NTM TN 14 – External cost from air pollutants and greenhouse gases
This Technical Notation aims to make recommendations on which values on the external cost of air pollutants and greenhouse gases should be implemented in NTMcalc, as well as describing the rationale behind the recommendations. The two sources discussed are the European source Handbook on the external cost of transport (van Essen et al. 2019) and the Swedish national guidelines for cost-benefit analysis in the transport sector Analysmetod och samhällekonomiska kalkylvärden för transportsektorn: ASEK 7.0 (Trafikverket 2020). It was asessed for NTM by Ander Bondemark from VTI (The Swedish National Road and Transport Research Institute) during 2021.
This notation is limited emissions from land transport sources. Emissions from air and sea sources are not included due to the differences in the impact of emissions from these sources. An important thing to remember when interpreting the values presented in this notation is that they are a description of reality, not reality itself. The costs presented here are better than ignorance, but they are not perfect descriptions of the costs associated with the emission. The notation begins with recommendations on which values to use and when to use them. Following the recommendations, the differences between van Essen et al. (2019) and Trafikverket (2020) are outlined, first general differences, then air pollution and finally climate gases.
Recommendations when using external costs
- Different values for different geographies
- Since Trafikverket (2020) is widely used when calculating the external costs in Sweden it is recommended to use these costs for trips primarily taking place in Sweden. This will ensure comparability with other Swedish sources.
- Since the health effects are heavily dependent on income and geography three different values are recommended for European transport, western Europe, eastern Europe and southern Europe, table 1 below.
- Since the health effects are dependent on geography and population density, i.e., if the transport is primarily urban or rural, the user must choose to use either urban or rural values dependent on the characteristics of the route.
- Include noise costs from the same sources
- Trafikverket (2020) regard the health costs associated with strokes as a cost arising from air pollution, van Essen et al. (2019) ascribe it to noise pollution. Since the cost of strokes is quite high it is recommended to include noise costs from the same source to make the costs more comparable.
- Do not include well to tank emissions by default
- Including emissions from other sectors i.e. fuel and electricity suppliers risks double counting emissions. However, there are cases when well to tank emissions are helpful, for example when performing an LCA analysis. There could therefore be an option to include these emissions in such assessments.
External costs for implementation into NTMcalc
Table 1 contains the values recommended for implementation in NTMcalc. All values are in 2020 Euro/kg. Table 1 – Recommended values Euro/kg (2020)
|TRAFIKVERKET (2020)||Van Essen et al. (2019)|
|Pollutant||Sweden||Western Europe||Eastern Europe||Southern Europe|
Differences between van Essen and ASEK
There are big differences in the cost presented for air pollutants by Trafikverket (2020) and van Essen et al. (2019). These differences are to a large extent a result of lack of sufficient knowledge on which pollutants give rise to which health effects on account of air pollutants being highly correlated, e.g., where there are NOx there are also PM2.5 pollution. The total costs per vehicle kilometre do, however, not differ very much. The average external cost per vehicle kilometre (vkm) for a regular car in 2020 prices is 0.013 Euro/vkm for Sweden (Trafikverket 2020) and 0.012 Euro/vkm for the EU average (van Essen et al. 2019). Comparing these values is somewhat like comparing apples and oranges since the societal damage costs vary with geography and income. They do however serve as an example of the fact that the differences in the valuation of individual pollutants can appear as a much larger problem than they are in total assessments.
The cost of air pollution is mainly driven by its adverse health effects. The total societal cost of negative health effects is heavily dependent on the number of people exposed to the pollutants. This is an important source of differences in societal costs between countries and are a result of population density and how much to total traffic takes place close to people, i.e., in cities. Additionally, geographical, and meteorological conditions such as natural ventilation linked to coastline, mountains, and wind play an important role. The other major source of differences is that health costs are based on valuation studies and are thus influenced by the income level in different countries.
The values presented by Trafikverket (2020) and van Essen et al. (2019) are presented for a specific year, 2017 and 2016, respectively. These values are adjusted with respect to income and price inflation to reflect the values a certain year. The two reports deal with the increased willingness to pay following an increase in income in different ways. Van Essen et al. (2019) recommends an income elasticity of 0.8 and Trafikverket (2020) an elasticity of 1. This means that when income rises with 10 per cent the valuation of pollutants rises with 8 and 10 per cent respectively. This has very little impact short term but a larger impact long term as the difference will increase exponentially.
There are significant differences in the individual external costs reported by van Essen et al. (2019) and Trafikverket (2020). These differences are not due to method used. Both sources use the damage cost approach to value the effect of air pollution. They also use an effect chain consisting of emission à dispersion à dose-response function à monetary valuation. The emission stage describes where the emissions take place. The dispersion step describes how these emissions disperse in the atmosphere and quantifies peoples/crops/habitats/buildings exposure to the emissions. The dose-response function describes how much damage the pollutant incurs to people/crops/habitats/buildings given the dose. The final stage, monetary valuation, consists of the monetary valuation which is how much individuals are willing to pay to avoid the damage incurred by the dose or the compensation they require if they were exposed to the dose. Table 2 contains a comparison of the external costs for each pollutant included in van Essen et al. (2019) and Trafikverket (2020). All values have been converted to the same year (2020) in (Euro). Trafikverket (2020) assumes/allocate a higher value on particles but a lower value on other pollutants compared to van Essen et al. (2019). Below is a review of differences and similarities in the valuation of pollutants between van Essen et al. (2019) and Trafikverket (2020). Table 2 – Comparison of external costs for Sweden, Euro per kilo 2020
|Pollutant||Van Essen et al. (2019)||Trafikverket (2020)|
Emissions of ammonia (NH3), give rise to a direct societal cost through their negative impact on ecosystems. Trafikverket bases its valuation on Söderqvist et al. (2019). The valuation of Söderqvist et al. (2019) bases its valuation of NH3 on its impact on overfertilization of the Baltic Sea. However, they only include Swedes’ valuation of Swedish emissions. This demarcation is consistent with the system boundaries used by ASEK 7 but is not consistent with a European perspective. According to van Essen et al. (2019) NH3 also reacts with NOx and SO2 to form other particles. Because of the non-linear relationship between NH3 and the formation of these particles, where NH3 has to decrease faster than NOx and SO2 in order to achieve the positive effect of the NOx and SO2 reduction. Van Essen et al. (2019) includes this second order effect in their valuation of these pollutants. Söderqvist et al. (2019) instead argues that since local nitrogen emissions are highly correlated with particle emissions it is hard, if not impossible, to disentangle the health effects of nitrogen emissions from particle emissions. They therefore regard all health effects to solely be a consequence of particle emissions.
The main cost arising from NMVOC, or non-methane volatile organic compounds, are crop losses arising from formation of ground level ozone. NMVOC is not the only pollutant resulting in ground level ozone, NOx also contributes to the formation. Van Essen et al. (2019) attributes some of the damage caused by ozone to NMVOC and some to NOx. Additionally, van Essen et al. (2019) includes damages to materials. Trafikverket (2020) on the other hand, does not have any costs related to NMVOC. In the underlying report (Söderqvist et al. 2019), the authors recognise that both NMVOC and NOx contribute to ground level ozone but they only present costs for NOx. Trafikverket (2020), argue that background levels of NMVOC are so low that there is no marginal cost of additional emissions.
The damages associated with SO2, sulfur dioxide, in van Essen et al. (2019) are crop losses, material and building damages and biodiversity loss through acidification. Trafikverket (2020) present no societal cost associated with SO2 and NMVOC since the background levels are too low, and therefore negligible. However, the underlying report, Söderqvist et al. (2019), presents a value of roughly 0,1 Euro/kg for SO2.
In van Essen et al. (2019), NOx, nitrogen oxides, give rise to a wide array of costs in the form of health, implications, crop losses, material and building damages and biodiversity loss. The main cost is the adverse health effect arising through the formation of ozone. The cost of NO2 emissions is thus heavily dependent on number of individuals exposed which is motivation for the different valuations for rural and urban areas. Trafikverket (2020) only include costs from crop losses and biodiversity losses due to overfertilization of the Baltic Sea. Since the mechanism for overfertilization is the same for NOx as for NH3 the costs are impacted by the same system boundaries.
The primary costs arising from emissions of particles, PM10 and PM2.5, are health costs. However, particles, primarily PM10 also damages buildings. Trafikverket (2020) attributes all building damage to PM10. According to Trafikverket (2020), 19 per cent of the total societal costs due to PM10 are costs attributed to damage to buildings. The primary reason why the external costs of particles differ between van Essen et al. (2019) and Trafikverket (2020) is likely an effect of the differences in how health effects are attributed to different pollutants. The authors of the underlying report to Trafikverket (2020), Söderqvist et al. (2019), argue that since the spatial correlation between particles and other pollutants are high, it is impossible to say which pollutant give rise to which proportion of the health effects. The methodology used in van Essen et al. (2019) differentiates the health effects to specific pollutants. There are also differences in included health effects. These differences are presented in Table 3. Table 3 – Health effects included in van Essen et al. (2019) and Trafikverket (2020)
|Van Essen et al. (2019)||Trafikverket (2020)|
|Cardiac hospital admissions||X||X|
|Net restricted activity days||X||X|
|Minor restricted activity days|
|Working day loss||X||X|
|Respiratory hospital admissions||X|
|Medication use and lower respiratory symptoms because of asthma||X||X|
|New cases of chronic bronchitis and COPD for adults||X||X|
The main difference in terms of included health effects is the cost of strokes which are included by Trafikverket (2020) but not by van Essen et al (2019). In van Essen et al. (2019) the cost of strokes is attributed to noise pollution. Since noise is highly correlated with air pollution, this discrepancy could also be an “accounting effect”. However, because of this, it is important to use air pollution costs and noise costs from the same source, i.e. use noise costs from van Essen et al. (2019) along with air pollution costs from van Essen et al. (2019). The same applies when using values from Trafikverket (2020).
Emissions impacting global warming are measured in CO2e, carbon dioxide equivalents, according to their global warming potential. The primary emission from the transport sector is carbon dioxide arising from burning fossil fuels. Both van Essen et al. (2019) and Trafikverket (2020) use the avoidance cost approach due to the large uncertainties related to the damage costs of climate change. The avoidance cost is the cost associated with avoiding a certain climate change outcome. In the case of carbon emissions this is typically the cost associated with avoiding an excess of x degrees warmer climate. Despite similarities in method the cost diverges. Van Essen et al. (2019) base their valuation on a large number of estimates of the cost of not exceeding 2 degrees Celsius above pre-industrial levels. They present confidence intervals for short and medium run costs as well as long run costs for CO2. In addition, they present their most probable estimate. Long run costs are significantly higher than short run costs since the willingness to pay for emission reductions are considered higher in the future and since emission reductions are more costly when there are fewer sources of emissions left. Since NTMcalc is used to calculate emissions from transport in the present, short run costs are more appropriate. The values are presented in Table 4. Trafikverket (2020) does not differentiate between long run and short run societal costs. Additionally, the societal cost is determined in a different way. The cost by Trafikverket (2020) is based on the highest politically determined price in Sweden, i.e., the shadow price reflecting the highest willingness to pay for CO2 in the transport sector. Since CO2 emissions are not limited to the transport sector nor to Sweden it is more logical to use the external costs presented in van Essen et al. (2019) for international transport. However, this will lead to strange effects when crossing borders. For example, a trip Stockholm-Malmö will have a higher external cost than for example Stockholm-Paris if values from Trafikverket (2020) are used for the domestic trip and values from van Essen et al. (2019) are used for international transport despite the international trip being 3 times as long. On the other hand, it does not make sense to impose the political value in Sweden on emissions abroad. Table 4 – Cost of CO2e Euro 2020
|Van Essen et al. (2019), short run||Van Essen et al. (2019), long run||Trafikverket (2020), short and long run|
Trafikverket (2020). Analysmetod och samhällsekonomiska kalkylvärden för transportsektorn: ASEK 7.0. Available: https://www.trafikverket.se/contentassets/4b1c1005597d47bda386d81dd3444b24/asek-7-hela-rapporten_210129.pdf [2021-10-18] Söderqvist, T., Bennet, C., Katre, Kriit, H., Tidbald, J., Andersson, J., Jansson, S-A., Svensson, M., Wallström, J., Andersson, C., Orru, H., Sommar, J. & Forsberg, B. (2019). Underlag för reviderade ASEK-värden för luftföroreningar: Slutrapport från projektet REVSEK. Borlänge: Trafikverket. Van Essen, H., van Wijngaarden, L., Schroten, A., Sutter, D., Bieler, C., Maffii, S., Brambilla, M., Fiorello, D., Fermi, F., Parolin, R. & El Beyrouty, K. (2019). Handbook on the external costs of transport. Luxembourg: Publications Office of the European Union.  By definition, there is no market for external effect. As such, we do not know the price of them. To solicit the price, researchers and consultants resort to valuation studies. These can either be through hedonic pricing, i.e. implicit valuation of external costs, or various choice experiments in which they can reveal their valuation of the external effect. Both of these methods are sensitive to the wealth/income of the people responding and the external costs, as such, vary with income level between countries.  The carbon cost in sectors included in EU-ETS are dictated by the cost of emission rights.