JUNE 2021 WORLD BANK GROUP SOCIO-ECONOMIC COSTS AND HUMAN IMPACTS OF ROAD ACCIDENTS IN AZERBAIJAN ACKNOWLEDGEMENTS The report was prepared by a World Bank team led by Nijat Valiyev (Senior Infrastructure Specialist) and Wim Wijnen (Consultant). The report benefited from the comments of Antonio Nunez (Senior Transport Specialist), Dipan Bose (Senior Transport Specialist), Sophia Georgieva (Senior Social Development Specialist), Jamie Lee Brown (ETC), and Alan Fuchs (Senior Economist). The team is grateful for the guidance provided by Karla Gonzalez Carvajal (Practice Manager) and Sarah Michael (Country Manager). The authors are also thankful to National Automobile Club of Azerbaijan (AMAK), Sabina Rustamova (Consultant), and other consultants for their inputs and contributions. The World Bank team would like to thank its counterparts in Azerbaijan including Ministry of Economy, Economic Research Institute of the Ministry of Economy, and others for fruitful discussions, consultations, ideas and access to data and information. The findings, interpretations, and conclusions expressed herein are those of the authors and do not necessarily reflect the views of the Board of Executive Directors of the World Bank or the governments they represent. This report, created by the World Bank, is available under the Creative Commons Attribution 3.0 Unported (CCBY3.0) license. 2|P a g e CONTENTS Abbreviations…………………………………………………………………………………………………………………………….. .5 Executive summary……………………………………………………………………………………………………………………...6 1 Introduction………………………………………………………………………………………………………………..………10 2 Methodology for Identification of Socio-economic Costs of Road Accidents………………………. 13 2.1 Socio-economic costs of road accidents……………………………………………………………………………… 13 2.1.1 The socio-economic perspective……………………………………………………………………………………….. 13 2.1.2 Socio-economic cost elements………………………………………………………………………………………….. 14 2.1.3 Valuation methods……………………………………………………………………………………………………………. 16 2.1.4 Estimating the number of road crashes and casualties……………………………………………………… 19 2.1.5 Calculation method per cost element……………………………………………………………………………….. 20 2.1.6 Data requirements……………………………………………………………………………………………………………. 27 2.2 Impact of road accidents on poverty and shared prosperity………………………………………………. 27 2.3 Data collection…………………………………………………………………………………………………………………… 29 3 Assessment of the socio-economic costs of road crashes …………………………………………………..32 3.1 Estimation of the number of road crashes and casualties …………………………………………………..32 3.2 Medical costs……………………………………………………………………………………………………………………… 36 3.3 Production loss…………………………………………………………………………………………………………………..40 3.4 Human costs………………………………………………………………………………………………………………………. 41 3.5 Property damage……………………………………………………………………………………………………………….. 43 3.6 Administrative costs…………………………………………………………………………………………………………… 45 3.7 Other costs……………………………………………………………………………………………………………………..…. 47 3.8 Total costs………………………………………………………………………………………………………………………….. 48 4 Human and Poverty Impacts of Road Accidents…………………………………………………………………. 52 4.1 The poverty and shared prosperity situation in Azerbaijan………………………………………………… 52 4.2 Impact of road accidents on economic wellbeing and poverty…………………………………………… 53 4.3 Medical Treatment Impacts……………………………………………………………………………………………….. 57 4.4 Vehicle damage impacts………………………………………………………………………… ………………………....57 4.5 Impact on quality of life………………………………………………………………………………………………..…...58 4.6 Impacts on vulnerable groups……………………………… …………………………………………………………..61 4.7 Gender aspects…..………………………………………………………………………………………………………………64 3|P a g e 5 Discussion and policy implications……………………………………………………………………………………… 67 5.1 Implications of the results on socio-economic costs for policy making…………………………….... 67 5.2 Results of the study in international perspective……………………………………………………………….. 69 5.3 Data availability and data collection…………………………………………………………………………………… 70 5.4 Key policy recommendations……………………………………………………………………………………………… 71 Annex 1: International evidence on the socio-economic and poverty impact of road crash costs……………………………………………………………………………………… ………………………………………74 Annex 2: Review of the government methodology for assessing road crash costs…..………………….78 Annex 3: Socio-economic cost items and valuation methods………………………………………………………81 Annex 4: Data requirements……………………………………………………………………………………………………….84 4|P a g e Abbreviations ADB Asian Development Bank AMAK National Automobile Club of Azerbaijan AZN Azerbaijani New Manat B40 Bottom 40% of population in terms of income distribution CASCO Casualty and Collision Insurance EaP Eastern Partnership EU European Union GDP Gross Domestic Product GRSP Global Road Safety Partnership HBS Household Budget Survey HC Human Capital HH Household HIC High Income Country ICU Intensive Care Unit IDP Internally Displaced Persons iRAP International Road Assessment Programme ITMC Intelligent Transport Management Centre LMIC Low and Middle Income Country MLSP Ministry of Labor and Social Protection MOH Ministry of Health NMT Non-Motorized Transport PDO Property Damage Only PPP Purchasing Power Parity RC Restitution Cost RP Revealed Preference SAAAR State Agency for Azerbaijan Automobile Roads SCD Systematic Country Diagnostic SRP State Road Police SP Stated Preference U60 Upper 60% of population in terms of income distribution UK United Kingdom UN United Nations USD United States Dollars VSL Value of a Statistical Life WTP Willingness to Pay WHO World Health Organization 5|P a g e Executive summary Summary of key study findings 1. Road accidents represent a growing worldwide human and socio-economic problem, in particular in low and middle-income countries. 821 people died due to road crashes in Azerbaijan in 2019, according to the official statistics, and a much larger number of people were seriously injured. Despite investments in road safety, the number of fatalities is still high and even increased in 2019. Moreover, the road fatality rate in Azerbaijan is much higher than in most of the European countries and about twice as high as the EU average. 2. Unlike many other middle-income countries, Azerbaijan currently lacks information on the estimated cost of road accidents from economic and social perspective and does not have the established methodology for undertaking such task. This represents a large knowledge and capacity gap for handling the road safety problems in the country. The report aims at implementation of the first professional level study in Azerbaijan for identification of socio-economic costs of road accidents and building awareness about magnitude of economic and social damage resulting from road accidents. 3. In 2018, the Government of Azerbaijan has launched the State Road Safety Program for 2019- 2023, which is aimed at substantially improving the road safety situation. Among other actions, the Program recognizes the need to assess the socio-economic costs of road crashes. The lack of information on the socio-economic costs of road crashes represents an important knowledge and capacity gap for handling road safety problems in Azerbaijan. The study aims at raising awareness of the socio-economic burden of road crashes and highlighting the benefits of road safety improvements. Information on the socio-economic costs of road crashes is also expected to become an important input for setting priorities in government investments and gaining insight into the cost- effectiveness of road safety measures. 4. Along with the socio-economic costs of road accidents, the study also assessed the impact of road crashes on the social and poverty situation at the household level. Information on these impacts is key to raise awareness of the importance of improving road safety from the perspective of poverty reduction and reaching shared prosperity. This also highlights the role of road safety in meeting the global development targets of reducing extreme poverty and fostering income growth of the bottom 40 percent (B40) in terms of income distribution. 5. Based on international standards and best practices, the study developed a methodology to assess the socio-economic costs of road crashes in Azerbaijan. Five main cost components have been distinguished: medical costs, production loss, human costs, property damage and administrative costs. The costs are estimated using three economic valuation approaches. The restitution costs are applied to estimate medical costs, property damage and administrative costs, while the human capital method is used to estimate production loss. The estimation of human costs, which are the intangible costs of loss of life years and quality of life, is based on the willingness to pay approach. 6. A large variety of data needed for the cost calculations was collected through four data collection methods. First, road safety and other relevant stakeholders were consulted and interviewed. These consultations concentrated on general data, such as data on road safety, macroeconomy, demography and insurance, as well as on data and expert opinions on specific cost elements. Second, a survey among about 200 road casualties or their relatives was conducted, aimed at collecting detailed information on injuries and crashes, including medical treatment, consequences of injuries and damage to vehicles. Third, in-depth interviews with several road 6|P a g e casualties or their relatives were conducted. These interviews were aimed at obtaining more detailed and personal information on the impact of a road crash on people’s daily life, livelihood, financial situation and quality of life. Finally, value transfer was applied with respect to data which is not available in Azerbaijan and could not be collected through other data collection methods. This means that data from other countries available in the literature was collected and adapted to the situation in Azerbaijan. This method was used in particular to estimate the human costs of road crashes and for estimation of the number of non-fatal casualties and crashes. 7. The socio-economic cost of road crashes in Azerbaijan has been conservatively estimated by the study at 1.6 billion AZN in 2019, which is the equivalent of 2.0% of GDP. Human costs account for a major proportion of the total costs (66%). The other main cost item is damage to vehicles, which accounts for 20% of the total costs. Other costs items are estimated to be relatively small. These costs include production loss due to the fact that road casualties are unable to work, temporarily or lifelong (9%), costs of medical treatment of road casualties (2%) and administrative costs (3%). About half of the costs (51%) are related to injuries, in particular serious injuries (45%). Fatalities and property damage only crashes account for 33% and 16% the costs respectively. 8. Sensitivity analysis tested the influence of several choices of input parameters, in particular with respect to the number of serious and slight injuries and the human costs and showed a cost range of 1.6-4.2% of GDP. The results are consistent with international evidence on the socio- economic cost of road crashes, which are mostly found to be in a similar range. A high proportion of human costs, typically around 50% or more, is in line with the findings in other countries as well. 9. The study assessed the impact of road accidents on poverty and shared prosperity and other human impacts using the survey among a representative number of road casualties as well as the in-depth interviews with casualties. Two key direct impacts of crashes include challenges of medical treatment and negative impacts on the ability to carry out daily activities. Half of the survey respondents were injured, and half of the injured respondents were admitted to hospital (in-patient). A substantial proportion of the injured casualties needed non-hospital care, in particular home care (81%), treatment by a general practitioner (57%) and physiotherapy (37%). A large majority of the respondents reported having difficulties with daily activities, in particular jobs, household work, education, sports, and other leisure activities. 10. The casualty survey showed that road accidents have important financial impacts on the casualties and their relatives. More than half of the survey respondents reported problems with respect to carrying out their job and 10% of them reported lifelong inability to work. About a third (38%) of the respondents reported an income loss after the accident, including 15% substantial income loss. These impacts are stronger for the bottom 40% of the population in terms of income distribution (B40), which means that road accidents have a negative impact on shared prosperity. About half (47%) of the B40 households suffer from income loss, which is much more than for the upper 60% population in terms of income distribution (U60) households (28%). The negative impact on shared prosperity is also reflected in income loss: B40 households lost 30% of their income, versus a 12% income loss for U60 households. 26% of the respondents in the B40 group were in the U60 group before the crash and fell into the B40 group due to the loss of income, as a likely consequence of the road crash. 11. The study found that the injuries and the loss of lives have a strong impact on the quality of life of the casualties and their relatives. Almost half of the respondents (44%) reported that their quality of life worsened after the accident, including 12% who reported a large quality of life loss. Substantially more people from B40 households face quality of life loss (57%) as compared to U60 households (31%). The proportion of B40 respondents who assessed their quality of life as satisfactory or better decreased from 92% to 67%, versus from 98% to 88% for U60 respondents. 7|P a g e Consequently, the impact of accidents on quality of life is felt stronger by B40 households. 12. According to the study findings, the majority of reported road fatalities and injuries (74%) in Azerbaijan are men, similar to the experience of most of other countries. Meanwhile, females spend more time on providing home care for the injured persons in the household. Also, since most road casualties are men, this burden is likely to be borne mostly by women. The need to provide home care and to earn income sometimes put a large pressure on females and limits their capabilities to conduct other duties such as taking care of the children, household activities and other activities. 13. The in-depth interviews with selected group of road casualties provided a personal and more detailed picture of the impact of road accidents on households. Interviews illustrated the consequences of income loss for the households due to inability to work. Some respondents reported loss of ability to afford the goods and services they were used to buy before the crash, or even to fulfill their basic needs. Some of them became financially and physically dependent on relatives and friends. Some casualties suffered from mental problems in addition to physical limitations. Policy implications 14. The findings of the study are recommended to be used by the stakeholders in Azerbaijan to raise awareness of road safety as an important human and socio-economic issue and justification of road safety investments. By launching the State Road Safety Program 2019-2023, the Government of Azerbaijan has committed to substantially improving the road safety situation. The study findings indicate that improving road safety in Azerbaijan and preventing road casualties can yield high socio- economic benefits (cost savings). An annual cost reduction of roughly 500 million AZN can be achieved if the targets established the State Road Safety Program are met. 15. Social and poverty impacts should be considered as important dimensions of the road safety policy making. Road safety investments could specifically focus on implementing measures aimed at improving the safety of the poor, including measures targeting vulnerable road users. In addition, road safety policies could concentrate more on areas where the concentration of poor people is higher, such as certain rural areas, and where the poor are at more risk. Identification and designing of such policy measures are recommended for further research. 16. The findings of the study on poverty impacts are recommended to be used in social policies. Financial impacts were found to be stronger for B40 households and access to compensations to be more limited. Social policies could concentrate on providing income loss compensation for both the casualties and the relatives who provide care for them, or compensation for medical expenditures which were found to be more financially challenging for B40 households. In addition, improving access to insurance could help to reduce the financial impacts on the B40 households. 17. The results of the study are proposed to be used in cost-benefit analysis of road safety investments. To improve road safety, efficiency in allocation of the available budgets is essential. In this regard, regular cost-benefit analysis of road safety investments can help policy makers in identification of the most cost-effective road safety measures and programs. Implementation of such analysis in some of the countries have shown that benefits of such investments often outweigh the costs. This is likely to be the case for Azerbaijan as well, given the large potential costs savings (benefits). Cost-benefit analysis for implementation of road safety measures in Azerbaijan, in particular those specified in the State Program on Road Safety, will help to justify road safety investments and ensure efficient use of resources. 18. The socio-economic cost of road crashes is recommended to be used as a high-level outcome 8|P a g e indicator for road safety performance. This implies that cost calculations should be performed on a regular basis, for example along with the publication of the yearly road safety statistics using yearly updates of the costs based on a few several key indicators. In that way, the indicator will serve to monitor road safety performance in terms of socio-economic cost savings. It is also suggested to use socio-economic costs as an indicator to monitor the progress made through the implementation of the State Program on Road Safety. 19. Poverty reduction is proposed to be used as an outcome indicator of road safety policy making. This indicator can be used to measure the poverty reduction achieved by the implementation of the State Program on Road Safety. For that purpose, an indicator such as (the reduction in) the number of people below a poverty line due to road injuries could be developed. Recommendations for follow-up research 20. Addressing the issue of data availability and quality is an important pre-condition for successful future studies. Major efforts were made under this study to collect the large variety of data which are needed to calculate the socio-economic costs of road crashes. Still, all data gaps and limitations could not be overcome. This particularly concerns statistics on road crashes and casualties, public health, public administration, and some others. To streamline the data collection process for the future exercises, which are expected to become regular, the required data and statistics (provided in this report) need to be routinely produced. In the case of crash data, underreporting is a common issue with respect to these statistics. On the other hand, the crash data collection system needs to be improved and modernized for completeness of recording details, such as on injury severity. Linking police data with hospital data has proved to be a successful approach in some other countries to cope with underreporting so is recommended to be tested. 21. Full update of calculations on a regular basis, e.g. once every 5 years, is recommended. In the meantime, methodological improvements and additional data collection need to be done. These improvements could concentrate on collecting country-specific data, which were not available for the current calculations. This concerns in particular the human costs, which were estimated on the basis of international evidence. Conducting a country-specific study on human costs is recommended for future research, since this is a major cost item. Also adding some missing cost items is recommended, such as congestion costs and environmental damage. 22. Assigning a lead institution is important for maintaining the methodology, data collection and knowledge transfer to other stakeholders. In addition, the efforts of this institution could concentrate on embedding research results on socio-economic and poverty impact of road crashes in the road safety policy making. 9|P a g e 1. Introduction 1. Road accidents represent a growing worldwide problem. The World Health Organization (WHO) estimates that road crashes result in about 1.35 million deaths and 50 million injuries every year. Road injuries are the eighth leading causes of death for all age groups and they are the leading cause of death for children and young adults aged 5–29 years1. An estimated 93% of these deaths take place in low and middle-income countries, while these countries only have 60% of the registered motor vehicles. The road fatality rate (number of road deaths per 100,000 inhabitants) is significantly higher in low and middle-income countries than in high-income countries. 2. According to the official statistics, 696 people died and 1,410 were seriously injured in road crashes in Azerbaijan in 2020. Since 2012, the number of road fatalities in the country had a trend for reducing. However, this decrease flattened since 2016 and in 2019 the number of fatalities (821) was 14% higher than in 2018 (Figure 1). Given that 2020 statistics is not representative as it was seriously impacted by COVİD-19 restrictions and lockdowns, the study is using 2019 data for the analysis purposes. The number of (mostly seriously) injured people was 1,702 in 2019, based on the official data. However, the actual number of injuries is probably much higher due to underreporting and deficiencies of data collection system which is a common phenomenon, particularly in low and middle-income countries. Figure 1: Annual number of reported road fatalities in Azerbaijan2 3. Similar to most of other post-Soviet countries, the road fatality rate in Azerbaijan is higher than in most European countries and about twice as high as the EU average (Figure 2). In that respect, the country still has a long way to go, particularly compared to advanced European countries, although Azerbaijan has made progress in improving the road safety situation over recent years in terms of enforcement and infrastructure. For example, the length of divided roads has increased from 115 km in 2003 to 890 km in 2019, by upgrading two-lane roads into four-lane (divided) roads. 48 grade separated pedestrian crossings were built since 2003 and the ambition is to build 53 more.3Technical equipment of traffic police has also significantly improved over the recent years. Still, despite the stabilizing number of road fatalities in recent years, the increase in the number of fatalities in 2019 is worrying and indicate on further actions needed to improve the road 1 WHO (2018). Global Status Report on Road Safety 2018. World Health Organization, Geneva. 2 State Statistics Committee of Azerbaijan 3 Source: State Agency for Azerbaijan Automobile Roads (SAAAR). 10 | P a g e safety situation in Azerbaijan. Figure 2: Road Crash Fatality Rates in EU and Azerbaijan4 4. The launch of the State Road Safety Program for 2019 – 2023 is an important step towards improving road safety in Azerbaijan. By adopting the program in 2018 the Government of Azerbaijan committed itself to substantially improving road safety.5 The strategic program aims to improve Azerbaijan’s road safety performance and to reduce the number of road crashes and casualties. It is an ambitious program which includes targets of reducing the number of fatalities, injuries (severe or slight) and crashes by 30% in 2023 as compared to the period 2013-2017. The program presents a broad package of measures aimed at improving road safety, including road safety management, infrastructure, traffic management, vehicle safety, road user behavior and post-crash measures. Also, the program aims to establish an effective financial system for road safety investments as well as a mobility strategy to ensure efficient, safe, and convenient mobility for all road users. The Program also includes an action plan which specifies actions and measures to be implemented, time frames and key responsible stakeholders. 5. The State Road Safety Program explicitly states that in addition to reducing the number of accidents and casualties, decreasing the socio-economic costs that result from road crashes is one of its aims. It is recognized that, road crashes impose huge costs on society, including impacts on quality of life and emotional impacts, as shown by numerous studies on the socio-economic costs of road crashes elsewhere in the world. Several types of costs result from road crashes, such as medical costs, loss of productive capacity, damage to property and administrative costs. The socio-economic costs of road crashes have been estimated to be at least 2-3% of GDP in many low and middle income countries and much higher (up to 9% of GDP) in some other countries (See Appendix 1 for a brief overview). These figures indicate that road crashes have a very high socio-economic burden. 6. Given the high road fatality rate as compared to most European countries, the economic costs of road crashes are expected to be very considerable in Azerbaijan. Quantifying these socio- economic costs helps to raise awareness of the socio-economic burden of road crashes and to justify investments in road safety. Information about the costs of road crashes is essential to highlight the benefits (in terms of cost savings) that result from allocating (government) budget to road safety investments. In that respect, information on the socio-economic costs of road crashes is helpful for 4 Based on data from “Annual Road Safety Performance Index (PIN) Report, ETSC– National Statistics and State Statistics Committee of Azerbaijan 5 State Program of Azerbaijan Republic on Road Safety for 2019-2023 11 | P a g e setting priorities in government investments and gaining insight into the cost-effectiveness of road safety measures. It is an essential input for cost-benefit calculations of road safety investments and other investments with impacts on road safety. Drawing on information on the costs of road crashes, several cost-benefit studies have shown that the benefits of road safety programs often outweigh the costs required for their implementation6. 7. The lack of information on the socio-economic costs of road crashes is an important knowledge and capacity gap for handling road safety problems in Azerbaijan. Although there is general awareness of the government and general public about the socio-economic damage resulting from road accidents, supplementing this knowledge by country-specific information and analysis is expected to have a significant positive effect on policies and resources allocated to road safety. This will help to improve the road safety situation in the country in a medium to long term perspective. 8. Road crashes are also known to have important impacts on the social and poverty situation at the household level. For example, 68% of road fatalities and 70% of the injuries (reported numbers) in Azerbaijan in 2019 were people in the age category of 25-64 years old (source: SRP), i.e. the people in the active labor age. In addition to a direct economic impact felt by the road casualties and their households, road accidents have significant non-financial impacts on quality of life, such as impacts of pain, grief and sorrow. Several studies show that road crashes often have important negative economic and social consequences on households, as a result of the death of the main income earner, the loss of a job and/or income due to injuries or expenses on medical treatment. Information on these impacts is key to raise awareness of the importance of improving road safety from the perspective of poverty reduction and reaching shared prosperity. This also highlights the role of road safety in meeting the global development targets of reducing extreme poverty to less than 3 percent by 2030 and fostering income growth of the bottom 40 percent (B40) of the population in each country. 9. The aim of the study is to assess the national socio-economic costs and the impact of road accidents on poverty, shared prosperity, and vulnerable groups. Concerning the socio-economic costs, all relevant cost items are included and quantified to the extent possible, following international standards and best practices. The poverty and shared prosperity analysis concentrates on the financial impacts and impacts on quality of life, and in particular on whether B40 households and other vulnerable groups are disproportionately affected by road crashes. 10. More specifically, the study aims at the following: - Development of a methodology for identification and quantification of socio-economic costs of road crashes in Azerbaijan; - Estimating the socio-economic costs of road accidents in Azerbaijan; - Identification of key human/moral costs and impacts of road accidents in Azerbaijan; - Assessing the impact of road crashes on poverty, shared prosperity and vulnerable groups. 11. The report is structured as follows. Chapter 2 presents the methodology for the assessment of the socio-economic costs and poverty impacts of road crashes in Azerbaijan, as well as the data collection methods, including a survey among road casualties. Chapter 3 presents the assessment of the socio-economic costs of road crashes and Chapter 4 - of the human and poverty impacts. Chapter 5 discusses policy implications of study results for policy making and presents key recommendations. 6 For example: Daniels, S. et al. (2019). A systematic cost-benefit analysis of 29 road safety measures. Accident Analysis and Prevention, 133, 105292. See also Section 6.4. 12 | P a g e 2. Methodology for Identification of Socio-economic Costs and Poverty Impacts of Road Accidents 12. This chapter presents the methodology for the identification and quantification of the socio- economic costs and impacts of road crashes on poverty and shared prosperity. These include general principles of the cost assessment as well as classification of cost items and the methodology used for estimating each of them. Further on, the methods applied for estimating the number of road casualties and crashes, which are used in the cost calculations, will be discussed. Finally, the methodology for the assessment of the impacts of road crashes on poverty and shared prosperity will be presented as well as the data collection methods, which are used both for the socio-economic cost and poverty impact assessments. 13. The methodology used for the estimation of the socio-economic costs of road crashes is based on international best practices and guidelines. Studies on the costs of road crashes have been conducted in many countries and a number of reviews of the methodologies and findings have been published (see Appendix 1 for an overview). In addition, several guidelines are available which provide guidance on the cost items to be included and the economic valuation methods.7 These best practices and guidelines follow the general principles and theories of socio-economic assessment of welfare gains and losses (in particular, welfare economic theory and the related theories on cost- benefit analysis).8 Adopting these principles in the current study on road crash costs in Azerbaijan ensures that the methodology and results are consistent with the international standards.9 2.1 Socio-Economic Costs of Road Accidents 2.1.1 The socio-economic perspective 14. Road accident costs can be studied from different perspectives, such as the financial perspective of stakeholders (e.g. government expenditures), a macro-economic perspective (e.g. impact on national income or GDP) or from a broader welfare perspective (e.g. including intangible costs). Road accident cost studies, however, usually adopt a broad socio-economic perspective. This means that the costs are determined for the whole society, irrespective of the question who pays for the costs. The socio-economic cost perspective is rooted in economic welfare theory and a main reason for choosing this perspective is the fact that social cost-benefit analysis is also based on this theory.10 Consistency with this theory ensures that the crash cost estimates can be used in cost- benefit analysis of road safety projects and broader transport projects. 15. One of the fundamentals of economic theory is that welfare is determined by the ‘utility’ that 7 Wijnen, W., Weijermars, W., Vanden Berghe, W., Schoeters, A., Bauer, R., Carnis, L., Elvik, R., Theofilatos, A., Perez, C. and Martensen, H. (2017). Crash cost estimates for European countries, D3.2 of the H2020 project SafetyCube. World Bank (2005). Valuation of accident reduction. Transport Note No. TRN-16. World Bank, Washington. BRS&TRL (2003). Guidelines for estimating the costs of road crashes in developing countries. Babtie Ross Silcock & Transport Research Laboratory. Alfaro, J. L., Chapuis, M., Fabre, F. (Eds.) (1994). Socio-economic cost of road accidents: final report of action COST 313. Commission of the European Community, Brussels. 8 See for example Boardman, A.E., Greenberg, D.H., Vining, A.R. & Weimer, D.L. (2011). Cost-benefit analysis. Concepts and practice. Fourth edition. Prentice Hall, Upper Saddle River, New Jersey. 9 In 2019, the Government of Azerbaijan has developed draft guidelines for identification of socio-economic and environmental costs of road accidents. The methodology applied in this study includes a broader range of cost items than the methodology developed by the government. Also, there are some methodological differences. See Annex 2 for more details. 10 Boardman, A.E., Greenberg, D.H., Vining, A.R. & Weimer, D.L. (2011). Cost-benefit analysis. Concepts and practice. Fourth edition. Prentice Hall, Upper Saddle River, New Jersey. 13 | P a g e people derive from consuming goods and services, which may or may not be traded on a market, and also from intangible factors that affect quality of life. Consequently, intangible impacts of road crashes, such as loss of quality of life, and other losses which are not reflected by market transactions (e.g. loss of ability to do household work) are usually included in road crash cost assessments. 16. The socio-economic perspective implies that costs representing only a transfer of money between stakeholders are not regarded as socio-economic costs. An example is a disability benefit that a road casualty receives from the government. This is a cost from the financial perspective of the government, but a benefit for the recipient. On the level of the society there is no net benefit, however. Nevertheless, there are socio-economic costs related to disability in terms of loss of productivity and human costs, which should be included in the cost assessment. However, the fact that a casualty receives a benefit from the government does not affect these socio-economic costs and, although costs of road crashes are often expressed as a percentage of GDP, these percentages do not necessarily represent a loss of GDP. 17. The fact that certain costs may not represent a loss of GDP is a consequence of adopting the broad socio-economic (or welfare) concept. For example, resources that are used to repair cars, such as labor and new car parts, are treated as a cost from this perspective. The reasoning is that these resources cannot be used to produce other goods or services (‘opportunity costs’).11 However, repairing cars contributes positively to GDP, as this is production delivered by garages. This illustrates that some impacts can be regarded as economic gains from a macro-economic (GDP) perspective, while they are losses from a broader welfare perspective. In addition, intangible costs such as loss of quality of life are often included in road crash cost studies. This is another reason why the socio- economic costs are not the same as impact on GDP. The World Bank (2018) studied the impact of reducing road casualties on GDP, which was mainly focused to the impact of the loss of human capital on the economy.12 This is one of the very few examples of a road crash cost study from a different perspective than the broad socio-economic perspective. 2.1.2 Socio-economic cost elements 18. Road accidents result in several types of socio-economic costs. Several international guidelines on road crash cost assessments13 and international reviews of crash studies in individual countries14 give an overview of these cost elements. Based on this literature, the socio-economic costs of road crashes can be classified into the following six cost components:15 11See for example Boardman, A.E., Greenberg, D.H., Vining, A.R. & Weimer, D.L. (2011). Cost-benefit analysis. Concepts and practice. Fourth edition. Prentice Hall, Upper Saddle River, New Jersey. 12 World Bank (2018). High toll of traffic injuries: Unacceptable and Preventable. The Macro-Economic And Welfare Benefits of Reducing Road Traffic Injuries in Low-& Middle-Income Countries. World Bank, Washington. 13 For example: World Bank (2005). Valuation of accident reduction. Transport Note No. TRN-16. World Bank, Washington. ADB (2003). Road Safety Guidelines for the Asian and Pacific Region. Asian Development Bank, Manila. Alfaro, J. L., Chapuis, M., Fabre, F. (Eds.) (1994). Socio-economic cost of road accidents: final report of action COST 313. Commission of the European Community, Brussels. 14For example: Wijnen, W., Weijermars, W., Vanden Berghe, W., Schoeters, A., Bauer, R., Carnis, L., Elvik, R., Martensen, H. (2019). An analysis of official road crash cost estimates in European countries. Safety Science 113, 318–327. Wijnen, W. & Stipdonk, H. (2016). Social costs of road crashes: an international analysis. Accident Analysis and Prevention, 94, 97–106. Trawén, A., Maraste, P. & Persson, U. (2002). International comparison of costs of a fatal casualty of road accidents in 1990 and 1999. Accident Analysis and Prevention, vol. 34, nr. 3, p. 323-332. 15 Wijnen, W., Weijermars, W., Vanden Berghe, W., Schoeters, A., Bauer, R., Carnis, L., Elvik, R., Martensen, H. (2019). An analysis of official road crash cost estimates in European countries. Safety Science 113, 318 –327. Kasnatscheew, A., Heinl, F., Schoenebeck, S., Lerner, M., & Hosta, P. (2016). Review of European Accident Cost Calculation Methods – With Regard to Vulnerable Road Users. Deliverable 5.1 of the Horizon 2020 InDeV project. 14 | P a g e - Medical costs, such as the costs of medical treatment, rehabilitation following injuries resulting from road crashes and home care provided by relatives; - Production loss: loss of human productive capacities and loss of consumption; - Human costs: the intangible costs of quality of life and life years; - Property damage, including damage to vehicles, roads, roadside objects and freight; - Administrative costs: costs of insurance, police, fire service and legal costs; - Other cost, such as funeral costs, congestion costs and costs of vehicle unavailability. 19. A distinction can be made between injury-related costs and crash-related costs, as summarized by Figure 3. Injury-related costs are estimated on a per-casualty basis (e.g. cost of medical treatment per casualty), while crash-related are determined on a per-crash basis (e.g. police costs per crash). This distinction is needed to be able to calculate the (total) cost per casualty and cost per crash, using information on the number of casualties per crash. Figure 3: Classification of the socio-economic costs of road crashes16 20. Each main cost component includes several more detailed cost items. Wijnen et al. (2017) made an overview of the cost items within each cost component and they distinguished between main, minor, and other cost items. Main cost items are known to be relatively large; they are recommended by international guidelines and they are usually included in national costs studies. Minor and other costs items are not always included in cost studies or recommended by guidelines. Minor costs items are relatively small and other cost items might be sizeable, although the evidence about most of these cost items is limited. 21. An overview of each cost item within the main cost components is provided based on the overview made in the SafetyCube project17. This overview covers all cost items that are expected to be included in this study on the costs in Azerbaijan, such as several medical cost items, production loss of fatalities and serious injuries (including disabled) and damage to property (vehicles and infrastructure). The only exception is environmental costs. This is not a common road crash cost item but will be taken into account in this study. On the other hand, the SafetyCube overview incudes some items which are not relevant for Azerbaijan (e.g. administrative costs of health insurance) or which are very low or negligible. Section 0Error! Reference source not found. discusses each cost item in more detail. 16 Source: Wijnen et al. (2017) 17 Wijnen, W., Weijermars, W., Vanden Berghe, W., Schoeters, A., Bauer, R., Carnis, L., Elvik, R., Theofilatos, A., Perez, C. and Martensen, H. (2017). Crash cost estimates for European countries, D3.2 of the H2020 project SafetyCube. 15 | P a g e 22. The costs of activities aimed at preventing road crashes, such as infrastructural measures, enforcement, and education, are not included in the classification of cost items. This follows the logic of cost-benefit analysis: the costs of preventive measures enter at the side of the costs in a cost- benefit analysis, whereas the prevented costs of road crashes enter at the benefit side. Consequently, the road crash cost studies are aimed at only the costs that result from road crashes. In a cost-benefit analysis, the reduction of these costs is weighted against the investments (costs of countermeasures) needed to achieve this reduction. 23. The indirect costs related to lack of safety and behavioral changes are also not included. People may change their travel mode, route, destinations or may choose not to travel at all because of road safety risks, which may prevent them from using some services or carrying out activities. Following the international standards, these costs are not included because they are not a direct consequence of road crashes. However, they are relevant costs and could be considered for inclusion in future updates of guidelines on road crash costs assessments.18 2.1.3 Valuation methods 24. Several methods for the monetary valuation of road crash costs are discussed and applied in the literature. International guidelines generally recommend three types of valuation19 - restitution costs, human capital, and willingness to pay. 25. Restitution costs (RC) approach: in this approach the direct costs resulting from a crash are estimated, such as the costs of medical treatment and vehicle repair. In general, market prices are used to estimate these costs, for example the price of medical treatment and vehicle repair. These prices are assumed to reflect the costs of resources that are needed to restore road casualties and their relatives and friends to the situation which would exist if they had not been involved in a road crash. 26. Human capital (HC) approach: in this approach the costs of the loss of productive capacities of road casualties are measured. The calculation is based on (working) time lost due to death, disability or injury, and monetary valuation of this time. Common indicators for the valuation of production per person are gross national/domestic product per capita, income and wage. The HC approach can also be applied to non-market production, such as household work. Also, the costs of home care provided by relatives are estimated using the human capital approach. In those cases proxy prices are used, e.g. based on wages or costs of hiring household personnel. 27. Willingness to pay (WTP) approach: in this approach costs are estimated on the basis of the amount individuals are willing to pay for a risk reduction. There are in general two approaches to determine the WTP, either using a ‘stated preference’ (SP) method or a ‘revealed preference’ method (RP) method20. RP methods value risk reductions on the basis of actual behavior, for example 18 See Wee, B. v., Hagenzieker, M. & Wijnen, W., 2014. Which indicators to include in the ex ante evaluations of safety effects of policy options? Gaps in evaluations and a discussion based on a ethical perspective. Transport Policy, 31, 19-26. 19 Wijnen, W., Weijermars, W., Vanden Berghe, W., Schoeters, A., Bauer, R., Carnis, L., Elvik, R., Theofilatos, A., Perez, C. and Martensen, H. (2017). Crash cost estimates for European countries, D3.2 of the H2020 project SafetyCube. DaCoTA (2012). Cost-benefit analysis. Deliverable 4.8d of the EC FP7 project DaCoTA. Alfaro, J. L., Chapuis, M., Fabre, F. (Eds.) (1994). Socio-economic cost of road accidents: final report of action COST 313. Commission of the European Community, Brussels. 20 See Blaeij, A.T. de (2003). The value of a statistical life in road safety; Stated preference methodologies and empirical estimates for the Netherlands. Tinbergen Institute Research Series, Vrije Universiteit, Amsterdam. Bahamonde-Birke, F.J., 16 | P a g e purchasing behavior regarding safety provisions (e.g. airbags), while SP methods use questionnaires in which people, directly or indirectly, are asked how much they are willing to pay for more safety. The WTP for a fatal crash risk reduction can be used to determine the Value of a Statistical Life (VSL). This value does not refer to individual lives, but statistical lives saved, because the VSL is based on the willingness to pay for reducing the probability of dying in a road crash. Valuation of individual lives would never ethically acceptable nor possible, because people would most probably pay everything, they have to avoid their death. 28. Each of the above three methods is aimed at the monetary valuation of specific cost components. In general, the RC approach is the most appropriate method to calculate medical costs, property damage and administrative cost. These cost components have in common that they are related to usage of resources for restoring losses as much as possible, such as health losses and vehicle damage. Moreover, market prices are mostly available for these costs. The HC approach is typically aimed at calculating production loss. The WTP approach is aimed at calculating human costs, since there is no market price for such intangible losses. The WTP approach provides the opportunity to express such costs also in monetary terms. It is a common method that is also broadly applied to determine monetary valuations of environmental impacts. 29. The HC and WTP methods are sometimes regarded as substitutes. However, these two approaches are aimed at different cost components (production loss and human costs respectively), which means that they are complementary.21 Consequently, both methods should be applied in a comprehensive road crash costs study. However, the WTP method is complex and not undebated in the literature. One of the issues discussed in the literature is, for example, whether people are capable of understanding small (crash) risks and stating their willingness to pay for reducing such small risks.22 Despite such complexities, it is recognized that the WTP method is the theoretically most sound methods for the monetary valuation of human costs, both in general23 and with respect to road crashes24. Therefore, this method is widely applied in road crash cost studies in Europe25 and in high income countries in other parts of the world26. In low and middle-income countries, the WTP method is less common, although there are several examples of successful application, e.g. in India27 Kunert, U. & Link, H. (2015). The Value of a Statistical Life in a Road Safety Context — A Review of the Current Literature. Transport Reviews, 35 (4), pp. 488-511. 21See for example Wijnen, W., Wesemann, P. & de Blaeij (2009). Valuation of road safety effects in cost-benefit analysis. Evaluation and Program Planning, 32, pp. 326-331. 22See for example Hauer, E. (2010). Computing what the public wants: Some issues in road safety cost –benefit analysis. Accident Analysis and Prevention, 43, pp. 151–164. 23 Freeman, A.M., Herriges, J.A., & Kling, C.L. (2014). The measurement of environmental and resource values. Theory and methods. Third edition. Resources for the future, New York. Boardman, A.E., Greenberg, D.H., Vining, A.R. & Weimer, D.L. (2011). Cost-benefit analysis. Concepts and practice. Fourth edition. Prentice Hall, Upper Saddle River, New Jersey. 24 Wijnen, W., Weijermars, W., Vanden Berghe, W., Schoeters, A., Bauer, R., Carnis, L., Elvik, R., Theofilatos, A., Perez, C. and Martensen, H. (2017). Crash cost estimates for European countries, D3.2 of the H2020 project SafetyCube. World Bank (2005). Valuation of accident reduction. Transport Note No. TRN-16. World Bank, Washington. BRS&TRL (2003). Guidelines for estimating the costs of road crashes in developing countries. Babtie Ross Silcock & Transport Research Laboratory. Alfaro, J. L., Chapuis, M., Fabre, F. (Eds.) (1994). Socio-economic cost of road accidents: final report of action COST 313. Commission of the European Community, Brussels. 25 Wijnen, W., Weijermars, W., Vanden Berghe, W., Schoeters, A., Bauer, R., Carnis, L., Elvik, R., Martensen, H. (2019). An analysis of official road crash cost estimates in European countries. Safety Science 113, 318–327. 26 Wijnen, W. & Stipdonk, H. (2016). Social costs of road crashes: an international analysis. Accident Analysis and Prevention, 94, 97–106. 27 Bhattacharya, S., A. Alberini and M.L. Cropper. 2006. The value of mortality risk reductions in Delhi, India. World Bank Policy Research Paper 3995. Washington: World Bank. 17 | P a g e and Iran28. 30. In addition to the above three approaches, several other methods are discussed in the literature.29 One of these methods is the compensation payment approach. In this approach compensation payments to road casualties or their relatives in courts or by law (statutory values) are used as monetary valuation of human costs. The compensation payments are seen as the value attributed to loss of life. Another approach is the life insurance method, which implies that the monetary valuation of lives is derived from the premiums people pay for life insurances. Finally, the costs-per-life-saved method is aimed at determining the (government) expenditures on saving lives. A major limitation of these methods is their inconsistency with economic welfare theory. One of the fundamentals of this theory is that valuation should be based on individual preferences. With respect to valuation of lives, the values should refer to statistical lives instead of specific lives, which means that the valuation of a risk reduction should occur ex ante30. The three methods do not meet these criteria, because they do not reflect individual preference (compensation payment and cost-per-life- methods), do not have an ex-ante basis (compensation payment and cost-per-life-methods) or are not based on risk reduction (compensation payment and life insurance methods).31 Nevertheless, these methods are sometimes applied. In general, the values found by those studies are much lower than the values that result from WTP studies.32 31. For the assessment of the socio-economic costs of road crashes in Azerbaijan, the three internationally recommended methods are used: restitution costs, human capital and willingness to pay. Concerning the latter method, results from willingness to pay studies in other countries are used instead of collecting country-specific information for Azerbaijan. Other methods are not recommended, given their theoretical limitations discussed above. The restitution costs and human capital methods are well applicable to the situation in Azerbaijan, as they are relatively simple from a theoretical perspective. Data requirements complicate the use of these methods, but several data gaps are filled by conducting stakeholder interviews and a casualty survey (see Section 730). As discussed above, the WTP method has several limitations, which restrict its applicability, particularly in low and middle-income countries.33 Nevertheless, several studies have demonstrated that it is feasible to determine the willingness to pay in these countries, but it requires a very well-designed and well-organized survey and sufficient resources. Conducting a country-specific WTP-study is beyond the scope of this study, but evidence from other countries as reflected in ‘value transfer’ 28 Ainy, E., Soori, H., Ganjali, M., Le, H., & Baghfalaki, T. (2014) "Estimating Cost of Road Traffic Injuries in Iran Using Willingness to Pay (WTP) Method". PLoS ONE 9(12) 29 See e.g. Wijnen, W., Weijermars, W., Vanden Berghe, W., Schoeters, A., Bauer, R., Carnis, L., Elvik, R., Theofilatos, A., Perez, C. and Martensen, H. (2017). Crash cost estimates for European countries, D3.2 of the H2020 project SafetyCube. World Bank (2005). Valuation of accident reduction. Transport Note No. TRN-16. World Bank, Washington. 30Mishan, E. J. (1971). Evaluation of life and limb: a theoretical approach. Journal of Political Economy, 79, 687-705. Schelling, T.C., (1968). The life you save may be your own. In: Chase, S.B. (Ed.), Problems in Public Expenditure Analysis. The Brookings Institution, Washington, DC, pp. 127–176. 31 See the discussion in Wijnen, W., Weijermars, W., Vanden Berghe, W., Schoeters, A., Bauer, R., Carnis, L., Elvik, R., Theofilatos, A., Perez, C. and Martensen, H. (2017). Crash cost estimates for European countries, D3.2 of the H2020 project SafetyCube, and Schoeters, A., Wijnen ,W., Carnis, L., Weijermars, W., Elvik, R., Johanssen, H., Vanden Berghe, W., Reed, S. and Daniels, S. (2018). Costs related to serious road injuries. Proceedings of 7th Transport Research Arena TRA 2018, April 16-19, 2018, Vienna, Austria.. 32Wijnen, W., Weijermars, W., Vanden Berghe, W., Schoeters, A., Bauer, R., Carnis, L., Elvik, R., Martensen, H. (2019). An analysis of official road crash cost estimates in European countries. Safety Science 113, 318 –327. Wijnen, W. & Stipdonk, H. (2016). Social costs of road crashes: an international analysis. Accident Analysis and Prevention, 94, 97–106. Trawén, A., Maraste, P. & Persson, U. (2002). International comparison of costs of a fatal casualty of road accidents in 1990 and 1999. Accident Analysis and Prevention, vol. 34, nr. 3, p. 323-332. Elvik, R. (1995). An analysis of official economic valuations of traffic accident fatalities in 20 motorized countries. Accident Analysis and Prevention, vol. 27, nr. 2, pp. 237-347. 33 World Bank (2005). Valuation of accident reduction. Transport Note No. TRN-16. World Bank, Washington. 18 | P a g e functions will be used as a second-best option. This ensures that all costs, including human costs, are taken into account in the socio-economic costs. Section 730 provides further details. 2.1.4 Estimating the number of road crashes and casualties 32. Accurate information on the number of casualties and crashes of all severities is essential for the cost calculations. Both the numbers of casualties and crashes are needed, since some costs are casualty-related (e.g. medical costs) and some are crash-related (e.g. police costs). In addition, information on the number of casualties per crash is needed to be able to convert costs per crash into costs per casualty. For the crash cost calculations, we distinguish between: - Fatality: a person killed immediately or who has died within 7 days as a result of a road crash.34 - Fatal crash: a crash resulting in at least one fatality; - Serious injury: an injured person who has been treated in hospital requiring an overnight stay. - Serious injury crash: a crash resulting in at least one serious injury (but no fatalities); - Slight injury: a person who has been injured and treated in hospital without overnight stay or and who has not been treated in hospital; - Slight injury crash: a crash resulting in at least one slight injury (but no fatalities or serious injuries); - Property Damage Only (PDO) crash: a road crash causing damage to at least one vehicle, but no fatalities or injuries. 33. In Azerbaijan, data of road crashes and casualties are collected by the State Road Police. However, the official road statistics is likely to suffer from underreporting and under-recording for various objective and subjective reasons, similar to many other countries. This is a well-known and serious problem regarding official road safety statistics, particularly in low and middle-income countries. For example, GRSF35 reports that the underreporting rate of road fatalities is estimated at 84% in low-income countries and 51% in middle-income countries. The underreporting rate for injuries is likely to be even higher, which affects estimates of the socio-economic costs of road crashes significantly. The underreporting is a very serious issue for developed countries as well. For example, calculations in the Netherlands and UK show that the costs are more than twice as high if adjustments are made for underreporting of road casualties.36 34. For road crash costs calculations, it is important to estimate the actual number of road casualties by severity, including unreported casualties, to avoid underestimating the costs. Ideally, the underreporting rate is estimated by linking different databases on road casualty statistics, using the capture-recapture method.37 In particular, police and hospital data can be linked to estimate 34 The study uses the criterion of 7 days after the crash, as this criterion is used by the State Road Police. There is no information on the number of casualties who die within 30 days, which is the international standard. 35 GRSF (2020). Guide for Road Safety Opportunities and Challenges: Low and middle-income Country Profiles. World Bank, Washington. 36 Wijnen, W., Weijermars, W., Vanden Berghe, W., Schoeters, A., Bauer, R., Carnis, L., Elvik, R., Theofilatos, A., Perez, C. and Martensen, H. (2017). Crash cost estimates for European countries, D3.2 of the H2020 project SafetyCube. 37 Zavareh, D.K., Mohammadi, R., Laflamme, L., Naghavi, M., Zarei, A. & Haglund, B.J.A. (2008). Estimating road traffic mortality more accurately: Use of the capture–recapture method in the West Azerbaijan Province of Iran. International Journal of Injury Control and Safety Promotion, 15 (1), 9-17. Derriks, H. and Mak, P. (2007). Underreporting of road traffic casualties. IRTAD Special report. Paris: OECD / International Transport Forum. 19 | P a g e underreporting rates. However, this is a complex method and hospital data on road casualties are not systematically available in Azerbaijan. For that reason, applying this method is not feasible for this study. As an alternative, a combination of national road safety data and data from other countries is used to estimate the actual number of road casualties (including unreported). In addition to international reviews, such as the GRFS report mentioned above, results of a survey on road crashes and injuries among a representative sample of the population of Kazakhstan are used. This survey revealed information on road crashes and injuries which is not available in most countries (particularly not in low and middle-income countries and post-Soviet countries), such as people’s involvement in road crashes, injuries resulting from these crashes and injury severity. 35. WHO estimated the actual number of road fatalities in Azerbaijan on the basis of a regression model. The method used several variables such as income, motorization, alcohol consumption and number of motorcycles as a proportion of all vehicles.38 In general, the WHO estimates of the number of road fatalities are higher than the reported numbers, in particular in low and middle-income countries. Likely explanations include underreporting and differences in definitions of road fatalities. Therefore, the WHO estimates are suitable for indicating the underreporting rate of fatalities and correcting for differences in definitions.39 In this study, the WHO estimates of the number of road fatalities in Azerbaijan is used for that purpose. This approach takes into account that only fatalities who died within 7 days after the crash are included in the official statistics in Azerbaijan, while the international standard is 30 days. Unfortunately, similar estimates which take into account underreporting are not available for the number of injuries and crashes. Therefore, evidence is used from other countries that have estimated the actual number of serious and slight injuries including unreported cases, e.g. based on the capture-recapture method. In particular, ratios of the number of fatalities, serious and slight injuries are used to (roughly) estimate the numbers in Azerbaijan. In addition, results from the survey and expert opinions of doctors working in hospitals in Azerbaijan are used for the estimation of the number of injuries by injury severity. 36. Concerning the number of crashes, information on the number of casualties per crash as found in other countries is used by WHO. The estimation of the number of PDO-crashes is based on the ratio of the number of PDO-crashes to the number of fatalities excluding non-motorized road users (pedestrians and cyclists). By excluding non-motorized road users, differences in motorization rate between Azerbaijan and other countries are accounted for. 2.1.5. Calculation method per cost element 37. To estimate the socio-economic costs of road crashes in Azerbaijan, the study followed the international state-of-the-art and best practices as much as possible. This implies that the analysis is conducted from a broad socio-economic perspective, to ensure that the results are consistent with the theory of cost-benefit analysis and can be applied in cost-benefit analysis of investments in road safety and other transport projects. Furthermore, the classifications of cost items as recommended in international guidelines on road crash costing (see Figure 1 and Appendix 1) and the internationally recommended valuation methods for each cost item are applied as much as possible. 38. To calculate the costs, a bottom-up or top-down approach can be applied. In the bottom-up approach, the costs per casualty or crash are determined firstly and the total cost is calculated by multiplying the cost per casualty/crash by the number of casualties/crashes. In the top-down 38 WHO (2018). Global Status Report on Road Safety 2018. World Health Organization, Geneva. 39 For example, GRSF estimated the fatality underreporting rate for a large number of countries using the WHO-estimates. GRSF (2020). Guide for Road Safety Opportunities and Challenges: Low and middle-income Country Profiles. World Bank, Washington. 20 | P a g e approach, the total costs are calculated firstly and the costs per casualty or crash are derived from the total costs. In this study, the bottom-up approach is used for all cost items, except insurance costs and legal costs. The top-down approach is more appropriate for these latter costs because data on total costs are available instead of costs per casualty. 39. The study uses market prices, or proxies of market prices to estimate several cost items, including medical costs, production loss, vehicle damage and administrative costs. These prices include, for example, prices of medical treatment, wages and vehicle repair. Following economic theory and international best practices, it is assumed that these market prices reflect the ‘opportunity cost’ which is defined as the value of resources (e.g. labor or equipment) in its best alternative use.40 According to economic welfare theory, the concept of opportunity costs should be used to determine socio-economic costs. The methodology for each cost item is discussed below in more detail. Medical costs 40. While medical costs usually account for a relatively small share of total costs of road accidents41, they can take significant health sector resources and are often the first and most tangible economic burden experienced by the persons involved in road accidents. Medical costs often consist of both private and public sector costs. The medical costs included in the analysis are the costs related to the following. - Transporting casualties to a hospital; - Hospital treatment, which includes emergency care directly after the road crash, treatment for casualties who stay overnight in hospital (in-patient) and follow-up hospital visits after hospital treatment (out-patient); - Treatment in a rehabilitation center; - Non-hospital treatment, such as treatment by primary health care clinics, physiotherapists and psychologists; - aids and appliances, for example crutches and wheelchairs; - medicines; - home care, provided by medical professionals or by relatives and friends; - house adaptations, for example to make it accessible for wheelchairs; - visiting patients in hospital. 41. The estimation of the medical costs is based on the RC method which means that the actual prices of medical care are used, such as the prices per ambulance ride, per hospital day, etc. Following a bottom-up approach, the total medical cost is calculated by multiplying the unit cost (e.g. costs per ambulance ride) by the number of ‘units’ (e.g. number of ambulance rides). The unit costs include all costs related to the medical treatment, both direct costs (e.g. costs of medical personnel, medicines, fuel for the ambulance) and indirect costs (equipment, management and administration, housing). For home care by relatives or friends and time costs of visits to the hospital, the price of professional home care is used as a proxy. In addition, data on the medical treatment is used, such 40Boardman, A.E., Greenberg, D.H., Vining, A.R. & Weimer, D.L. (2011). Cost-benefit analysis. Concepts and practice. Fourth edition. Prentice Hall, Upper Saddle River, New Jersey. 41Wijnen, W., Weijermars, W., Vanden Berghe, W., Schoeters, A., Bauer, R., Carnis, L., Elvik, R., Martensen, H. (2019). An analysis of official road crash cost estimates in European countries. Safety Science 113, 318–327. Wijnen, W. & Stipdonk, H. (2016). Social costs of road crashes: an international analysis. Accident Analysis and Prevention, 94, 97–106. 21 | P a g e as the average duration of hospital treatment, the proportion of road casualties transported by ambulance (by injury severity), etc. Production loss 42. The human capital method is applied to estimate production loss. This means that production loss of a casualty is calculated by multiplying the period of time they are not able to work due to the crash with a valuation of the production per person per unit of time. Regarding fatalities, the period of time taken into consideration is the remaining number of productive years until retirement. For injured casualties the relevant period of time ranges from a few days’ absence from work, to all the remaining working years until retirement if someone is permanently disabled and unable to work. Also, there are time losses related to finding a new job if a casualty got unemployed due to the injury. On the employer side, there are costs for finding new employees (known as ‘friction costs’), which are usually very small compared to production loss.42 A common indicator for the valuation of production and time loss is wage. The calculation takes into account gender differences. 43. The potential production loss has been estimated following international best practices43. This means that all productive capacities of people are included, regardless of whether they are employed or not. The reasoning is human resources are lost and that the valuation of this loss should not depend on employments. Moreover, this approach takes into account the valuation of non-market production such as household and voluntary work. 44. The gross potential production loss is calculated, which means that loss of consumption is included as well. By using wage as the valuation indicator, both production and consumption are included. On the one hand, wage reflects the valuation of human production. On the other hand, wage is the income that people spend on consumption. This implies that production and consumption loss should not be included separately, because this would imply a double counting. 45. A social discount rate applies to future production loss, following the practice of cost-benefit analysis of assigning a lower weight to costs and benefits if they occur further in the future. The social discount reflects the fact that people prefer to have benefits rather sooner than later (‘time preference’) and vice versa with respect to costs. In addition, the discount rate reflects the opportunity costs of capital, which is the fact that capital used in investment projects can yield a return if used in for other purposes.44 46. Key data needed for the calculation of production loss include the age distribution of fatalities and disabled casualties, duration of absence from work due to injuries, yearly number of working hours and wage (by gender). Human costs 47. Identification of human costs are should ideally be based on a country-specific WTP-study. Conducting a thorough WTP-study, however, is complex and demanding in terms of resources 42 Wijnen, W. & Stipdonk, H. (2016). Social costs of road crashes: an international analysis. Accident Analysis and Prevention, 94, 97–106. 43 Wijnen, W., Weijermars, W., Vanden Berghe, W., Schoeters, A., Bauer, R., Carnis, L., Elvik, R., Theofilatos, A., Perez, C. and Martensen, H. (2017). Crash cost estimates for European countries, D3.2 of the H2020 project SafetyCube. 44 See for example Boardman, A.E., Greenberg, D.H., Vining, A.R. & Weimer, D.L. (2011). Cost-benefit analysis. Concepts and practice. Fourth edition. Prentice Hall, Upper Saddle River, New Jersey 22 | P a g e needed to for implementation of study. Stated preference studies require development of a well- designed and tested questionnaire and a survey among a representative sample of population. Revealed preference studies are very demanding with respect to data availability, and for that reason they are not often applied in the context of road safety.45 For these reasons, conducting a country- specific WTP-study is beyond the scope of this study. However, given the fact that human costs constitute a major cost component (see Section 0), omitting these costs would dramatically underestimate the full socio-economic costs. 48. To avoid the underestimation, human costs have been be included based on the evidence from other countries. For this particular purpose, value transfer functions that relate the Value of a Statistical Life (VSL) to GDP per capita was developed by the World Bank study46 and the International Road Assessment Programme (iRAP)47. Using regression analysis, they specified functions that relate GDP per capita to the VSL. The World Bank study developed separate value transfer functions for high-income countries and for low and middle income countries. The function for low and middle income countries is based on 123 VSL estimates found in stated preference studies in many countries around the world. The function specifies the VSL as follows: VSL = 1.3732E−4 * (GDP per capita)ˆ2.478. This function is specified in 2005 international dollars. The iRAP function is a more simple linear function: VSL = 70 * (GDP per capita). This function is based on official VSLs in seven countries, including two low and middle income countries, which are based on a WTP method. If GDP per capita is above roughly 9,000 USD (2019 international dollars), the iRAP function results in lower values than the World Bank study function. 49. This study applies the iRAP value transfer function to estimate the VSL in Azerbaijan. This results in values which are probably more in line with values that are commonly used in policy research and it can be regarded as a conservative approach. The World Bank study function is based on a much larger number of VSL estimates found in scientific studies, which are usually higher than the official values as determined by government bodies and as used in economic appraisal. However, the scientific foundation of the World Bank study function is more solid and for that reason this study uses this function in the sensitivity analysis. Clearly, the values resulting from both functions should be interpreted as an indication, since they are not based on the WTP of the population of Azerbaijan. 50. The VSL includes both human costs and consumption loss, and therefore it overlaps with gross production loss, which also includes consumption loss (discussed above). The VSL is assumed to reflect the utility an individual gains during people’s lifetime. This utility is derived from consuming goods and services as well as from intangible issues. Consumption of goods and services overlaps with production loss because an individual would spend his/her wage (which is used as the indicator for production loss) on this consumption. This is illustrated by 51. Figure 44. For that reason, lifetime consumption is deducted from the VSL to avoid double counting. This is the standard practice in road crash cost studies.48 45Blaeij, A.T. de, Florax, R.J.G.M., Rietveld, P. & Verhoef, E. (2003). The value of statistical life in road safety; A meta- analysis. In: Accident Analysis and Prevention, vol. 35, nr. 6, p. 973-986. Lindhjem, H., Navrud, S., Braathen, N.A. & Biausque, V. (2011). Valuing mortality risk reductions from environmental, transport and health policies; a global meta-analysis of stated preference studies. In: Risk Analysis, vol. 31, nr. 9, p. 1381-1407. 46 Milligan, C., Kopp, A., Dahdah, S. & Montufar, J. (2014). Value of a statistical life in road safety: A benefit-transfer function with risk analysis guidance based on developing country data. Accident Analysis and Prevention, 71, pp. 236-247. 47 McMahon, K. & Dahdah, S. (2008). The true costs of road crashes. International Road Assessment Programme iRAP. 48 Wijnen, W. & Stipdonk, H. (2016). Social costs of road crashes: an international analysis. Accident Analysis and Prevention, 94, 97–106. 23 | P a g e Figure 4: Relation between VSL, human costs, production loss and consumption loss49 52. Consumption loss consists of the value of the consumption that a person would have gained if he/she was not killed in a road crash. This means that the average future lifetime consumption per person needs to be calculated. This is done on the basis of single age categories (e.g. all 1 year old, 2 years old, etc. people) in a few steps: - The future lifetime consumption per person, by age category and gender, is calculated by multiplying yearly consumption by the remaining life years (which is equal to the life expectancy at each age). Future consumption is discounted using the social discount rate (discussed above); - Total future consumption by age category is calculated by multiplying future lifetime consumption per person by the number of populations in each age category; - Total future lifetime consumption of the entire population is calculated as the future lifetime consumption of all age categories; - Average future lifetime consumption is calculated by dividing total future lifetime consumption by the populations size. 53. A value transfer approach is also applied to estimate the human costs related to injuries. Studies on the WTP for non-fatal crash risk reduction have been conducted in several countries, including European countries.50 In these studies the monetary valuation of preventing serious and/or slight injuries is determined as a proportion of the VSL. The human cost of serious and slight injuries is calculated using the VSL in Azerbaijan and the value of an injury as a proportion of the VSL as found in the international literature. Following international common practice, it is assumed that there is no consumption loss related to injuries.51 Property damage 54. Property damage due to road crashes is mainly related to the damage to vehicles, usually 49Based on Wijnen, W., Wesemann, P. & de Blaeij (2009). Valuation of road safety effects in cost-benefit analysis. Evaluation and Program Planning, 32, pp. 326-331. 50 Schoeters, A., Wijnen ,W., Carnis, L., Weijermars, W., Elvik, R., Johanssen, H., Vanden Berghe, W., Reed, S. and Daniels, S. (2018). Costs related to serious road injuries. Proceedings of 7th Transport Research Arena TRA 2018, April 16-19, 2018, Vienna, Austria. 51 See for example Van Essen, H., Wijngaarden, L. van,. Schroten, A., Sutter, D., Bieler, C., Maffii, S., Brambilla, M., Fiorello, D., Fermi, F., Parolin, R., El Beyrouty, K. (2019). Handbook of external costs of transport; 2019 version. CE Delft, Delft. 24 | P a g e more than 90% of total property damage.52 Vehicle damage is estimated using the restitution costs approach, which means that actual cost of resources needed to repair damage or replace property is measured on the basis of market prices. Information on the average damage per vehicle and number of vehicles damaged in road crashes (by crash severity) is being used for this calculation. 55. Insurance business is a good potential source for data on property damage. However, insurance data is not representative for the total property damage caused by road crashes, because only limited proportion of vehicle damages are reported to insurance companies. Studies in high income countries show that insurance statistics may cover only 50% of total property damage53, and this proportion may be even lower in low and middle-income countries. Therefore, the total vehicle damage cannot be estimated accurately on the basis of the total damage reported to insurance companies (‘top down’). Instead, a bottom-up approach is applied using information on the average damage per vehicle, the number of vehicles per crash and the number of crashes (by crash severity). For the estimation of vehicle damage, information on the number of crashes is essential. Lower severity crashes should also be included, as their impact on total costs can be very significant (about 20-25% of the total costs54). In cases when the road accidents involve structures adjacent to the roads (especially in the urban setup), damage to infrastructure can also be significant. Such damage can be estimated using information on repair costs (restitution costs method) as collected by the State Road Agency. In addition, there might be damage to personal belongings in the vehicle and damage to freight carried by trucks. The costs are usually very small or negligible as compared to vehicle damage and for that reason they are not included in this study calculations. Administrative costs 56. Administrative costs include police costs, costs of other emergency services (fire service, towing, etc.), legal costs and insurance costs. These costs can be calculated on the basis of the actual costs of resources spent by the organizations involved, which includes the costs of time spending by personnel and the costs of equipment (restitution costs method). 57. Police costs concern the costs of time police officers spend on road crashes and their investigation (excluding prevention of road crashes) as well as material costs such as equipment costs. To estimate these costs, information on the number of police officers coming to the crash location and the time they spend is used (by crash severity), as well as their wage. Indirect costs, such as equipment and housing costs, should be added to that, e.g. based on financial accounts of the police organization. A similar approach can be used for estimating the costs of fire service. 58. Costs of towing can be estimated on the basis of the price of towing services (restitution costs method). In addition, information on the average number of vehicles towed away per crash is needed (by crash severity). 59. Insurance costs are the administrative costs of insurers related to vehicle insurances, which include costs of personnel for handling claims and indirect costs such as office and equipment costs. All administrative costs related to vehicle insurances are included, instead of only costs related 52 Wijnen, W. & Stipdonk, H. (2016). Social costs of road crashes: an international analysis. Accident Analysis and Prevention, 94, 97–106. Wijnen, W., Weijermars, W., Vanden Berghe, W., Schoeters, A., Bauer, R., Carnis, L., Elvik, R., Martensen, H. (2019). An analysis of official road crash cost estimates in European countries. Safety Science 113, 318–327. 53 Wijnen, W. & Stipdonk, H. (2016). Social costs of road crashes: an international analysis. Accident Analysis and Prevention, 94, 97–106. 54 Wijnen, W. & Stipdonk, H. (2016). Social costs of road crashes: an international analysis. Accident Analysis and Prevention, 94, 97–106. 25 | P a g e to handling claims, which is common international practice.55 The underlying idea is that all vehicle insurance costs can be attributed to road crashes, because insurance for damage would not be needed if there were no crashes. Note that insurance costs do not include damage payments, as they are included in the calculation of property damage. In addition, there are administrative costs related to other insurances, in particular health insurance. These costs are not commonly included in road crash cost studies and probably they are negligible in Azerbaijan because health care is mostly financed by the government and mandatory health insurance is just being introduced. The calculation of vehicle insurance costs is based on a top-down approach, using accounting information of insurance companies, in particular the income from vehicle insurance premiums and the administrative costs as a proportion of this income. Costs per crash are calculated using the number of vehicles per crash (Table 6). 60. Legal costs include costs of prosecuting offenders who caused a road crash, costs of court cases resulting from road crashes, and costs of imprisonment. The study uses a top-down approach, because detailed information on the costs per road crash case is not available. This means that the cost estimate is based on the costs (budgets) of the institutions involved in these legal issues, in particular the Prosecutor General's Office and the Ministry of Justice. The costs related to road crash cases are estimated using the number of court cases related to road crashes as a proportion of the total number of court cases. This assumes that the costs per road crash case are equal to the average costs of all cases. The advantage of this top-down approach is that it includes both the direct costs of prosecution, court cases and imprisonment as well as the indirect cost of all organizations involved, including administrative costs, management costs, office costs, equipment, etc. Other costs 61. Congestion cost is another important cost item. Road accidents can cause delays in traffic, especially on congested roads of capital city of Baku. These delays are associated with additional socio-economic costs of road accidents, which have several elements. Travel time loss is one of the largest categories of transport costs and any delays in traffic can lead to significant time costs. Traffic delays also inevitably result in extra fuel consumption. Another form of the associated cost is environmental externality, which refers to the costs due to increased emissions of pollutants due to slower speed. Most countries that have estimated these costs use congestion models, which calculate the time losses resulting from crashes using several parameters such as traffic volumes, road capacities, capacity reductions due to a crash and emergency response times.56 Developing such a detailed model is complex and beyond the scope of this project. Alternatively, a value transfer approach could be applied. This implies that congestion costs per crash from studies in other countries are used and adapted to the situation in Azerbaijan, using data on traffic volumes and crash severity among others. Unfortunately, systematic traffic and congestion data are not available in Azerbaijan, and for that reason congestion costs cannot be included in the cost calculations. However, an indication of the size of the congestion costs will be given based on studies in other countries. 62. Funeral costs typically have a small share in total crash costs, but they may have substantial financial impact on the individual or household level. They are a common cost item in road crash cost studies. The cost refers to the fact that a funeral takes place (much) earlier than expected. Consequently, and following international best practices, the costs are calculated as the difference between the actual costs of a funeral and the discounted future costs of the funeral if the person was not killed in a road crash. This accounts for the fact that if somebody dies in a road crash at an 55Wijnen, W., Weijermars, W., Vanden Berghe, W., Schoeters, A., Bauer, R., Carnis, L., Elvik, R., Theofilatos, A., Perez, C. and Martensen, H. (2017). Crash cost estimates for European countries, D3.2 of the H2020 project SafetyCube. 56 Wijnen, W. (2021). Costs of congestion related to road crashes in Ireland. Road Safety Authority, Ballina. 26 | P a g e older age, the person would have died sooner because of another death cause and funeral costs would have occurred sooner anyway. The costs are calculated using the average costs per funeral, the number of fatalities by age, life expectancy (by age) and the social discount rate. 63. Road crashes can also result in environmental costs, as they may result in petrol and fluid leaks, emission of chemicals into the environment that pollute grass and plants and harm wildlife (in addition to environmental costs related to congestion). Environmental damage is a particular problem for crashes that happen near natural and manmade water resources. When vehicles or their parts are destroyed in road crashes, some elements are recycled, but other remain at road sides, open air, or end up in landfills where they harm nature and have negative effects on soil, water, and air influencing the entire ecosystem. According to international reviews, these environmental costs have not been taken into account yet in road crash cost studies in other countries, but in principle the costs can be estimated using the restitution costs method. This means that the costs of restoring the damage, such as removing waste from roadside areas and cleaning up oil spills from roads, are estimated. However, data on these costs is not systematically collected in Azerbaijan and for that reason these costs cannot be included in the calculations. 2.1.6 Data requirements 64. An analysis of the socio-economic costs of road crashes is very demanding with respect to data requirements. The extent to which each cost item can be included, and the accuracy of the estimates depend to a large extent on the availability and quality of data on a large variety of parameters. 65. Besides road safety data, a large variety of other data is needed for the cost calculations including data on medical costs, labor market, economy, demography and insurance, as indicated in Section 0. Appendix 4 gives an overview of the data that are needed for the calculation of each cost item. The data are retrieved from several data sources, including existing databases or publications, interviews with stakeholders such as the State Road Police, ministries, other governmental agencies and NGOs, a survey among road casualties and their relatives and data from other countries. 66. The base year of this study is 2019, which means that all data and outcomes refer to 2019 unless stated otherwise. Monetary values from other years (including the results of the casualty survey) are transferred into 2019 price level using consumer price index data as published by the IMF.57 2.2. Impact of road accidents on poverty and shared prosperity 67. Although poverty has many dimensions, common indicators for poverty are based on consumption or income.58 Poverty is usually measured as the proportion of people below a certain consumption or income level (poverty line). Other issues, such as deprivations with respect to education and health, are also relevant for poverty measurement, although multidimensional indicators including such issues are not commonly applied. 68. In Azerbaijan, the poverty level is defined based on the per capita consumption expenditure. Annually, the poverty line, or the minimum subsistence level, is established by law for key socio- demographic population groups, based on the Household Budget Survey (HBS) conducted by the 57 https://www.imf.org/en/Countries/AZE 58 World Bank (2015). A Measured Approach to Ending Poverty and Boosting Shared Prosperity. 27 | P a g e State Statistical Committee (AZSTAT).59 The subsistence minimum is determined based on average market prices for products, goods and services included in a consumer basket. For 2020, the general monthly subsistence level was set at 190 AZN. For able-bodied persons it is 201 AZN, for pensioners 157 AZN and for children 170 AZN60. Individuals with a per capita consumption below the poverty (nominal) line are considered poor. 69. In addition to poverty level, policy targets are aimed at improving shared prosperity, which means increasing the average income of the bottom 40 percent (B40) of the population. Shared prosperity is not a measure of inequality in itself but provides insight into the changes in inequality by comparing the development of the income of B40 with the upper 60 percent (U60) of the population.61 70. To assess the impact of road crashes on poverty and shared prosperity, the study conducted a survey among road casualties. The questionnaire used in the survey included questions on the (i) economic status of the casualty households, (ii) the impact of the road crash on their economic position, (iii) and the impact on livelihood and quality of life. The economic status of respondents was identified based on their current consumption expenses and household incomes. This mostly included indirect questions to assess the financial position of the households, such as the proportion of income spent on food and the ability of the household to buy basic and luxury goods and services. The impact of the road crashes on economic status of casualties was measured by gauging change in income, impacts of expenses related to the crash, such as medical expenses and vehicle repair, on the household budget, any financial compensations received from insurance companies and other sources, the need to borrow money or sell assets, etc. To identify the change in the quality of life respondents were asked to rate the extent to which the crash affected their quality of life. 71. In-depth interviews with several road casualties were conducted, aimed at providing more detailed and personalized picture of the impact of road crashes on people’s lives. The topics covered by the interviews, inter alia, included consequences of accidents for subsequent life, short- and long-term health impacts, financial burdens and the impacts on economic status, poverty, and income levels of casualties. 72. To assess the impact of road crashes on poverty and shared prosperity, the impacts of the road crash are examined separately for B40 and U60 households. The impact on consumption and income provides insight into the poverty impact. Furthermore, the impacts on B40 and U60 households are compared, to see whether road crashes have an adverse impact on shared prosperity. The respondents were classified as either B40 or U60 based on their self-reported household income and the minimum subsistence thresholds mentioned above. Household income of the respondents was translated into income per capita using the number of household members. Furthermore, the impacts for some vulnerable groups are considered separately: IDP families, large households (5 members or more), elderly (above 60 years old) and low educated people (general secondary or less). 73. In addition, a comparison is made between respondents in Baku and rural regions, because the traffic and transport situation as well as the poverty situation is expected to be different for the rural regions as compared to Baku. All regions except Baku and the cities of Ganja, Sumgayit and 59 This is a nationally representative survey which collects information on household income and expenditure, housing conditions, ownership of consumer durables, access to agricultural land and demographic characteristics of household members. See AZSTAT (2020): https://www.stat.gov.az/source/budget_households/?lang=en, and the International Household Survey Network (2019): https://catalog.ihsn.org/index.php/catalog/2162. 60 https://apa.az/en/finance-news/Minimum-subsistence-level-for-2020-defined-in-Azerbaijan-298259 61 World Bank (2015). A Measured Approach to Ending Poverty and Boosting Shared Prosperity. 28 | P a g e Khirdalan have been considered as rural. According to official statistics, the proportion of the population under the poverty line in Baku (2%) is lower than in the rural regions, where this proportion ranges from 3% and 4% in Absheron and Ganja-Gazakh respectively to 14% in Lankaran.62 74. The statistical program SPSS was used for data entering, data cleaning, data processing and producing numerical and text outputs. The survey data are analyzed and presented in a Chapter 4. 2.3. Data collection 75. Identification of the socio-economic costs of road crashes and assessment of their social and poverty impacts normally require a large variety of data and different data collection methods. Four data collection methods used in implementation of this study, including (i) stakeholder consultations and interviews, (ii) road casualty survey, (iii) in-depth interviews, and (iv) value transfer methods are discussed below. Method of Stakeholder Consultations and Interviews 76. Stakeholder consultations and interviews mostly concentrated on collection of data and opinions on road safety policy, specific road accident cost elements, and social impact issues. Based on the list of data and information required, stakeholders from government ministries, police, insurance companies, health facilities and institutions, academia, and NGOs were identified and interviewed. Particular attention was paid to collecting medical data as this information was mostly found to be not readily available. To fill this information gap, interviews were conducted with a number of health specialists from different departments of several public or private hospitals. Method of Road Casualty Surveys 77. A survey was conducted among 200 road casualties (or their relatives) who were involved in road accidents in the years 2018-2020. The survey aimed at collecting detailed information on crashes on all road types and on injuries, such as circumstances of accidents, medical treatment requirements, consequences of the injuries with respect to jobs and daily activities, damage to vehicles, medical and other expenses, impact on livelihood and income, and others. Interviews also included collection of information on socio-demographic characteristics of the respondents, such as gender, age and employment. Separate questionnaires were developed for respondents who were injured and for relatives of fatalities. 78. To ensure representativeness, target numbers of respondents were used with respect to gender, age category and transport mode, based on the national official road casualty statistics. In addition, quotas for injury severity (fatal, disabled/serious injury, slight injury and property damage only) and regions (Baku versus rural regions) were used. Due to travel restriction related to the Covid- 19 pandemic, the majority of the interviews (195) was conducted through phone calls. 79. The gender and age of surveyed participants were close to distribution of the number of reported fatalities and injuries in the official statistics. Thus, 74.4% of the respondents (148) were male and the rest were females. Error! Reference source not found. shows the age distribution of the sample in comparison to reported casualties (fatalities and injuries), using the categories as used in the official statistics. The share of the age groups 25-65 and over 65 years old corresponds to the official statistics, whereas the age group 18-24 years old is slightly overrepresented (20.7% versus 12.5%), while the group 0-17 years old is slightly underrepresented (6.1% versus 10.5%). 62 World Bank (2015), Azerbaijan Systematic Country Diagnostic. 29 | P a g e Figure 3: Age distribution sample and reported casualties 80.0% 60.0% 40.0% 20.0% 0.0% 0-17 18-24 25-65 65-over Sample Reported casualties 80. Most respondents were car drivers, car passengers or pedestrians when they were involved in the crash, which corresponds to the transport mode distribution of the reported fatalities (Error! Reference source not found.). However, the sample includes more car drivers and less car passengers as compared to the fatality statistics. The influence of this difference on injury severity, and thereby on costs and other crash impacts, is assumed to be small because the transport mode (car) is the same. Figure 6: Transport mode distribution sample and reported fatalities63 60% 50% 40% 30% 20% 10% 0% Car drivers Motorized Bicyclist Pedestrian Bus and and 2-wheeler truck passengers drivers and passengers Sample Reported fatalities 81. Half of the respondents were injuries (52 serious, 48 slight), while there were 31 fatal cases (Figure 7). A third of the respondents were not injured but were exposed to property damage only. As explained in Section 0, the injury severity distribution was not intended to be consistent with the official casualty statistics. Instead, the aim was to have sufficient numbers of respondents in each severity category, to enable making cost estimates for each severity category. 63 The source of the reported number of fatalities in 2016 is the WHO report Global Status on Road Safety 2018, based on SRP statistics. 30 | P a g e Figure 7: Number of respondents by injury severity 31 67 52 48 Fatal Serious Slight PDO 82. The main causes of the crashes were ignoring traffic rules, distraction, and speeding. Speeding was reported more often by respondents from rural regions than from Baku (23.9% versus 15.5%), while ignoring traffic rules was reported more by respondents from Baku (31.6% versus 21.0% in rural regions). Most of the concerned crashes happened on urban roads (65.3%), while rural roads and motorways account for 19.1% and 13.6% of the crashes respectively (2.0% happened on other roads). In-Depth Interview Method 83. The study was also informed by several in-depth interviews conducted with road casualties or their relatives. The cases were selected from the respondents of the casualty survey, taking into account diversity with respect to injury severity, transport mode, region, age and gender. While the structure of the interview was open a checklist was used by the interviewers to ensure that all relevant topics are addressed. Value Transfer Method 84. Value transfer method implies that available data from other countries is collected and adapted to the situation in Azerbaijan while taking into account differences in income, purchasing power, etc. This method is known as ‘value transfer’ or ‘benefit transfer’64. The approach is only used as a fallback option if some specific data on costs of road crashes were not available in Azerbaijan and could not be collected through interviews with stakeholders or road casualties. This approach was in particular used to estimate the human costs of road crashes and to estimate the number of non-fatal casualties and crashes. 64 Freeman, A.M., Herriges, J.A., & Kling, C.L. (2014). The measurement of environmental and resource values. Theory and methods. Third edition. Resources for the future, New York. 31 | P a g e 3. Assessment of the socio-economic costs of road accidents 85. This chapter presents the assessment of the socio-economic costs of road accidents in Azerbaijan. As per the methodology described in Section 0, the assessment of road accident costs requires prior estimation of the number of road crashes and casualties of various severity level in the country. Based on this critical data, the calculation of each cost item will be completed. The chapter concludes with an overview of the total associated costs. 3.1. Estimation of the number of road crashes and casualties Fatalities 86. According to the official road safety statistics, the reported number of road fatalities in Azerbaijan was 821 in 2019. WHO conservatively estimated the actual number of fatalities in 2016 at 845, which is 11% higher than the reported number in 2016 (759), to account for likely underreporting. Assuming this underreporting rate also applies to the 2019 data, the actual number of fatalities (including unreported cases) is estimated at 914. 87. Concerning the age distribution of fatalities, the age categories in the State Road Police data are used as a basis. The broad age category 25-64 years old is split into 5-year categories using population statistics (source: Ministry of Labor and Social Protection), assuming that the number of fatalities per 5-year age category is proportional to the population size per category. The same approach is used for the category 65 years and over. A maximum age of 79 years is assumed, and the underreporting rate is assumed to be equal for each age category. Table shows the resulting number of fatalities per age category. Table 1: Number of fatalities by age (including unreported)65 Age Reported Including unreported Male Female Total Male Female Total 0-5 14 7 21 16 8 23 6-9 9 3 12 10 3 13 10-14 13 3 16 14 3 18 15-17 16 4 20 18 4 22 18-20 20 2 22 22 2 24 21-24 52 13 65 58 14 72 25-64 25-29 n/a n/a n/a 80 21 101 30-34 n/a n/a n/a 81 21 102 35-39 n/a n/a n/a 69 18 87 40-44 n/a n/a n/a 57 15 72 45-49 n/a n/a n/a 55 14 69 50-54 n/a n/a n/a 57 15 72 55-59 n/a n/a n/a 57 15 71 60-64 n/a n/a n/a 41 11 51 subtotal 446 115 561 497 128 625 65 -over 65-69 n/a n/a n/a 30 17 47 70 - over n/a n/a n/a 44 24 68 subtotal 67 37 104 74 41 116 Total 637 184 821 709 205 914 65 Sources: State Road Police (reported) and study calculations 32 | P a g e Serious injuries 88. According to the official statistics, the number of injuries in 2019 was 1,702. As discussed in Section 0, the actual number of injuries is likely to be largely underestimated. For example, an analysis of the number of road casualties in 23 OECD-countries by GRSF report (2020) shows that the ratio of serious injuries (defined as hospitalized injuries or using other national definitions) to fatalities ranges from 2:1 to 31:1 with 9:1 as the median.66 In the survey among a representative sample of the population of Kazakhstan on road crashes and injuries in Kazakhstan a ratio of 4:1 was found. These ratios are much higher than the ratio of all reported injuries to fatalities to in Azerbaijan, which is about 2:1, and the ratio of serious injuries to fatalities will be even lower. Due to underreporting the actual ratio is likely to be much higher, as indicated by the international evidence. To roughly and conservatively estimate the actual number of serious injuries in Azerbaijan, the study applied a ratio of 6.5:1, which is in between the ratios found in the above GRSF study and the study in Kazakhstan.67 This results in approximately 5,900 serious injuries. This figure refers to serious injuries who have been treated in hospital with an overnight stay, since this is the most frequently used definition internationally (including the study in Kazakhstan). In a sensitivity analysis, alternative ratios of the number of serious injuries to fatalities are used: 4:1 and 9:1 as found in the study in Kazakhstan and the GRSF study. 89. The age distribution is determined using the approach as for fatalities. Table 2 shows the resulting age distribution. Table 2: Number of serious injuries by age (including unreported)68 Age Reported Including unreported Male Female Total Male Female Total 0-5 20 13 33 70 45 115 6-9 33 3 36 115 10 126 10-14 47 20 67 164 70 234 15-17 45 15 60 157 52 209 18-20 54 11 65 188 38 227 21-24 134 29 163 468 101 569 25-64 25-29 140 54 194 487 189 676 30-34 140 54 194 489 190 678 35-39 120 47 166 419 162 581 40-44 99 38 138 346 134 480 45-49 95 37 132 331 128 459 50-54 99 38 137 345 134 478 55-59 98 38 136 342 133 475 60-64 71 28 98 248 96 344 subtotal 861 334 1,195 3,005 1,166 4,171 65 -over 65-69 19 15 34 66 53 118 70 - over 27 22 49 95 76 171 subtotal 46 37 83 161 129 290 Total 462 1,702 1,240 4,328 1,613 5,941 66 GRSF (2020). Guide for Road Safety Opportunities and Challenges: Low and middle-income Country Profiles. World Bank, Washington. 67 According to the GRSF report, the median ratio is about 1:15 if a fatality underreporting rate of 30-50% is accounted for. However, the fatality underreporting rate is much lower. For that reason, we apply the uncorrected median ratio of 1:9. 68 Sources: State Road Police (reported) and study estimates. 33 | P a g e Slight injuries 90. According to data provided by doctors working in public hospitals and insurance data, about 65% of all road casualties treated in hospital stay overnight. Based on this information and the estimated number of serious (hospitalized) injuries (5,900), it is estimated that the number of road casualties treated in hospital without overnight stay is 3,200. 91. Information on the number of slight injuries who are not treated in hospital is not available in Azerbaijan, like in most other countries. However, the above referenced survey in Kazakhstan showed that a quarter of the slight injuries are treated in hospital (without overnight stay). Assuming that a similar proportion applies to Azerbaijan, the number of slight injuries who are not treated in hospital is roughly estimated at 8,900. 92. The above estimated number of slight injuries is low as compared to the numbers used in road crash cost studies in other countries, where the ratio of slight to serious injuries ranges from 5 to 15.69 This is significantly higher than the ratio that results from the calculation above (12,100 slight injuries : 5,900 serious injuries). Therefore, we make an alternative calculation using the ratio of the number of all injuries (serious and slight) to the number of fatalities. The GRSF study mentioned above found that this ratio ranges from 5:1 to 96:1 with 46:1 as the median. Applying the median ratio to the number of fatalities in Azerbaijan results in 42,000 injuries. If 5,900 of them are serious injuries (as estimated above), the number of slight injuries is roughly 36,000 and the ratio of slight to serious injuries is 6. This number of slight injuries is used in a sensitivity analysis. Crashes 93. The official statistics reported 1,870 road crashes in 2019. This refers to crashes of all severities and a breakdown into crash severity is not available. To estimate the number of crashes by severity, the study used the ratios of the number of crashes to the number of casualties as applied in other road crash costs studies. Table provides an overview of these ratios. The study uses the mean ratios to estimate the number of crashes in Azerbaijan: 1.10 fatalities per fatal crash, 1.15 serious injuries per serious injury crash and 1.35 slight injuries per slight injury crash. Table 3: Ratio number of casualties to number of crashes by crash severity70 Fatal Serious injury Slight injury Australia 1.1 n.a. n.a. Austria 1.06 n.a. n.a. Egypt 1.4 1.0 1.2 Germany 1.08 1.15 1.35 Myanmar 1.10 1.12 1.4 New Zealand 1.07 1.18 1.29 Thailand 1.16 1.25 1.72 Median 1.14 1.14 1.39 Mean 1.10 1.15 1.35 69Wijnen, W. (2021). Socio-economic costs of road crashes in middle-income countries: applying a hybrid approach to Kazakhstan. IATSS Research. 70Wijnen, W. (2021). Socio-economic costs of road crashes in middle-income countries: applying a hybrid approach to Kazakhstan. IATSS Research. Ahadi, M.R. and Razi-Ardakani, H. Estimating the cost of road traffic accidents in Iran using human capital method. International Journal of Transpotation Engineering, 2 (3). 34 | P a g e 94. Accurate information on the actual number of crashes with property damage only (PDO) is very scarce and, like in almost all countries, not available in Azerbaijan. Only five countries included in a review of road crash cost studies worldwide (Australia, Austria, Germany, the Netherlands and the US) have estimated the number of PDO crashes including a correction for underreporting.71 Table shows the ratio of PDO crashes to the number of fatalities according to these studies72, as well as the proportion of fatalities among motorized road users (that means excluding pedestrians and cyclists) as estimated by the WHO73 and the adjusted ratio PDO-crashes/fatalities using this proportion. We estimate the number of PDO-crashes in Azerbaijan using the median adjusted ratio (490:1) and the number of fatalities among motorized road users in Azerbaijan. The proportion of motorized road users in the number of fatalities in Azerbaijan is 58% according to the WHO and the corresponding number of fatalities is 530 (914*58%). This results in roughly 260,000 PDO crashes. Table 4: Ratio of PDO crashes to the number of fatalities (both corrected for underreporting)74 Ratio PDO Proportion of Adjusted ratio PDO crashes/fatalities motorized road users crashes/fatalities in fatalities (motorized) Australia 274 72% 380 Austria 1,232 86% 1,467 Germany 358 73% 490 Netherlands 1,418 61% 2,325 US 323 83% 389 Median 408 490 Summary 95. Table summarizes the estimated numbers of casualties and crashes which are used for the crash cost calculations. Table 5: Number of road casualties and crashes used in the crash-cost calculations Severity of Injuries Casualties Crashes Fatal 914 1,005 Serious injury (overnight stay in hospital) 5,900 6,800 Slight injury 12,100 16,300 - treated in hospital (without overnight stay) 3,200 - not treated in hospital 8,900 Property damage only - 260,000 71 Wijnen, W. & Stipdonk, H. (2016). Social costs of road crashes: an international analysis. Accident Analysis and Prevention, 94, 97–106. 72The severity of included PDO crashes is not very clear from these estimates. In some countries, most damages were reported to the insurance, which are usually more severe damages, although they may also include, e.g., parking damages. In other countries, a large proportion of crashes was not reported to the police, which could mainly include minor damages. 73 WHO (2018). Global Status Report on Road Safety 2018. World Health Organization, Geneva. 74 Wijnen (2021), WHO (2018) and study calculations 35 | P a g e 3.2. Medical costs Ambulance costs 96. The costs per ambulance ride are estimated in the range 60-150 AZN. This is based on prices stated by emergency departments of hospitals, insurance companies and the 103 dispatch center. The amounts include costs of equipment, personnel, and the vehicle (fuel, depreciation, etc.). The median value is 110 AZN, which is used to calculate the total costs. These are the costs of ambulances of public hospitals, which are mostly used for road casualties. Ambulances of private hospitals are mostly used for transporting patients between hospitals. 97. Table 66 shows the proportions of casualties who are taken to hospital by ambulance, based on the casualty survey. As expected, this proportion increases if injuries are more severe. Also, some casualties who indicated that they were not treated in hospital were transported to hospital by an ambulance for medical checks without further treatment. The total costs of ambulance trips are estimated at 0.7 million AZN, which are mainly related to serious injuries. Table 6: Proportion of casualties transported by ambulance and ambulance costs (source: study survey) Severity of injuries Proportion ambulance Costs (million transportation AZN) Fatalities 87.1% 0.1 Serious injuries 69.1% 0.5 Slight injuries with hospital treatment 31.8% 0.1 Slight injuries without hospital treatment 3.8% 0.04 Total 0.7 Hospital admission with overnight stay 98. According to the survey, the average duration of hospitalization is 24 days (N= 51) for serious injuries. This is on the higher end of the duration as stated by health professionals, who said that the hospital stay of road casualties varies from about 2-7 days for slight injuries up to about one month for severe injuries. Concerning fatalities, 45% was hospitalized with an average duration of 4 days (N=25). This is consistent with the information received from health specialists, who mentioned a hospitalization of 3-5 days. 99. The daily costs of admission to a public hospital is estimated to range from 30-50 AZN per day for traumatology and general surgery/ICU to 50-70 AZN per day for neurosurgery. This includes the costs of medical personnel and materials and medicines. Indirect costs are not included, which we assume to be 50 AZN per day (the lower boundary of indirect cost estimates in private hospitals). One-time cost of diagnostics ranges from 10 AZN (traumatology) to 100-150 AZN (general surgery and neurosurgery). In addition, there are costs of surgery and related diagnostic treatments, which is needed in 30-50% (general surgery and neurosurgery) of the cases up to 50-70% (traumatology). The costs of relatively ordinary surgeries range from 200-500 AZN per surgery (traumatology departments of public hospitals) to about 600-1,000 AZN (general surgery departments and traumatology institutes), including diagnostic treatments. The majority of traumatology patients (estimated at 95% or more) need one surgery. For more complicated surgery, which is needed in about 5% of the cases, the costs are usually 2,000-3,000 AZN. The majority of patients (estimated at 97%) need one surgery. Costs of neurological surgery start at about 1,500 AZN, up to 10,000 AZN. In 36 | P a g e these cases, a higher proportion of patients (estimated at about 50%) need more than one surgery. 100. In private hospitals, the total cost of hospitalization at a traumatology department range from about 400 to 700 AZN per day. This includes indirect costs, which are estimated at about 50-100 AZN per day. Average costs of diagnostic treatments are about 250 AZN. Surgery costs are estimated at about 600-2000 AZN, up to more than 10,000 AZN for complicated surgeries. The costs of neurological surgery start at about 2,600 AZN (including diagnostics). The costs in private hospitals are higher because they are equipped with modern technology, better conditions for the patients and more expensive treatments and medicines. 101. Table7 summarizes the cost information including a best estimate of the average cost per hospitalized casualty (serious injury), assuming hospitalization during 24 days. The average costs for treatment in public hospitals range from about 2,500 AZN (traumatology and general surgery/ICU) to 6,400 AZN (neurosurgery). The costs in private hospitals are much higher: approximately 14,400 to 17,400 AZN per casualty. The daily costs account for a major proportion of the costs, which are around 2,200 AZN in public hospitals for example (traumatology and general surgery/ICU). Surgery at neurosurgery departments is more costly and has a larger share in the costs than surgery at other departments. Based on the interviews with health professionals on the type of injuries of road casualties, it is assumed that roughly a third of the casualties is treated at neurosurgery departments, a third at general surgery/ICU and a third at specialized traumatology hospitals or departments (of which 50% at specialized traumatology hospitals). For private hospitals, it is assumed that a third is treated at neurosurgery and two thirds at traumatology. Furthermore, based on the interviews it is assumed that 75% of the casualties are treated in public hospitals and 25% in private hospitals. The resulting weighted average costs per casualty are 6,700 AZN. Table 7: Costs of hospitalization (Source: interviews with health specialists, study analysis) Costs Public Private Specialized Traumatology General Neurosurgery Traumatology Neurosurgery traumatology departments surgery / ICU Daily costs 80-100 80-100 80-100 100-120 400-700 400-450 (direct and indirect) Diagnostics 10 10 130-160 100-150 250 N.a. general Diagnostics 10-120 10-80 140-150 100-180 70-170 600-700 for surgery Surgery 600-800 200-500 600-900 1,500-10,000 600-2,000 2,000-10,000 Proportion 50-70% 50-70% 30% 30-50% 60% 70% surgery Average 2,600 2,400 2,500 6,400 14,400 17,400 costs Hospital treatment without overnight stay 102. The costs of hospital treatment at specialized public traumatology hospitals without overnight stay range from about 50 to 160 AZN per casualty. This includes direct costs of medical personnel and materials and medicines. At general surgery and neurosurgery departments the costs are higher, 140-240 AZN, because more costly diagnostic methods are needed, in particular CT scans. It is assumed that the indirect costs are about 15% (equal to indirect costs of hospitalization). In private 37 | P a g e hospitals, the costs range from 120 to 360 AZN per casualty (including indirect costs), among others depending on the type of diagnostic treatment (X-ray, MRI, or CT scan). Using the same weights as for hospitalization, the average costs per casualty are estimated at 160 AZN. Follow-up hospital visits 103. According to the casualty survey, slight injuries (treated in hospital without overnight stay) visited the hospital on average 4 times for follow-up treatment or check-ups. For serious injuries, the average number of follow-up visits is 9. This is consistent with the information received from doctors, who stated that slightly injured road casualties (no overnight stay in hospital) visit the hospital for follow-up treatment about once a month during about 2 months. For seriously injured casualties, doctors stated that the number of visits is about two times a month (traumatology) up to 3 times a month (neurosurgery), during 2 to 8 months. The costs per visit range from 20-70 AZN in public hospitals to 40-80 AZN in private hospitals (traumatology). For neurosurgery patients, there may be additional costs for medication and MRI scans (up to about 300 AZN). Based on this information, and the same weights as for hospitalization, the average costs per visit are 50 AZN, the costs per slight injury 200 AZN and the costs per serious injury 450 AZN. Non-hospital care 104. About half of the road casualties visit a general practitioner (46% for slight injuries to 63% for serious injuries), according to the casualty survey. About half of the serious injuries and a quarter of the slight injuries treated in hospital visit a physiotherapist or rehabilitation center (Table 8). The latter proportions are in line with the expert opinions of the interviewed doctors, who stated that about 10-30% of the road casualties treated in public hospitals (traumatology, with overnight stay) visit a rehabilitation center. For private hospitals, this proportion is higher, 50-70%, which may reflect both higher abilities to pay for rehabilitation and higher injury severity. For neurosurgery patients the proportion ranges from 30% to 70%. Only one respondent (serious injury) reported a visit to a psychologist. The average number of visits ranges from 2.4 (slight injuries, general practitioner) to 17 (physiotherapy for serious injuries). 105. It is assumed that the costs of a visit to the general practitioner are 17 AZN, based on a list of standard prices compiled by TABIB (the price of a ‘qualified medical examination’). According to rehabilitation centers, the costs of physiotherapy range from about 20 to 45 AZN per visit, up to 100 AZN for very severe injuries. For the cost calculation an average of 30 AZN per day is used. Table 8: Non-hospital treatment (source: study survey) Proportion of casualties treated Average number of visits General Physiotherapist Psychologist General Physiotherapist Psychologist practitioner /rehabilitation practitioner / rehabilitation center center Serious injuries 63% 60% 2% 6.4 17.0 10.0 (N=52) Slight injuries 55% 23% 0% 2.4 5.6 - with hospital treatment (N=22) Slight injuries 46% 4% 0% 2.4 14.0 - without hospital treatment (N=26) 106. The survey also shows that 16% of the serious injuries (N=51) are admitted to a rehabilitation center with overnight stay, after hospitalization. None of the respondents in other severity 38 | P a g e categories was admitted to a rehabilitation center. According to the rehabilitation centers and experts from insurance companies, the number of days of treatment ranges from about 10 to 25. For the cost calculation we use 18 days of treatment. The costs per day (in-patient) are estimated at 80 AZN per day (50 AZN higher than the costs per day of out-patient treatment). 107. Furthermore, 98% of the serious injuries receive home care. This proportion is 86% and 42% for slight injuries treated and not treated in hospital respectively (source: study survey). On average, serious injuries received home care during 4.3 months and slight injuries during 3.6 months. In the case of serious injuries, 49% of the home care was provided by professionals and the remainder by relatives, friends, or other people. For slight injuries, this proportion of professional home care is 20%. The expenditures respondents made for (professional) home care range from 193 AZN for slight injuries to 1,271 AZN for serious injuries. The study uses these expenditures to estimate the total cost, assuming that most of the costs are paid by the casualties. Furthermore, 50% of the costs of professional home care is used as an indication of the costs of non-professional home care (e.g. forgone income of and time loss of relatives). 108. Based on the above information on non-hospital treatment, the total costs of non-hospital treatment are estimated at are estimated at 11 million AZN. This amount consists of 3.3 million AZN rehabilitation and physiotherapy costs, 1.1 million AZN general practitioner costs, and 6.7 million AZN home care costs. Costs of visitors 109. According to the casualty survey, 78% of the hospitalized casualties (serious injuries) received visitors. On average, they were visited 23 times during their stay in hospital and the visitors stayed for eight hours on average (N=40). Fatalities who were treated in hospital were visited four times with an average duration of seven hours (N=12). The visitors travelled 11 km per visit on average and most of them (69%) travelled by car, while 19% used a taxi, 7% public transport and 6% came by foot. The costs per visit by car and taxi are estimated at about 1 AZN and 8 AZN respectively, and the weighted average costs at 2.25 AZN. 110. To estimate the time costs, the value of time is estimated at 1.7 AZN per hour. This is based on an average gross hourly wage of 3.8 AZN (source: MLSP). Following the World Bank study guidelines, non-working time is valued at 30% of the wage. Based on the average number of working hours per year (1926, source: MLSP), we assume that 20% of the visitor’s time is working time (valued at 3.8 AZN per hour) and 80% is non-working time (valued at 30% of 3.8 AZN per hour). Based on this information, the total costs of visitors (transport and time) are estimated at 1.9 million AZN. House adaptation 111. According to the survey, adaptation to the house such as making it accessible for wheelchairs, was needed for 13% of the serious injuries. The average costs were 327 AZN. Based on these data, the total costs of house adaptations are estimated at 250,000 AZN. 112. The total medical costs are thus estimated at about 40 million AZN (see 113. 114. 115. Table 9). Hospitalization accounts for more than half of the costs. More than 90% of the costs are related to serious injuries. 39 | P a g e Table 9: Medical costs resulting from road crashes (million AZN) Cost item Costs (million AZN) Serious Fatalities injuries Slight injuries Total - Ambulance 0.1 0.5 0.1 0.7 - Hospital treatment without - - 0.5 0.5 overnight stay - Hospitalization 0.3 21.7 - 21.9 - Follow-up visits - 2.7 0.6 3.3 - Rehabilitation/physiotherapy - 3.2 0.1 3.3 - General practitioner - 0.6 0.4 1.0 - Home care - 5.5 1.2 6.7 - Visitors 0.0 1.9 - 1.9 - House adaptations - 0.3 - 0.3 Total 0.3 36.3 2.9 39.5 3.3. Production loss 116. In the case of fatalities, production is lost during the life years that would remain if the person was not killed in the crash. The number of lost productive life years is based on the age of the fatalities and the average retirement age, which is 63.5 for men and 60.5 for women (source: MLSP). Concerning casualties who have not entered the labor market yet, the number of lost productive life years is calculated using the average age of entering labor market instead of age. Table 1 shows the proportion of the population of 15 years or older by education level (source: MLSP) and the assumed age of entering the labor market. The weighted average age was identified as 17.3 years. Table 1: Age of entering the labor market by education level. Education level Proportion of Age entering population (15 years labor market and older) Higher education 13% 22 Completed or specialized secondary 73% 17 General secondary 13% 15 Average 17.3 117. In the case of injuries, the study survey showed that 9.6% of the serious injuries are lifelong disabled and not able to work anymore (Table 2). The calculation of the production loss related to these disabled casualties is similar to the calculation for fatalities. It is assumed that the age distribution of serious injuries (see Section 5.3.1) also applies to disabled casualties. Table 2: Proportions of road casualties unable to work (source: study survey) Serious (N=52) Slight (hospital; Slight (no hospital; All N=22) N=26) (N=100) 40 | P a g e Inability to work None 28.8% 59.1% 80.8% 49.5% Temporarily 61.5% 40.9% 19.2% 45.5% Permanently 9.6% 0.0% 0.0% 5.0% 118. The study survey estimated the duration of absence from work for casualties who are temporarily unable to work (Table 12). To calculate the production loss, the mean durations are used, which is 76 days for serious injuries and 27 days for slight injuries. Given the low numbers of slight injuries, slight injuries treated in hospital and those who are not treated in hospital are not distinguished. Table 3: Duration of absence from work (days) (source: study survey) Serious (N=32) Slight (N=14) Mean 76 27 Median 50 35 Minimum 2 1 Maximum 300 90 119. Average gross wage is used as the indicator for production loss per person, which is 7,370 AZN per year (source: MLSP).75 This corresponds to 3.83 AZN per hour, as the average number of working hours per employed person is 1,926 (source: MLSP). 120. The future production loss is discounted using a discount rate. The discount rate usually applied for the World Bank financed investment projects is 6%, which is also used in this study. 121. The total production loss has been estimated at 151 million AZN (see Table 4). Fatalities and serious injuries account for a major part of this production loss. In the case of fatalities, this is due to the large number of productive life years lost, which is reflected in the high cost per fatality. For serious injuries, the large number of injuries is the main explanation for the large share in the costs. Table 4: Production loss per casualty and total (AZN) Severity of Costs per casualty Total costs (million AZN) Injury (AZN) Fatality 68,111 62.3 Serious injury 13,191 78.4 Slight injury 827 10.0 Total 150.6 3.4. Human costs 122. The human costs related to fatalities in Azerbaijan have been estimated using the value transfer function developed within iRAP: VSL = 70 * (GDP per capita). Based on the GDP per capita in Azerbaijan in 2019, which is 8,149 AZN (Source: World Development Indicators, 14-12-2020), this results in a VSL of 0.57 million AZN. For the sensitivity analysis, the value transfer function is used as 75 Both gross and net wage are used road crash costs calculations in other countries, as well as other indicators such as GDP per employed person. In this study, gross wage is used because it reflects the value of the goods or services someone produces better than net wage. This is related to the fact that indirect costs (e.g. social benefit payments and labor taxes) are included in the value (price) of the goods and services. 41 | P a g e developed by the World Bank study76: VSL = 1.3732E−4 * (GDP per capita)ˆ2.478. The resulting VSL is 1.15 million AZN.77 The first VSL is lower because it is based on (a small sample of) values used by governments in different countries. These values are usually lower than values found in scientific studies on which the World Bank study function is based (see Section 0). 123. Arrival at the human costs requires deduction of consumption loss from the VSL. The calculation of average lifetime future consumption is based on consumption expenditures and demographic data. The consumption expenditures per capita for men and women are 3,397 and 3,602 AZN per year respectively (source: Ministry of Economy). For the calculation of the average lifetime future consumption per person, population data (5-year categories) and life expectancy at birth provided by the Ministry of Labor and Social Protection are used. The calculation is based on single ages and the population is assumed to be equal for each single age within a 5-year category. Furthermore, data on life expectancy by age in Kazakhstan is used, as this information is not available in Azerbaijan.78 Following the calculation steps explained in Section 2.1.5, the average lifetime future consumption per person is 35,800 AZN. This implies that the human costs per fatality are estimated at 0.52 million AZN (1.10 million AZN in the sensitivity analysis). 124. Human costs of serious and slight injuries are calculated as a proportion of the VSL as determined in the international literature. A recent overview of 13 studies shows a wide range of WTP values for preventing a serious road injury relative to the WTP for preventing a fatality, 1.3% up to 47%.79 For slight injuries this proportion ranges from 0.3% to 2.4%.80 Different WTP-methods have been used in these studies. The literature review includes an assessment of these methods using several criteria such as theoretical basis and reliability. This assessment shows that the stated choice method is regarded as the most appropriate method, which is applied in all recent studies. Therefore, the study uses the results of the six stated choice studies included in the review. The median value of a serious injury found in these studies is 17% of the VSL. For slight injuries the study uses the results of all studies, because there is just one stated choice which included slight injuries. The median value of a slight injury is 0.9% of the VSL. These proportions are in line with the standard proportions recommended in European studies, for example, the Handbook of External Costs of Transport and the HEATCO project.81 They recommend 13% and 1% of the VSL for serious and slight injuries respectively. The proportions of 17% and 0.9% of the VSL result in human costs of approximately 97,000 and 5,000 AZN per serious and slight injury respectively (195,000 and 10,000 AZN respectively in the sensitivity analysis using the lower VSL). The study applies the value for human costs of slight injuries only to injuries treated in hospital, which is likely for most examples of 76 Milligan, C., Kopp, A., Dahdah, S. & Montufar, J. (2014). Value of a statistical life in road safety: A benefit-transfer function with risk analysis guidance based on developing country data. Accident Analysis and Prevention, 71, pp. 236-247. 77The function is modelled for the GDP per capita and VSL expressed in 2005 international dollars. To calculate the VSL in Azerbaijan, the study follows the calculation procedure as described by names are missing here in the World Bank paper. Firstly, the GDP per capita in national currency 2019 transferred into 2005 international dollars using the purchasing power parity in 2019 and GDP deflator indices for the US dollar in 2005 and 2019 (source: World Development Indicators, 16-12- 2020). Next, the VSL in international dollars 2005 resulting from the value transfer function has been transferred back to national currency 2019. 78The Kazakhstan data are taken from Wijnen (2021) Socio-economic costs of road crashes in middle-income countries: applying a hybrid approach to Kazakhstan. IATSS Research. The data refer to 2012 and they are adjusted for the fact that life expectancy at birth in Azerbaijan in 2019 is higher (73.3 and 78.2 years for men and women respectively, versus 64.8 and 74.3 years in Kazakhstan in 2012). 79Wijnen, W. (2021). Economic valuation of preventing non-fatal road injuries: a literature review. Presentation at the Society for Benefit-Cost Analysis Annual Conference 2021. 80 In one study a value of 32.1% of the VSL was found, which is clearly an outlier. 81 Van Essen, H. et al. (2019). Handbook of external costs of transport; 2019 version. CE Delft, Delft. Bickel, P. et al. (2006). Proposal for harmonized guidelines. EU project HEATCO Deliverable 5. University of Stuttgart, Stuttgart. 42 | P a g e injuries used in the valuation studies (whiplash, pulse fracture, concussion). Table 5 summarizes the human costs per casualty and the resulting total costs. Table 5: Human costs per casualty and total (AZN) Severity of Costs per casualty Total costs (million Injury (AZN) AZN) Fatality 517,153 472.7 Serious injury 96,973 576.1 Slight injury 5,134 16.4 Total 1,065.2 3.5 Property damage 125. The calculation of vehicle damage is based on the average damage per vehicle, the number of vehicles per crash and the number of crashes (by severity). Table 6 shows the average vehicle damage per vehicle and the number of vehicles per crash based on the survey.82 These survey data do not include non-motorized road users (pedestrian and cyclists), for which the number of vehicles per crash is usually 1. Consequently, the number of vehicles per crash should be adjusted for that. For fatalities, the proportion of non-motorized road users is estimated at 58%. According to the official road casualty statistics, this proportion is a factor 0.74 lower for injuries. Therefore, the study assumes that the proportion of non-motorized road users among serious and slight injuries is 43%. The average costs of a PDO crash is likely to be overestimated, because people whose car was only slightly damaged (dents and scratches) are probably underrepresented in the sample (59% of respondents involved in PDO crashes reported serious damage or total loss). A likely explanation is that people whose car was only slightly damaged may have assumed that they were not the target group of the survey. The cost of average slight damages (dents and scratches) amounted to 374 AZN (N=23).83 Assuming that about 90% of the PDO damages was slight, the average PDO damage is estimated at 600 AZN per vehicle. The resulting average vehicle damage, weighted by crash severity (see Table 15), is, therefore, equals to, approximately, 1,000 AZN. Table 6: Average damage per vehicle and number of vehicles per crash by crash severity source: study survey and estimates) Crash Damage per Vehicles per crash N Vehicles per crash (adjusted severity vehicle (excluding non-motorized for non-motorized (AZN) respondents) respondents) Fatal 5,533 2.1 8 1.5 Serious 3,491 1.9 17 1.4 Slight 3,150 1.9 31 1.4 PDO 600 1.8 1.3 126. According to the statistics of the Compulsory Insurance Bureau, the total amount of damage 82 One respondent involved in a slight injury crash stated that the cost was 25,000 AZN, which is clear outlier which raises the average costs per slight injury crash considerably. This study has excluded this case from the calculation of the average damage. 83 One respondent reported a slight damage of 7,000 AZN. This outlier was removed from the sample. 43 | P a g e payments, related to vehicle insurance, by insurance companies was 80 million AZN. Most of this amount concerns compulsory liability insurance (67 million AZN), while voluntary insurances account for 13 million AZN. These amounts may also include other damage than vehicle damage resulting from road crashes, such as damage from theft and crime. Data from one of the interviewed insurance companies showed, however, that this is a minor part of the total damage. 127. The damage payments made by insurance companies are much lower than the total vehicle damage as calculated above. This is explained by the fact that vehicle insurances do not cover the full damage. In particular, liability insurance only covers third-party damage, but not the damage to the client’s own vehicle. Furthermore, in many cases people may not submit a claim to their insurance company because of no-claim premiums. Instead, the road users involved in the crash may settle the case among themselves. The casualty survey showed that 38% of the respondents (N=199) received a financial compensation for damage resulting from the crash they were involved in. In 60% of these cases, the payment was made by other road users who were involved in the crash, while vehicle insurance companies paid for damage in 29% of the cases. In other cases, non-vehicle insurance companies and the government provided financial compensations. 128. Data from four insurance companies, provided by the Azerbaijan Insurers Association, shows that the average vehicle damage payment amounts to 1,021 AZN per vehicle for compulsory liability insurance and 1,562 AZN per vehicle for CASCO insurance (Table 7). The weighted average is 1,060 AZN per vehicle. This value is in line with the values resulting from the survey, taking into account the adjusted value of PDO damages (see above). Table 7: Vehicle damage as reported by four insurance companies Total amount (1000 Damage cases AZN) Average damage (AZN) Liability Casco Total Liability Casco Total Liability Casco Total Passenger 22,777 2033 24,810 23,069 3,100 26,168 1,013 1,524.76 1,055 cars Cargo 3,073 26 3,099 3,342 3,447 1,088 4,018.42 1,112 vehicles 104 Buses 905 14 919 890 20 919 994 1,402.57 1,000 Trailer 104 2 106 101 12 113 974 5,950.00 1,067 vehicles Other 22 3 25 31 11 42 1,412 3,767.00 1,694 heavy vehicles Motorcycle 3 1 4 1 448 2 382 448.84 398 Total 26,884 2079 28,963 27,444 3,247 30,692 1,021 1,562.10 1,060 129. The total vehicle damage is estimated at 321 million AZN (Table 8). Most of the costs concern PDO crashes, due to their large number. Table 8: Total vehicle damage by crash severity (million AZN) 44 | P a g e Crash Total cots (million AZN) severity Costs per crash (AZN) Fatal 8,147 8.2 Serious 4,873 33.3 Slight 4,305 70.1 PDO 808 209.2 Total 320.8 130. In addition to vehicle damage, there is damage to infrastructure. According to the State Agency of Azerbaijan Automobile Roads, 547 road crashes occurred in 2019 which resulted in damage to infrastructure, including damage to guard rails, lamp posts, information boards and bridge railing. This damage amounts to 680,459 AZN in 2019. Insurance companies paid 62% of this damage. Also, road users pay for the damage they caused to infrastructure. The average cost per crash is 1,244 AZN. While the above statistics is likely to be significantly underreported, the study is using them in final calculations given the lack of other data and following a conservative approach. 3.6. Administrative costs Police and other emergency services 131. The results of the casualty survey show that the proportion of road crashes attended by the police ranges from 64% in case of slight injury accidents to 97% for fatal crashes ( 132. Table 9). The result of the survey that the police attends crashes without any injuries more frequently than slight injury crashes is surprising, but not unusual in relatively small surveys given the insignificant margin of deviation. Table 9: Police attendance by crash severity (source: study survey) Crash Proportion attended by N severity police Fatal 97% 31 Serious 83% 53 Slight 64% 53 PDO 72% 61 133. According to the stakeholder interviews, the time the police officers spend on a crash ranges from about half an hour to three hours, depending on the crash severity. The number of police officers ranges 1-2 in slight cases up to 4-6 for more severe crashes. Based on this information, Table 10 shows the assumed time spending and number of police offers by crash severity. Table 10: Police time spending and number of police officers per crash. (source: assumptions based on interviews with stakeholders) Crash Time spent per crash Number of police officers per severity (hours) crash Fatal 3 6 Serious 2 4 Slight 1 2 PDO 0.5 1 45 | P a g e 134. Based on the interviews with the stakeholders, the study assumes that the average monthly gross wage of police officers is about 500 AZN which translates into about 3 AZN per hour (assuming 40 working hours per week). Information on the indirect costs (equipment, management, administration, housing, etc.) is not available. As an indication, data on the indirect costs of the special risk rescue service of the Ministry of Emergency Situations (see below) are used. According to these data, wages have a proportion of 30% in the total cost which means that that indirect cost are 70% of the total costs. Consequently, the wages are raised by factor 3.3 to account for indirect costs. 135. According the Ministry of Emergency Situations, fire service is needed in 37% of the severe crashes. The study assumes that this proportion applies to fatal and serious injury crashes. Depending on the crash severity, a small (5-6 persons) or large (10-12 persons) team is sent to the place of the crash and they spend 0.5 to 1.5 hour per crash. In the calculations it is assumed that 5 officers spend 45 minutes for serious injury crashes and 11 officers spend 75 minutes for fatal crashes. The average monthly wage is 715 AZN and they work about 300 hours a month (3-day shifts of 24 hours a day), which translates into 2.4 AZN per hour. 136. The special risk rescue service of the Ministry of Emergency Situations also provides emergency assistance. However, the number of road crashes this service was involved in is relatively limited (26 crashes). Given that these costs are negligible as compared to the costs of other emergency services they are not included in the calculations. According to the State Road Police, towing service is needed in about 70% of the severe crashes. For slight injury crashes and PDO crashes proportions of 35% and 10% are assumed respectively. The towing costs per vehicle are estimated at 20 AZN. 137. The resulting emergency services costs per crash and the total costs are presented in Table 11. Total police costs are estimated at about 2 million AZN. Fire service costs are much lower, 235,000 AZN, because fire service is needed in fewer cases and time spending is less. Table 11: Emergency service costs, average per crash and total Crash Costs per crash (AZN) Total costs (1,000 AZN) severity Police Fire Towing Police Fire Towing service service Fatal 161 24 21 162 25 21 Serious 61 22 20 420 152 133 Slight 12 - 10 194 - 156 PDO 3 - 3 867 - 697 Total 1,643 176 1,007 Insurance costs 138. Based on the statistics from the Compulsory Insurance Bureau, the total income of all insurance companies in Azerbaijan from vehicle insurance premiums in 2019 was 138 million AZN. The most of this amount (100 million AZN) relates to compulsory liability insurance. Voluntary insurance accounts for 37 million AZN, 98% of which relates to CASCO insurance and 2% to liability insurance. Data from two insurance companies indicate that the administrative costs of vehicle insurances are 30-35% of the income from insurance premiums. This includes personnel costs, equipment costs, overhead costs, and fees to be paid to the Compulsory Insurance Bureau and the 46 | P a g e Central Bank. This is in line with the data from the Compulsory Insurance Bureau. According to these data, vehicle damage payments amount to 80 million AZN, which is 58% of the income from vehicle insurances. Adding 30-35% administrative costs leaves a small proportion of the income which can be allocated for profits and possibly some other costs. Based on these figures, the total administrative costs are estimated at 45 million AZN (32.5% of 138 million AZN). Judicial costs 139. The survey indicates that judicial consequences are important for crash participants and casualties. With regard to fatal crashes, 68% (N=31) of the respondents caused road crashes were prosecuted, while this proportion is 38% (N=53) for serious injury crashes. For slight crashes these proportions are much lower (slight injury crash: 8%, N=53; PDO crash: 7%, N=61). 140. According to the statistics of the Ministry of Justice, in 2019 the number of lawsuits related to road users who caused a crash was 1,009, 835 of which were proceeded, and 644 people were sentenced. A total number of 275,623 court cases were proceeded in 2019, which means that cases related to road crashes have of a proportion of 0.3% in the total number of cases. 141. The total judicial costs include the costs prosecution, court cases and imprisonment. The costs of the Prosecutor General's Office were 56.4 million AZN (source: Ministry of Justice, database of legal acts84). The costs of the Ministry of Justice for the judicial system was 70.0 million AZN in 2019 (source: Ministry of Finance), consisting of the budget for the ministry itself as well as for subordinate bodies (32.2 million AZN), such as local and other notary bodies, regional justice departments, executive and probation services, state registry service, local and other registration bodies, as well as financial support for courts maintained by the ministry (37.7 million AZN). The costs of penitentiary services of the Ministry of Justice were 127.7 million AZN in 2019 (source: Ministry of Justice, database of legal acts). Based on these costs, the total judicial costs amount to 254.1 million AZN. This includes direct costs of prosecution, court cases and sentencing as well as indirect cost of all organizations involved, including administrative costs, management costs, housing, equipment, etc. The average costs per court case are estimated at 922 AZN (based on 275,623 court cases in 2019). Assuming that this average cost estimate applies to road crash cases, the judicial costs of the 835 cases are estimated at 770.000 AZN. Based on the survey results (see above) and the number of fatal and serious injury crashes the study assumes that fatal crashes account for 16% of the costs and serious injury crashes for 84% (proportions weighted by probabiliy of prosecution and the number of crashes). Judicial costs related to slight injury and PDO crashes are probabliy negligible because of the lower proportion of cases and probably (much) lower costs per case. 3.7 Other costs Funeral costs 142. The costs of funerals depend, among others, on the socio-economic status of the person who died and the number ceremonies, and therefore the costs vary from person to person. 143. Table 12 shows the costs per funeral element as found in research by three news websites.85 Based on the costs it is estimated that the total costs range from about 2,500 to 15,000 AZN (assuming that 30 to 200 person attend the ceremonies; source: study survey). This includes the costs of all traditional ceremonies starting from the first day of the death up to 40 days, as well as the ceremony for a 1 year commemoration of the death, the venue of the cemetery and the price of the 84 http://www.e-qanun.az/framework/41117 85 www.aznews.az/news/cemiyyet/102311.html, news.milli.az/society/250684.html, report.az/din/azerbaycanda-defn- xercleri-arasdirma/ 47 | P a g e grave stone and burial service costs. Table 12: Costs of a funeral (AZN) Components Costs Graveyard worker 50 Tent for funeral ceremony 600-1500 Tea set per person 5 Food per person 10 Hearse 150 Service fee for the 120-150 undertaker Service fee for priest 50 Gravestone 800-2500 Place in the graveyard 63 (official fee), 300-9000 depending on the location 144. According to the survey among relatives of fatalities, the average cost per funeral that were covered by the family amount to 2,772 AZN and the median is 2,913 AZN (N=31). Furthermore, the survey shows that on average three to four ceremonies are organized which are attended by 95 people on average. Based on the above information, the average total funeral costs are estimated at 3,000 AZN. The costs are calculated as the difference between the actual costs of a funeral and the discounted future costs of the funeral if the person was not killed in a road crash (see Section 2.1.5). The discounted future costs depend on the age of the casualty: the younger the casualty, the lower the future costs due to discounting, and the higher the resulting net costs. Using the number of fatalities by age, life expectancy by age (see Section 0) and a discount rate of 6%, the total funeral costs are estimated at 2.0 million AZN. Congestion costs 145. According to the Baku Transport Agency and the State Road Police, road crashes have a considerable impact on the urban transport system due to traffic jams after crashes. The agencies regard road crashes as a major cause of congestion on some of the road arteries in Baku. Data on traffic flows and congestion are collected by the Intelligent Traffic Management Centre using cameras on the road network of Baku, but, unfortunatelty, the data are not suitable for estimating congestion costs resulting from road crashes. Annual systemic data on traffic jams in Baku are not available (nor in other parts of the country), and particulary on the proportion of traffic jams which is caused by road crashes. For that reason, the congestion costs related to road crashes in Azerbaijan cannot be estimated. 146. A review of crash-related congestion costs in four countries showed that congestion costs range from 7 to 92 Euro per motor vehicle (price level 2014, 28-360 AZN price level 2019), or 1-5% of the total road crash costs.86 Two of these countries included environmental impacts, which have a minor share (2-6%) in the congestion costs. Environmental costs 147. The Ministry of Ecology and Natural Resources does not collect systematic information on 86 Source: Wijnen, W. (2021). Costs of congestion related to road crashes in Ireland. Road Safety Authority, Ballina. The figures of a fifth country (Austria) are not included because they are extremely low (1 Euro per motor vehicle, 0.04% of total road crash costs) and not plausible. 48 | P a g e environmental damage due to road crashes. In some cases, the Ministry investigates environmental damage and it reports that five road crashes have been recorded which caused damage to trees in 2019 (15,300 AZN). Information on other damage is not available. According to the Ministry, emissions is the main environmental problem related to road traffic, but there is no information about the proportion which can be attributed to road crashes. 3.8. Total costs 148. Based on the above cost elements, the total cost of road accidents in Azerbaijan in 2019 is estimated at 1.6 billion AZN (Table 22). This is the equivalent of 2.0% of GDP. Human costs have a major share in the total costs (66%,), followed by property damage (20%). Serious injuries account for a large proportion of the human costs (54%) due to the high number of serious injuries as compared to fatalities who account for 44% of the human costs. Concerning property damage, PDO crashes have a large share in the total costs, because the estimated number of PDO crashes is much higher than injury crashes which dominates the effect of lower average damage per crash. Production loss, medical costs and administrative costs are relatively small cost components. Medical costs are mainly related to serious injuries, since medical care and in particular hospitalization is mostly needed for serious injuries. Table 13: Total costs by cost component and injury/crash severity (million AZN) Fatal Serious injury Slight injury PDO Total Injury-related costs Medical costs - Ambulance 0.1 0.5 0.1 - 0.7 - Out-patient - - 0.5 - 0.5 - Hospitalization 0.3 21.7 - - 21.9 - Follow-up visits - 2.7 0.6 - 3.3 - Non-hospital care - 9.3 1.6 - 11.0 - Other 0.0 2.1 - - 2.1 - Subtotal 0.3 36.3 2.9 - 39.5 Production loss 62.3 78.4 10.0 - 150.6 Human costs 472.7 576.1 16.4 - 1,065.2 Funeral costs 2.0 - - - 2.0 Crash-related costs Property damage - Vehicles 8.2 33.3 70.1 209.2 320.8 - Infrastructure 0.0 0.0 0.0 0.6 0.7 - Subtotal 8.2 33.3 70.1 209.8 321.4 Administrative costs - Police 0.2 0.4 0.2 0.9 1.6 - Fire service 0.0 0.2 - - 0.2 - Towing service 0.0 0.1 0.2 0.7 1.0 - Insurance 0.2 1.1 2.6 40.8 44.7 - Judicial 0.2 0.6 0.8 - Subtotal 0.5 2.4 3.0 42.4 48.3 Total 544.0 726.5 102.4 252.2 1,625.1 Figure 8: Proportions of costs components in total costs 49 | P a g e 3.0% 2.4% 9.3% 19.8% 65.5% Medical costs Production loss Human costs Property damage Administrative costs 149. A large proportion of the total costs (45%) is related to serious injuries, while fatalities account for 33% of costs and PDO crashes for 16% ( 150. Figure ). Slight injuries account for only 5% of the costs. This is due to much lower costs per slight injury as compared to serious injuries (in particular costs of hospitalization, production loss and human costs), while the estimated number of slight injuries is only about two times higher than the number of serious injuries. Figure 9: Proportions of severity categories in total costs 16% 33% 6% 45% Fatalities Serious injuries Slight injuries PDO 151. The study has derived the injury-related costs per casualty and the crash-related costs per crash ( 152. Table 14). The injury-related costs range from 587,798 AZN per fatality to 114,338 and 6,186 AZN per serious and slight injury respectively. These costs are dominated by the human costs and therefore they strongly depend on the estimates of the VSL and the value per injury as a proportion of the VSL. Crash-related cost range from 971 AZN per PDO crash to 8,689 AZN per fatal crash. These costs mainly consist of vehicle damage. Administrative costs are relatively low. Table 14: Costs per injury and costs per crash by cost component and injury/crash severity (AZN) Fatal Serious injury Slight injury PDO Injury-related costs 50 | P a g e Medical costs - Ambulance 96 76 12 - - Out-patient - - 160 - - Hospitalization 274 3,648 - - - Follow-up visits - 450 53 - - Non-hospital care - 1,573 126 - - Other 109 355 - - - Subtotal 479 6,102 351 - Production loss 68,111 13,191 827 - Human costs 517,153 96,973 5,134 - Funeral costs 2,164 - - - Total costs per 587,907 116,266 6,311 - injury Crash-related costs Property damage - Vehicles 8,147 4,871 4,302 805 - Infrastructure 2 2 2 2 - Subtotal 8,150 4,873 4,305 808 Administrative costs - Police 161 61 12 3 - Fire service 24 22 - - - Towing service 21 20 10 3 - Insurance 172 163 160 157 - Judicial 160 89 - - - Subtotal 539 356 181 163 Total costs per 8,689 5,229 4,486 971 crash 153. Sensitivity analysis has been conducted with respect to the several input parameters used in the calculations and showed a cost range of 1.6-4.2% of GDP. Figure 10 presents the total costs if other parameters are used, as discussed in this chapter. A relatively low VSL was used in the baseline scenario. If a higher VSL is applied based the World Bank study value transfer function (1.15 instead of 0.57 million AZN), the costs raise to 2.8 billion AZN (3.4% of GDP). Changing the assumptions on the ratio of serious injuries to fatalities lower the costs to 1.3 billion AZN (ratio 4:1 instead of 6.5:1, see Section 0) or increase the costs to 1.9 billion AZN (ratio 9:1). Then the costs are equal to 1.6% and 2.3% of GDP respectively. The number of slight injuries in the baseline scenario is low as compared to other countries. A higher estimate of the number of slight injuries (36,000 instead of 12,100) results in higher costs (1.8 billion AZN, 2.2% of GDP). A combination of a higher VSL and higher number of serious and slight injuries results in a cost estimate of 3.5 billion AZN (4.2% of GDP), which can be considered as a maximum. These results show that the costs are in particular sensitive to the VSL, although also the (assumed) number of serious and slight injuries influences the results. Assumptions on the number of slight injuries does not have significant influence on the costs, which is explained by the relatively low costs per slight injury. The distribution of the cost over cost components does not change significantly in these scenarios. Figure 10: Sensitivity of the total cost to key parameters 51 | P a g e 4,000 3,500 3,000 2,500 2,000 1,500 1,000 500 - baseline higher VSL lower serious higher serious higher slight higher VSL, injuries injuries injuries higher serious and slight injuries Medical costs Production loss Human costs Property damage Administrative costs 4. Human and Poverty Impacts of Road Accidents 154. Road accidents have an enormous impact on human lives in a modern mobility age. Losing beloved persons or getting seriously injured in road accidents change people’s lives dramatically. Being involved in a road crash has major impacts on physical and mental health as well as economic impacts for the casualty, his/her household members as well as other relatives and friends. This chapter discusses the human and poverty impact aspect of road crashes, based on the results of the casualty survey and the in-depth interviews with victims of such accidents. 155. The disproportionate impact of road accidents on poorer households and vulnerable groups is particularly important from social and developmental perspective. The empirical evidence from elsewhere in the world shows bigger effects of road accidents on poorer households because they are often more vulnerable to deprivations and lack safety nets, especially if the sole income- generating source is affected by the accident. For example, the medical treatment of a victim may continue over many years, or even for their entire lifetime. Without a secure and sustainable source of revenue, poor and vulnerable households find it very difficult to overcome the damage and trauma. While the linkage between the road accident outcomes and poverty as well as impacts on socially vulnerable groups are not extensively researched, several recent studies have illustrated the strong relationship between road accidents and poverty. 156. The studies on the experience from different parts of the world highlight various correlations between the road accidents, poverty, and social vulnerability (see Annex 1 for more detailed review of literature). For example, the report on the experience from European countries87 indicates that the road accident injury rate in the UK is almost twice as high in the deprived areas compared to the most affluent areas. The difference in the injury incidence rate is particularly large with respect to pedestrian injury. Findings from Norway show that despite the fact that people with higher social status tend to travel more they use much safer vehicles than those used by the poor. A study also reveals that the risk of repeated injuries in Sweden was substantially higher among those with only basic education than among those who had higher education (university degrees). The study from 87 “Social and Economic consequences of road traffic injury in Europe� , European Transport Safety Council, 2007 52 | P a g e India and Bangladesh88 shows that “vulnerable road users accounted for the vast majority of all road deaths and serious injuries to both the poor and non-poor, with pedestrians as the most common casualty�. The study on Myanmar’s experience89 also showed that lower social economic groups are more involved in road crashes compared to higher social economic groups. The findings of the study from Albania90 indicate that road crashes disproportionally affect those in B40 compared to the rest of the population. The study also found that road crashes victims are most likely to be rural poor and vulnerable groups such as pedestrian, cyclists and motorcyclists. The cumulative effect of crashes has the most impact on the bottom 40 percent, thus contributing to poverty. 4.1. The poverty and shared prosperity situation in Azerbaijan 157. According to the official government statistics, 4.8% of the population lived under the poverty line of 180 AZN monthly in 2019. This proportion was 29.3% in 2005, which indicates that Azerbaijan made important progress in terms of reducing poverty. Poverty reduction was driven by strong economic growth, a rise in wages, and social protection measures91. Also, poverty was reduced by high public spending in the construction and service industries92. 158. Despite the recent declines in poverty and improvements in shared prosperity, many low- income households can still be classified as vulnerable. These households accounted for 65 percent of the population in 2012.93 This means that the vulnerable exceeded the size of the bottom 40. Within the B40 population in Azerbaijan, more households are vulnerable (34%) than poor (5%). 159. Although the employment situation in Azerbaijan is considered to be reasonably good94, many people are working in low-paid sectors. According to the International Labor Organization, 44% of the active population works in the informal economy95. These are mostly low-educated people who work, for example, at markets or own a small business, or are seasonally employed in the construction sector.96 Other sectors of economy which employ people are agriculture (36%), trade (14.5%), education (7.8%), construction (7.3%), public administration and defense (5.8%) and others97. Only 1% of the workforce holds jobs in the well-paid oil sector, which generates about half of gross domestic product. 160. The impact of various factors contributing to poverty in Azerbaijan, and particularly road accidents, is not well researched. Insight into these impacts and better understanding of critical linkages is expected to raise awareness of the issue, help design policy measures, and provide extra 88 “The involvement and impact of road crashes on the poor: Bangladesh and India case studies� Ms A Aeron-Thomas (TRL), Dr G D Jacobs (TRL) Mr B Sexton (TRL), Dr G Gururaj (NIMHANS), and Dr F Rahman (ICMH), 2004. 89 Thwe, Ph., Kanitpong, K., Jiwattanakulpaisarn, P, 2013, "Road Crashes and Poverty in Myanmar: Yangon Case Study", Proceedings of the Eastern Asia Society for Transportation Studies, Vol.9, 2013. https://socialscienceresearch.org/index.php/GJHSS/article/view/2771/2660 90 ASSESSING THE SOCIAL COSTS AND BENEFITS OF BETTER AND SAFER ROADS FOR THE BOTTOM 40 PERCENT AND OTHER VULNERABLE GROUPS: THE ALBANIAN CASE, Tirana Institute of Economic Research, 2015 91 World Bank (2009). Azerbaijan: Country Economic Memorandum. 92 ADB (2014). Country Partnership Strategy Azerbaijan 2014-2018. 93 World Bank (2015). Azerbaijan Systematic Country Diagnostic. 94 According to the State Statistics Committee, in 2019 the unemployment rate in Azerbaijan was 4.8%. The unemployment rate in Azerbaijan is defined as the number of unemployed people as percent of the labor force. The labor force includes the people who are either employed or unemployed, i.e. who don't have a job but are actively looking for one. 95 ADB (2014). Country Partnership Strategy Azerbaijan 2014-2018. 96 Guliyev, F. (2015). The informal economy in Azerbaijan. Caucasus Analytical Digest. 97 AZSTAT, 2017 , https://www.stat.gov.az/source/labour/?lang=en 53 | P a g e stimulus to further improve road safety. 4.2. Impact of road accidents on economic wellbeing and poverty 161. The results of the survey among road casualties and in-depth interviews with them revealed significant negative impact of road accidents on economic wellbeing and poverty status of crash casualties. In the case of fatalities, all or a main part of the household income is lost if the person killed in the road crash was the main income earner. A lot of road casualties, and especially those who work in an informal sector, face income loss because they lose the ability to work temporarily or lifelong. Both serious and slight crash injuries frequently have important consequences for the income and livelihood of the casualty households, particularly if the casualty is the main bread earner and the family is financially dependent on the casualty. 162. The economic wellbeing and poverty impacts are reflected in the following findings of the survey. About a third (38% ) of the respondents reported a decrease in income after the road crash they were involved in, including 15% of the respondents who reported a substantial income loss (Error! Reference source not found.11). Relatively more people in the B40 group suffer from income loss (47%) as compared to the U60 group (28%). The proportion of substantial income loss is twice as high in the B40 group (21%) as compared to U60 (10%). The income of a small proportion of respondents (3% in B40 and 8% in U60) increased after the crash. This might be due to influences which are not related to the crash, such as a household member starting new job. The average monthly household income decreased by 17% after the road crash for all interviewed households. The income loss is higher for the B40 group (30%) as compared to the U60 group (12%). Fuad (male, 25 years old) is a courier from Baku. His moped collided with a car on a rural road. He was wearing a helmet so was not seriously injured. However, due to an arm injury he couldn’t carry out his work as a courier for long time, which caused serious problems for maintaining the three members of his family. He describes the consequences as follows: “I cannot explain you how difficult it was not being able to work and earn money. We could only buy basic things like food and got help from our relatives to pay for other things.� 163. The difference between Baku and the rural regions in terms of impact of road accidents on incomes is slight: 36% of the respondents from Baku reported an income decrease (16% substantial) versus 40% in regions (16% substantial). The income loss was 16% in Baku versus 18% in the rural regions. 164. The casualties’ contribution to the household’s income decreased by 27% after crash and this decline is much higher for B40 than for U60 (52% versus 17% decrease). This difference might be related to the fact that B40 households are more dependent on a single wage-earner. Furthermore, 26% of the respondents in the B40 group were in the U60 group before the crash. They fell into the B40 group due to the loss of income, (most probably) as a consequence of the road crash. Figure 11: Impact of road crashes on household income (N=197) (Source: study survey) 54 | P a g e 70% 60% 50% 40% 30% 20% 10% 0% Increased Remained Decreased Decreased Don't know unchanged to some substantially extent B40 U60 All 165. The financial impact of road crashes is also reflected by fact that 29% (N=199) of the households needed to borrow money as a consequence of the road crash (Error! Reference source not found.). This proportion is higher for the B40 group (34%; N=95) than for the U60 group (26%; N= 99). It is also higher in the rural regions (36%) than in Baku (23%). Nazakat (female, 42 years old) was hit by a car while crossing a motorway where there was no pedestrian crossing facility. She was seriously injured, stayed 20 days in hospital and got permanently disabled. She is from a low-income family and she is the main income earner. She receives a disability benefit from the state which is not sufficient to cover the needs of her family. She had to borrow 4,000 AZN from the bank to cover medical expenses and also sold jewelry for the family needs. 166. About a third of the respondents who borrowed, borrowed money from a bank and two thirds from relatives and friends. The latter proportion is slightly higher for the B40 group (69%) than for the U60 group (62%). In the rural regions, relatively fewer people borrow from family and friends (62%) than in Baku (68%). The reasons for borrowing money from relatives and friends may include the lack of a permanent job, lack of collateral and lower (or no) interest rates. 167. Some casualty families sold their assets to cope with the consequences of the road accident. More people in the B40 needed to sell assets as a result of the road crash: 7.4% versus 2.0% in the U60 group. Figure 12: Proportion of people who needed to borrow money (N=199). (Source: study survey) 55 | P a g e 40.0% 35.0% 30.0% 25.0% 20.0% 15.0% 10.0% 5.0% 0.0% B40 U60 All Bank Relatives/friends Other 168. Besides the loss of income, expenditures resulting from road crashes affect the financial situation of households. For almost 30% of the households, medical expenditures had a substantial impact on the household budget. Vehicle damage affected substantially the budget of 26% of the B40 households and 35% of the U60 households (Error! Reference source not found.). This higher proportion for U60 household is likely to be related to the fact that car ownership among U60 households is higher. The road accidents have an impact not only on the casualty and his/her household, but also other relatives and friends. Relatives and friends often provide home care to the casualty which is time consuming and may come at the cost of their own income. The case of Ilhama (female, 35 years old) is a good illustration for such cases. She was hit by a taxi in Baku and got seriously injured. She had serious brain trauma and several fractures. She was not able to work for more than 6 months and lost her income during that period. Her parents had to get long leaves without pay from their works to be able to take care of their daughter also bearing serious financial losses. Figure 13: The impact of crash-related expenditures on household budget (N=194) (Source: study survey) 40% 35% 30% 25% 20% 15% 10% 5% 0% Slight Substantial Slight Substantial Slight Substantial Medical costs Vehicle damage Judicial cost U40 B60 169. The larger financial impact of road crashes on B40 households may also be related to the fact that fewer of these households receive damage compensation payments from insurance companies or other road users involved in the crash: 31% of the B40 households received compensation payments, while this proportion is 46% for U60 households (Error! Reference source 56 | P a g e not found.). Surprisingly, relatively more respondents from rural regions (49%) received compensation as compared to respondents from Baku (32%). Possibly, it is easier to negotiate about compensation in rural areas than in Baku. In Baku, relatively more respondents received compensation from vehicle insurances companies. Relatively very few compensations came from other non-vehicle insurance or the government. B40 respondents received relatively more payments from other road users than U60 respondents received, which may be related to the fact that most of B40 respondents live in rural areas. 170. The average damage amount received is the same for B40 and U60 households (2,253 and 2,234 AZN respectively), but the insurance amounts are lower for B40 households (1,488 versus 1,837 AZN) and the amount received from other road users is higher for B40 households (2,814 versus 2,616). Possible explanations for the fact that fewer B40 households receive financial compensations include a lower education level, lack of higher education and lack of good insurance or inability to claim damage. However, people in the B40 group receive slightly higher compensation amounts from other road users. Figure 14: Proportion of respondents receiving financial compensation (Source: study survey) 60.0% 40.0% 20.0% 0.0% B40 U60 Baku Rural areas Government Both vehicle insurance and other road user Other road user Non-vehicle insurance Vehicle insurance Casualties are often not able to pay for their regular financial bills, such as rent, electricity, taxes, etc. In such cases, financial losses are sometimes born by third parties, as in the example of Adil (male, 31 years old), a motorcycle rider who was hit by a car in the center of Baku and got seriously injured (14 days in hospital, severe pain and serious mobility problems). At the time of the interview he was unable to work for 4 months and he was dependent on friends to pay for basic needs such as food. Also, he was not able to pay his rent and was dependent on the courtesy of the landlord to be able to stay in his home. 4.3. Medical Treatment Impacts 171. Medical treatment related impacts are direct impacts experienced by all road crash casualties getting injured. Medical treatment normally includes hospital treatment (in-patient or out-patient) as well as non-hospital medical care as provided by general practitioners, physiotherapists, psychologists, and rehabilitation centers. Half of the respondents of the casualty survey was injured (non-fatal) and half of them (51 respondents) was admitted to hospital. They stayed in hospital for 24 days in average. In addition, a substantial proportion of the respondents was treated by a general practitioner (57%) or needed physiotherapy (37%). Furthermore, most of the injured casualties (81%) needed home care during four months on average. The home care was mostly provided by family or friends (57%) but also by professionals (39%). Finally, 8% of the respondents were admitted to a 57 | P a g e rehabilitation center with overnight stay and one respondent was treated by a psychologist. Gunel (girl, 10 years old) was hit by a car in Ujar region when crossing rural road and got seriously injured (two weeks in hospital, several fractures, two operations, mobility problems). After the operations, she spent several months bedridden in plaster, suffering a lot from heat in summer months as they did not have air conditioner. She could not go to school for one year and a teacher was assigned by school for home-based tutoring, because they could not afford internet. Her grandmother takes care of her, because mother died, and her father is in prison. Her income from the state (pension and child benefit) is too low to buy basic needs. She gets occasionally help in terms of cash, clothes and other basic goods. The grandmother said: ‘The road crash affected our life very seriously. We cannot afford going to check-ups and other medical procedures. My grandchild does not get the required rehabilitation treatment and physiotherapy and does not get enough calcium and vitamin-rich food and she is underweight. The road crash involved additional costs for us and made us live in a desperate condition’. 172. Medical treatment also affects relatives and friends because they spend time and money on visiting the casualty. Most of the hospitalized casualties in the casualty survey (78%) were visited by them, mostly daily. The visitors stayed from two hours to up to 24 hours a day. 173. The medical treatment had serious financial consequences for the respondents. Respondents who had to pay for medical care, paid 908 AZN on average for home care (50% of the injured respondents paying), 771 AZN for medicines (44% paying), 627 AZN for physiotherapy (16% paying). These expenses constitute significant part of the household budgets. 4.4. Vehicle damage impacts 174. Another direct impact of being involved in a road crash concerns damage to the vehicle. Almost all motorized respondents (not including bus and taxi passengers) reported damage to their vehicle (98%). Most of the damage was serious (57%) and 19% concerned total loss of the vehicle. In 24% of cases the damage to the vehicle was slight (dents and scratches). The costs of repairing or replacing the vehicle ranged from zero (no repair) to 25,000 AZN, with an average of 2.378 AZN. The costs were partly paid by the casualties themselves, but 65% also received compensation payments from insurance companies or other road users involved in the crash. 175. In addition to the direct costs, 31% of the respondents reported financial consequences due to the unavailability of their vehicle. This includes the cost of renting a car or using other transport modes (14%), income loss (4%) and other non-specified cost (14%). The costs of renting a car or using alternative transports, as reported by the casualties, range from 250 to 7,000 AZN (median value 250 AZN), while the income loss ranges from 15 to 4,200 AZN (median value 500 AZN). 4.5. Impact on quality of life 176. In addition to economic impacts (but likely related to) the road accidents also have an impact on quality of life. Quality of life loss primarily relates to the injuries. Casualties suffer from pain and discomfort because of their injuries and the medical treatment. In addition to physical problems, some casualties suffer from mental problems, for example anxieties. In the case of fatalities, the grief and sorrow of losing a beloved person also impacts the quality of life of the family members, relatives, and friends. 58 | P a g e A seriously injured casualty (Amina, female, pedestrian, 35 year old) told that she had both physical and mental traumas after the road crash and that it impacted her life for a long time. After she recovered, she was afraid of crossing streets alone and was running away from cars. 177. Loss of quality of life is also related to the indirect impacts of getting injured. The financial impacts discussed above also affect the quality of life. The loss of income, due to inability to work or the loss of a household member who earned (part of) the household come, may imply that road casualties and their households cannot afford the goods and services they were used to buy before the crash, or even not be able to fulfill their basic needs. They may become (partially) dependent on relatives and friends, financially or with respect to home care and other types of assistance, which may have an additional negative impact on the quality of life of the casualty as well as the people who take care of them. 178. The injuries lead to inconveniences related to changes in daily life order. For example, the survey shows that more than half of the injured respondents experience problems with carrying out work and leisure activities (Error! Reference source not found.). Most respondents who reported problems with jobs, were temporarily unable to work (90%) and 10% of them are lifelong unable to work. Figure 15: Proportion of respondents reporting problems with daily activities (Source: study survey) 70% 60% 50% 40% 30% 20% 10% 0% Job Household Education Sports Leisure work activities 179. Overall, more than half of the road casualties reported problems related to jobs, while 37% and 10% had problems with household work and education respectively (Error! Reference source not found.). Given the fact that 35% of the respondents were not employed and 11% of the respondents were school students, this means that almost all respondents had problems related to carrying out their daily activities. Table 15: Proportion of respondents reporting problems with daily activities (N=101) (Source: study survey) B40 U60 Total Job 58.8% 50.0% 54.5% Household 43.1% 30.0% 36.6% work Education 15.7% 6.0% 10.9% 59 | P a g e 180. Because of their physical limitations, people may also need to change their traditional life space and adapt their living conditions. This includes adapting their houses (reported by seven respondents) or moving to another house because of their disabilities (reported by one respondents). Also, casualties might need to move to another job due their injuries. This was reported by one respondent. A slightly injured moped rider (Fuad, male, 25 years old) said: “Mentally I was so down, not because of my injury, but because of the hard situation that my family was passing through. I couldn’t stay without work for a long time so with my injury I took orders as a photographer and worked with my injured arm.� 181. The impact on quality of life is confirmed by the casualty survey, as shown by Error! Reference source not found.6. 44% of the respondents reported that their quality of life worsened after the road crash, including 12% who reported a large quality of life loss. Substantially more people in the B40 group face quality of life loss (57%) as compared to the U60 group (31%). In particular the number of respondents who reported that their quality of life worsened a lot after the road crash is higher in the B40 group (19% versus 6% in the U60 group). The number of households belonging to the bottom 40 who assessed their quality of living conditions as poor has significantly increased after crash (32.7%). Figure 16: Impact of road crashes on quality of life (Source: study survey) 80.0% 70.0% 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0% Worsened a lot Somewhat No change Improved worsened B40 U60 All 182. In both the B40 and U60 the proportion of respondents with a poor or very poor quality of life increased considerably after the road crash, from 7% to 33% and from 2% to 11% respectively. The impact on quality of life is further illustrated by Error! Reference source not found.7. The large majority of the respondents in the B40 group (92%) assessed the quality of their living conditions before crash as satisfactory or better (good, excellent). This number has decreased to 67% after the crash. In the U60 group this decline is much smaller: from 98% to 88%. In particular, the number of B40 respondents who reported a good quality of life has declined after the road crash (from 32% to 13%). Figure 17: Impact of road crashes on quality of life, B40 versus U60 (Source: study survey) 60 | P a g e 70% 60% 50% 40% 30% 20% 10% 0% Before After Before After Before After Before After Before After Excellent Good Satisfactory Poor Very poor B40 U60 183. Relatively more people in the rural regions (45%) reported a quality of life loss as compared to Baku (38%). However, the quality of life loss in Baku is stronger than in the rural regions: 16% of the respondents from Baku reported that their quality of life worsened a lot, versus 9% in the rural regions. The proportion of respondents in each of the categories ‘excellent, good and satisfactory’ declines in Baku after the crash, while in the rural areas the proportion of respondents reporting a ‘satisfactory’ quality of life increases. In particular respondents who reported a very poor quality of life after the crash are from Baku (Error! Reference source not found.8). Figure 18: Impact of road crashes on quality of life, Baku versus rural areas (Source: study survey) 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0% Before After Before After Before After Before After Before After Excellent Good Satisfactory Poor Very poor Baku Rural areas 4.6. Impacts on vulnerable groups 184. As discussed in earlier sections, the impact of being involved in a road crash is likely to be more severe for some vulnerable groups. This section concentrates on assessment of the impact on household income and quality of life for four groups: larger households (five or more members), elderly (above 60 years old), lower educated people (general secondary or lower) and IDP households. Larger households 61 | P a g e 185. Respondents from larger households did not report a disproportionate impact on household income than smaller households (Figure 9). The average income of smaller households decreased by 17.4% versus 16.0% for larger households. Also, the impact on quality of life does not differ substantially. The proportions of respondents reporting an excellent or good quality of life decreased at the same rate for larger and smaller households (from 50.0% to 34.1% for larger household and from 49.9% to 34.3% for smaller households). Figure 19: Impact of road crashes on household income by household size. (Source: study survey) 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0% Increased Remained Decreased Decreased unchanged to some substantially extent Larger households Smaller Households 186. The proportion of respondents reporting poor or very poor quality of life increased substantially for both smaller and larger households: from 6.8% to 29.6% for smaller households and from 4.0% to 19.1% for larger households ( 187. Figure 420). Although this proportion increased at a slightly higher rate for smaller households, the absolute percentage increase of the number of respondents reporting poor or very poor quality of life is higher for larger households. This means that the impact in terms of the number of people reporting poor or very poor quality of life is relatively stronger for larger households. Figure 40: Impact on quality life by household size. (Source: study survey) 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0% Before Before Before Before Before After After After After After Excellent Good Satisfactory Poor Very poor Larger households Smaller households Lower educated people 188. Survey did not reveal a stronger income loss for lower educated people (Figure 5). The income of lower educated people decreased by 17.4% versus 16.9% for higher educated people. However, this income loss can be more problematic for lower educated people because their income is lower (on average 947 AZN versus 1,410 AZN for higher educated people, according to the casualty survey). Figure 5: Impact of road crashes on household income by education level. (Source: study survey) 62 | P a g e 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0% Increased Remained Decreased to Decreased unchanged some extent substantially Lower education Higher education 189. Lower educated people reported a lower quality of life (Error! Not a valid bookmark self- reference.2) and the proportion of respondents reported poor or very poor quality of life increased from 6.9% to 32.9%. Although this proportion for higher educated people increased at a about the same rate (from 3.3% to 14.6% poor or very poor), the number of people reporting poor or very poor quality of life increased more in terms of percentage points for lower educated people. Figure 6: Impact of road crashes on quality life by education level. (Source: study survey) 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0% Before After Before After Before After Before After Before After Excellent Good Satisfactory Poor Very poor Lower educated Higher educated Elderly 190. Casualties of 60 years old and above faced a smaller income loss than younger casualties (13.1% versus 17.5% income loss). However, a larger proportion of elderly reported an income loss (55.5% versus 36.4% for people below 60 years old). The average income loss is lower because more elderly reported an income increase (Figure 73). Figure 73: Impact of road crashes on income by age category. (Source: study survey) 63 | P a g e 70.0% 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0% Increased Remained Decreased to Decreased unchanged some extent substantially Below 60 60 and above 191. None of the elderly reported a poor or very poor quality of life before the road crash, but after the road crash the proportion is 15.8% (Figure 84). For respondents under 60 years old, the proportion of people with poor or very poor quality of life increases from 5.1% to 22.0%. The proportion of people reporting a good or very good quality of life decreases at about the same rate for both groups. Figure 84: Impact of road crashes on quality of life by age category. (Source: study survey) 70.0% 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0% Before After Before After Before After Before After Before After Excellent Good Satisfactory Poor Very poor 60 and above below 60 IDP households 192. The income loss of IDP respondents is smaller than for non-IDP respondents (10.1% versus 18.3% decrease). Fewer IDP respondents reported a substantial income loss (4.4% versus 17.0% for non-IDP households) and more of them reported an income increase ( 193. Figure 5). Figure 25: Impact of road crashes on income, IDP and non-IDP. 64 | P a g e (Source: study survey) 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0% Increased Remained Decreased to Decreased unchanged some extent substantially IDP non-IDP 194. Concerning quality of life, more IDP respondents reported a poor or very poor quality of life before the crash as compared to non-IDP families (Figure 9). This proportion increased both for IDP and non-IDP respondents, but at a lower rate for IDP respondents (from 8.7% to 26.1% versus from 4.0% to 20.8% for non-IDP respondents.). Figure 9: Impact of road crashes on quality of life, IDP and non-IDP. (Source: study survey) 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0% Before After Before After Before After Before After Before After Excellent Good Satisfactory Poor Very poor IDP non-IDP 4.7. Gender aspects 195. The number of male road casualties in Azerbaijan is much higher than the number of female casualties. 78% of the road fatalities and 73% of the injured casualties are men. A high proportion of male casualties is a common international phenomenon. For example, 74% of the road fatalities in the European Union are men. European research shows that a large proportion of male fatalities are car drivers, while female fatalities are mostly car occupants and pedestrians.98 These findings are also reflected in the survey sample of the 200 road casualties in Azerbaijan. 59% of the female casualties were pedestrians, while 32% of the male casualties were pedestrians. 54% of the male 98 ETSC (2013). Back on track to reach the EU 2020 Road Safety Target? 7th Road Safety PIN Report. European Traffic Safety Council, Brussels. 65 | P a g e casualties were car drivers or occupants versus 31% of the female casualties (Table 27). Table 27: Proportion of survey respondents by transport mode and gender (source: study survey) Male (N=148) Female (N=51) Car 54% 31% Motorcycle 3% 0% Moped 1% 0% Car/Van passenger 2% 4% Taxi passenger 0% 4% Bus passenger 1% 2% Bicycle 7% 0% Pedestrian 32% 59% 100% 100% 196. A likely explanation for the higher proportion of male casualties is the fact that men behave riskier in traffic. Studies have shown that men are more prone to speeding, driving under influence of alcohol and drugs and violating other traffic rules. The higher proportion of male casualties is also related to the fact that men travel more.99 Women are found to be more vulnerable to injuries as car drivers or occupants. For example, women have a higher risk of getting whiplash injuries than men. One of the reasons of this is that the crashworthiness of cars is mainly based on males. 100 197. The study survey shows that a larger proportion of male casualties report problems with jobs, while relatively more women have problems with doing household work (Table8). This is related to the fact that a larger proportion of the male respondents were employed (68%, versus 40% for female respondents). Income loss poses financial problems on the household and forces the women or other members of the household to work more to compensate the income loss. According to the survey, 80% of the injured casualties need home care. In many cases females spend more time on providing home care for the injured persons in the household. Also, since most road casualties are men, this burden is likely to be borne mostly by women. The need to provide home care and to earn income can put a large pressure on females and limits their capabilities to conduct other duties such as taking care of the children, household activities and other activities. Table 28: Proportion of respondents reporting problems with daily activities (source: study survey) Men (N=71) Women (N=30) Job 61% 40% Household work 25% 63% Education 10% 13% Sports 0% 0% Leisure activities 58% 60% 198. The need to provide home care may imply that other household members, particularly women, have less opportunities to work. This difficulty to compensate income loss further complicates the financial situation. In some cases, household members need to take a leave from their job to be able to provide home care. Concerning male casualties, in 3% of the cases a household member took leave from a job, according to the casualty survey. This proportion is higher in case of female casualties (10%), which may reflect that the employment rate of women is lower and so they don’t need to take a leave. However, the need to provide home care in addition to their household 99 See ETSC (2013) and Turner, J. & Fletcher, J. (2008). Gender and Road Safety. TI-UP Enquiry. 100 See ETSC (2013) and, for example, Linder, A.. Carlsson, A., Svensson, M., & Siegmund, G. (2008). Dynamic Responses of Female and Male Volunteers in Rear Impacts. Traffic Injury Prevention v9 n6 (20081201): 592-599 66 | P a g e work and other duties puts an extra pressure on them (as discussed above). Nazakat (female, 42 years old) was seriously injured as a pedestrian when she was hit by a car. The driver took her to the hospital himself. She had nose and leg fractures and was not able to walk for several months. She was not able to take care of her family during this period. Her sister took care of her and her family. Her son and husband are disabled, so it was very hard for her sister to take care of three disabled people. The family also ran into financial problems, because Nazakat is the income earner of the household and they had to cover the high costs of her medical treatment and medicines. 199. More male respondents reported a household income decrease due to the road crash. There are, in particular, more substantial income losses among male respondents ( 200. Figure 7). However, in households where the man is the main income earner, these financial impacts will affect all members of the household. Possibly other household members, in particular women, need to spend time on home care for the casualty, which puts an extra (financial) pressure on the household. Figure 27: Impact of road crashes on income by gender (Source: study survey) 70% 60% 50% 40% 30% 20% 10% 0% Increased Remained Decreased to Decreased unchanged some extent substantially Male Female 5. Discussion and policy implications 201. This chapter discusses the results of the socio-economic costs and poverty assessments, in particular with respect to policy implications. The study results will be reviewed in the context of international evidence on the socio-economic costs of road crashes. The issue of data collection and availability will be also discussed. Finally, the relevance of study results for policy making will be 67 | P a g e discussed and recommendations made from several perspectives, including the implementation of the State Program on Road Safety. 5.1. Implications of the study results for policy making 202. In this study the socio-economic costs of road crashes in Azerbaijan are estimated at 1.6 billion AZN in 2019, which is equal to 2.0% of GDP.101 About 65% of these costs are the intangible costs of losing years and quality of life (human costs) and the other 35% are economic costs in terms of damage to vehicles, loss of productivity, medical and administrative costs. This implies that road crashes pose a very considerable burden on the country, both from a human point of view and from an economic perspective. 203. The large size of the socio-economic costs implies that preventing road crashes will result in large socio-economic benefits (cost savings). Investments in road safety measures are needed to materialize these benefits, which requires, among others, a good organization of road safety management. An important step towards improving road safety in Azerbaijan was taken by adopting the first strategic road safety plan (State Program on Road Safety) in 2018. The program sets the objective to reduce the number of fatalities and serious injuries by 30% in 2023 as compared to 2017. According to the study, this is expected to result in a yearly socio-economic cost reduction of about 500 million AZN (assuming that slight injuries and PDO crashes will be prevented proportionally). This study has shown that serious injuries account for a major share (45%) of the total costs, which means that preventing serious injuries can yield large benefits. Including a target for reducing the number of serious injuries, as has been done in the State Program, is therefore important from a socio- economic and human point of view. 204. The study shows that road crashes have strong financial impacts on households, in particular poor households. About a third (38%) of the respondents reported an income loss after the crash, including 15% substantial income loss, and the households lost 17% of their income on average. In addition, important negative financial impacts were identified due to increased expenditures of casualties, such as medical expenditures and recovery of vehicle damage. These financial impacts were found to be much stronger for the B40 group, which means that road accidents have significant negative impact on shared prosperity. Thus, almost half of the B40 households suffered from income loss and their income loss averaged 30%. This might partly be related to the fact that the people belonging to the B40 category are more frequently make their earnings out of physical labour and road crashes primarily damage physical capabilities of people. Road crashes were also found to have a substantial impact on quality of life. The household income of about a quarter of the respondents went below the threshold for the B40 population due to the road accidents and almost half of the respondents reported that their quality of life worsened after the road crash, including 12% who reported a large quality of life loss. Substantially more people from B40 households face quality of life loss (57%) as compared to U60 households (31%), which is likely to be related to their financial vulnerability. 205. The impacts on poverty and shared prosperity have important implications for road safety policy making. They suggest that improving road safety will not only yield socio-economic benefits at the country level, but also help avoid financial challenges at the household level, prevent households from getting into a poverty situation and deteriorating quality of life. For example, the survey results imply that reducing the number of fatalities and serious injuries by 30%, as aimed for 101 The fact that the costs are equal to 2.7% of GDP does not mean that the GDP would be 2.7% in the (hypothetical) situation without road crashes. This is because the welfare economic perspective of the cost assessment is different than the perspective of GDP (as discussed in Section Error! Reference source not found.). The comparison with GDP is made to scale the size of the costs and to make comparisons with other countries. 68 | P a g e in the State Program on Road Safety, could prevent about 800 households from financial constraints or falling into poverty. Meanwhile, the poverty dimension is often not taken into account in road safety policy making and, hence, increased awareness of this dimension should be built thus creating an additional motivation for road safety policy makers to invest more in improving road safety. Moreover, road safety programs could specifically focus on implementing measures aimed at improving the safety of vulnerable road users. For example, at least 40% of the road fatalities in Azerbaijan are pedestrians, and poor people are likely to be overrepresented among these fatalities as a larger proportion of them cannot afford motorized vehicles. Road safety policies could also concentrate more on areas where the concentration of poor people is higher (e.g. rural areas) and/or where the poor are more at risk. Identification and designing of such policy measures are recommended for further research. 206. The study findings on the negative economic impacts of road accidents on casualties have implications for social and poverty reduction policies. These policies may concentrate on providing income loss compensation for both the casualties and the relatives who provide care for them, or compensation for medical expenditures. This study shows that the financial impacts are felt stronger by poorer households and that they have more limited access to financial compensations. As a consequence, a larger proportion needs to borrow money (34% of the B40 households) or sell assets (7%), which may relieve the financial problems in the short term but worsen the situation in the longer term. For some casualties, the costs of medical treatment may be an obstacle to get appropriate medical care, and social policies may concentrate on removing this barrier. Furthermore, improving access to insurance could help to reduce the financial impacts on the poorer households. 207. To obtain the potential socio-economic benefits, road safety policies need to be implemented, which requires sufficient funding and efficient spending of the available budgets. This is raising an important question about the level of investments needed to achieve the targets set by the State Road Safety Program for 2023 (and later). To estimate the required budgets, cost-effectiveness analysis of road safety measures is critical. Cost-effectiveness analysis allows identification and selection of measures to save road casualties among competing road safety projects, which ensures optimization of the total budget needed to achieve the targets. Such an analysis requires data on the costs and effectiveness of the road safety measures to be implemented in terms of casualty reduction. 208. A cost-effectiveness analysis was not part of this study, but some indications of the required road safety budget for Azerbaijan can be provided based on the international experience. Within iRAP (International Road Assessment Program), calculations have been made on the investments in safe infrastructure aimed at upgrading the road network.102 iRAP uses a star-rating system, ranging from 1 (unsafe) to 5-star (safest) roads. An assessment was made on reaching the UN global road safety target of 75% of travel on 3-star roads. According to the iRAP calculations for Azerbaijan, an annual investment of 0.33% of GDP (270 million AZN in 2019) would result in a 33% fatality reduction (282 fatalities, taking a number of 845 fatalities in 2016 as the baseline, based on WHO estimates). This investment is about 20% of the budget for road investments in Azerbaijan (1.4 billion AZN in 2020)103. Furthermore, iRAP calculated a benefit-cost ratio of 3, which means that the benefits of this investment exceed the costs by a factor 3. It should be noted that these calculations are based on international rules of thumb on the effectiveness of road safety improvements (infrastructure and speed measures) and economic valuations. Therefore, the results of the iRAP should be regarded as rough indications and detailed country-specific assessments are needed for more precise estimates of the required budgets to meet the road safety targets in Azerbaijan. 102 https://www.vaccinesforroads.org/business-case-for-safer-roads/ 103 Source: Ministry of Finance 69 | P a g e 209. The following example from Netherlands provides further insight into the issue. The exercise was undertaken for calculation of the investment needed to reduce the number of fatalities in the Netherlands from 613 in 2017 to zero in 2050, based on cost-effectiveness of improving infrastructure, enforcement and campaigns and increasing vehicle safety.104 The required annual investments were estimated to be at least 480 million Euro (340 million AZN, based on PPP). The total required investments over the period 2020-2050 were estimated at 15 billion Euro (11 billion AZN), which is in the same order of magnitude as the annual socio-economic costs of road crashes (17 billion Euro). 210. The above estimates indicate that, roughly speaking, a few hundreds of million AZN are needed annually to improve road safety in Azerbaijan and to meet the targets set by the government. Still these costs are likely to be much lower than the expected benefits. 5.2 Results of the study in international perspective 211. The size of the road accident costs as conservatively estimated in this study is in line with the costs identified for other countries. This is shown in the 212. Figure 28 , which includes the costs as found in an international review and a study for Kazakhstan. In a review of the official cost estimates used by the national governments in 31 European countries, costs were found to range from 0.4% to 4.1%. Figure 28: Road crash costs proportion of GDP105 213. Regarding the distribution of the costs over severity categories, the most remarkable difference with other countries is the low proportion of slight injuries (Figure 29). PDO crashes have a smaller share in the costs, whereas the proportion of fatalities and serious injuries is a bit higher than in other countries. The latter might be related to the underestimation of the costs of slight injuries. Figure 29: Proportions of severity categories in total costs106 104 Wijnen, W. (2019). Costs of Vision Zero in the Netherlands. Summary. W2Economics, Utrecht. 105 Wijnen, W. & Stipdonk, H. (2016). Social costs of road crashes: an international analysis. Accident Analysis and Prevention, 94, 97–106 & Wijnen, W. (2021). Socio-economic costs of road crashes in middle-income countries: applying a hybrid approach to Kazakhstan. IATSS Research 106 Wijnen & Stipdonk (2016) and results of this study 70 | P a g e 214. The high proportion of human costs in the total costs is in line with estimates in other countries if a WTP method is used. Figure30 shows the distribution over cost components in high income countries, separately for all countries included in the source study and those using a WTP method, low and middle-income countries (none of which used a WTP method) and Azerbaijan. Clearly, human costs are relatively low if the (human) costs are not based on WTP. The proportion of property damage is in the same order of magnitude as in other countries, while the proportions of medical costs, production loss and administrative costs are relatively low. Figure 30: Proportions of cost components in total costs107 5.3. Data availability and data collection 215. As discussed above, an assessment of the socio-economic costs of road crashes requires a large variety of data, including detailed road safety statistics and other data such as medical, demographic, economic, emergency service, insurance, and judicial data. In many cases, the available general statistics do not provide enough detail and specific information related to road crashes, as needed. Consequently, most of the required data is not readily available. This applies not only to Azerbaijan, but also to all low and middle-income countries as well as many high-income 107 Wijnen & Stipdonk (2016) and results of this study 71 | P a g e countries. In-depth research is needed to collect the data needed for road crash costs assessments, which has been done in this study by conducting a detailed survey among road casualties and by conducting interviews with all relevant stakeholders. The casualty survey provided detailed data on road crashes and injuries and on the impact on daily life, personal financial situations, livelihood, quality of life and poverty, while the stakeholder interviews provided details on additional aspects such as medical costs, emergency costs and damage to infrastructure. 216. Despite the extensive data collection efforts under this study, several data gaps remained. This in particularly concerns road safety data. The State Road Police collects basic data on road crashes and casualties, such as the (reported) number of fatalities and injuries by gender and age category and the number of crashes by type of crash and road type. A main limitation of these data is underreporting, which is a well-known issue in most countries and particularly low and middle- income countries. The police is not able to report all crashes, for example because in many cases the police is not called and does not come to the crash location. This applies particularly to lower severity crashes, although a comparison with data from other countries indicates that also the number of serious (non-fatal) injuries is likely to be largely underreported. Another data limitation is the fact that detailed data on road crashes, which are needed for the cost calculations, are not available in Azerbaijan. This concern, for example, data on the severity of road crashes, the number of slight injuries and PDO-crashes and the number of persons and vehicles involved in a crash. Some of these issues will, hopefully, be resolved within the ongoing modernization of the crash data collection system in Azerbaijan. 217. To cope with the lack of data and to avoid underestimating the socio-economic costs, the official statistics have been supplemented with data from other, as much as possible, similar countries (value transfer). Obviously, the value transfer approach introduces uncertainties and therefore the results should be interpreted as best approximations given the data limitations. In general, conservative assumptions were made, for example with respect to the estimation of the number of serious and slight injuries and the human costs (value of a statistical life). For that reason, a sensitivity analysis was conducted to indicate the results if higher values are used, which are still very plausible given the international evidence. 218. Road safety statistics of good quality are not only essential for cost calculations, but also for road safety research and policy making in general. Therefore, it is recommended to explore opportunities to improve road safety data collection, as per the actions established by the State Road Safety Program for 2019-2023. Concerning underreporting, linking police data with hospital data has proved to be a successful approach to mitigate underreporting. 219. Information on human costs in Azerbaijan is one of the critical missing elements, which have a major share in the total costs. Within the scope of this study, it was not feasible to establish country-specific estimates for Azerbaijan, as this requires a complex and well-designed willingness to pay study. Conducting such a study is recommended for future research. 5.4. Key policy recommendations 220. The results of the study can be used for several policy purposes. First, they can help to raise awareness about road safety as a major socio-economic and poverty issue. The socio-economic costs of road crashes in Azerbaijan are huge and road accidents are an important contributor to poverty. The study results show that improvements in road safety are needed and that the gains from preventing road crashes and injuries are expected to be very large in terms of socio-economic cost savings and poverty reduction. It is recommended to use the results of this study in road safety and poverty reduction policy discussions and strategic policy documents to emphasize the importance of road safety improvements. Furthermore, the evidence on the socio-economic costs and poverty 72 | P a g e impacts can be used in communication, education, and public campaigns on road safety, to increase the awareness of road safety professionals as well as road users on the impacts of road crashes. 221. The socio-economic costs and poverty impacts of road crashes can be used as a high-level outcome indicator for road safety management. The road safety management pyramid provided in 222. Figure 311031 distinguishes between institutional management functions, interventions and results. The primary result of road safety management is reduction in the number of casualties, which translates into reduced socio-economic cost. Therefore, socio-economic costs are included in the top of the pyramid as a high-level road safety outcome indicator. Using socio-economic costs as an indicator for policy making implies that cost calculations should be performed on a regular basis, for example along with the publication of the yearly road safety statistics. A relatively simple calculation method could be developed to calculate the costs on a yearly basis, by updating the costs as presented in this report on the basis of several key indicators such as the number of fatalities and serious injuries, GDP increase and inflation. In that way, this indicator serves to monitor road safety performance in terms of socio-economic cost savings. Similarly, poverty reduction can be used as an outcome indicator of road safety policy making (which is not yet included in the pyramid). For that purposes, an indicator such as (the reduction of) the number of people below a poverty line due to road injuries could be developed. Figure 3110: Road safety management structure108 223. The socio-economic costs and poverty reduction are recommended to be used as indicators to monitor the progress made through the implementation of the State Program on Road Safety. The impacts of the state program on the number of fatalities and injuries can be translated into cost reductions and poverty reduction, using the findings of this study. In addition to monitoring the overall progress made on preventing road casualties, the impact of specific road safety measures can be monitored. This requires monitoring and evaluation studies aimed at the effectiveness of the measures with respect to preventing road casualties. This monitoring and evaluation will provide more detailed information on the (type of) road safety measures that are most effective to reduce 108 Source: Bliss, T. & Breene, J. (2009). Country guidelines for the conduct of road safety management capacity reviews and the specification of lead agency reforms, investment strategies and safe system projects. World Bank, Washington. 73 | P a g e the number of casualties and on the groups of road users who benefit from the investments (including vulnerable road users). 224. The study results can be used for planning of government budget allocations. The study provides evidence that allocating (more) budget to road safety will result, if spent on (cost-) effective programs, in major socio-economic cost savings and poverty reduction. Several studies have shown that the socio-economic benefits of road safety investments (largely) outweigh the costs, in particular due to the high socio-economic cost (and thus opportunities for cost savings).109 This is likely to apply to Azerbaijan as well, given the high socio-economic costs and the high fatality rate as compared to high-income countries. This evidence provides arguments for increasing budgets for road safety improvements. In addition, allocating more budget to road safety measures that are specifically aimed at reducing the crash risk of the poor (for example aimed at vulnerable road users and rural regions, see Section 6.1) would help to reduce poverty related to road crashes. With respect to selecting road safety measures to be implemented, it is important to assess the risks of getting injured or killed in a road crash (by location, road type, transport mode, etc.), the related needs of road users to reduce these risks and the effectiveness of road safety measures. All these would ensure that government budgets are spent effectively. 225. Related to the previous point, the results of the socio-economic cost calculations can be used for cost-benefit analysis of road safety measures. In a cost-benefit analysis, the costs associated with implementing a road safety measures, such as the costs of construction and maintenance of infrastructure or enforcement costs, are calculated and weighted against its benefits. These benefits consist of the saved road crash costs when casualties are prevented by road safety measures, and they can be calculated using the costs estimates in this report. A cost-benefit analysis shows whether the benefits (saved crash costs) of road safety investments are higher or lower than the costs. The results can be used for prioritization of road safety investments on the basis of their benefit-cost ratio. In this way cost-benefit analysis can support decision makers in ensuring efficiency of expenditures. An assessment of baseline risks and the effectiveness of road safety measures (as discussed above) is an important input for cost-benefit analysis. There is evidence from other countries that the number of casualties can decrease dramatically if the most cost-effective measures are implemented.110 Conducting cost-benefit analysis of some of the actions included in the State Program on Road Safety and follow-up road safety investments would provide evidence on their cost-effectiveness. Capacity building and training on economic assessment tools and their use in policy making and research is also recommended. 109 Daniels, S. et al. (2019). A systematic cost-benefit analysis of 29 road safety measures. Accident Analysis and Prevention, 133, 105292. ROSEBUD. 2006. Examples of assessed road safety measures - a short handbook. Deliverable of the European project ‘Road Safety and Environmental Benefit-Cost and Cost-Effectiveness Analysis for Use in Decision-Making’. 110 Elvik. 2003. How would setting policy priorities according to cost-benefit analysis affect the provision of road safety? Accident Analysis and Prevention. 35. pp. 557-570. 74 | P a g e Annex 1: International evidence on the socio-economic and poverty impact of road crash costs Socio-economic costs 1. Studies into the socio-economic costs of road crashes have been conducted in many countries. Wijnen & Stipdonk111 reviewed crash cost studies in ten high-income countries and seven low and middle-income countries. The costs in high-income countries ranged from 0.5% to 6.0% of GDP and in low and middle-income countries from 1.1% to 2.9% of GDP. The cost differences were mainly explained by methodological differences, in particular whether or not a willingness to pay method was applied. Non-fatal injuries were found to account for about half of the costs, both in high-income and low and middle-income countries. The main cost elements were human cost, vehicle damage and loss of production. Human costs, which are the intangible costs of loss of quality of life and life years account for 40% of total costs on average in high-income countries (and 55% in countries using a willingness to pay method), while property damage was found to be the most important cost item in low and middle-income countries (39% of total costs on average). Production loss accounted for about 20% of total costs in all countries. Other cost elements, such as medical costs and administrative costs (e.g. police and insurance costs) were found to be relatively low. 2. A detailed review of crash cost estimates in 31 European countries was conducted in the EU- project SafetyCube112. This review concentrated on the official estimates as determined by the government. The costs ranged from 0.4% to 4.1% of GDP (Figure 32). Again, the cost differences were mainly explained by the methods that countries use to estimate the costs, not by differences in road safety performance. In particular the choice of whether or not to use a willingness to pay method was found to have major influence on the size of the costs. This result is in line with earlier international reviews of road crash costs113. 3. A few post-Soviet countries have estimated the socio-economic costs of road crashes, including Latvia, Estonia and Lithuania. Estonia and Lithuania do not apply a willingness to pay method, for Latvia this is not known114. A detailed cost study was conducted in Kazakhstan115. The costs were estimated at 3.5% of GDP, including an estimate human costs on the basis of willingness to pay. Other countries have estimated the costs using rules of thumb on the valuation per fatality and serious injury developed within the international road assessment program (iRAP), for example Georgia where the costs are estimated at 5.2% of GDP in 2015116. 4. The socio-economic cost of road crashes has also been estimated at the regional or global level. Wijnen et al. (2017) estimated the costs of road crashes in the EU-28 to be at least 500 billion 111Wijnen, W. & Stipdonk, H. (2016). Social costs of road crashes: an international analysis. Accident Analysis and Prevention, 94, 97–106. 112Wijnen, W., Weijermars, W., Vanden Berghe, W., Schoeters, A., Bauer, R., Carnis, L., Elvik, R., Martensen, H. (2019). An analysis of official road crash cost estimates in European countries. Safety Science 113, 318–327. 113Elvik, R. (1995). An analysis of official economic valuations of traffic accident fatalities in 20 motorized countries. Accident Analysis and Prevention, vol. 27, nr. 2, pp. 237-347. Elvik, R. (2000). How much do road accidents cost the national economy? Accident Analysis and Prevention, vol. 32, nr. 6, pp. 849-851. Trawén, A., Maraste, P. & Persson, U. (2002). International comparison of costs of a fatal casualty of road accidents in 1990 and 1999. Accident Analysis and Prevention, vol. 34, nr. 3, p. 323-332. 114 Wijnen, W., Weijermars, W., Vanden Berghe, W., Schoeters, A., Bauer, R., Carnis, L., Elvik, R., Martensen, H. (2019). An analysis of official road crash cost estimates in European countries. Safety Science 113, 318–327. 115 Wijnen, W. Costs of road crashes in middle-income countries: A hybrid approach applied to Kazakhstan (forthcoming). 116 UN (2018). Road Safety Performance Review Georgia. United Nations, New York and Geneva. 75 | P a g e Euro in 2015.117 This estimate was based on the official cost estimates (per cost item) in individual countries, in combination with a ‘value transfer’ approach. The value transfer implied that values that were missing in some countries, or that were not estimated using an internationally recommended method, were added or replaced by standardized European values. These standard values were determined on the basis of best practices in Europe. 5. The World Bank estimated the costs of road crashes in low and middle-income countries at 4.8% of GDP in Europe and Central Asia, up to 9.0% of GDP in Africa.118 The total costs in all low and middle-income countries were estimated at 1.7 trillion USD. A relatively simple method was applied, based on the number of fatalities as estimated by the WHO, a standard ratio of fatalities to serious injuries and standardized values per fatality and injury. Figure 32: Costs of road crashes as percentage of GDP in European countries119 Impact of road crashes on poverty 6. Several previous studies have illustrated the strong relationship between road accidents and poverty. Thwe, Ph. et al (2013) in their paper revealed the impact of road accidents on poor people in Myanmar. The study’s results showed that lower social economic groups are more involved in road crashes compared to higher social economic groups. The costs and impacts of road crashes are higher burden for poor people120. The study was based on questionnaire survey among households involved 117 Wijnen, W., Weijermars, W., Vanden Berghe, W., Schoeters, A., Bauer, R., Carnis, L., Elvik, R., Theofilatos, A., Perez, C. and Martensen, H. (2017). Crash cost estimates for European countries, D3.2 of the H2020 project SafetyCube. 118 GRSF (2020). Guide for Road Safety Opportunities and Challenges: Low and middle-income Country Profiles. World Bank, Washington. 119 Wijnen et al., 2019. 120Thwe, Ph., Kanitpong, K., Jiwattanakulpaisarn, P, 2013, "Road Crashes and Poverty in Myanmar: Yangon Case Study", Proceedings of the Eastern Asia Society for Transportation Studies, Vol.9, 2013. https://socialscienceresearch.org/index.php/GJHSS/article/view/2771/2660 76 | P a g e in road crashes. The study used hypothesis tests and a linear regression model to identify involvement in road crashes and impacts on poor and non-poor households. 7. The report on “Social and Economic consequences of road traffic injury in Europe� (2007) 121 provided findings from desk research of various countries’ road crashes studies and determined which variables are the strongest predictors of social disparities in road accident risk. In the study conducted in the UK it was shown that the road accident injury rate is almost twice as high in the deprived areas as in the most affluent areas. The difference in the injury incidence rate is particularly large with respect to pedestrian injury. In other study in Norway it was found out that those who have a high social status tend to travel more than those who have a low social status and they might be more at risk of road crashes. However, the cars that rich use could be much safer than those used by the poor. A study in Sweden revealed that the risk of repeated injuries was substantially higher among those with only basic education than among those who had higher education (university degrees). The study investigated the social background of impaired drivers involved in accidents. Drivers in the lowest social status group were 1.24 times more often involved in injury accidents when impaired (i.e. influenced by alcohol or other drugs) than drivers in the highest ranked group. In the poorest areas of the USA, the rate of motor vehicle occupant death rates is about 2.5 times higher than in the richest areas. For pedestrian deaths, the rate ratio between the poorest and richest areas is about 2. Overall, the report stated evidence on decline in quality of life and standard of living among families of dead and disabled victims of road crashes. The incidence of traffic injury is higher in low status groups than in high status groups. 9. A particularly interesting findings of the study “Assessing the social costs and benefits of better and safer roads for the bottom 40 percent and other vulnerable groups in Albania� reported that road crashes disproportionally affect those in bottom 40 from the rest of the population 122. Road crashes victims are most likely to be rural poor and vulnerable groups such as pedestrian, cyclists and motorcyclists. The cumulative effect of crashes has the most impact on the bottom 40 percent, thus contributing to poverty. The study used mix of quantitative (survey among road crash affected people) and qualitative techniques (focus group discussions with affected families and in depth interviews with stakeholders) to collect and analyze data on road crashes and evaluate their economic and social impact. The data were analyzed using descriptive statistics, logistic regression, and other modeling methods. 10. A GRSP (Global Road Safety Partnership) paper talks about correlation between poverty and road traffic injuries123. Poor people are more likely to be involved in road traffic accidents as they are more likely to live in less safe areas next to major magistral roads and using less safe traffic modes. Road accidents impact on poor families more due to the potential loss of family income earner and cost of medical treatment. The paper suggests a series of solutions and recommendations in improving of road safety management. Among solutions and interventions are improving road safety management, designing safer roads, strengthening vehicle safety standards, implementation, and enforcement of comprehensive laws to protect road users, and implementing effective post-crash care. 11. Another study 124 analyzed findings from household surveys on road casualties conducted in Bangladesh, and India that compared the poor with the non-poor when involved in fatal and 121 “Social and Economic consequences of road traffic injury in Europe�, European Transport Safety Council, 2007 122 ASSESSING THE SOCIAL COSTS AND BENEFITS OF BETTER AND SAFER ROADS FOR THE BOTTOM 40 PERCENT AND OTHER VULNERABLE GROUPS: THE ALBANIAN CASE, Tirana Institute of Economic Research, 2015 123Poverty and Road Safety: A GRSP positioning paper, Global Road Safety Partnership 124“The involvement and impact of road crashes on the poor: Bangladesh and India case studies� Ms A Aeron-Thomas (TRL), Dr G D Jacobs (TRL) Mr B Sexton (TRL), Dr G Gururaj (NIMHANS), and Dr F Rahman (ICMH), 2004 . 77 | P a g e serious road crashes in urban and rural areas. Both studies relied on official definitions of poverty and used household income as the key indicator, as the information collected on household assets was unable to differentiate the poor from the non-poor. The poverty definitions used by this study are official government estimates of poverty; household per capita income; post-crash household income. The study’s findings showed that “vulnerable road users accounted for the vast majority of all road deaths and serious injuries to both the poor and non-poor, with pedestrians as the most common casualty�. In addition, the study report presented coping strategies that affected households adopted and the consequences of road crashes (loss of job, reduced HH income, etc.). The data were analyzed using SPSS program with chi square, ANOVA and other tests conducted. 78 | P a g e Annex 2: Review of the government methodology for assessing road crash costs In 2019 the Government of Azerbaijan developed a methodology for the estimation of the socio- economic costs and environmental damage resulting from road crashes. The development of this methodology (referred to as ‘government methodology’ in this annex) wa s embedded as an action in the State Program on road safety. This annex reviews the government methodology, compares it to the methodology as developed and applied in this study (referred to as the ‘current study’), and provides some recommendations. The government methodology The government methodology is aimed at estimating the economic loss as measured by loss of gross domestic product (GDP). Several cost components are identified, and the calculations are specified by formulas. The government methodology includes the following cost components: 1. Costs of death and injuries These costs particularly refer to loss of GDP as a result of the fact that road casualties cannot work temporarily (non-fatal injuries) or lifelong (disabled casualties and fatalities). A distinction is made between married and single persons. Also, the costs of young people who are going to school are calculated separately. In the latter case, future education costs are distracted from the future productivity loss. In addition to GDP loss, disability payments, hospital costs (in-patient), funeral costs are included in this cost component. 2. Damage to vehicles and cargo This component includes damage of total losses, vehicle repair costs and also costs of some emergency services (towing and crash investigation) and court costs. Damage to cars, trucks and buses is included, with a distinction between local and imported vehicles. These costs are calculated on the basis of the costs needed to repair the vehicles or as the value of the vehicle in case of total loss. With respect to trucks, also the costs related to the fact that goods cannot be delivered on time, or not at all, are included. 3. Damage to roads and road infrastructure This cost components includes damage to particular road signs and road fences or obstacles. The methodology specifies types of road signs (e.g. metal road signs, metal frame constructions, traffic lights) and road fences or obstacles (e.g. barriers, bus stops, pedestrian facilities). The calculation of this damage is based on the cost of repairing the damage. 4. Environmental costs This component includes costs of air, water and land pollution due to oil spills, fires and electromagnetic radiation as well as costs of destruction of nature (e.g. trees). The method for calculating the costs of damage is mostly based on costs needed to restore the damage as much as possible. The costs components included in the government methodology are very relevant costs resulting from road crashes. It should be noted that particularly the first cost component is aimed at estimating the loss of GDP. A human capital approach is applied which means that the loss of productivity of road casualties is measured, which represents a loss of GDP. However, the other 79 | P a g e three costs components are calculated on the basis of the costs which are needed to repair or restore damage, which is known as the restitution costs approach. The restitution costs do not necessary reflect GDP loss. For example, repairing vehicles contributes positively to GDP, as this is production delivered by garages. However, logically vehicle damage should be regarded as a cost because material resources (vehicles in this case) are damaged, despite the fact that repairing the damage yields income for some businesses. Another issue to consider is whether disability benefit payments (or other social benefit payments) can be regarded as a cost. From the perspective of the government budget it is clearly a cost. However, from the perspective of the casualty it is an income, and at the national level the cost for the government and the benefit for the casualty cancel out. The perspective of the cost assessment In general, for the assessment of road crash costs, it is important to define the perspective of the assessment. Legitimate perspectives include the perspective of economic welfare as measured by GDP (that is income, consumption and production), the government budget or a broader socio- economic perspective, which also includes intangible issues such as cost of loss of life years and quality of life. The international standard in road crash cost studies is the broad socio-economic perspective, and this approach is used in almost all studies on cost of road crashes. This approach is based on economic welfare theory and the basis for defining and valuing socio-economic costs is the loss of resources, such as labor and materials, and ‘utility’. People derive utility from consumption but also from more intangible issues such as enjoying leisure activities. Thereby the welfare economic approach implies that a broader perspective is used by including, for example, intangible costs of loss of life years and quality of life (human costs). The advantage of this approach is the fact that in principle all impacts which are relevant to society are included. Also, the welfare economic approach is consistent with the theory of social cost-benefit analysis, which has the same theoretical basis (welfare economic theory). Adopting the broad socio-economic approach in road crash cost assessments ensures that the resulting cost figures can be applied in cost-benefit analysis of road safety programs. Also, the impact of road crashes on GDP and the consequences for the government budget are relevant perspective for policy makers. However, this requires a different type of analysis. An analysis of the impact on GDP is particularly focused on the loss of human capital. In such an analysis, the impact on labor productivity, labor supply, education and savings and investments are assessed. An assessment of the impact of road crashes on governments concentrates on financial impacts for the government, such as disability payment and tax incomes. Given these different options for the perspective of the analysis, it is recommended to choose the perspective which best reflects the needs of the stakeholders who will be using the results, and to apply the chosen approach consistently for all relevant cost items. A comparison of the government methodology and the methodology used in the current study The government methodology mainly adopts the socio-economic approach, since the restitution costs methods and the human capital method are used. These are the valuation methods, along with the willingness to pay method, which are used in the socio-economic approach. This methodological approach is the same as the approach adopted in the current study. A main difference is the fact that the current study includes several additional costs, based on international guidelines and best practices. These costs are in particular: 80 | P a g e - medical costs other than costs of in-patient hospital treatment, such as ambulance costs, out-patient hospital costs, non-hospital costs (e.g. rehabilitation) and home care; - human costs of fatalities, serious and slight injuries; - administrative costs, in particular police costs and insurance costs. In addition, the costs of congestion resulting from road crashes should be considered. The government study, on the other hand, includes some costs which are not included in the current study, such as social benefit payments. The relevance of including such costs depends on the chosen perspective, as discussed above. If the broad socio-economic approach is chosen for the government methodology, it is recommended to consider including more cost items to ensure that the methodology is as complete as possible. The willingness to pay approach, which is the international standard for assessing intangible impacts, may be added to the methodology. 81 | P a g e Annex 3: Socio-economic cost items and valuation methods Component Sub-component Proposed Explanation valuation method Medical costs Main a) First aid at crash Restitution - Actual costs of medical location and costs resources (labor, equipment, transportation etc.), Calculation: costs per b) Treatment at the Restitution ‘unit’ (per ambulance trip, accident and emergency costs per day, per treatment, etc.) department of hospitals times the number of ‘units’ c) In-patient hospital Restitution (number of ambulance trips, treatment costs average duration of hospital d) Out-patient hospital Restitution stay, frequency of non- treatment costs hospital treatment, etc.) e) Non-hospital treatment Restitution (rehabilitation centres, costs general practitioners, etc.) Minor f) Aids and appliances Restitution costs g) Home care Restitution costs h) Visiting people in Restitution Actual costs of visits, in particular hospital costs travel costs and time costs i) Moving and house Restitution Actual cost for moving and for adaption cost costs adaptations (equipment, labor) Production loss Main a) Lost market production Human capital - production per person per year (e.g. GDP/capita or income) * lost productive years - Gross production loss: including consumption loss - Potential production loss - Discounting future losses Other b) Lost non-market Human capital - time spent on non-market production (household production times value of work, taking care of time (e.g. wage as indicator children, voluntary work, - Discounting future losses etc.) Minor c) Friction costs Restitution - Actual costs of recruiting and costs training new employees and actual costs of vocational rehabilitation Human costs Main a) Fatalities (lost life Willingness to - VSL-consumption loss * years) pay number of fatalities 82 | P a g e - Individual WTP from stated or revealed preference studies b) Injuries (loss of quality Willingness to - Serious and slight injuries of life) pay - %VSL *number of injuries Other c) Human costs for Willingness to - Not calculated separately: relatives and friends pay included in WTP fatalities/injuries Property damage Main a) Vehicles Restitution - Actual costs to repair costs damage or replace vehicles - Preferably cars, motorcycles and trucks/vans; optionally buses, mopeds and bicycles - Two calculation approaches: 1.Bottom-up: average damage per vehicle * number of damaged vehicles (including non-reported crashes) 2. Top-down: total vehicle damage (including estimate of non-reported damage) Minor b) Infrastructure, fixed Restitution - Actual costs to repair roadside objects and costs damage or replace property buildings c) Freight carried by Restitution lorries costs d) Personal property Restitution costs Administrative costs Main a) Police costs Restitution - Actual costs of resources of costs police assistance (labor, equipment) - Excluding costs of prevention - Two calculation approaches: 1.Bottom-up: time spent on road crashes * costs per unit of time 2. Top-down: total police costs * time share road crashes b) Fire service costs Restitution - Actual costs of resources of costs fire service assistance (labor, equipment) - Bottom-up or top-down calculation (similar to police cost) c) Towing costs Restitution - Actual costs of resources costs towing (labor, equipment) d) Vehicle insurance Restitution - All administrative costs costs costs related to vehicle insurances e) Legal costs Restitution - Actual costs of prosecution, costs lawsuits and imprisonment 83 | P a g e - Bottom-up or top-down calculation Other f) Other insurance Restitution - All administrative costs costs costs related to other insurances (e.g. health) Other costs Main a) Funeral costs - difference between the actual funeral costs and (discounted) future costs of the funeral if the person was not killed in a crash b) Congestion costs - Time loss due to traffic jams resulting from road crashes - Calculation: time loss *value of time, or total congestion costs * share road crashes - Cost of travel time unreliability of travel times, adapting travel behavior, extra fuel costs and pollution may be included Minor c) Vehicle Restitution - Actual costs of replacing the unavailability costs vehicle (e.g. renting car and time costs) d) Environmental Restitution - Actual costs to restore costs costs damage 84 | P a g e Annex 4: Data requirements Road safety Number of reported - Fatalities, by road casualties o gender o age (preferably each age: number of 0, 1, 2, 3 – 99 years old road fatalities; if not available age categories) o by time of dying after the crash (e.g. within 24h after the crash, 72h after the crash, 5 days, 10 days, 20- 30 days ) - Serious injuries, by o gender o age (preferably each age: number of 0, 1, 2, 3 – 99 years old serious injuries; if not available age categories) o injury severity (preferably including permanent disability) - Slight injuries, by gender and age Number of reported - Fatal crashes, by road type (eg. motorways, urban roads, crashes rural roads) - Serious injury crashes, by road type - Slight injury crashes, by road type - Property damage only crashes, by road type Number of - Fatal crashes: number of fatalities, serious injuries and slight casualties per crash injuries per fatal crash - Serious injury crashes: number of serious injuries and slight injuries per serious injury crash - Slight injury crashes: number of slight injuries per slight injury crash Percentage of - Fatalities underreporting of - Serious injuries the number of - Slight injuries casualties and - Fatal crashes crashes - Serious injury crashes - Slight injury crashes - Property damage only crashes Medical Transportation - Proportion of road casualties transported by ambulance (or helicopter if applicable) by severity (fatal, serious injury, slight injury). Or: total number of ambulance rides (and helicopter flights) for road casualties, by severity. - Costs of an ambulance trip (and helicopter flight) Hospitalization - Average number of days of hospital treatment for hospitalized injuries (excluding casualties who die in hospital within 30 days) - Average number of days of hospital treatment for casualties who die in hospital within 30 days - Average costs of in-patient hospital treatment (with overnight stay) per patient per day 85 | P a g e Out-patient - Average number of out-patient hospital visits per patient after treatment hospitalization - Average costs of out-patient hospital visit Emergency - Number of injured people treated at the hospital emergency department department (without overnight stay) - Average costs of hospital emergency treatment (without overnight stay) Rehabilitation - Number of road casualties treated in a rehabilitation centre - Average number of days of treatment in a rehabilitation centre - Average costs of treatment in a rehabilitation centre per day Other non-hospital - Number of casualties who visit other medical practitioners treatment (e.g. general practitioner, physiotherapist) - Average costs per visit Appliances, - Average cost of medical appliances (e.g. wheelchair) and medicines medicines per casualty by severity Home care - Proportion of casualties (by severity) who received home care, split into professional and in kind care - Duration of home after the crash - Average number of hours of care per week - Average wage of professional homecare Visitors - Average number of visitors per hospital day - Average duration per visit - Average travel costs per visit House adaptations - Proportion of casualties (by severity) who need house adaptions or moving - Average costs of house adaptions or moving Inability to work - Average number of days a seriously injured casualty is unable to work after the crash - Average number of days a slightly injured casualty is unable to work after the crash Demography - Population by o gender o age (preferably each age: number of 0, 1, 2, 3 – 99 years) o education level - Life expectancy by gender and age Labour market - Age of entering the labor market by gender and education level - Average retirement age by gender - Yearly number of working hours per laborer, by gender - Gross wage by gender and age (or age groups) - Time spend by employers to find new personnel to replace road casualties Economy - Gross Domestic Product per capita - Private consumption per capita, by gender - Official discount rate used in economic assessments of governments investments (for example used in cost-benefit analyses) 86 | P a g e Vehicle damage - Number of insurance compensation payments for vehicle damage resulting from road crashes, preferably by o vehicle type: car, truck, bus, motorcycle, moped o insurance type: full insurance (including own vehicle) vs. third party insurance only - Average damage amount paid, preferable by vehicle type and insurance type (see above) - Total amount of damage payments, preferable by vehicle type and insurance type (see above) - Percentage vehicles that are insured, by insurance type - Percentage of damage cases for which a claim is submitted to an insurance company, by insurance type Insurance - Administrative costs of insurance companies related to vehicle administration insurances (personnel costs, overhead) costs - Profits insurance companies from vehicle insurances - Income from vehicle insurance premiums - Any costs related to vehicle insurances other than compensation payments and administrative costs Infrastructure - Number of crashes that caused damage to the road or roadside infrastructure - Average infrastructure damage per crash, by crash severity Environment - Number of crashes that caused environmental damage (e.g. petrol or oil leaks, vehicle parts at roadsides) - Average costs of cleaning up environmental damage per case Traffic and - Traffic volume (motor vehicles per 24h) by road type (e.g. congestion motorways, urban road, rural roads) - Proportion of traffic jams, by length and duration, caused by road crashes - Average value of time of motor vehicle road users Police - Proportion of crashes (by severity) the police attends - Average number of policemen coming to the place of a road crash, by crash severity - Average time a police officer spends per crash, by crash severity - Average wage of a police officer OR: - Annual national traffic police budget - Proportion of personnel costs in national police budget - Time share traffic police officers spend on road crashes, on- site and administration (percentage of daily or weekly working hours) Fire service - Proportion of crashes (by severity) attended by fire service - Average number of fire service officers coming to the place of a road crash, by crash severity - Average time of fire service officer spends per crash, by crash severity - Average wage of a fire service officer OR: - Annual national fire service budget 87 | P a g e - Proportion of personnel costs in national fire service budget - Time share traffic fire service officers spend on road crashes, on-site and administration (percentage of daily or weekly working hours) Towing - Proportion of crashes (by severity) that require towing - Number of vehicles towed away per crash (by severity) - Towing costs per vehicle Legal - Time spending of police on prosecution of offenders who caused a crash - Number of court cases concerning road users who have caused a crash - Average costs per trial - Number of road users that go to prison due to causing a road crash - Average number of days in prison - Prison costs per prisoner per day Vehicle - Average number of days a vehicle is unavailable after a crash unavailability - Additional costs of using other transport modes Funerals - Average costs of a funeral 88 | P a g e