the scenarios presented here. Incorporating different forms of mobility (e.g. shared mobility, micromobility) into the analysis would require future extensions of the current study.

The approach taken in this report can be a springboard to explore the impact of other urban policies on the economy and the environment. It can also be a point of departure for similar work that examines different environmental implications of land use and transport policies. In line with these extensions, future work may assess possible synergies or trade-offs between reducing greenhouse gas emissions and tackling air pollution, or focus on the effect of policies aiming to promote more energy efficient housing.

References

[6] Ellingsen, L., B. Singh and A. Strømman (2016), “The size and range effect: lifecycle greenhouse gas emissions of electric vehicles”, Environmental Research Letters, Vol. 11/5, p. 054010, https://doi.org/10.1088/1748-9326/11/5/054010.

[1] IEA (2019), “Emissions per kWh of electricity and heat output”, IEA CO2 Emissions from Fuel Combustion Statistics (database), https://dx.doi.org/10.1787/data-00432-en (accessed on 20 August 2019).

[3] IEA (2018), Energy prices in national currency per unit.

[2] IEA (2018), Energy prices in national currency per unit.

[8] ITF (2017), “Transition to Shared Mobility: How large cities can deliver inclusive transport services”, International Transport Forum Policy Papers, No. 33, OECD Publishing, Paris, https://dx.doi.org/10.1787/b1d47e43-en.

[7] ITF (2015), “Urban Mobility System Upgrade: How shared self-driving cars could change city traffic”, International Transport Forum Policy Papers, No. 6, OECD Publishing, Paris, https://dx.doi.org/10.1787/5jlwvzdk29g5-en.

[4] Statistics New Zealand (2017), Disposable income per person, http://archive.stats.govt.nz/browse_for_stats/snapshots-of-nz/nz-social-indicators/Home/Standard%20of%20living/disp-income-pp.aspx.

[5] Wu, Z. et al. (2018), “Life cycle greenhouse gas emission reduction potential of battery electric vehicle”, Journal of Cleaner Production, Vol. 190, pp. 462-470, https://doi.org/10.1016/j.jclepro.2018.04.036.

Notes

← 1. See Chapter 5 for more details on how the relative advantage of ICE vehicles over other modes of transport gradually declines over time.

← 2. This assumption does not affect the modal split between ICE vehicles and EVs. That is because the price of electricity is not linked to the use of EVs and, therefore, their emissions.

Metadata, Legal and Rights

This document, as well as any data and map included herein, are without prejudice to the status of or sovereignty over any territory, to the delimitation of international frontiers and boundaries and to the name of any territory, city or area. Extracts from publications may be subject to additional disclaimers, which are set out in the complete version of the publication, available at the link provided.

https://doi.org/10.1787/095848a3-en

The use of this work, whether digital or print, is governed by the Terms and Conditions to be found at http://www.oecd.org/termsandconditions.

╳

: Decarbonising Urban Mobility with Land Use and Transport Policies: The Case of Auckland, New Zealand

Continue to open the publication page on OECD iLibrary to find:

All available formats

Buy option for print or PDF copies

All publication details?

Cancel